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, 94, 95, 96, 97, 98, 99 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
34 /* Windows does not provide unistd.h, which is required for abort() */
39 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
40 # define PARAMS(args) args
42 # define PARAMS(args) ()
44 #endif /* Not PARAMS. */
46 #if defined STDC_HEADERS && !defined emacs
49 /* We need this for `regex.h', and perhaps for the Emacs include files. */
50 # include <sys/types.h>
53 #define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
55 /* For platform which support the ISO C amendement 1 functionality we
56 support user defined character classes. */
57 #if defined _LIBC || WIDE_CHAR_SUPPORT
58 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
64 /* We have to keep the namespace clean. */
65 # define regfree(preg) __regfree (preg)
66 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
67 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
68 # define regerror(errcode, preg, errbuf, errbuf_size) \
69 __regerror(errcode, preg, errbuf, errbuf_size)
70 # define re_set_registers(bu, re, nu, st, en) \
71 __re_set_registers (bu, re, nu, st, en)
72 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
73 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
74 # define re_match(bufp, string, size, pos, regs) \
75 __re_match (bufp, string, size, pos, regs)
76 # define re_search(bufp, string, size, startpos, range, regs) \
77 __re_search (bufp, string, size, startpos, range, regs)
78 # define re_compile_pattern(pattern, length, bufp) \
79 __re_compile_pattern (pattern, length, bufp)
80 # define re_set_syntax(syntax) __re_set_syntax (syntax)
81 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
82 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
83 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
88 /* This is for other GNU distributions with internationalized messages. */
89 #if HAVE_LIBINTL_H || defined _LIBC
92 # define gettext(msgid) (msgid)
96 /* This define is so xgettext can find the internationalizable
98 # define gettext_noop(String) String
101 /* The `emacs' switch turns on certain matching commands
102 that make sense only in Emacs. */
109 #else /* not emacs */
111 /* If we are not linking with Emacs proper,
112 we can't use the relocating allocator
113 even if config.h says that we can. */
116 # if defined STDC_HEADERS || defined _LIBC
123 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
124 If nothing else has been done, use the method below. */
125 # ifdef INHIBIT_STRING_HEADER
126 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
127 # if !defined bzero && !defined bcopy
128 # undef INHIBIT_STRING_HEADER
133 /* This is the normal way of making sure we have a bcopy and a bzero.
134 This is used in most programs--a few other programs avoid this
135 by defining INHIBIT_STRING_HEADER. */
136 # ifndef INHIBIT_STRING_HEADER
137 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
141 # define bzero(s, n) (memset (s, '\0', n), (s))
143 # define bzero(s, n) __bzero (s, n)
147 # include <strings.h>
149 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
152 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
157 /* Define the syntax stuff for \<, \>, etc. */
159 /* This must be nonzero for the wordchar and notwordchar pattern
160 commands in re_match_2. */
165 # ifdef SWITCH_ENUM_BUG
166 # define SWITCH_ENUM_CAST(x) ((int)(x))
168 # define SWITCH_ENUM_CAST(x) (x)
171 /* How many characters in the character set. */
172 # define CHAR_SET_SIZE 256
176 extern char *re_syntax_table;
178 # else /* not SYNTAX_TABLE */
180 static char re_syntax_table[CHAR_SET_SIZE];
191 memset (re_syntax_table, 0, sizeof re_syntax_table);
193 for (c = 'a'; c <= 'z'; c++)
194 re_syntax_table[c] = Sword;
196 for (c = 'A'; c <= 'Z'; c++)
197 re_syntax_table[c] = Sword;
199 for (c = '0'; c <= '9'; c++)
200 re_syntax_table[c] = Sword;
202 re_syntax_table['_'] = Sword;
207 # endif /* not SYNTAX_TABLE */
209 # define SYNTAX(c) re_syntax_table[c]
211 #endif /* not emacs */
213 /* Get the interface, including the syntax bits. */
214 #include <regex-gnu.h>
216 /* isalpha etc. are used for the character classes. */
219 /* Jim Meyering writes:
221 "... Some ctype macros are valid only for character codes that
222 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
223 using /bin/cc or gcc but without giving an ansi option). So, all
224 ctype uses should be through macros like ISPRINT... If
225 STDC_HEADERS is defined, then autoconf has verified that the ctype
226 macros don't need to be guarded with references to isascii. ...
227 Defining isascii to 1 should let any compiler worth its salt
228 eliminate the && through constant folding."
229 Solaris defines some of these symbols so we must undefine them first. */
232 #if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
233 # define ISASCII(c) 1
235 # define ISASCII(c) isascii(c)
239 # define ISBLANK(c) (ISASCII (c) && isblank (c))
241 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
244 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
246 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
250 #define ISPRINT(c) (ISASCII (c) && isprint (c))
251 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
252 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
253 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
254 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
255 #define ISLOWER(c) (ISASCII (c) && islower (c))
256 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
257 #define ISSPACE(c) (ISASCII (c) && isspace (c))
258 #define ISUPPER(c) (ISASCII (c) && isupper (c))
259 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
262 # define TOLOWER(c) _tolower(c)
264 # define TOLOWER(c) tolower(c)
268 # define NULL (void *)0
271 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
272 since ours (we hope) works properly with all combinations of
273 machines, compilers, `char' and `unsigned char' argument types.
274 (Per Bothner suggested the basic approach.) */
275 #undef SIGN_EXTEND_CHAR
277 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
278 #else /* not __STDC__ */
279 /* As in Harbison and Steele. */
280 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
283 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
284 use `alloca' instead of `malloc'. This is because using malloc in
285 re_search* or re_match* could cause memory leaks when C-g is used in
286 Emacs; also, malloc is slower and causes storage fragmentation. On
287 the other hand, malloc is more portable, and easier to debug.
289 Because we sometimes use alloca, some routines have to be macros,
290 not functions -- `alloca'-allocated space disappears at the end of the
291 function it is called in. */
295 # define REGEX_ALLOCATE malloc
296 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
297 # define REGEX_FREE free
299 #else /* not REGEX_MALLOC */
301 /* Emacs already defines alloca, sometimes. */
304 /* Make alloca work the best possible way. */
306 # define alloca __builtin_alloca
307 # else /* not __GNUC__ */
310 # endif /* HAVE_ALLOCA_H */
311 # endif /* not __GNUC__ */
313 # endif /* not alloca */
315 # define REGEX_ALLOCATE alloca
317 /* Assumes a `char *destination' variable. */
318 # define REGEX_REALLOCATE(source, osize, nsize) \
319 (destination = (char *) alloca (nsize), \
320 memcpy (destination, source, osize))
322 /* No need to do anything to free, after alloca. */
323 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
325 #endif /* not REGEX_MALLOC */
327 /* Define how to allocate the failure stack. */
329 #if defined REL_ALLOC && defined REGEX_MALLOC
331 # define REGEX_ALLOCATE_STACK(size) \
332 r_alloc (&failure_stack_ptr, (size))
333 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
334 r_re_alloc (&failure_stack_ptr, (nsize))
335 # define REGEX_FREE_STACK(ptr) \
336 r_alloc_free (&failure_stack_ptr)
338 #else /* not using relocating allocator */
342 # define REGEX_ALLOCATE_STACK malloc
343 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
344 # define REGEX_FREE_STACK free
346 # else /* not REGEX_MALLOC */
348 # define REGEX_ALLOCATE_STACK alloca
350 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
351 REGEX_REALLOCATE (source, osize, nsize)
352 /* No need to explicitly free anything. */
353 # define REGEX_FREE_STACK(arg)
355 # endif /* not REGEX_MALLOC */
356 #endif /* not using relocating allocator */
359 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
360 `string1' or just past its end. This works if PTR is NULL, which is
362 #define FIRST_STRING_P(ptr) \
363 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
365 /* (Re)Allocate N items of type T using malloc, or fail. */
366 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
367 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
368 #define RETALLOC_IF(addr, n, t) \
369 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
370 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
372 #define BYTEWIDTH 8 /* In bits. */
374 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
378 #define MAX(a, b) ((a) > (b) ? (a) : (b))
379 #define MIN(a, b) ((a) < (b) ? (a) : (b))
381 typedef char boolean;
385 static int re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
386 const char *string1, int size1,
387 const char *string2, int size2,
389 struct re_registers *regs,
392 /* These are the command codes that appear in compiled regular
393 expressions. Some opcodes are followed by argument bytes. A
394 command code can specify any interpretation whatsoever for its
395 arguments. Zero bytes may appear in the compiled regular expression. */
401 /* Succeed right away--no more backtracking. */
404 /* Followed by one byte giving n, then by n literal bytes. */
407 /* Matches any (more or less) character. */
410 /* Matches any one char belonging to specified set. First
411 following byte is number of bitmap bytes. Then come bytes
412 for a bitmap saying which chars are in. Bits in each byte
413 are ordered low-bit-first. A character is in the set if its
414 bit is 1. A character too large to have a bit in the map is
415 automatically not in the set. */
418 /* Same parameters as charset, but match any character that is
419 not one of those specified. */
422 /* Start remembering the text that is matched, for storing in a
423 register. Followed by one byte with the register number, in
424 the range 0 to one less than the pattern buffer's re_nsub
425 field. Then followed by one byte with the number of groups
426 inner to this one. (This last has to be part of the
427 start_memory only because we need it in the on_failure_jump
431 /* Stop remembering the text that is matched and store it in a
432 memory register. Followed by one byte with the register
433 number, in the range 0 to one less than `re_nsub' in the
434 pattern buffer, and one byte with the number of inner groups,
435 just like `start_memory'. (We need the number of inner
436 groups here because we don't have any easy way of finding the
437 corresponding start_memory when we're at a stop_memory.) */
440 /* Match a duplicate of something remembered. Followed by one
441 byte containing the register number. */
444 /* Fail unless at beginning of line. */
447 /* Fail unless at end of line. */
450 /* Succeeds if at beginning of buffer (if emacs) or at beginning
451 of string to be matched (if not). */
454 /* Analogously, for end of buffer/string. */
457 /* Followed by two byte relative address to which to jump. */
460 /* Same as jump, but marks the end of an alternative. */
463 /* Followed by two-byte relative address of place to resume at
464 in case of failure. */
467 /* Like on_failure_jump, but pushes a placeholder instead of the
468 current string position when executed. */
469 on_failure_keep_string_jump,
471 /* Throw away latest failure point and then jump to following
472 two-byte relative address. */
475 /* Change to pop_failure_jump if know won't have to backtrack to
476 match; otherwise change to jump. This is used to jump
477 back to the beginning of a repeat. If what follows this jump
478 clearly won't match what the repeat does, such that we can be
479 sure that there is no use backtracking out of repetitions
480 already matched, then we change it to a pop_failure_jump.
481 Followed by two-byte address. */
484 /* Jump to following two-byte address, and push a dummy failure
485 point. This failure point will be thrown away if an attempt
486 is made to use it for a failure. A `+' construct makes this
487 before the first repeat. Also used as an intermediary kind
488 of jump when compiling an alternative. */
491 /* Push a dummy failure point and continue. Used at the end of
495 /* Followed by two-byte relative address and two-byte number n.
496 After matching N times, jump to the address upon failure. */
499 /* Followed by two-byte relative address, and two-byte number n.
500 Jump to the address N times, then fail. */
503 /* Set the following two-byte relative address to the
504 subsequent two-byte number. The address *includes* the two
508 wordchar, /* Matches any word-constituent character. */
509 notwordchar, /* Matches any char that is not a word-constituent. */
511 wordbeg, /* Succeeds if at word beginning. */
512 wordend, /* Succeeds if at word end. */
514 wordbound, /* Succeeds if at a word boundary. */
515 notwordbound /* Succeeds if not at a word boundary. */
518 ,before_dot, /* Succeeds if before point. */
519 at_dot, /* Succeeds if at point. */
520 after_dot, /* Succeeds if after point. */
522 /* Matches any character whose syntax is specified. Followed by
523 a byte which contains a syntax code, e.g., Sword. */
526 /* Matches any character whose syntax is not that specified. */
531 /* Common operations on the compiled pattern. */
533 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
535 #define STORE_NUMBER(destination, number) \
537 (destination)[0] = (number) & 0377; \
538 (destination)[1] = (number) >> 8; \
541 /* Same as STORE_NUMBER, except increment DESTINATION to
542 the byte after where the number is stored. Therefore, DESTINATION
543 must be an lvalue. */
545 #define STORE_NUMBER_AND_INCR(destination, number) \
547 STORE_NUMBER (destination, number); \
548 (destination) += 2; \
551 /* Put into DESTINATION a number stored in two contiguous bytes starting
554 #define EXTRACT_NUMBER(destination, source) \
556 (destination) = *(source) & 0377; \
557 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
561 static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
563 extract_number (dest, source)
565 unsigned char *source;
567 int temp = SIGN_EXTEND_CHAR (*(source + 1));
568 *dest = *source & 0377;
572 # ifndef EXTRACT_MACROS /* To debug the macros. */
573 # undef EXTRACT_NUMBER
574 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
575 # endif /* not EXTRACT_MACROS */
579 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
580 SOURCE must be an lvalue. */
582 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
584 EXTRACT_NUMBER (destination, source); \
589 static void extract_number_and_incr _RE_ARGS ((int *destination,
590 unsigned char **source));
592 extract_number_and_incr (destination, source)
594 unsigned char **source;
596 extract_number (destination, *source);
600 # ifndef EXTRACT_MACROS
601 # undef EXTRACT_NUMBER_AND_INCR
602 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
603 extract_number_and_incr (&dest, &src)
604 # endif /* not EXTRACT_MACROS */
608 /* If DEBUG is defined, Regex prints many voluminous messages about what
609 it is doing (if the variable `debug' is nonzero). If linked with the
610 main program in `iregex.c', you can enter patterns and strings
611 interactively. And if linked with the main program in `main.c' and
612 the other test files, you can run the already-written tests. */
616 /* We use standard I/O for debugging. */
619 /* It is useful to test things that ``must'' be true when debugging. */
620 # include "zassert.h"
624 # define DEBUG_STATEMENT(e) e
625 # define DEBUG_PRINT1(x) if (debug) printf (x)
626 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
627 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
628 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
629 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
630 if (debug) print_partial_compiled_pattern (s, e)
631 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
632 if (debug) print_double_string (w, s1, sz1, s2, sz2)
635 /* Print the fastmap in human-readable form. */
638 print_fastmap (fastmap)
641 unsigned was_a_range = 0;
644 while (i < (1 << BYTEWIDTH))
650 while (i < (1 << BYTEWIDTH) && fastmap[i])
666 /* Print a compiled pattern string in human-readable form, starting at
667 the START pointer into it and ending just before the pointer END. */
670 print_partial_compiled_pattern (start, end)
671 unsigned char *start;
676 unsigned char *p = start;
677 unsigned char *pend = end;
685 /* Loop over pattern commands. */
688 printf ("%d:\t", p - start);
690 switch ((re_opcode_t) *p++)
698 printf ("/exactn/%d", mcnt);
709 printf ("/start_memory/%d/%d", mcnt, *p++);
714 printf ("/stop_memory/%d/%d", mcnt, *p++);
718 printf ("/duplicate/%d", *p++);
728 register int c, last = -100;
729 register int in_range = 0;
731 printf ("/charset [%s",
732 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
734 assert (p + *p < pend);
736 for (c = 0; c < 256; c++)
738 && (p[1 + (c/8)] & (1 << (c % 8))))
740 /* Are we starting a range? */
741 if (last + 1 == c && ! in_range)
746 /* Have we broken a range? */
747 else if (last + 1 != c && in_range)
776 case on_failure_jump:
777 extract_number_and_incr (&mcnt, &p);
778 printf ("/on_failure_jump to %d", p + mcnt - start);
781 case on_failure_keep_string_jump:
782 extract_number_and_incr (&mcnt, &p);
783 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
786 case dummy_failure_jump:
787 extract_number_and_incr (&mcnt, &p);
788 printf ("/dummy_failure_jump to %d", p + mcnt - start);
791 case push_dummy_failure:
792 printf ("/push_dummy_failure");
796 extract_number_and_incr (&mcnt, &p);
797 printf ("/maybe_pop_jump to %d", p + mcnt - start);
800 case pop_failure_jump:
801 extract_number_and_incr (&mcnt, &p);
802 printf ("/pop_failure_jump to %d", p + mcnt - start);
806 extract_number_and_incr (&mcnt, &p);
807 printf ("/jump_past_alt to %d", p + mcnt - start);
811 extract_number_and_incr (&mcnt, &p);
812 printf ("/jump to %d", p + mcnt - start);
816 extract_number_and_incr (&mcnt, &p);
818 extract_number_and_incr (&mcnt2, &p);
819 printf ("/succeed_n to %d, %d times", p1 - start, mcnt2);
823 extract_number_and_incr (&mcnt, &p);
825 extract_number_and_incr (&mcnt2, &p);
826 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
830 extract_number_and_incr (&mcnt, &p);
832 extract_number_and_incr (&mcnt2, &p);
833 printf ("/set_number_at location %d to %d", p1 - start, mcnt2);
837 printf ("/wordbound");
841 printf ("/notwordbound");
853 printf ("/before_dot");
861 printf ("/after_dot");
865 printf ("/syntaxspec");
867 printf ("/%d", mcnt);
871 printf ("/notsyntaxspec");
873 printf ("/%d", mcnt);
878 printf ("/wordchar");
882 printf ("/notwordchar");
894 printf ("?%d", *(p-1));
900 printf ("%d:\tend of pattern.\n", p - start);
905 print_compiled_pattern (bufp)
906 struct re_pattern_buffer *bufp;
908 unsigned char *buffer = bufp->buffer;
910 print_partial_compiled_pattern (buffer, buffer + bufp->used);
911 printf ("%ld bytes used/%ld bytes allocated.\n",
912 bufp->used, bufp->allocated);
914 if (bufp->fastmap_accurate && bufp->fastmap)
916 printf ("fastmap: ");
917 print_fastmap (bufp->fastmap);
920 printf ("re_nsub: %d\t", bufp->re_nsub);
921 printf ("regs_alloc: %d\t", bufp->regs_allocated);
922 printf ("can_be_null: %d\t", bufp->can_be_null);
923 printf ("newline_anchor: %d\n", bufp->newline_anchor);
924 printf ("no_sub: %d\t", bufp->no_sub);
925 printf ("not_bol: %d\t", bufp->not_bol);
926 printf ("not_eol: %d\t", bufp->not_eol);
927 printf ("syntax: %lx\n", bufp->syntax);
928 /* Perhaps we should print the translate table? */
933 print_double_string (where, string1, size1, string2, size2)
946 if (FIRST_STRING_P (where))
948 for (this_char = where - string1; this_char < size1; this_char++)
949 putchar (string1[this_char]);
954 for (this_char = where - string2; this_char < size2; this_char++)
955 putchar (string2[this_char]);
966 #else /* not DEBUG */
971 # define DEBUG_STATEMENT(e)
972 # define DEBUG_PRINT1(x)
973 # define DEBUG_PRINT2(x1, x2)
974 # define DEBUG_PRINT3(x1, x2, x3)
975 # define DEBUG_PRINT4(x1, x2, x3, x4)
976 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
977 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
979 #endif /* not DEBUG */
981 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
982 also be assigned to arbitrarily: each pattern buffer stores its own
983 syntax, so it can be changed between regex compilations. */
984 /* This has no initializer because initialized variables in Emacs
985 become read-only after dumping. */
986 reg_syntax_t re_syntax_options;
989 /* Specify the precise syntax of regexps for compilation. This provides
990 for compatibility for various utilities which historically have
991 different, incompatible syntaxes.
