1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993-1999, 2000, 2001 Free Software Foundation, Inc.
6 This file is part of the GNU C Library.
8 The GNU C Library is free software; you can redistribute it and/or
9 modify it under the terms of the GNU Lesser General Public
10 License as published by the Free Software Foundation; either
11 version 2.1 of the License, or (at your option) any later version.
13 The GNU C Library is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 Lesser General Public License for more details.
18 You should have received a copy of the GNU Lesser General Public
19 License along with the GNU C Library; if not, write to the Free
20 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
23 /* AIX requires this to be the first thing in the file. */
24 #if defined _AIX && !defined REGEX_MALLOC
36 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
37 # define PARAMS(args) args
39 # define PARAMS(args) ()
41 #endif /* Not PARAMS. */
43 #ifndef INSIDE_RECURSION
45 # if defined STDC_HEADERS && !defined emacs
48 /* We need this for `regex.h', and perhaps for the Emacs include files. */
49 # include <sys/types.h>
52 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
54 /* For platform which support the ISO C amendement 1 functionality we
55 support user defined character classes. */
56 # if defined _LIBC || WIDE_CHAR_SUPPORT
57 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
63 /* We have to keep the namespace clean. */
64 # define regfree(preg) __regfree (preg)
65 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
66 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
67 # define regerror(errcode, preg, errbuf, errbuf_size) \
68 __regerror(errcode, preg, errbuf, errbuf_size)
69 # define re_set_registers(bu, re, nu, st, en) \
70 __re_set_registers (bu, re, nu, st, en)
71 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
72 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
73 # define re_match(bufp, string, size, pos, regs) \
74 __re_match (bufp, string, size, pos, regs)
75 # define re_search(bufp, string, size, startpos, range, regs) \
76 __re_search (bufp, string, size, startpos, range, regs)
77 # define re_compile_pattern(pattern, length, bufp) \
78 __re_compile_pattern (pattern, length, bufp)
79 # define re_set_syntax(syntax) __re_set_syntax (syntax)
80 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
81 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
82 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
84 # define btowc __btowc
86 /* We are also using some library internals. */
87 # include <locale/localeinfo.h>
88 # include <locale/elem-hash.h>
89 # include <langinfo.h>
90 # include <locale/coll-lookup.h>
93 /* This is for other GNU distributions with internationalized messages. */
94 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
98 # define gettext(msgid) __dcgettext ("libc", msgid, LC_MESSAGES)
101 # define gettext(msgid) (msgid)
104 # ifndef gettext_noop
105 /* This define is so xgettext can find the internationalizable
107 # define gettext_noop(String) String
110 /* The `emacs' switch turns on certain matching commands
111 that make sense only in Emacs. */
118 # else /* not emacs */
120 /* If we are not linking with Emacs proper,
121 we can't use the relocating allocator
122 even if config.h says that we can. */
125 # if defined STDC_HEADERS || defined _LIBC
132 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
133 If nothing else has been done, use the method below. */
134 # ifdef INHIBIT_STRING_HEADER
135 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
136 # if !defined bzero && !defined bcopy
137 # undef INHIBIT_STRING_HEADER
142 /* This is the normal way of making sure we have a bcopy and a bzero.
143 This is used in most programs--a few other programs avoid this
144 by defining INHIBIT_STRING_HEADER. */
145 # ifndef INHIBIT_STRING_HEADER
146 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
150 # define bzero(s, n) (memset (s, '\0', n), (s))
152 # define bzero(s, n) __bzero (s, n)
156 # include <strings.h>
158 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
161 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
166 /* Define the syntax stuff for \<, \>, etc. */
168 /* This must be nonzero for the wordchar and notwordchar pattern
169 commands in re_match_2. */
174 # ifdef SWITCH_ENUM_BUG
175 # define SWITCH_ENUM_CAST(x) ((int)(x))
177 # define SWITCH_ENUM_CAST(x) (x)
180 # endif /* not emacs */
182 # if defined _LIBC || HAVE_LIMITS_H
187 # define MB_LEN_MAX 1
190 /* Get the interface, including the syntax bits. */
193 /* isalpha etc. are used for the character classes. */
196 /* Jim Meyering writes:
198 "... Some ctype macros are valid only for character codes that
199 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
200 using /bin/cc or gcc but without giving an ansi option). So, all
201 ctype uses should be through macros like ISPRINT... If
202 STDC_HEADERS is defined, then autoconf has verified that the ctype
203 macros don't need to be guarded with references to isascii. ...
204 Defining isascii to 1 should let any compiler worth its salt
205 eliminate the && through constant folding."
206 Solaris defines some of these symbols so we must undefine them first. */
208 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
209 # define IN_CTYPE_DOMAIN(c) 1
211 # define IN_CTYPE_DOMAIN(c) isascii(c)
215 # define ISBLANK(c) (IN_CTYPE_DOMAIN (c) && isblank (c))
217 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
220 # define ISGRAPH(c) (IN_CTYPE_DOMAIN (c) && isgraph (c))
222 # define ISGRAPH(c) (IN_CTYPE_DOMAIN (c) && isprint (c) && !isspace (c))
226 # define ISPRINT(c) (IN_CTYPE_DOMAIN (c) && isprint (c))
227 # define ISDIGIT(c) (IN_CTYPE_DOMAIN (c) && isdigit (c))
228 # define ISALNUM(c) (IN_CTYPE_DOMAIN (c) && isalnum (c))
229 # define ISALPHA(c) (IN_CTYPE_DOMAIN (c) && isalpha (c))
230 # define ISCNTRL(c) (IN_CTYPE_DOMAIN (c) && iscntrl (c))
231 # define ISLOWER(c) (IN_CTYPE_DOMAIN (c) && islower (c))
232 # define ISPUNCT(c) (IN_CTYPE_DOMAIN (c) && ispunct (c))
233 # define ISSPACE(c) (IN_CTYPE_DOMAIN (c) && isspace (c))
234 # define ISUPPER(c) (IN_CTYPE_DOMAIN (c) && isupper (c))
235 # define ISXDIGIT(c) (IN_CTYPE_DOMAIN (c) && isxdigit (c))
238 # define TOLOWER(c) _tolower(c)
240 # define TOLOWER(c) tolower(c)
244 # define NULL (void *)0
247 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
248 since ours (