993 The argument SYNTAX is a bit mask comprised of the various bits
994 defined in regex.h. We return the old syntax. */
997 re_set_syntax (syntax)
1000 reg_syntax_t ret = re_syntax_options;
1002 re_syntax_options = syntax;
1004 if (syntax & RE_DEBUG)
1006 else if (debug) /* was on but now is not */
1012 weak_alias (__re_set_syntax, re_set_syntax)
1015 /* This table gives an error message for each of the error codes listed
1016 in regex.h. Obviously the order here has to be same as there.
1017 POSIX doesn't require that we do anything for REG_NOERROR,
1018 but why not be nice? */
1020 static const char re_error_msgid[] =
1022 #define REG_NOERROR_IDX 0
1023 gettext_noop ("Success") /* REG_NOERROR */
1025 #define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success")
1026 gettext_noop ("No match") /* REG_NOMATCH */
1028 #define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match")
1029 gettext_noop ("Invalid regular expression") /* REG_BADPAT */
1031 #define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression")
1032 gettext_noop ("Invalid collation character") /* REG_ECOLLATE */
1034 #define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character")
1035 gettext_noop ("Invalid character class name") /* REG_ECTYPE */
1037 #define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name")
1038 gettext_noop ("Trailing backslash") /* REG_EESCAPE */
1040 #define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash")
1041 gettext_noop ("Invalid back reference") /* REG_ESUBREG */
1043 #define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference")
1044 gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */
1046 #define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^")
1047 gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */
1049 #define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(")
1050 gettext_noop ("Unmatched \\{") /* REG_EBRACE */
1052 #define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{")
1053 gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */
1055 #define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}")
1056 gettext_noop ("Invalid range end") /* REG_ERANGE */
1058 #define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end")
1059 gettext_noop ("Memory exhausted") /* REG_ESPACE */
1061 #define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted")
1062 gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */
1064 #define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression")
1065 gettext_noop ("Premature end of regular expression") /* REG_EEND */
1067 #define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression")
1068 gettext_noop ("Regular expression too big") /* REG_ESIZE */
1070 #define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big")
1071 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1074 static const size_t re_error_msgid_idx[] =
1095 /* Avoiding alloca during matching, to placate r_alloc. */
1097 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1098 searching and matching functions should not call alloca. On some
1099 systems, alloca is implemented in terms of malloc, and if we're
1100 using the relocating allocator routines, then malloc could cause a
1101 relocation, which might (if the strings being searched are in the
1102 ralloc heap) shift the data out from underneath the regexp
1105 Here's another reason to avoid allocation: Emacs
1106 processes input from X in a signal handler; processing X input may
1107 call malloc; if input arrives while a matching routine is calling
1108 malloc, then we're scrod. But Emacs can't just block input while
1109 calling matching routines; then we don't notice interrupts when
1110 they come in. So, Emacs blocks input around all regexp calls
1111 except the matching calls, which it leaves unprotected, in the
1112 faith that they will not malloc. */
1114 /* Normally, this is fine. */
1115 #define MATCH_MAY_ALLOCATE
1117 /* When using GNU C, we are not REALLY using the C alloca, no matter
1118 what config.h may say. So don't take precautions for it. */
1123 /* The match routines may not allocate if (1) they would do it with malloc
1124 and (2) it's not safe for them to use malloc.
1125 Note that if REL_ALLOC is defined, matching would not use malloc for the
1126 failure stack, but we would still use it for the register vectors;
1127 so REL_ALLOC should not affect this. */
1128 #if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1129 # undef MATCH_MAY_ALLOCATE
1133 /* Failure stack declarations and macros; both re_compile_fastmap and
1134 re_match_2 use a failure stack. These have to be macros because of
1135 REGEX_ALLOCATE_STACK. */
1138 /* Number of failure points for which to initially allocate space
1139 when matching. If this number is exceeded, we allocate more
1140 space, so it is not a hard limit. */
1141 #ifndef INIT_FAILURE_ALLOC
1142 # define INIT_FAILURE_ALLOC 5
1145 /* Roughly the maximum number of failure points on the stack. Would be
1146 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1147 This is a variable only so users of regex can assign to it; we never
1148 change it ourselves. */
1152 # if defined MATCH_MAY_ALLOCATE
1153 /* 4400 was enough to cause a crash on Alpha OSF/1,
1154 whose default stack limit is 2mb. */
1155 long int re_max_failures = 4000;
1157 long int re_max_failures = 2000;
1160 union fail_stack_elt
1162 unsigned char *pointer;
1166 typedef union fail_stack_elt fail_stack_elt_t;
1170 fail_stack_elt_t *stack;
1171 unsigned long int size;
1172 unsigned long int avail; /* Offset of next open position. */
1175 #else /* not INT_IS_16BIT */
1177 # if defined MATCH_MAY_ALLOCATE
1178 /* 4400 was enough to cause a crash on Alpha OSF/1,
1179 whose default stack limit is 2mb. */
1180 int re_max_failures = 20000;
1182 int re_max_failures = 2000;
1185 union fail_stack_elt
1187 unsigned char *pointer;
1191 typedef union fail_stack_elt fail_stack_elt_t;
1195 fail_stack_elt_t *stack;
1197 unsigned avail; /* Offset of next open position. */
1200 #endif /* INT_IS_16BIT */
1202 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1203 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1204 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1207 /* Define macros to initialize and free the failure stack.
1208 Do `return -2' if the alloc fails. */
1210 #ifdef MATCH_MAY_ALLOCATE
1211 # define INIT_FAIL_STACK() \
1213 fail_stack.stack = (fail_stack_elt_t *) \
1214 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1216 if (fail_stack.stack == NULL) \
1219 fail_stack.size = INIT_FAILURE_ALLOC; \
1220 fail_stack.avail = 0; \
1223 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1225 # define INIT_FAIL_STACK() \
1227 fail_stack.avail = 0; \
1230 # define RESET_FAIL_STACK()
1234 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1236 Return 1 if succeeds, and 0 if either ran out of memory
1237 allocating space for it or it was already too large.
1239 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1241 #define DOUBLE_FAIL_STACK(fail_stack) \
1242 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1244 : ((fail_stack).stack = (fail_stack_elt_t *) \
1245 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1246 (fail_stack).size * sizeof (fail_stack_elt_t), \
1247 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1249 (fail_stack).stack == NULL \
1251 : ((fail_stack).size <<= 1, \
1255 /* Push pointer POINTER on FAIL_STACK.
1256 Return 1 if was able to do so and 0 if ran out of memory allocating
1258 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1259 ((FAIL_STACK_FULL () \
1260 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1262 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1265 /* Push a pointer value onto the failure stack.
1266 Assumes the variable `fail_stack'. Probably should only
1267 be called from within `PUSH_FAILURE_POINT'. */
1268 #define PUSH_FAILURE_POINTER(item) \
1269 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1271 /* This pushes an integer-valued item onto the failure stack.
1272 Assumes the variable `fail_stack'. Probably should only
1273 be called from within `PUSH_FAILURE_POINT'. */
1274 #define PUSH_FAILURE_INT(item) \
1275 fail_stack.stack[fail_stack.avail++].integer = (item)
1277 /* Push a fail_stack_elt_t value onto the failure stack.
1278 Assumes the variable `fail_stack'. Probably should only
1279 be called from within `PUSH_FAILURE_POINT'. */
1280 #define PUSH_FAILURE_ELT(item) \
1281 fail_stack.stack[fail_stack.avail++] = (item)
1283 /* These three POP... operations complement the three PUSH... operations.
1284 All assume that `fail_stack' is nonempty. */
1285 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1286 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1287 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1289 /* Used to omit pushing failure point id's when we're not debugging. */
1291 # define DEBUG_PUSH PUSH_FAILURE_INT
1292 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1294 # define DEBUG_PUSH(item)
1295 # define DEBUG_POP(item_addr)
1299 /* Push the information about the state we will need
1300 if we ever fail back to it.
1302 Requires variables fail_stack, regstart, regend, reg_info, and
1303 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1306 Does `return FAILURE_CODE' if runs out of memory. */
1308 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1310 char *destination; \
1311 /* Must be int, so when we don't save any registers, the arithmetic \
1312 of 0 + -1 isn't done as unsigned. */ \
1313 /* Can't be int, since there is not a shred of a guarantee that int \
1314 is wide enough to hold a value of something to which pointer can \
1316 active_reg_t this_reg; \
1318 DEBUG_STATEMENT (failure_id++); \
1319 DEBUG_STATEMENT (nfailure_points_pushed++); \
1320 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1321 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1322 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1324 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1325 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1327 /* Ensure we have enough space allocated for what we will push. */ \
1328 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1330 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1331 return failure_code; \
1333 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1334 (fail_stack).size); \
1335 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1338 /* Push the info, starting with the registers. */ \
1339 DEBUG_PRINT1 ("\n"); \
1342 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1345 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1346 DEBUG_STATEMENT (num_regs_pushed++); \
1348 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1349 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1351 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1352 PUSH_FAILURE_POINTER (regend[this_reg]); \
1354 DEBUG_PRINT2 (" info: %p\n ", \
1355 reg_info[this_reg].word.pointer); \
1356 DEBUG_PRINT2 (" match_null=%d", \
1357 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1358 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1359 DEBUG_PRINT2 (" matched_something=%d", \
1360 MATCHED_SOMETHING (reg_info[this_reg])); \
1361 DEBUG_PRINT2 (" ever_matched=%d", \
1362 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1363 DEBUG_PRINT1 ("\n"); \
1364 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1367 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1368 PUSH_FAILURE_INT (lowest_active_reg); \
1370 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1371 PUSH_FAILURE_INT (highest_active_reg); \
1373 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1374 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1375 PUSH_FAILURE_POINTER (pattern_place); \
1377 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1378 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1380 DEBUG_PRINT1 ("'\n"); \
1381 PUSH_FAILURE_POINTER (string_place); \
1383 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1384 DEBUG_PUSH (failure_id); \
1387 /* This is the number of items that are pushed and popped on the stack
1388 for each register. */
1389 #define NUM_REG_ITEMS 3
1391 /* Individual items aside from the registers. */
1393 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1395 # define NUM_NONREG_ITEMS 4
1398 /* We push at most this many items on the stack. */
1399 /* We used to use (num_regs - 1), which is the number of registers
1400 this regexp will save; but that was changed to 5
1401 to avoid stack overflow for a regexp with lots of parens. */
1402 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1404 /* We actually push this many items. */
1405 #define NUM_FAILURE_ITEMS \
1407 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1411 /* How many items can still be added to the stack without overflowing it. */
1412 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1415 /* Pops what PUSH_FAIL_STACK pushes.
1417 We restore into the parameters, all of which should be lvalues:
1418 STR -- the saved data position.
1419 PAT -- the saved pattern position.
1420 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1421 REGSTART, REGEND -- arrays of string positions.
1422 REG_INFO -- array of information about each subexpression.
1424 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1425 `pend', `string1', `size1', `string2', and `size2'. */
1427 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1429 DEBUG_STATEMENT (unsigned failure_id;) \
1430 active_reg_t this_reg; \
1431 const unsigned char *string_temp; \
1433 assert (!FAIL_STACK_EMPTY ()); \
1435 /* Remove failure points and point to how many regs pushed. */ \
1436 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1437 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1438 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1440 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1442 DEBUG_POP (&failure_id); \
1443 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1445 /* If the saved string location is NULL, it came from an \
1446 on_failure_keep_string_jump opcode, and we want to throw away the \
1447 saved NULL, thus retaining our current position in the string. */ \
1448 string_temp = POP_FAILURE_POINTER (); \
1449 if (string_temp != NULL) \
1450 str = (const char *) string_temp; \
1452 DEBUG_PRINT2 (" Popping string %p: `", str); \
1453 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1454 DEBUG_PRINT1 ("'\n"); \
1456 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1457 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1458 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1460 /* Restore register info. */ \
1461 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1462 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1464 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1465 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1468 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1470 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1472 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1473 DEBUG_PRINT2 (" info: %p\n", \
1474 reg_info[this_reg].word.pointer); \
1476 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1477 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1479 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1480 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1484 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1486 reg_info[this_reg].word.integer = 0; \
1487 regend[this_reg] = 0; \
1488 regstart[this_reg] = 0; \
1490 highest_active_reg = high_reg; \
1493 set_regs_matched_done = 0; \
1494 DEBUG_STATEMENT (nfailure_points_popped++); \
1495 } /* POP_FAILURE_POINT */
1499 /* Structure for per-register (a.k.a. per-group) information.
1500 Other register information, such as the
1501 starting and ending positions (which are addresses), and the list of
1502 inner groups (which is a bits list) are maintained in separate
1505 We are making a (strictly speaking) nonportable assumption here: that
1506 the compiler will pack our bit fields into something that fits into
1507 the type of `word', i.e., is something that fits into one item on the
1511 /* Declarations and macros for re_match_2. */
1515 fail_stack_elt_t word;
1518 /* This field is one if this group can match the empty string,
1519 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1520 #define MATCH_NULL_UNSET_VALUE 3
1521 unsigned match_null_string_p : 2;
1522 unsigned is_active : 1;
1523 unsigned matched_something : 1;
1524 unsigned ever_matched_something : 1;
1526 } register_info_type;
1528 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1529 #define IS_ACTIVE(R) ((R).bits.is_active)
1530 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1531 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1534 /* Call this when have matched a real character; it sets `matched' flags
1535 for the subexpressions which we are currently inside. Also records
1536 that those subexprs have matched. */
1537 #define SET_REGS_MATCHED() \
1540 if (!set_regs_matched_done) \
1543 set_regs_matched_done = 1; \
1544 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1546 MATCHED_SOMETHING (reg_info[r]) \
1547 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1554 /* Registers are set to a sentinel when they haven't yet matched. */
1555 static char reg_unset_dummy;
1556 #define REG_UNSET_VALUE (®_unset_dummy)
1557 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1559 /* Subroutine declarations and macros for regex_compile. */
1561 static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
1562 reg_syntax_t syntax,
1563 struct re_pattern_buffer *bufp));
1564 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1565 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1566 int arg1, int arg2));
1567 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1568 int arg, unsigned char *end));
1569 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1570 int arg1, int arg2, unsigned char *end));
1571 static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
1572 reg_syntax_t syntax));
1573 static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
1574 reg_syntax_t syntax));
1575 static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr,
1578 reg_syntax_t syntax,
1581 /* Fetch the next character in the uncompiled pattern---translating it
1582 if necessary. Also cast from a signed character in the constant
1583 string passed to us by the user to an unsigned char that we can use
1584 as an array index (in, e.g., `translate'). */
1586 # define PATFETCH(c) \
1587 do {if (p == pend) return REG_EEND; \
1588 c = (unsigned char) *p++; \
1589 if (translate) c = (unsigned char) translate[c]; \
1593 /* Fetch the next character in the uncompiled pattern, with no
1595 #define PATFETCH_RAW(c) \
1596 do {if (p == pend) return REG_EEND; \
1597 c = (unsigned char) *p++; \
1600 /* Go backwards one character in the pattern. */
1601 #define PATUNFETCH p--
1604 /* If `translate' is non-null, return translate[D], else just D. We
1605 cast the subscript to translate because some data is declared as
1606 `char *', to avoid warnings when a string constant is passed. But
1607 when we use a character as a subscript we must make it unsigned. */
1609 # define TRANSLATE(d) \
1610 (translate ? (char) translate[(unsigned char) (d)] : (d))
1614 /* Macros for outputting the compiled pattern into `buffer'. */
1616 /* If the buffer isn't allocated when it comes in, use this. */
1617 #define INIT_BUF_SIZE 32
1619 /* Make sure we have at least N more bytes of space in buffer. */
1620 #define GET_BUFFER_SPACE(n) \
1621 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1624 /* Make sure we have one more byte of buffer space and then add C to it. */
1625 #define BUF_PUSH(c) \
1627 GET_BUFFER_SPACE (1); \
1628 *b++ = (unsigned char) (c); \
1632 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1633 #define BUF_PUSH_2(c1, c2) \
1635 GET_BUFFER_SPACE (2); \
1636 *b++ = (unsigned char) (c1); \
1637 *b++ = (unsigned char) (c2); \
1641 /* As with BUF_PUSH_2, except for three bytes. */
1642 #define BUF_PUSH_3(c1, c2, c3) \
1644 GET_BUFFER_SPACE (3); \
1645 *b++ = (unsigned char) (c1); \
1646 *b++ = (unsigned char) (c2); \
1647 *b++ = (unsigned char) (c3); \
1651 /* Store a jump with opcode OP at LOC to location TO. We store a
1652 relative address offset by the three bytes the jump itself occupies. */
1653 #define STORE_JUMP(op, loc, to) \
1654 store_op1 (op, loc, (int) ((to) - (loc) - 3))
1656 /* Likewise, for a two-argument jump. */
1657 #define STORE_JUMP2(op, loc, to, arg) \
1658 store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
1660 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1661 #define INSERT_JUMP(op, loc, to) \
1662 insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
1664 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1665 #define INSERT_JUMP2(op, loc, to, arg) \
1666 insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
1669 /* This is not an arbitrary limit: the arguments which represent offsets
1670 into the pattern are two bytes long. So if 2^16 bytes turns out to
1671 be too small, many things would have to change. */
1672 /* Any other compiler which, like MSC, has allocation limit below 2^16
1673 bytes will have to use approach similar to what was done below for
1674 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1675 reallocating to 0 bytes. Such thing is not going to work too well.
1676 You have been warned!! */
1677 #if defined _MSC_VER && !defined _WIN32
1678 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
1679 The REALLOC define eliminates a flurry of conversion warnings,
1680 but is not required. */
1681 # define MAX_BUF_SIZE 65500L
1682 # define REALLOC(p,s) realloc ((p), (size_t) (s))
1684 # define MAX_BUF_SIZE (1L << 16)
1685 # define REALLOC(p,s) realloc ((p), (s))
1688 /* Extend the buffer by twice its current size via realloc and
1689 reset the pointers that pointed into the old block to point to the
1690 correct places in the new one. If extending the buffer results in it
1691 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1692 #define EXTEND_BUFFER() \
1694 unsigned char *old_buffer = bufp->buffer; \
1695 if (bufp->allocated == MAX_BUF_SIZE) \
1697 bufp->allocated <<= 1; \
1698 if (bufp->allocated > MAX_BUF_SIZE) \
1699 bufp->allocated = MAX_BUF_SIZE; \
1700 bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
1701 if (bufp->buffer == NULL) \
1702 return REG_ESPACE; \
1703 /* If the buffer moved, move all the pointers into it. */ \
1704 if (old_buffer != bufp->buffer) \
1706 b = (b - old_buffer) + bufp->buffer; \
1707 begalt = (begalt - old_buffer) + bufp->buffer; \
1708 if (fixup_alt_jump) \
1709 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1711 laststart = (laststart - old_buffer) + bufp->buffer; \
1712 if (pending_exact) \
1713 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1718 /* Since we have one byte reserved for the register number argument to
1719 {start,stop}_memory, the maximum number of groups we can report
1720 things about is what fits in that byte. */
1721 #define MAX_REGNUM 255
1723 /* But patterns can have more than `MAX_REGNUM' registers. We just
1724 ignore the excess. */
1725 typedef unsigned regnum_t;
1728 /* Macros for the compile stack. */
1730 /* Since offsets can go either forwards or backwards, this type needs to
1731 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1732 /* int may be not enough when sizeof(int) == 2. */
1733 typedef long pattern_offset_t;
1737 pattern_offset_t begalt_offset;
1738 pattern_offset_t fixup_alt_jump;
1739 pattern_offset_t inner_group_offset;
1740 pattern_offset_t laststart_offset;
1742 } compile_stack_elt_t;
1747 compile_stack_elt_t *stack;
1749 unsigned avail; /* Offset of next open position. */
1750 } compile_stack_type;
1753 #define INIT_COMPILE_STACK_SIZE 32
1755 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1756 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1758 /* The next available element. */
1759 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1762 /* Set the bit for character C in a list. */
1763 #define SET_LIST_BIT(c) \
1764 (b[((unsigned char) (c)) / BYTEWIDTH] \
1765 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1768 /* Get the next unsigned number in the uncompiled pattern. */
1769 #define GET_UNSIGNED_NUMBER(num) \
1773 while (ISDIGIT (c)) \
1777 num = num * 10 + c - '0'; \
1785 #if defined _LIBC || WIDE_CHAR_SUPPORT
1786 /* The GNU C library provides support for user-defined character classes
1787 and the functions from ISO C amendement 1. */
1788 # ifdef CHARCLASS_NAME_MAX
1789 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1791 /* This shouldn't happen but some implementation might still have this
1792 problem. Use a reasonable default value. */
1793 # define CHAR_CLASS_MAX_LENGTH 256
1797 # define IS_CHAR_CLASS(string) __wctype (string)
1799 # define IS_CHAR_CLASS(string) wctype (string)
1802 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1804 # define IS_CHAR_CLASS(string) \
1805 (STREQ (string, "alpha") || STREQ (string, "upper") \
1806 || STREQ (string, "lower") || STREQ (string, "digit") \
1807 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1808 || STREQ (string, "space") || STREQ (string, "print") \
1809 || STREQ (string, "punct") || STREQ (string, "graph") \
1810 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1813 #ifndef MATCH_MAY_ALLOCATE
1815 /* If we cannot allocate large objects within re_match_2_internal,
1816 we make the fail stack and register vectors global.
1817 The fail stack, we grow to the maximum size when a regexp
1819 The register vectors, we adjust in size each time we
1820 compile a regexp, according to the number of registers it needs. */
1822 static fail_stack_type fail_stack;
1824 /* Size with which the following vectors are currently allocated.
1825 That is so we can make them bigger as needed,
1826 but never make them smaller. */
1827 static int regs_allocated_size;
1829 static const char ** regstart, ** regend;
1830 static const char ** old_regstart, ** old_regend;
1831 static const char **best_regstart, **best_regend;
1832 static register_info_type *reg_info;
1833 static const char **reg_dummy;
1834 static register_info_type *reg_info_dummy;
1836 /* Make the register vectors big enough for NUM_REGS registers,
1837 but don't make them smaller. */
1840 regex_grow_registers (num_regs)
1843 if (num_regs > regs_allocated_size)
1845 RETALLOC_IF (regstart, num_regs, const char *);
1846 RETALLOC_IF (regend, num_regs, const char *);
1847 RETALLOC_IF (old_regstart, num_regs, const char *);
1848 RETALLOC_IF (old_regend, num_regs, const char *);
1849 RETALLOC_IF (best_regstart, num_regs, const char *);
1850 RETALLOC_IF (best_regend, num_regs, const char *);
1851 RETALLOC_IF (reg_info, num_regs, register_info_type);
1852 RETALLOC_IF (reg_dummy, num_regs, const char *);
1853 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1855 regs_allocated_size = num_regs;
1859 #endif /* not MATCH_MAY_ALLOCATE */
1861 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
1865 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1866 Returns one of error codes defined in `regex.h', or zero for success.
1868 Assumes the `allocated' (and perhaps `buffer') and `translate'
1869 fields are set in BUFP on entry.
1871 If it succeeds, results are put in BUFP (if it returns an error, the
1872 contents of BUFP are undefined):
1873 `buffer' is the compiled pattern;
1874 `syntax' is set to SYNTAX;
1875 `used' is set to the length of the compiled pattern;
1876 `fastmap_accurate' is zero;
1877 `re_nsub' is the number of subexpressions in PATTERN;
1878 `not_bol' and `not_eol' are zero;
1880 The `fastmap' and `newline_anchor' fields are neither
1881 examined nor set. */
1883 /* Return, freeing storage we allocated. */
1884 #define FREE_STACK_RETURN(value) \
1885 return (free (compile_stack.stack), value)
1887 static reg_errcode_t
1888 regex_compile (pattern, size, syntax, bufp)
1889 const char *pattern;
1891 reg_syntax_t syntax;
1892 struct re_pattern_buffer *bufp;
1894 /* We fetch characters from PATTERN here. Even though PATTERN is
1895 `char *' (i.e., signed), we declare these variables as unsigned, so
1896 they can be reliably used as array indices. */
1897 register unsigned char c, c1;
1899 /* A random temporary spot in PATTERN. */
1902 /* Points to the end of the buffer, where we should append. */
1903 register unsigned char *b;
1905 /* Keeps track of unclosed groups. */
1906 compile_stack_type compile_stack;
1908 /* Points to the current (ending) position in the pattern. */
1909 const char *p = pattern;
1910 const char *pend = pattern + size;
1912 /* How to translate the characters in the pattern. */
1913 RE_TRANSLATE_TYPE translate = bufp->translate;
1915 /* Address of the count-byte of the most recently inserted `exactn'
1916 command. This makes it possible to tell if a new exact-match
1917 character can be added to that command or if the character requires
1918 a new `exactn' command. */
1919 unsigned char *pending_exact = 0;
1921 /* Address of start of the most recently finished expression.
1922 This tells, e.g., postfix * where to find the start of its
1923 operand. Reset at the beginning of groups and alternatives. */
1924 unsigned char *laststart = 0;
1926 /* Address of beginning of regexp, or inside of last group. */
1927 unsigned char *begalt;
1929 /* Place in the uncompiled pattern (i.e., the {) to
1930 which to go back if the interval is invalid. */
1931 const char *beg_interval;
1933 /* Address of the place where a forward jump should go to the end of
1934 the containing expression. Each alternative of an `or' -- except the
1935 last -- ends with a forward jump of this sort. */
1936 unsigned char *fixup_alt_jump = 0;
1938 /* Counts open-groups as they are encountered. Remembered for the
1939 matching close-group on the compile stack, so the same register
1940 number is put in the stop_memory as the start_memory. */
1941 regnum_t regnum = 0;
1944 DEBUG_PRINT1 ("\nCompiling pattern: ");
1947 unsigned debug_count;
1949 for (debug_count = 0; debug_count < size; debug_count++)
1950 putchar (pattern[debug_count]);
1955 /* Initialize the compile stack. */
1956 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1957 if (compile_stack.stack == NULL)
1960 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1961 compile_stack.avail = 0;
1963 /* Initialize the pattern buffer. */
1964 bufp->syntax = syntax;
1965 bufp->fastmap_accurate = 0;
1966 bufp->not_bol = bufp->not_eol = 0;
1968 /* Set `used' to zero, so that if we return an error, the pattern
1969 printer (for debugging) will think there's no pattern. We reset it
1973 /* Always count groups, whether or not bufp->no_sub is set. */
1976 #if !defined emacs && !defined SYNTAX_TABLE
1977 /* Initialize the syntax table. */
1978 init_syntax_once ();
1981 if (bufp->allocated == 0)
1984 { /* If zero allocated, but buffer is non-null, try to realloc
1985 enough space. This loses if buffer's address is bogus, but
1986 that is the user's responsibility. */
1987 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1990 { /* Caller did not allocate a buffer. Do it for them. */
1991 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1993 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1995 bufp->allocated = INIT_BUF_SIZE;
1998 begalt = b = bufp->buffer;
2000 /* Loop through the uncompiled pattern until we're at the end. */
2009 if ( /* If at start of pattern, it's an operator. */
2011 /* If context independent, it's an operator. */
2012 || syntax & RE_CONTEXT_INDEP_ANCHORS
2013 /* Otherwise, depends on what's come before. */
2014 || at_begline_loc_p (pattern, p, syntax))
2024 if ( /* If at end of pattern, it's an operator. */
2026 /* If context independent, it's an operator. */
2027 || syntax & RE_CONTEXT_INDEP_ANCHORS
2028 /* Otherwise, depends on what's next. */
2029 || at_endline_loc_p (p, pend, syntax))
2039 if ((syntax & RE_BK_PLUS_QM)
2040 || (syntax & RE_LIMITED_OPS))
2044 /* If there is no previous pattern... */
2047 if (syntax & RE_CONTEXT_INVALID_OPS)
2048 FREE_STACK_RETURN (REG_BADRPT);
2049 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2054 /* Are we optimizing this jump? */
2055 boolean keep_string_p = false;
2057 /* 1 means zero (many) matches is allowed. */
2058 char zero_times_ok = 0, many_times_ok = 0;
2060 /* If there is a sequence of repetition chars, collapse it
2061 down to just one (the right one). We can't combine
2062 interval operators with these because of, e.g., `a{2}*',
2063 which should only match an even number of `a's. */
2067 zero_times_ok |= c != '+';
2068 many_times_ok |= c != '?';
2076 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2079 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2081 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2084 if (!(c1 == '+' || c1 == '?'))
2099 /* If we get here, we found another repeat character. */
2102 /* Star, etc. applied to an empty pattern is equivalent
2103 to an empty pattern. */
2107 /* Now we know whether or not zero matches is allowed
2108 and also whether or not two or more matches is allowed. */
2110 { /* More than one repetition is allowed, so put in at the
2111 end a backward relative jump from `b' to before the next
2112 jump we're going to put in below (which jumps from
2113 laststart to after this jump).
2115 But if we are at the `*' in the exact sequence `.*\n',
2116 insert an unconditional jump backwards to the .,
2117 instead of the beginning of the loop. This way we only
2118 push a failure point once, instead of every time
2119 through the loop. */
2120 assert (p - 1 > pattern);
2122 /* Allocate the space for the jump. */
2123 GET_BUFFER_SPACE (3);
2125 /* We know we are not at the first character of the pattern,
2126 because laststart was nonzero. And we've already
2127 incremented `p', by the way, to be the character after
2128 the `*'. Do we have to do something analogous here
2129 for null bytes, because of RE_DOT_NOT_NULL? */
2130 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2132 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2133 && !(syntax & RE_DOT_NEWLINE))
2134 { /* We have .*\n. */
2135 STORE_JUMP (jump, b, laststart);
2136 keep_string_p = true;
2139 /* Anything else. */
2140 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2142 /* We've added more stuff to the buffer. */
2146 /* On failure, jump from laststart to b + 3, which will be the
2147 end of the buffer after this jump is inserted. */
2148 GET_BUFFER_SPACE (3);
2149 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2157 /* At least one repetition is required, so insert a
2158 `dummy_failure_jump' before the initial
2159 `on_failure_jump' instruction of the loop. This
2160 effects a skip over that instruction the first time
2161 we hit that loop. */
2162 GET_BUFFER_SPACE (3);
2163 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2178 boolean had_char_class = false;
2180 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2182 /* Ensure that we have enough space to push a charset: the
2183 opcode, the length count, and the bitset; 34 bytes in all. */
2184 GET_BUFFER_SPACE (34);
2188 /* We test `*p == '^' twice, instead of using an if
2189 statement, so we only need one BUF_PUSH. */
2190 BUF_PUSH (*p == '^' ? charset_not : charset);
2194 /* Remember the first position in the bracket expression. */
2197 /* Push the number of bytes in the bitmap. */
2198 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2200 /* Clear the whole map. */
2201 memset (b, 0, (1 << BYTEWIDTH) / BYTEWIDTH);
2203 /* charset_not matches newline according to a syntax bit. */
2204 if ((re_opcode_t) b[-2] == charset_not
2205 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2206 SET_LIST_BIT ('\n');
2208 /* Read in characters and ranges, setting map bits. */
2211 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2215 /* \ might escape characters inside [...] and [^...]. */
2216 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2218 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2225 /* Could be the end of the bracket expression. If it's
2226 not (i.e., when the bracket expression is `[]' so
2227 far), the ']' character bit gets set way below. */
2228 if (c == ']' && p != p1 + 1)
2231 /* Look ahead to see if it's a range when the last thing
2232 was a character class. */
2233 if (had_char_class && c == '-' && *p != ']')
2234 FREE_STACK_RETURN (REG_ERANGE);
2236 /* Look ahead to see if it's a range when the last thing
2237 was a character: if this is a hyphen not at the
2238 beginning or the end of a list, then it's the range
2241 && !(p - 2 >= pattern && p[-2] == '[')
2242 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2246 = compile_range (&p, pend, translate, syntax, b);
2247 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2250 else if (p[0] == '-' && p[1] != ']')
2251 { /* This handles ranges made up of characters only. */
2254 /* Move past the `-'. */
2257 ret = compile_range (&p, pend, translate, syntax, b);
2258 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2261 /* See if we're at the beginning of a possible character
2264 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2265 { /* Leave room for the null. */
2266 char str[CHAR_CLASS_MAX_LENGTH + 1];
2271 /* If pattern is `[[:'. */
2272 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2277 if ((c == ':' && *p == ']') || p == pend)
2279 if (c1 < CHAR_CLASS_MAX_LENGTH)
2282 /* This is in any case an invalid class name. */
2287 /* If isn't a word bracketed by `[:' and `:]':
2288 undo the ending character, the letters, and leave
2289 the leading `:' and `[' (but set bits for them). */
2290 if (c == ':' && *p == ']')
2292 #if defined _LIBC || WIDE_CHAR_SUPPORT
2293 boolean is_lower = STREQ (str, "lower");
2294 boolean is_upper = STREQ (str, "upper");
2298 wt = IS_CHAR_CLASS (str);
2300 FREE_STACK_RETURN (REG_ECTYPE);
2302 /* Throw away the ] at the end of the character
2306 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2308 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2311 if (__iswctype (__btowc (ch), wt))
2314 if (iswctype (btowc (ch), wt))
2318 if (translate && (is_upper || is_lower)
2319 && (ISUPPER (ch) || ISLOWER (ch)))
2323 had_char_class = true;
2326 boolean is_alnum = STREQ (str, "alnum");
2327 boolean is_alpha = STREQ (str, "alpha");
2328 boolean is_blank = STREQ (str, "blank");
2329 boolean is_cntrl = STREQ (str, "cntrl");
2330 boolean is_digit = STREQ (str, "digit");
2331 boolean is_graph = STREQ (str, "graph");
2332 boolean is_lower = STREQ (str, "lower");
2333 boolean is_print = STREQ (str, "print");
2334 boolean is_punct = STREQ (str, "punct");
2335 boolean is_space = STREQ (str, "space");
2336 boolean is_upper = STREQ (str, "upper");
2337 boolean is_xdigit = STREQ (str, "xdigit");
2339 if (!IS_CHAR_CLASS (str))
2340 FREE_STACK_RETURN (REG_ECTYPE);
2342 /* Throw away the ] at the end of the character
2346 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2348 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2350 /* This was split into 3 if's to
2351 avoid an arbitrary limit in some compiler. */
2352 if ( (is_alnum && ISALNUM (ch))
2353 || (is_alpha && ISALPHA (ch))
2354 || (is_blank && ISBLANK (ch))
2355 || (is_cntrl && ISCNTRL (ch)))
2357 if ( (is_digit && ISDIGIT (ch))
2358 || (is_graph && ISGRAPH (ch))
2359 || (is_lower && ISLOWER (ch))
2360 || (is_print && ISPRINT (ch)))
2362 if ( (is_punct && ISPUNCT (ch))
2363 || (is_space && ISSPACE (ch))
2364 || (is_upper && ISUPPER (ch))
2365 || (is_xdigit && ISXDIGIT (ch)))
2367 if ( translate && (is_upper || is_lower)
2368 && (ISUPPER (ch) || ISLOWER (ch)))
2371 had_char_class = true;
2372 #endif /* libc || wctype.h */
2381 had_char_class = false;
2386 had_char_class = false;
2391 /* Discard any (non)matching list bytes that are all 0 at the
2392 end of the map. Decrease the map-length byte too. */
2393 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2401 if (syntax & RE_NO_BK_PARENS)
2408 if (syntax & RE_NO_BK_PARENS)
2415 if (syntax & RE_NEWLINE_ALT)
2422 if (syntax & RE_NO_BK_VBAR)
2429 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2430 goto handle_interval;
2436 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2438 /* Do not translate the character after the \, so that we can
2439 distinguish, e.g., \B from \b, even if we normally would
2440 translate, e.g., B to b. */
2446 if (syntax & RE_NO_BK_PARENS)
2447 goto normal_backslash;
2453 if (COMPILE_STACK_FULL)
2455 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2456 compile_stack_elt_t);
2457 if (compile_stack.stack == NULL) return REG_ESPACE;
2459 compile_stack.size <<= 1;
2462 /* These are the values to restore when we hit end of this
2463 group. They are all relative offsets, so that if the
2464 whole pattern moves because of realloc, they will still
2466 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2467 COMPILE_STACK_TOP.fixup_alt_jump
2468 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2469 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2470 COMPILE_STACK_TOP.regnum = regnum;
2472 /* We will eventually replace the 0 with the number of
2473 groups inner to this one. But do not push a
2474 start_memory for groups beyond the last one we can
2475 represent in the compiled pattern. */
2476 if (regnum <= MAX_REGNUM)
2478 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2479 BUF_PUSH_3 (start_memory, regnum, 0);
2482 compile_stack.avail++;
2487 /* If we've reached MAX_REGNUM groups, then this open
2488 won't actually generate any code, so we'll have to
2489 clear pending_exact explicitly. */
2495 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2497 if (COMPILE_STACK_EMPTY)
2499 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2500 goto normal_backslash;
2502 FREE_STACK_RETURN (REG_ERPAREN);
2507 { /* Push a dummy failure point at the end of the
2508 alternative for a possible future
2509 `pop_failure_jump' to pop. See comments at
2510 `push_dummy_failure' in `re_match_2'. */
2511 BUF_PUSH (push_dummy_failure);
2513 /* We allocated space for this jump when we assigned
2514 to `fixup_alt_jump', in the `handle_alt' case below. */
2515 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2518 /* See similar code for backslashed left paren above. */
2519 if (COMPILE_STACK_EMPTY)
2521 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2524 FREE_STACK_RETURN (REG_ERPAREN);
2527 /* Since we just checked for an empty stack above, this
2528 ``can't happen''. */
2529 assert (compile_stack.avail != 0);
2531 /* We don't just want to restore into `regnum', because
2532 later groups should continue to be numbered higher,
2533 as in `(ab)c(de)' -- the second group is #2. */
2534 regnum_t this_group_regnum;
2536 compile_stack.avail--;
2537 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2539 = COMPILE_STACK_TOP.fixup_alt_jump
2540 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2542 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2543 this_group_regnum = COMPILE_STACK_TOP.regnum;
2544 /* If we've reached MAX_REGNUM groups, then this open
2545 won't actually generate any code, so we'll have to
2546 clear pending_exact explicitly. */
2549 /* We're at the end of the group, so now we know how many
2550 groups were inside this one. */
2551 if (this_group_regnum <= MAX_REGNUM)
2553 unsigned char *inner_group_loc
2554 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2556 *inner_group_loc = regnum - this_group_regnum;
2557 BUF_PUSH_3 (stop_memory, this_group_regnum,
2558 regnum - this_group_regnum);
2564 case '|': /* `\|'. */
2565 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2566 goto normal_backslash;
2568 if (syntax & RE_LIMITED_OPS)
2571 /* Insert before the previous alternative a jump which
2572 jumps to this alternative if the former fails. */
2573 GET_BUFFER_SPACE (3);
2574 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2578 /* The alternative before this one has a jump after it
2579 which gets executed if it gets matched. Adjust that
2580 jump so it will jump to this alternative's analogous
2581 jump (put in below, which in turn will jump to the next
2582 (if any) alternative's such jump, etc.). The last such
2583 jump jumps to the correct final destination. A picture:
2589 If we are at `b', then fixup_alt_jump right now points to a
2590 three-byte space after `a'. We'll put in the jump, set
2591 fixup_alt_jump to right after `b', and leave behind three
2592 bytes which we'll fill in when we get to after `c'. */
2595 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2597 /* Mark and leave space for a jump after this alternative,
2598 to be filled in later either by next alternative or
2599 when know we're at the end of a series of alternatives. */
2601 GET_BUFFER_SPACE (3);
2610 /* If \{ is a literal. */
2611 if (!(syntax & RE_INTERVALS)
2612 /* If we're at `\{' and it's not the open-interval
2614 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2615 || (p - 2 == pattern && p == pend))
2616 goto normal_backslash;
2620 /* If got here, then the syntax allows intervals. */
2622 /* At least (most) this many matches must be made. */
2623 int lower_bound = -1, upper_bound = -1;
2625 beg_interval = p - 1;
2629 if (syntax & RE_NO_BK_BRACES)
2630 goto unfetch_interval;
2632 FREE_STACK_RETURN (REG_EBRACE);
2635 GET_UNSIGNED_NUMBER (lower_bound);
2639 GET_UNSIGNED_NUMBER (upper_bound);
2640 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2643 /* Interval such as `{1}' => match exactly once. */
2644 upper_bound = lower_bound;
2646 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2647 || lower_bound > upper_bound)
2649 if (syntax & RE_NO_BK_BRACES)
2650 goto unfetch_interval;
2652 FREE_STACK_RETURN (REG_BADBR);
2655 if (!(syntax & RE_NO_BK_BRACES))
2657 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2664 if (syntax & RE_NO_BK_BRACES)
2665 goto unfetch_interval;
2667 FREE_STACK_RETURN (REG_BADBR);
2670 /* We just parsed a valid interval. */
2672 /* If it's invalid to have no preceding re. */
2675 if (syntax & RE_CONTEXT_INVALID_OPS)
2676 FREE_STACK_RETURN (REG_BADRPT);
2677 else if (syntax & RE_CONTEXT_INDEP_OPS)
2680 goto unfetch_interval;
2683 /* If the upper bound is zero, don't want to succeed at
2684 all; jump from `laststart' to `b + 3', which will be
2685 the end of the buffer after we insert the jump. */
2686 if (upper_bound == 0)
2688 GET_BUFFER_SPACE (3);
2689 INSERT_JUMP (jump, laststart, b + 3);
2693 /* Otherwise, we have a nontrivial interval. When
2694 we're all done, the pattern will look like:
2695 set_number_at <jump count> <upper bound>
2696 set_number_at <succeed_n count> <lower bound>
2697 succeed_n <after jump addr> <succeed_n count>
2699 jump_n <succeed_n addr> <jump count>
2700 (The upper bound and `jump_n' are omitted if
2701 `upper_bound' is 1, though.) */
2703 { /* If the upper bound is > 1, we need to insert
2704 more at the end of the loop. */
2705 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2707 GET_BUFFER_SPACE (nbytes);
2709 /* Initialize lower bound of the `succeed_n', even
2710 though it will be set during matching by its
2711 attendant `set_number_at' (inserted next),
2712 because `re_compile_fastmap' needs to know.
2713 Jump to the `jump_n' we might insert below. */
2714 INSERT_JUMP2 (succeed_n, laststart,
2715 b + 5 + (upper_bound > 1) * 5,
2719 /* Code to initialize the lower bound. Insert
2720 before the `succeed_n'. The `5' is the last two
2721 bytes of this `set_number_at', plus 3 bytes of
2722 the following `succeed_n'. */
2723 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2726 if (upper_bound > 1)
2727 { /* More than one repetition is allowed, so
2728 append a backward jump to the `succeed_n'
2729 that starts this interval.
2731 When we've reached this during matching,
2732 we'll have matched the interval once, so
2733 jump back only `upper_bound - 1' times. */
2734 STORE_JUMP2 (jump_n, b, laststart + 5,
2738 /* The location we want to set is the second
2739 parameter of the `jump_n'; that is `b-2' as
2740 an absolute address. `laststart' will be
2741 the `set_number_at' we're about to insert;
2742 `laststart+3' the number to set, the source
2743 for the relative address. But we are
2744 inserting into the middle of the pattern --
2745 so everything is getting moved up by 5.
2746 Conclusion: (b - 2) - (laststart + 3) + 5,
2747 i.e., b - laststart.
2749 We insert this at the beginning of the loop
2750 so that if we fail during matching, we'll
2751 reinitialize the bounds. */
2752 insert_op2 (set_number_at, laststart, b - laststart,
2753 upper_bound - 1, b);
2758 beg_interval = NULL;
2763 /* If an invalid interval, match the characters as literals. */
2764 assert (beg_interval);
2766 beg_interval = NULL;
2768 /* normal_char and normal_backslash need `c'. */
2771 if (!(syntax & RE_NO_BK_BRACES))
2773 if (p > pattern && p[-1] == '\\')
2774 goto normal_backslash;
2779 /* There is no way to specify the before_dot and after_dot
2780 operators. rms says this is ok. --karl */
2788 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2794 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2800 if (syntax & RE_NO_GNU_OPS)
2803 BUF_PUSH (wordchar);
2808 if (syntax & RE_NO_GNU_OPS)
2811 BUF_PUSH (notwordchar);
2816 if (syntax & RE_NO_GNU_OPS)
2822 if (syntax & RE_NO_GNU_OPS)
2828 if (syntax & RE_NO_GNU_OPS)
2830 BUF_PUSH (wordbound);
2834 if (syntax & RE_NO_GNU_OPS)
2836 BUF_PUSH (notwordbound);
2840 if (syntax & RE_NO_GNU_OPS)
2846 if (syntax & RE_NO_GNU_OPS)
2851 case '1': case '2': case '3': case '4': case '5':
2852 case '6': case '7': case '8': case '9':
2853 if (syntax & RE_NO_BK_REFS)
2859 FREE_STACK_RETURN (REG_ESUBREG);
2861 /* Can't back reference to a subexpression if inside of it. */
2862 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
2866 BUF_PUSH_2 (duplicate, c1);
2872 if (syntax & RE_BK_PLUS_QM)
2875 goto normal_backslash;
2879 /* You might think it would be useful for \ to mean
2880 not to translate; but if we don't translate it
2881 it will never match anything. */
2889 /* Expects the character in `c'. */
2891 /* If no exactn currently being built. */
2894 /* If last exactn not at current position. */
2895 || pending_exact + *pending_exact + 1 != b
2897 /* We have only one byte following the exactn for the count. */
2898 || *pending_exact == (1 << BYTEWIDTH) - 1
2900 /* If followed by a repetition operator. */
2901 || *p == '*' || *p == '^'
2902 || ((syntax & RE_BK_PLUS_QM)
2903 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2904 : (*p == '+' || *p == '?'))
2905 || ((syntax & RE_INTERVALS)
2906 && ((syntax & RE_NO_BK_BRACES)
2908 : (p[0] == '\\' && p[1] == '{'))))
2910 /* Start building a new exactn. */
2914 BUF_PUSH_2 (exactn, 0);
2915 pending_exact = b - 1;
2922 } /* while p != pend */
2925 /* Through the pattern now. */
2928 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2930 if (!COMPILE_STACK_EMPTY)
2931 FREE_STACK_RETURN (REG_EPAREN);
2933 /* If we don't want backtracking, force success
2934 the first time we reach the end of the compiled pattern. */
2935 if (syntax & RE_NO_POSIX_BACKTRACKING)
2938 free (compile_stack.stack);
2940 /* We have succeeded; set the length of the buffer. */
2941 bufp->used = b - bufp->buffer;
2946 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2947 print_compiled_pattern (bufp);
2951 #ifndef MATCH_MAY_ALLOCATE
2952 /* Initialize the failure stack to the largest possible stack. This
2953 isn't necessary unless we're trying to avoid calling alloca in
2954 the search and match routines. */
2956 int num_regs = bufp->re_nsub + 1;
2958 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2959 is strictly greater than re_max_failures, the largest possible stack
2960 is 2 * re_max_failures failure points. */
2961 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2963 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2966 if (! fail_stack.stack)
2968 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2969 * sizeof (fail_stack_elt_t));
2972 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2974 * sizeof (fail_stack_elt_t)));
2975 # else /* not emacs */
2976 if (! fail_stack.stack)
2978 = (fail_stack_elt_t *) malloc (fail_stack.size
2979 * sizeof (fail_stack_elt_t));
2982 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2984 * sizeof (fail_stack_elt_t)));
2985 # endif /* not emacs */
2988 regex_grow_registers (num_regs);
2990 #endif /* not MATCH_MAY_ALLOCATE */
2993 } /* regex_compile */
2995 /* Subroutines for `regex_compile'. */
2997 /* Store OP at LOC followed by two-byte integer parameter ARG. */
3000 store_op1 (op, loc, arg)
3005 *loc = (unsigned char) op;
3006 STORE_NUMBER (loc + 1, arg);
3010 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
3013 store_op2 (op, loc, arg1, arg2)
3018 *loc = (unsigned char) op;
3019 STORE_NUMBER (loc + 1, arg1);
3020 STORE_NUMBER (loc + 3, arg2);
3024 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
3025 for OP followed by two-byte integer parameter ARG. */
3028 insert_op1 (op, loc, arg, end)
3034 register unsigned char *pfrom = end;
3035 register unsigned char *pto = end + 3;
3037 while (pfrom != loc)
3040 store_op1 (op, loc, arg);
3044 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
3047 insert_op2 (op, loc, arg1, arg2, end)
3053 register unsigned char *pfrom = end;
3054 register unsigned char *pto = end + 5;
3056 while (pfrom != loc)
3059 store_op2 (op, loc, arg1, arg2);
3063 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
3064 after an alternative or a begin-subexpression. We assume there is at
3065 least one character before the ^. */
3068 at_begline_loc_p (pattern, p, syntax)
3069 const char *pattern, *p;
3070 reg_syntax_t syntax;
3072 const char *prev = p - 2;
3073 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
3076 /* After a subexpression? */
3077 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
3078 /* After an alternative? */
3079 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
3083 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3084 at least one character after the $, i.e., `P < PEND'. */
3087 at_endline_loc_p (p, pend, syntax)
3088 const char *p, *pend;
3089 reg_syntax_t syntax;
3091 const char *next = p;
3092 boolean next_backslash = *next == '\\';
3093 const char *next_next = p + 1 < pend ? p + 1 : 0;
3096 /* Before a subexpression? */
3097 (syntax & RE_NO_BK_PARENS ? *next == ')'
3098 : next_backslash && next_next && *next_next == ')')
3099 /* Before an alternative? */
3100 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3101 : next_backslash && next_next && *next_next == '|');
3105 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3106 false if it's not. */
3109 group_in_compile_stack (compile_stack, regnum)
3110 compile_stack_type compile_stack;
3115 for (this_element = compile_stack.avail - 1;
3118 if (compile_stack.stack[this_element].regnum == regnum)
3125 /* Read the ending character of a range (in a bracket expression) from the
3126 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3127 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3128 Then we set the translation of all bits between the starting and
3129 ending characters (inclusive) in the compiled pattern B.
3131 Return an error code.
3133 We use these short variable names so we can use the same macros as
3134 `regex_compile' itself. */
3136 static reg_errcode_t
3137 compile_range (p_ptr, pend, translate, syntax, b)
3138 const char **p_ptr, *pend;
3139 RE_TRANSLATE_TYPE translate;
3140 reg_syntax_t syntax;
3145 const char *p = *p_ptr;
3146 unsigned int range_start, range_end;
3151 /* Even though the pattern is a signed `char *', we need to fetch
3152 with unsigned char *'s; if the high bit of the pattern character
3153 is set, the range endpoints will be negative if we fetch using a
3156 We also want to fetch the endpoints without translating them; the
3157 appropriate translation is done in the bit-setting loop below. */
3158 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
3159 range_start = ((const unsigned char *) p)[-2];
3160 range_end = ((const unsigned char *) p)[0];
3162 /* Have to increment the pointer into the pattern string, so the
3163 caller isn't still at the ending character. */
3166 /* If the start is after the end, the range is empty. */
3167 if (range_start > range_end)
3168 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3170 /* Here we see why `this_char' has to be larger than an `unsigned
3171 char' -- the range is inclusive, so if `range_end' == 0xff
3172 (assuming 8-bit characters), we would otherwise go into an infinite
3173 loop, since all characters <= 0xff. */
3174 for (this_char = range_start; this_char <= range_end; this_char++)
3176 SET_LIST_BIT (TRANSLATE (this_char));
3182 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3183 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3184 characters can start a string that matches the pattern. This fastmap
3185 is used by re_search to skip quickly over impossible starting points.
3187 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3188 area as BUFP->fastmap.
3190 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3193 Returns 0 if we succeed, -2 if an internal error. */
3196 re_compile_fastmap (bufp)
3197 struct re_pattern_buffer *bufp;
3200 #ifdef MATCH_MAY_ALLOCATE
3201 fail_stack_type fail_stack;
3203 #ifndef REGEX_MALLOC
3207 register char *fastmap = bufp->fastmap;
3208 unsigned char *pattern = bufp->buffer;
3209 unsigned char *p = pattern;
3210 register unsigned char *pend = pattern + bufp->used;
3213 /* This holds the pointer to the failure stack, when
3214 it is allocated relocatably. */
3215 fail_stack_elt_t *failure_stack_ptr;
3218 /* Assume that each path through the pattern can be null until
3219 proven otherwise. We set this false at the bottom of switch
3220 statement, to which we get only if a particular path doesn't
3221 match the empty string. */
3222 boolean path_can_be_null = true;
3224 /* We aren't doing a `succeed_n' to begin with. */
3225 boolean succeed_n_p = false;
3227 assert (fastmap != NULL && p != NULL);
3230 memset (fastmap, 0, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3231 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3232 bufp->can_be_null = 0;
3236 if (p == pend || *p == succeed)
3238 /* We have reached the (effective) end of pattern. */
3239 if (!FAIL_STACK_EMPTY ())
3241 bufp->can_be_null |= path_can_be_null;
3243 /* Reset for next path. */
3244 path_can_be_null = true;
3246 p = fail_stack.stack[--fail_stack.avail].pointer;
3254 /* We should never be about to go beyond the end of the pattern. */
3257 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3260 /* I guess the idea here is to simply not bother with a fastmap
3261 if a backreference is used, since it's too hard to figure out
3262 the fastmap for the corresponding group. Setting
3263 `can_be_null' stops `re_search_2' from using the fastmap, so
3264 that is all we do. */
3266 bufp->can_be_null = 1;
3270 /* Following are the cases which match a character. These end
3279 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3280 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3286 /* Chars beyond end of map must be allowed. */
3287 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3290 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3291 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3297 for (j = 0; j < (1 << BYTEWIDTH); j++)
3298 if (SYNTAX (j) == Sword)
3304 for (j = 0; j < (1 << BYTEWIDTH); j++)
3305 if (SYNTAX (j) != Sword)
3312 int fastmap_newline = fastmap['\n'];
3314 /* `.' matches anything ... */
3315 for (j = 0; j < (1 << BYTEWIDTH); j++)
3318 /* ... except perhaps newline. */
3319 if (!(bufp->syntax & RE_DOT_NEWLINE))
3320 fastmap['\n'] = fastmap_newline;
3322 /* Return if we have already set `can_be_null'; if we have,
3323 then the fastmap is irrelevant. Something's wrong here. */
3324 else if (bufp->can_be_null)
3327 /* Otherwise, have to check alternative paths. */
3334 for (j = 0; j < (1 << BYTEWIDTH); j++)
3335 if (SYNTAX (j) == (enum syntaxcode) k)
3342 for (j = 0; j < (1 << BYTEWIDTH); j++)
3343 if (SYNTAX (j) != (enum syntaxcode) k)
3348 /* All cases after this match the empty string. These end with
3368 case push_dummy_failure:
3373 case pop_failure_jump:
3374 case maybe_pop_jump:
3377 case dummy_failure_jump:
3378 EXTRACT_NUMBER_AND_INCR (j, p);
3383 /* Jump backward implies we just went through the body of a
3384 loop and matched nothing. Opcode jumped to should be
3385 `on_failure_jump' or `succeed_n'. Just treat it like an
3386 ordinary jump. For a * loop, it has pushed its failure
3387 point already; if so, discard that as redundant. */
3388 if ((re_opcode_t) *p != on_failure_jump
3389 && (re_opcode_t) *p != succeed_n)
3393 EXTRACT_NUMBER_AND_INCR (j, p);
3396 /* If what's on the stack is where we are now, pop it. */
3397 if (!FAIL_STACK_EMPTY ()
3398 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3404 case on_failure_jump:
3405 case on_failure_keep_string_jump:
3406 handle_on_failure_jump:
3407 EXTRACT_NUMBER_AND_INCR (j, p);
3409 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3410 end of the pattern. We don't want to push such a point,
3411 since when we restore it above, entering the switch will
3412 increment `p' past the end of the pattern. We don't need
3413 to push such a point since we obviously won't find any more
3414 fastmap entries beyond `pend'. Such a pattern can match
3415 the null string, though. */
3418 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3420 RESET_FAIL_STACK ();
3425 bufp->can_be_null = 1;
3429 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3430 succeed_n_p = false;
3437 /* Get to the number of times to succeed. */
3440 /* Increment p past the n for when k != 0. */
3441 EXTRACT_NUMBER_AND_INCR (k, p);
3445 succeed_n_p = true; /* Spaghetti code alert. */
3446 goto handle_on_failure_jump;
3463 abort (); /* We have listed all the cases. */
3466 /* Getting here means we have found the possible starting
3467 characters for one path of the pattern -- and that the empty
3468 string does not match. We need not follow this path further.
3469 Instead, look at the next alternative (remembered on the
3470 stack), or quit if no more. The test at the top of the loop
3471 does these things. */
3472 path_can_be_null = false;
3476 /* Set `can_be_null' for the last path (also the first path, if the
3477 pattern is empty). */
3478 bufp->can_be_null |= path_can_be_null;
3481 RESET_FAIL_STACK ();
3483 } /* re_compile_fastmap */
3485 weak_alias (__re_compile_fastmap, re_compile_fastmap)
3488 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3489 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3490 this memory for recording register information. STARTS and ENDS
3491 must be allocated using the malloc library routine, and must each
3492 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3494 If NUM_REGS == 0, then subsequent matches should allocate their own
3497 Unless this function is called, the first search or match using
3498 PATTERN_BUFFER will allocate its own register data, without
3499 freeing the old data. */
3502 re_set_registers (bufp, regs, num_regs, starts, ends)
3503 struct re_pattern_buffer *bufp;
3504 struct re_registers *regs;
3506 regoff_t *starts, *ends;
3510 bufp->regs_allocated = REGS_REALLOCATE;
3511 regs->num_regs = num_regs;
3512 regs->start = starts;
3517 bufp->regs_allocated = REGS_UNALLOCATED;
3519 regs->start = regs->end = (regoff_t *) 0;
3523 weak_alias (__re_set_registers, re_set_registers)
3526 /* Searching routines. */
3528 /* Like re_search_2, below, but only one string is specified, and
3529 doesn't let you say where to stop matching. */
3532 re_search (bufp, string, size, startpos, range, regs)
3533 struct re_pattern_buffer *bufp;
3535 int size, startpos, range;
3536 struct re_registers *regs;
3538 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3542 weak_alias (__re_search, re_search)
3546 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3547 virtual concatenation of STRING1 and STRING2, starting first at index
3548 STARTPOS, then at STARTPOS + 1, and so on.
3550 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3552 RANGE is how far to scan while trying to match. RANGE = 0 means try
3553 only at STARTPOS; in general, the last start tried is STARTPOS +
3556 In REGS, return the indices of the virtual concatenation of STRING1
3557 and STRING2 that matched the entire BUFP->buffer and its contained
3560 Do not consider matching one past the index STOP in the virtual
3561 concatenation of STRING1 and STRING2.
3563 We return either the position in the strings at which the match was
3564 found, -1 if no match, or -2 if error (such as failure
3568 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3569 struct re_pattern_buffer *bufp;
3570 const char *string1, *string2;
3574 struct re_registers *regs;
3578 register char *fastmap = bufp->fastmap;
3579 register RE_TRANSLATE_TYPE translate = bufp->translate;
3580 int total_size = size1 + size2;
3581 int endpos = startpos + range;
3583 /* Check for out-of-range STARTPOS. */
3584 if (startpos < 0 || startpos > total_size)
3587 /* Fix up RANGE if it might eventually take us outside
3588 the virtual concatenation of STRING1 and STRING2.
3589 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3591 range = 0 - startpos;
3592 else if (endpos > total_size)
3593 range = total_size - startpos;
3595 /* If the search isn't to be a backwards one, don't waste time in a
3596 search for a pattern that must be anchored. */
3597 if (bufp->used > 0 && range > 0
3598 && ((re_opcode_t) bufp->buffer[0] == begbuf
3599 /* `begline' is like `begbuf' if it cannot match at newlines. */
3600 || ((re_opcode_t) bufp->buffer[0] == begline
3601 && !bufp->newline_anchor)))
3610 /* In a forward search for something that starts with \=.
3611 don't keep searching past point. */
3612 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3614 range = PT - startpos;
3620 /* Update the fastmap now if not correct already. */
3621 if (fastmap && !bufp->fastmap_accurate)
3622 if (re_compile_fastmap (bufp) == -2)
3625 /* Loop through the string, looking for a place to start matching. */
3628 /* If a fastmap is supplied, skip quickly over characters that
3629 cannot be the start of a match. If the pattern can match the
3630 null string, however, we don't need to skip characters; we want
3631 the first null string. */
3632 if (fastmap && startpos < total_size && !bufp->can_be_null)
3634 if (range > 0) /* Searching forwards. */
3636 register const char *d;
3637 register int lim = 0;
3640 if (startpos < size1 && startpos + range >= size1)
3641 lim = range - (size1 - startpos);
3643 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3645 /* Written out as an if-else to avoid testing `translate'
3649 && !fastmap[(unsigned char)
3650 translate[(unsigned char) *d++]])
3653 while (range > lim && !fastmap[(unsigned char) *d++])
3656 startpos += irange - range;
3658 else /* Searching backwards. */
3660 register char c = (size1 == 0 || startpos >= size1
3661 ? string2[startpos - size1]
3662 : string1[startpos]);
3664 if (!fastmap[(unsigned char) TRANSLATE (c)])
3669 /* If can't match the null string, and that's all we have left, fail. */
3670 if (range >= 0 && startpos == total_size && fastmap
3671 && !bufp->can_be_null)
3674 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3675 startpos, regs, stop);
3676 #ifndef REGEX_MALLOC
3705 weak_alias (__re_search_2, re_search_2)
3708 /* This converts PTR, a pointer into one of the search strings `string1'
3709 and `string2' into an offset from the beginning of that string. */
3710 #define POINTER_TO_OFFSET(ptr) \
3711 (FIRST_STRING_P (ptr) \
3712 ? ((regoff_t) ((ptr) - string1)) \
3713 : ((regoff_t) ((ptr) - string2 + size1)))
3715 /* Macros for dealing with the split strings in re_match_2. */
3717 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3719 /* Call before fetching a character with *d. This switches over to
3720 string2 if necessary. */
3721 #define PREFETCH() \
3724 /* End of string2 => fail. */ \
3725 if (dend == end_match_2) \
3727 /* End of string1 => advance to string2. */ \
3729 dend = end_match_2; \
3733 /* Test if at very beginning or at very end of the virtual concatenation
3734 of `string1' and `string2'. If only one string, it's `string2'. */
3735 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3736 #define AT_STRINGS_END(d) ((d) == end2)
3739 /* Test if D points to a character which is word-constituent. We have
3740 two special cases to check for: if past the end of string1, look at
3741 the first character in string2; and if before the beginning of
3742 string2, look at the last character in string1. */
3743 #define WORDCHAR_P(d) \
3744 (SYNTAX ((d) == end1 ? *string2 \
3745 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3748 /* Disabled due to a compiler bug -- see comment at case wordbound */
3750 /* Test if the character before D and the one at D differ with respect
3751 to being word-constituent. */
3752 #define AT_WORD_BOUNDARY(d) \
3753 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3754 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3757 /* Free everything we malloc. */
3758 #ifdef MATCH_MAY_ALLOCATE
3759 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3760 # define FREE_VARIABLES() \
3762 REGEX_FREE_STACK (fail_stack.stack); \
3763 FREE_VAR (regstart); \
3764 FREE_VAR (regend); \
3765 FREE_VAR (old_regstart); \
3766 FREE_VAR (old_regend); \
3767 FREE_VAR (best_regstart); \
3768 FREE_VAR (best_regend); \
3769 FREE_VAR (reg_info); \
3770 FREE_VAR (reg_dummy); \
3771 FREE_VAR (reg_info_dummy); \
3774 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3775 #endif /* not MATCH_MAY_ALLOCATE */
3777 /* These values must meet several constraints. They must not be valid
3778 register values; since we have a limit of 255 registers (because
3779 we use only one byte in the pattern for the register number), we can
3780 use numbers larger than 255. They must differ by 1, because of
3781 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3782 be larger than the value for the highest register, so we do not try
3783 to actually save any registers when none are active. */
3784 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3785 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3787 /* Matching routines. */
3789 #ifndef emacs /* Emacs never uses this. */
3790 /* re_match is like re_match_2 except it takes only a single string. */
3793 re_match (bufp, string, size, pos, regs)
3794 struct re_pattern_buffer *bufp;
3797 struct re_registers *regs;
3799 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3801 # ifndef REGEX_MALLOC
3809 weak_alias (__re_match, re_match)
3811 #endif /* not emacs */
3813 static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
3815 register_info_type *reg_info));
3816 static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
3818 register_info_type *reg_info));
3819 static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
3821 register_info_type *reg_info));
3822 static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
3823 int len, char *translate));
3825 /* re_match_2 matches the compiled pattern in BUFP against the
3826 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3827 and SIZE2, respectively). We start matching at POS, and stop
3830 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3831 store offsets for the substring each group matched in REGS. See the
3832 documentation for exactly how many groups we fill.
3834 We return -1 if no match, -2 if an internal error (such as the
3835 failure stack overflowing). Otherwise, we return the length of the
3836 matched substring. */
3839 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3840 struct re_pattern_buffer *bufp;
3841 const char *string1, *string2;
3844 struct re_registers *regs;
3847 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3849 #ifndef REGEX_MALLOC
3857 weak_alias (__re_match_2, re_match_2)
3860 /* This is a separate function so that we can force an alloca cleanup
3863 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3864 struct re_pattern_buffer *bufp;
3865 const char *string1, *string2;
3868 struct re_registers *regs;
3871 /* General temporaries. */
3875 /* Just past the end of the corresponding string. */
3876 const char *end1, *end2;
3878 /* Pointers into string1 and string2, just past the last characters in
3879 each to consider matching. */
3880 const char *end_match_1, *end_match_2;
3882 /* Where we are in the data, and the end of the current string. */
3883 const char *d, *dend;
3885 /* Where we are in the pattern, and the end of the pattern. */
3886 unsigned char *p = bufp->buffer;
3887 register unsigned char *pend = p + bufp->used;
3889 /* Mark the opcode just after a start_memory, so we can test for an
3890 empty subpattern when we get to the stop_memory. */
3891 unsigned char *just_past_start_mem = 0;
3893 /* We use this to map every character in the string. */
3894 RE_TRANSLATE_TYPE translate = bufp->translate;
3896 /* Failure point stack. Each place that can handle a failure further
3897 down the line pushes a failure point on this stack. It consists of
3898 restart, regend, and reg_info for all registers corresponding to
3899 the subexpressions we're currently inside, plus the number of such
3900 registers, and, finally, two char *'s. The first char * is where
3901 to resume scanning the pattern; the second one is where to resume
3902 scanning the strings. If the latter is zero, the failure point is
3903 a ``dummy''; if a failure happens and the failure point is a dummy,
3904 it gets discarded and the next next one is tried. */
3905 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3906 fail_stack_type fail_stack;
3909 static unsigned failure_id;
3910 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3914 /* This holds the pointer to the failure stack, when
3915 it is allocated relocatably. */
3916 fail_stack_elt_t *failure_stack_ptr;
3919 /* We fill all the registers internally, independent of what we
3920 return, for use in backreferences. The number here includes
3921 an element for register zero. */
3922 size_t num_regs = bufp->re_nsub + 1;
3924 /* The currently active registers. */
3925 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3926 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3928 /* Information on the contents of registers. These are pointers into
3929 the input strings; they record just what was matched (on this
3930 attempt) by a subexpression part of the pattern, that is, the
3931 regnum-th regstart pointer points to where in the pattern we began
3932 matching and the regnum-th regend points to right after where we
3933 stopped matching the regnum-th subexpression. (The zeroth register
3934 keeps track of what the whole pattern matches.) */
3935 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3936 const char **regstart, **regend;
3939 /* If a group that's operated upon by a repetition operator fails to
3940 match anything, then the register for its start will need to be
3941 restored because it will have been set to wherever in the string we
3942 are when we last see its open-group operator. Similarly for a
3944 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3945 const char **old_regstart, **old_regend;
3948 /* The is_active field of reg_info helps us keep track of which (possibly
3949 nested) subexpressions we are currently in. The matched_something
3950 field of reg_info[reg_num] helps us tell whether or not we have
3951 matched any of the pattern so far this time through the reg_num-th
3952 subexpression. These two fields get reset each time through any
3953 loop their register is in. */
3954 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3955 register_info_type *reg_info;
3958 /* The following record the register info as found in the above
3959 variables when we find a match better than any we've seen before.
3960 This happens as we backtrack through the failure points, which in
3961 turn happens only if we have not yet matched the entire string. */
3962 unsigned best_regs_set = false;
3963 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3964 const char **best_regstart, **best_regend;
3967 /* Logically, this is `best_regend[0]'. But we don't want to have to
3968 allocate space for that if we're not allocating space for anything
3969 else (see below). Also, we never need info about register 0 for
3970 any of the other register vectors, and it seems rather a kludge to
3971 treat `best_regend' differently than the rest. So we keep track of
3972 the end of the best match so far in a separate variable. We
3973 initialize this to NULL so that when we backtrack the first time
3974 and need to test it, it's not garbage. */
3975 const char *match_end = NULL;
3977 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3978 int set_regs_matched_done = 0;
3980 /* Used when we pop values we don't care about. */
3981 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3982 const char **reg_dummy;
3983 register_info_type *reg_info_dummy;
3987 /* Counts the total number of registers pushed. */
3988 unsigned num_regs_pushed = 0;
3991 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3995 #ifdef MATCH_MAY_ALLOCATE
3996 /* Do not bother to initialize all the register variables if there are
3997 no groups in the pattern, as it takes a fair amount of time. If
3998 there are groups, we include space for register 0 (the whole
3999 pattern), even though we never use it, since it simplifies the
4000 array indexing. We should fix this. */
4003 regstart = REGEX_TALLOC (num_regs, const char *);
4004 regend = REGEX_TALLOC (num_regs, const char *);
4005 old_regstart = REGEX_TALLOC (num_regs, const char *);
4006 old_regend = REGEX_TALLOC (num_regs, const char *);
4007 best_regstart = REGEX_TALLOC (num_regs, const char *);
4008 best_regend = REGEX_TALLOC (num_regs, const char *);
4009 reg_info = REGEX_TALLOC (num_regs, register_info_type);
4010 reg_dummy = REGEX_TALLOC (num_regs, const char *);
4011 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
4013 if (!(regstart && regend && old_regstart && old_regend && reg_info
4014 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
4022 /* We must initialize all our variables to NULL, so that
4023 `FREE_VARIABLES' doesn't try to free them. */
4024 regstart = regend = old_regstart = old_regend = best_regstart
4025 = best_regend = reg_dummy = NULL;
4026 reg_info = reg_info_dummy = (register_info_type *) NULL;
4028 #endif /* MATCH_MAY_ALLOCATE */
4030 /* The starting position is bogus. */
4031 if (pos < 0 || pos > size1 + size2)
4037 /* Initialize subexpression text positions to -1 to mark ones that no
4038 start_memory/stop_memory has been seen for. Also initialize the
4039 register information struct. */
4040 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4042 regstart[mcnt] = regend[mcnt]
4043 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
4045 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
4046 IS_ACTIVE (reg_info[mcnt]) = 0;
4047 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
4048 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
4051 /* We move `string1' into `string2' if the latter's empty -- but not if
4052 `string1' is null. */
4053 if (size2 == 0 && string1 != NULL)
4060 end1 = string1 + size1;
4061 end2 = string2 + size2;
4063 /* Compute where to stop matching, within the two strings. */
4066 end_match_1 = string1 + stop;
4067 end_match_2 = string2;
4072 end_match_2 = string2 + stop - size1;
4075 /* `p' scans through the pattern as `d' scans through the data.
4076 `dend' is the end of the input string that `d' points within. `d'
4077 is advanced into the following input string whenever necessary, but
4078 this happens before fetching; therefore, at the beginning of the
4079 loop, `d' can be pointing at the end of a string, but it cannot
4081 if (size1 > 0 && pos <= size1)
4088 d = string2 + pos - size1;
4092 DEBUG_PRINT1 ("The compiled pattern is:\n");
4093 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
4094 DEBUG_PRINT1 ("The string to match is: `");
4095 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
4096 DEBUG_PRINT1 ("'\n");
4098 /* This loops over pattern commands. It exits by returning from the
4099 function if the match is complete, or it drops through if the match
4100 fails at this starting point in the input data. */
4104 DEBUG_PRINT2 ("\n%p: ", p);
4106 DEBUG_PRINT2 ("\n0x%x: ", p);
4110 { /* End of pattern means we might have succeeded. */
4111 DEBUG_PRINT1 ("end of pattern ... ");
4113 /* If we haven't matched the entire string, and we want the
4114 longest match, try backtracking. */
4115 if (d != end_match_2)
4117 /* 1 if this match ends in the same string (string1 or string2)
4118 as the best previous match. */
4119 boolean same_str_p = (FIRST_STRING_P (match_end)
4120 == MATCHING_IN_FIRST_STRING);
4121 /* 1 if this match is the best seen so far. */
4122 boolean best_match_p;
4124 /* AIX compiler got confused when this was combined
4125 with the previous declaration. */
4127 best_match_p = d > match_end;
4129 best_match_p = !MATCHING_IN_FIRST_STRING;
4131 DEBUG_PRINT1 ("backtracking.\n");
4133 if (!FAIL_STACK_EMPTY ())
4134 { /* More failure points to try. */
4136 /* If exceeds best match so far, save it. */
4137 if (!best_regs_set || best_match_p)
4139 best_regs_set = true;
4142 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4144 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4146 best_regstart[mcnt] = regstart[mcnt];
4147 best_regend[mcnt] = regend[mcnt];
4153 /* If no failure points, don't restore garbage. And if
4154 last match is real best match, don't restore second
4156 else if (best_regs_set && !best_match_p)
4159 /* Restore best match. It may happen that `dend ==
4160 end_match_1' while the restored d is in string2.
4161 For example, the pattern `x.*y.*z' against the
4162 strings `x-' and `y-z-', if the two strings are
4163 not consecutive in memory. */
4164 DEBUG_PRINT1 ("Restoring best registers.\n");
4167 dend = ((d >= string1 && d <= end1)
4168 ? end_match_1 : end_match_2);
4170 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4172 regstart[mcnt] = best_regstart[mcnt];
4173 regend[mcnt] = best_regend[mcnt];
4176 } /* d != end_match_2 */
4179 DEBUG_PRINT1 ("Accepting match.\n");
4181 /* If caller wants register contents data back, do it. */
4182 if (regs && !bufp->no_sub)
4184 /* Have the register data arrays been allocated? */
4185 if (bufp->regs_allocated == REGS_UNALLOCATED)
4186 { /* No. So allocate them with malloc. We need one
4187 extra element beyond `num_regs' for the `-1' marker
4189 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4190 regs->start = TALLOC (regs->num_regs, regoff_t);
4191 regs->end = TALLOC (regs->num_regs, regoff_t);
4192 if (regs->start == NULL || regs->end == NULL)
4197 bufp->regs_allocated = REGS_REALLOCATE;
4199 else if (bufp->regs_allocated == REGS_REALLOCATE)
4200 { /* Yes. If we need more elements than were already
4201 allocated, reallocate them. If we need fewer, just
4203 if (regs->num_regs < num_regs + 1)
4205 regs->num_regs = num_regs + 1;
4206 RETALLOC (regs->start, regs->num_regs, regoff_t);
4207 RETALLOC (regs->end, regs->num_regs, regoff_t);
4208 if (regs->start == NULL || regs->end == NULL)
4217 /* These braces fend off a "empty body in an else-statement"
4218 warning under GCC when assert expands to nothing. */
4219 assert (bufp->regs_allocated == REGS_FIXED);
4222 /* Convert the pointer data in `regstart' and `regend' to
4223 indices. Register zero has to be set differently,
4224 since we haven't kept track of any info for it. */
4225 if (regs->num_regs > 0)
4227 regs->start[0] = pos;
4228 regs->end[0] = (MATCHING_IN_FIRST_STRING
4229 ? ((regoff_t) (d - string1))
4230 : ((regoff_t) (d - string2 + size1)));
4233 /* Go through the first `min (num_regs, regs->num_regs)'
4234 registers, since that is all we initialized. */
4235 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
4238 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4239 regs->start[mcnt] = regs->end[mcnt] = -1;
4243 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4245 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4249 /* If the regs structure we return has more elements than
4250 were in the pattern, set the extra elements to -1. If
4251 we (re)allocated the registers, this is the case,
4252 because we always allocate enough to have at least one
4254 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
4255 regs->start[mcnt] = regs->end[mcnt] = -1;
4256 } /* regs && !bufp->no_sub */
4258 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4259 nfailure_points_pushed, nfailure_points_popped,
4260 nfailure_points_pushed - nfailure_points_popped);
4261 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4263 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4267 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4273 /* Otherwise match next pattern command. */
4274 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4276 /* Ignore these. Used to ignore the n of succeed_n's which
4277 currently have n == 0. */
4279 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4283 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4286 /* Match the next n pattern characters exactly. The following
4287 byte in the pattern defines n, and the n bytes after that
4288 are the characters to match. */
4291 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4293 /* This is written out as an if-else so we don't waste time
4294 testing `translate' inside the loop. */
4300 if ((unsigned char) translate[(unsigned char) *d++]
4301 != (unsigned char) *p++)
4311 if (*d++ != (char) *p++) goto fail;
4315 SET_REGS_MATCHED ();
4319 /* Match any character except possibly a newline or a null. */
4321 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4325 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4326 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4329 SET_REGS_MATCHED ();
4330 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4338 register unsigned char c;
4339 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4341 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4344 c = TRANSLATE (*d); /* The character to match. */
4346 /* Cast to `unsigned' instead of `unsigned char' in case the
4347 bit list is a full 32 bytes long. */
4348 if (c < (unsigned) (*p * BYTEWIDTH)
4349 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4354 if (!not) goto fail;
4356 SET_REGS_MATCHED ();
4362 /* The beginning of a group is represented by start_memory.
4363 The arguments are the register number in the next byte, and the
4364 number of groups inner to this one in the next. The text
4365 matched within the group is recorded (in the internal
4366 registers data structure) under the register number. */
4368 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4370 /* Find out if this group can match the empty string. */
4371 p1 = p; /* To send to group_match_null_string_p. */
4373 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4374 REG_MATCH_NULL_STRING_P (reg_info[*p])
4375 = group_match_null_string_p (&p1, pend, reg_info);
4377 /* Save the position in the string where we were the last time
4378 we were at this open-group operator in case the group is
4379 operated upon by a repetition operator, e.g., with `(a*)*b'
4380 against `ab'; then we want to ignore where we are now in
4381 the string in case this attempt to match fails. */
4382 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4383 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4385 DEBUG_PRINT2 (" old_regstart: %d\n",
4386 POINTER_TO_OFFSET (old_regstart[*p]));
4389 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4391 IS_ACTIVE (reg_info[*p]) = 1;
4392 MATCHED_SOMETHING (reg_info[*p]) = 0;
4394 /* Clear this whenever we change the register activity status. */
4395 set_regs_matched_done = 0;
4397 /* This is the new highest active register. */
4398 highest_active_reg = *p;
4400 /* If nothing was active before, this is the new lowest active
4402 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4403 lowest_active_reg = *p;
4405 /* Move past the register number and inner group count. */
4407 just_past_start_mem = p;
4412 /* The stop_memory opcode represents the end of a group. Its
4413 arguments are the same as start_memory's: the register
4414 number, and the number of inner groups. */
4416 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4418 /* We need to save the string position the last time we were at
4419 this close-group operator in case the group is operated
4420 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4421 against `aba'; then we want to ignore where we are now in
4422 the string in case this attempt to match fails. */
4423 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4424 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4426 DEBUG_PRINT2 (" old_regend: %d\n",
4427 POINTER_TO_OFFSET (old_regend[*p]));
4430 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4432 /* This register isn't active anymore. */
4433 IS_ACTIVE (reg_info[*p]) = 0;
4435 /* Clear this whenever we change the register activity status. */
4436 set_regs_matched_done = 0;
4438 /* If this was the only register active, nothing is active
4440 if (lowest_active_reg == highest_active_reg)
4442 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4443 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4446 { /* We must scan for the new highest active register, since
4447 it isn't necessarily one less than now: consider
4448 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4449 new highest active register is 1. */
4450 unsigned char r = *p - 1;
4451 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4454 /* If we end up at register zero, that means that we saved
4455 the registers as the result of an `on_failure_jump', not
4456 a `start_memory', and we jumped to past the innermost
4457 `stop_memory'. For example, in ((.)*) we save
4458 registers 1 and 2 as a result of the *, but when we pop
4459 back to the second ), we are at the stop_memory 1.
4460 Thus, nothing is active. */
4463 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4464 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4467 highest_active_reg = r;
4470 /* If just failed to match something this time around with a
4471 group that's operated on by a repetition operator, try to
4472 force exit from the ``loop'', and restore the register
4473 information for this group that we had before trying this
4475 if ((!MATCHED_SOMETHING (reg_info[*p])
4476 || just_past_start_mem == p - 1)
4479 boolean is_a_jump_n = false;
4483 switch ((re_opcode_t) *p1++)
4487 case pop_failure_jump:
4488 case maybe_pop_jump:
4490 case dummy_failure_jump:
4491 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4501 /* If the next operation is a jump backwards in the pattern
4502 to an on_failure_jump right before the start_memory
4503 corresponding to this stop_memory, exit from the loop
4504 by forcing a failure after pushing on the stack the
4505 on_failure_jump's jump in the pattern, and d. */
4506 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4507 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4509 /* If this group ever matched anything, then restore
4510 what its registers were before trying this last
4511 failed match, e.g., with `(a*)*b' against `ab' for
4512 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4513 against `aba' for regend[3].
4515 Also restore the registers for inner groups for,
4516 e.g., `((a*)(b*))*' against `aba' (register 3 would
4517 otherwise get trashed). */
4519 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4523 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4525 /* Restore this and inner groups' (if any) registers. */
4526 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4529 regstart[r] = old_regstart[r];
4531 /* xx why this test? */
4532 if (old_regend[r] >= regstart[r])
4533 regend[r] = old_regend[r];
4537 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4538 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4544 /* Move past the register number and the inner group count. */
4549 /* \<digit> has been turned into a `duplicate' command which is
4550 followed by the numeric value of <digit> as the register number. */
4553 register const char *d2, *dend2;
4554 int regno = *p++; /* Get which register to match against. */
4555 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4557 /* Can't back reference a group which we've never matched. */
4558 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4561 /* Where in input to try to start matching. */
4562 d2 = regstart[regno];
4564 /* Where to stop matching; if both the place to start and
4565 the place to stop matching are in the same string, then
4566 set to the place to stop, otherwise, for now have to use
4567 the end of the first string. */
4569 dend2 = ((FIRST_STRING_P (regstart[regno])
4570 == FIRST_STRING_P (regend[regno]))
4571 ? regend[regno] : end_match_1);
4574 /* If necessary, advance to next segment in register
4578 if (dend2 == end_match_2) break;
4579 if (dend2 == regend[regno]) break;
4581 /* End of string1 => advance to string2. */
4583 dend2 = regend[regno];
4585 /* At end of register contents => success */
4586 if (d2 == dend2) break;
4588 /* If necessary, advance to next segment in data. */
4591 /* How many characters left in this segment to match. */
4594 /* Want how many consecutive characters we can match in
4595 one shot, so, if necessary, adjust the count. */
4596 if (mcnt > dend2 - d2)
4599 /* Compare that many; failure if mismatch, else move
4602 ? bcmp_translate (d, d2, mcnt, translate)
4603 : memcmp (d, d2, mcnt))
4605 d += mcnt, d2 += mcnt;
4607 /* Do this because we've match some characters. */
4608 SET_REGS_MATCHED ();
4614 /* begline matches the empty string at the beginning of the string
4615 (unless `not_bol' is set in `bufp'), and, if
4616 `newline_anchor' is set, after newlines. */
4618 DEBUG_PRINT1 ("EXECUTING begline.\n");
4620 if (AT_STRINGS_BEG (d))
4622 if (!bufp->not_bol) break;
4624 else if (d[-1] == '\n' && bufp->newline_anchor)
4628 /* In all other cases, we fail. */
4632 /* endline is the dual of begline. */
4634 DEBUG_PRINT1 ("EXECUTING endline.\n");
4636 if (AT_STRINGS_END (d))
4638 if (!bufp->not_eol) break;
4641 /* We have to ``prefetch'' the next character. */
4642 else if ((d == end1 ? *string2 : *d) == '\n'
4643 && bufp->newline_anchor)
4650 /* Match at the very beginning of the data. */
4652 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4653 if (AT_STRINGS_BEG (d))
4658 /* Match at the very end of the data. */
4660 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4661 if (AT_STRINGS_END (d))
4666 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4667 pushes NULL as the value for the string on the stack. Then
4668 `pop_failure_point' will keep the current value for the
4669 string, instead of restoring it. To see why, consider
4670 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4671 then the . fails against the \n. But the next thing we want
4672 to do is match the \n against the \n; if we restored the
4673 string value, we would be back at the foo.
4675 Because this is used only in specific cases, we don't need to
4676 check all the things that `on_failure_jump' does, to make
4677 sure the right things get saved on the stack. Hence we don't
4678 share its code. The only reason to push anything on the
4679 stack at all is that otherwise we would have to change
4680 `anychar's code to do something besides goto fail in this
4681 case; that seems worse than this. */
4682 case on_failure_keep_string_jump:
4683 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4685 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4687 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
4689 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4692 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4696 /* Uses of on_failure_jump:
4698 Each alternative starts with an on_failure_jump that points
4699 to the beginning of the next alternative. Each alternative
4700 except the last ends with a jump that in effect jumps past
4701 the rest of the alternatives. (They really jump to the
4702 ending jump of the following alternative, because tensioning
4703 these jumps is a hassle.)
4705 Repeats start with an on_failure_jump that points past both
4706 the repetition text and either the following jump or
4707 pop_failure_jump back to this on_failure_jump. */
4708 case on_failure_jump:
4710 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4712 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4714 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
4716 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4719 /* If this on_failure_jump comes right before a group (i.e.,
4720 the original * applied to a group), save the information
4721 for that group and all inner ones, so that if we fail back
4722 to this point, the group's information will be correct.
4723 For example, in \(a*\)*\1, we need the preceding group,
4724 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4726 /* We can't use `p' to check ahead because we push
4727 a failure point to `p + mcnt' after we do this. */
4730 /* We need to skip no_op's before we look for the
4731 start_memory in case this on_failure_jump is happening as
4732 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4734 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4737 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4739 /* We have a new highest active register now. This will
4740 get reset at the start_memory we are about to get to,
4741 but we will have saved all the registers relevant to
4742 this repetition op, as described above. */
4743 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4744 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4745 lowest_active_reg = *(p1 + 1);
4748 DEBUG_PRINT1 (":\n");
4749 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4753 /* A smart repeat ends with `maybe_pop_jump'.
4754 We change it to either `pop_failure_jump' or `jump'. */
4755 case maybe_pop_jump:
4756 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4757 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4759 register unsigned char *p2 = p;
4761 /* Compare the beginning of the repeat with what in the
4762 pattern follows its end. If we can establish that there
4763 is nothing that they would both match, i.e., that we
4764 would have to backtrack because of (as in, e.g., `a*a')
4765 then we can change to pop_failure_jump, because we'll
4766 never have to backtrack.
4768 This is not true in the case of alternatives: in
4769 `(a|ab)*' we do need to backtrack to the `ab' alternative
4770 (e.g., if the string was `ab'). But instead of trying to
4771 detect that here, the alternative has put on a dummy
4772 failure point which is what we will end up popping. */
4774 /* Skip over open/close-group commands.
4775 If what follows this loop is a ...+ construct,
4776 look at what begins its body, since we will have to
4777 match at least one of that. */
4781 && ((re_opcode_t) *p2 == stop_memory
4782 || (re_opcode_t) *p2 == start_memory))
4784 else if (p2 + 6 < pend
4785 && (re_opcode_t) *p2 == dummy_failure_jump)
4792 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4793 to the `maybe_finalize_jump' of this case. Examine what
4796 /* If we're at the end of the pattern, we can change. */
4799 /* Consider what happens when matching ":\(.*\)"
4800 against ":/". I don't really understand this code
4802 p[-3] = (unsigned char) pop_failure_jump;
4804 (" End of pattern: change to `pop_failure_jump'.\n");
4807 else if ((re_opcode_t) *p2 == exactn
4808 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4810 register unsigned char c
4811 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4813 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4815 p[-3] = (unsigned char) pop_failure_jump;
4816 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4820 else if ((re_opcode_t) p1[3] == charset
4821 || (re_opcode_t) p1[3] == charset_not)
4823 int not = (re_opcode_t) p1[3] == charset_not;
4825 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4826 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4829 /* `not' is equal to 1 if c would match, which means
4830 that we can't change to pop_failure_jump. */
4833 p[-3] = (unsigned char) pop_failure_jump;
4834 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4838 else if ((re_opcode_t) *p2 == charset)
4841 register unsigned char c
4842 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4846 if ((re_opcode_t) p1[3] == exactn
4847 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4848 && (p2[2 + p1[5] / BYTEWIDTH]
4849 & (1 << (p1[5] % BYTEWIDTH)))))
4851 if ((re_opcode_t) p1[3] == exactn
4852 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4853 && (p2[2 + p1[4] / BYTEWIDTH]
4854 & (1 << (p1[4] % BYTEWIDTH)))))
4857 p[-3] = (unsigned char) pop_failure_jump;
4858 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4862 else if ((re_opcode_t) p1[3] == charset_not)
4865 /* We win if the charset_not inside the loop
4866 lists every character listed in the charset after. */
4867 for (idx = 0; idx < (int) p2[1]; idx++)
4868 if (! (p2[2 + idx] == 0
4869 || (idx < (int) p1[4]
4870 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4875 p[-3] = (unsigned char) pop_failure_jump;
4876 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4879 else if ((re_opcode_t) p1[3] == charset)
4882 /* We win if the charset inside the loop
4883 has no overlap with the one after the loop. */
4885 idx < (int) p2[1] && idx < (int) p1[4];
4887 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4890 if (idx == p2[1] || idx == p1[4])
4892 p[-3] = (unsigned char) pop_failure_jump;
4893 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4898 p -= 2; /* Point at relative address again. */
4899 if ((re_opcode_t) p[-1] != pop_failure_jump)
4901 p[-1] = (unsigned char) jump;
4902 DEBUG_PRINT1 (" Match => jump.\n");
4903 goto unconditional_jump;
4905 /* Note fall through. */
4908 /* The end of a simple repeat has a pop_failure_jump back to
4909 its matching on_failure_jump, where the latter will push a
4910 failure point. The pop_failure_jump takes off failure
4911 points put on by this pop_failure_jump's matching
4912 on_failure_jump; we got through the pattern to here from the
4913 matching on_failure_jump, so didn't fail. */
4914 case pop_failure_jump:
4916 /* We need to pass separate storage for the lowest and
4917 highest registers, even though we don't care about the
4918 actual values. Otherwise, we will restore only one
4919 register from the stack, since lowest will == highest in
4920 `pop_failure_point'. */
4921 active_reg_t dummy_low_reg, dummy_high_reg;
4922 unsigned char *pdummy;
4925 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4926 POP_FAILURE_POINT (sdummy, pdummy,
4927 dummy_low_reg, dummy_high_reg,
4928 reg_dummy, reg_dummy, reg_info_dummy);
4930 /* Note fall through. */
4934 DEBUG_PRINT2 ("\n%p: ", p);
4936 DEBUG_PRINT2 ("\n0x%x: ", p);
4938 /* Note fall through. */
4940 /* Unconditionally jump (without popping any failure points). */
4942 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4943 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4944 p += mcnt; /* Do the jump. */
4946 DEBUG_PRINT2 ("(to %p).\n", p);
4948 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4953 /* We need this opcode so we can detect where alternatives end
4954 in `group_match_null_string_p' et al. */
4956 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4957 goto unconditional_jump;
4960 /* Normally, the on_failure_jump pushes a failure point, which
4961 then gets popped at pop_failure_jump. We will end up at
4962 pop_failure_jump, also, and with a pattern of, say, `a+', we
4963 are skipping over the on_failure_jump, so we have to push
4964 something meaningless for pop_failure_jump to pop. */
4965 case dummy_failure_jump:
4966 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4967 /* It doesn't matter what we push for the string here. What
4968 the code at `fail' tests is the value for the pattern. */
4969 PUSH_FAILURE_POINT (NULL, NULL, -2);
4970 goto unconditional_jump;
4973 /* At the end of an alternative, we need to push a dummy failure
4974 point in case we are followed by a `pop_failure_jump', because
4975 we don't want the failure point for the alternative to be
4976 popped. For example, matching `(a|ab)*' against `aab'
4977 requires that we match the `ab' alternative. */
4978 case push_dummy_failure:
4979 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4980 /* See comments just above at `dummy_failure_jump' about the
4982 PUSH_FAILURE_POINT (NULL, NULL, -2);
4985 /* Have to succeed matching what follows at least n times.
4986 After that, handle like `on_failure_jump'. */
4988 EXTRACT_NUMBER (mcnt, p + 2);
4989 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4992 /* Originally, this is how many times we HAVE to succeed. */
4997 STORE_NUMBER_AND_INCR (p, mcnt);
4999 DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
5001 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
5007 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
5009 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
5011 p[2] = (unsigned char) no_op;
5012 p[3] = (unsigned char) no_op;
5018 EXTRACT_NUMBER (mcnt, p + 2);
5019 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
5021 /* Originally, this is how many times we CAN jump. */
5025 STORE_NUMBER (p + 2, mcnt);
5027 DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
5029 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
5031 goto unconditional_jump;
5033 /* If don't have to jump any more, skip over the rest of command. */
5040 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
5042 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5044 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5046 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
5048 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
5050 STORE_NUMBER (p1, mcnt);
5055 /* The DEC Alpha C compiler 3.x generates incorrect code for the
5056 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
5057 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
5058 macro and introducing temporary variables works around the bug. */
5061 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
5062 if (AT_WORD_BOUNDARY (d))
5067 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
5068 if (AT_WORD_BOUNDARY (d))
5074 boolean prevchar, thischar;
5076 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
5077 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5080 prevchar = WORDCHAR_P (d - 1);
5081 thischar = WORDCHAR_P (d);
5082 if (prevchar != thischar)
5089 boolean prevchar, thischar;
5091 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
5092 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5095 prevchar = WORDCHAR_P (d - 1);
5096 thischar = WORDCHAR_P (d);
5097 if (prevchar != thischar)
5104 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5105 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
5110 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5111 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
5112 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
5118 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5119 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
5124 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5125 if (PTR_CHAR_POS ((unsigned char *) d) != point)
5130 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5131 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
5136 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
5141 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
5145 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5147 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
5149 SET_REGS_MATCHED ();
5153 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5155 goto matchnotsyntax;
5158 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5162 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5164 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
5166 SET_REGS_MATCHED ();
5169 #else /* not emacs */
5171 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5173 if (!WORDCHAR_P (d))
5175 SET_REGS_MATCHED ();
5180 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5184 SET_REGS_MATCHED ();
5187 #endif /* not emacs */
5192 continue; /* Successfully executed one pattern command; keep going. */
5195 /* We goto here if a matching operation fails. */
5197 if (!FAIL_STACK_EMPTY ())
5198 { /* A restart point is known. Restore to that state. */
5199 DEBUG_PRINT1 ("\nFAIL:\n");
5200 POP_FAILURE_POINT (d, p,
5201 lowest_active_reg, highest_active_reg,
5202 regstart, regend, reg_info);
5204 /* If this failure point is a dummy, try the next one. */
5208 /* If we failed to the end of the pattern, don't examine *p. */
5212 boolean is_a_jump_n = false;
5214 /* If failed to a backwards jump that's part of a repetition
5215 loop, need to pop this failure point and use the next one. */
5216 switch ((re_opcode_t) *p)
5220 case maybe_pop_jump:
5221 case pop_failure_jump:
5224 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5227 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5229 && (re_opcode_t) *p1 == on_failure_jump))
5237 if (d >= string1 && d <= end1)
5241 break; /* Matching at this starting point really fails. */
5245 goto restore_best_regs;
5249 return -1; /* Failure to match. */
5252 /* Subroutine definitions for re_match_2. */
5255 /* We are passed P pointing to a register number after a start_memory.
5257 Return true if the pattern up to the corresponding stop_memory can
5258 match the empty string, and false otherwise.
5260 If we find the matching stop_memory, sets P to point to one past its number.
5261 Otherwise, sets P to an undefined byte less than or equal to END.
5263 We don't handle duplicates properly (yet). */
5266 group_match_null_string_p (p, end, reg_info)
5267 unsigned char **p, *end;
5268 register_info_type *reg_info;
5271 /* Point to after the args to the start_memory. */
5272 unsigned char *p1 = *p + 2;
5276 /* Skip over opcodes that can match nothing, and return true or
5277 false, as appropriate, when we get to one that can't, or to the
5278 matching stop_memory. */
5280 switch ((re_opcode_t) *p1)
5282 /* Could be either a loop or a series of alternatives. */
5283 case on_failure_jump:
5285 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5287 /* If the next operation is not a jump backwards in the
5292 /* Go through the on_failure_jumps of the alternatives,
5293 seeing if any of the alternatives cannot match nothing.
5294 The last alternative starts with only a jump,
5295 whereas the rest start with on_failure_jump and end
5296 with a jump, e.g., here is the pattern for `a|b|c':
5298 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5299 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5302 So, we have to first go through the first (n-1)
5303 alternatives and then deal with the last one separately. */
5306 /* Deal with the first (n-1) alternatives, which start
5307 with an on_failure_jump (see above) that jumps to right
5308 past a jump_past_alt. */
5310 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5312 /* `mcnt' holds how many bytes long the alternative
5313 is, including the ending `jump_past_alt' and
5316 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5320 /* Move to right after this alternative, including the
5324 /* Break if it's the beginning of an n-th alternative
5325 that doesn't begin with an on_failure_jump. */
5326 if ((re_opcode_t) *p1 != on_failure_jump)
5329 /* Still have to check that it's not an n-th
5330 alternative that starts with an on_failure_jump. */
5332 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5333 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5335 /* Get to the beginning of the n-th alternative. */
5341 /* Deal with the last alternative: go back and get number
5342 of the `jump_past_alt' just before it. `mcnt' contains
5343 the length of the alternative. */
5344 EXTRACT_NUMBER (mcnt, p1 - 2);
5346 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5349 p1 += mcnt; /* Get past the n-th alternative. */
5355 assert (p1[1] == **p);
5361 if (!common_op_match_null_string_p (&p1, end, reg_info))
5364 } /* while p1 < end */
5367 } /* group_match_null_string_p */
5370 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5371 It expects P to be the first byte of a single alternative and END one
5372 byte past the last. The alternative can contain groups. */
5375 alt_match_null_string_p (p, end, reg_info)
5376 unsigned char *p, *end;
5377 register_info_type *reg_info;
5380 unsigned char *p1 = p;
5384 /* Skip over opcodes that can match nothing, and break when we get
5385 to one that can't. */
5387 switch ((re_opcode_t) *p1)
5390 case on_failure_jump:
5392 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5397 if (!common_op_match_null_string_p (&p1, end, reg_info))
5400 } /* while p1 < end */
5403 } /* alt_match_null_string_p */
5406 /* Deals with the ops common to group_match_null_string_p and
5407 alt_match_null_string_p.
5409 Sets P to one after the op and its arguments, if any. */
5412 common_op_match_null_string_p (p, end, reg_info)
5413 unsigned char **p, *end;
5414 register_info_type *reg_info;
5419 unsigned char *p1 = *p;
5421 switch ((re_opcode_t) *p1++)
5441 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5442 ret = group_match_null_string_p (&p1, end, reg_info);
5444 /* Have to set this here in case we're checking a group which
5445 contains a group and a back reference to it. */
5447 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5448 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5454 /* If this is an optimized succeed_n for zero times, make the jump. */
5456 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5464 /* Get to the number of times to succeed. */
5466 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5471 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5479 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5487 /* All other opcodes mean we cannot match the empty string. */
5493 } /* common_op_match_null_string_p */
5496 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5497 bytes; nonzero otherwise. */
5500 bcmp_translate (s1, s2, len, translate)
5501 const char *s1, *s2;
5503 RE_TRANSLATE_TYPE translate;
5505 register const unsigned char *p1 = (const unsigned char *) s1;
5506 register const unsigned char *p2 = (const unsigned char *) s2;
5509 if (translate[*p1++] != translate[*p2++]) return 1;
5515 /* Entry points for GNU code. */
5517 /* re_compile_pattern is the GNU regular expression compiler: it
5518 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5519 Returns 0 if the pattern was valid, otherwise an error string.
5521 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5522 are set in BUFP on entry.
5524 We call regex_compile to do the actual compilation. */
5527 re_compile_pattern (pattern, length, bufp)
5528 const char *pattern;
5530 struct re_pattern_buffer *bufp;
5534 /* GNU code is written to assume at least RE_NREGS registers will be set
5535 (and at least one extra will be -1). */
5536 bufp->regs_allocated = REGS_UNALLOCATED;
5538 /* And GNU code determines whether or not to get register information
5539 by passing null for the REGS argument to re_match, etc., not by
5543 /* Match anchors at newline. */
5544 bufp->newline_anchor = 1;
5546 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5550 return gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
5553 weak_alias (__re_compile_pattern, re_compile_pattern)
5556 /* Entry points compatible with 4.2 BSD regex library. We don't define
5557 them unless specifically requested. */
5559 #if defined _REGEX_RE_COMP || defined _LIBC
5561 /* BSD has one and only one pattern buffer. */
5562 static struct re_pattern_buffer re_comp_buf;
5566 /* Make these definitions weak in libc, so POSIX programs can redefine
5567 these names if they don't use our functions, and still use
5568 regcomp/regexec below without link errors. */
5578 if (!re_comp_buf.buffer)
5579 return gettext ("No previous regular expression");
5583 if (!re_comp_buf.buffer)
5585 re_comp_buf.buffer = (unsigned char *) malloc (200);
5586 if (re_comp_buf.buffer == NULL)
5587 return (char *) gettext (re_error_msgid
5588 + re_error_msgid_idx[(int) REG_ESPACE]);
5589 re_comp_buf.allocated = 200;
5591 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5592 if (re_comp_buf.fastmap == NULL)
5593 return (char *) gettext (re_error_msgid
5594 + re_error_msgid_idx[(int) REG_ESPACE]);
5597 /* Since `re_exec' always passes NULL for the `regs' argument, we
5598 don't need to initialize the pattern buffer fields which affect it. */
5600 /* Match anchors at newlines. */
5601 re_comp_buf.newline_anchor = 1;
5603 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5608 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5609 return (char *) gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
5620 const int len = strlen (s);
5622 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5625 #endif /* _REGEX_RE_COMP */
5627 /* POSIX.2 functions. Don't define these for Emacs. */
5631 /* regcomp takes a regular expression as a string and compiles it.
5633 PREG is a regex_t *. We do not expect any fields to be initialized,
5634 since POSIX says we shouldn't. Thus, we set
5636 `buffer' to the compiled pattern;
5637 `used' to the length of the compiled pattern;
5638 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5639 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5640 RE_SYNTAX_POSIX_BASIC;
5641 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5642 `fastmap' to an allocated space for the fastmap;
5643 `fastmap_accurate' to zero;
5644 `re_nsub' to the number of subexpressions in PATTERN.
5646 PATTERN is the address of the pattern string.
5648 CFLAGS is a series of bits which affect compilation.
5650 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5651 use POSIX basic syntax.
5653 If REG_NEWLINE is set, then . and [^...] don't match newline.
5654 Also, regexec will try a match beginning after every newline.
5656 If REG_ICASE is set, then we considers upper- and lowercase
5657 versions of letters to be equivalent when matching.
5659 If REG_NOSUB is set, then when PREG is passed to regexec, that
5660 routine will report only success or failure, and nothing about the
5663 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5664 the return codes and their meanings.) */
5667 regcomp (preg, pattern, cflags)
5669 const char *pattern;
5674 = (cflags & REG_EXTENDED) ?
5675 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5677 /* regex_compile will allocate the space for the compiled pattern. */
5679 preg->allocated = 0;
5682 /* Try to allocate space for the fastmap. */
5683 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
5685 if (cflags & REG_ICASE)
5690 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5691 * sizeof (*(RE_TRANSLATE_TYPE)0));
5692 if (preg->translate == NULL)
5693 return (int) REG_ESPACE;
5695 /* Map uppercase characters to corresponding lowercase ones. */
5696 for (i = 0; i < CHAR_SET_SIZE; i++)
5697 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
5700 preg->translate = NULL;
5702 /* If REG_NEWLINE is set, newlines are treated differently. */
5703 if (cflags & REG_NEWLINE)
5704 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5705 syntax &= ~RE_DOT_NEWLINE;
5706 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5707 /* It also changes the matching behavior. */
5708 preg->newline_anchor = 1;
5711 preg->newline_anchor = 0;
5713 preg->no_sub = !!(cflags & REG_NOSUB);
5715 /* POSIX says a null character in the pattern terminates it, so we
5716 can use strlen here in compiling the pattern. */
5717 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5719 /* POSIX doesn't distinguish between an unmatched open-group and an
5720 unmatched close-group: both are REG_EPAREN. */
5721 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5723 if (ret == REG_NOERROR && preg->fastmap)
5725 /* Compute the fastmap now, since regexec cannot modify the pattern
5727 if (re_compile_fastmap (preg) == -2)
5729 /* Some error occurred while computing the fastmap, just forget
5731 free (preg->fastmap);
5732 preg->fastmap = NULL;
5739 weak_alias (__regcomp, regcomp)
5743 /* regexec searches for a given pattern, specified by PREG, in the
5746 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5747 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5748 least NMATCH elements, and we set them to the offsets of the
5749 corresponding matched substrings.
5751 EFLAGS specifies `execution flags' which affect matching: if
5752 REG_NOTBOL is set, then ^ does not match at the beginning of the
5753 string; if REG_NOTEOL is set, then $ does not match at the end.
5755 We return 0 if we find a match and REG_NOMATCH if not. */
5758 regexec (preg, string, nmatch, pmatch, eflags)
5759 const regex_t *preg;
5762 regmatch_t pmatch[];
5766 struct re_registers regs;
5767 regex_t private_preg;
5768 int len = strlen (string);
5769 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5771 private_preg = *preg;
5773 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5774 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5776 /* The user has told us exactly how many registers to return
5777 information about, via `nmatch'. We have to pass that on to the
5778 matching routines. */
5779 private_preg.regs_allocated = REGS_FIXED;
5783 regs.num_regs = nmatch;
5784 regs.start = TALLOC (nmatch * 2, regoff_t);
5785 if (regs.start == NULL)
5786 return (int) REG_NOMATCH;
5787 regs.end = regs.start + nmatch;
5790 /* Perform the searching operation. */
5791 ret = re_search (&private_preg, string, len,
5792 /* start: */ 0, /* range: */ len,
5793 want_reg_info ? ®s : (struct re_registers *) 0);
5795 /* Copy the register information to the POSIX structure. */
5802 for (r = 0; r < nmatch; r++)
5804 pmatch[r].rm_so = regs.start[r];
5805 pmatch[r].rm_eo = regs.end[r];
5809 /* If we needed the temporary register info, free the space now. */
5813 /* We want zero return to mean success, unlike `re_search'. */
5814 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5817 weak_alias (__regexec, regexec)
5821 /* Returns a message corresponding to an error code, ERRCODE, returned
5822 from either regcomp or regexec. We don't use PREG here. */
5825 regerror (err, preg, errbuf, errbuf_size)
5827 const regex_t *preg;
5835 || err >= (int) (sizeof (re_error_msgid_idx)
5836 / sizeof (re_error_msgid_idx[0])))
5837 /* Only error codes returned by the rest of the code should be passed
5838 to this routine. If we are given anything else, or if other regex
5839 code generates an invalid error code, then the program has a bug.
5840 Dump core so we can fix it. */
5843 msg = gettext (re_error_msgid + re_error_msgid_idx[err]);
5845 msg_size = strlen (msg) + 1; /* Includes the null. */
5847 if (errbuf_size != 0)
5849 if (msg_size > errbuf_size)
5851 #if defined HAVE_MEMPCPY || defined _LIBC
5852 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
5854 memcpy (errbuf, msg, errbuf_size - 1);
5855 errbuf[errbuf_size - 1] = 0;
5859 memcpy (errbuf, msg, msg_size);
5865 weak_alias (__regerror, regerror)
5869 /* Free dynamically allocated space used by PREG. */
5875 if (preg->buffer != NULL)
5876 free (preg->buffer);
5877 preg->buffer = NULL;
5879 preg->allocated = 0;
5882 if (preg->fastmap != NULL)
5883 free (preg->fastmap);
5884 preg->fastmap = NULL;
5885 preg->fastmap_accurate = 0;
5887 if (preg->translate != NULL)
5888 free (preg->translate);
5889 preg->translate = NULL;
5892 weak_alias (__regfree, regfree)
5895 #endif /* not emacs */
projects / quagga-debian.git / blob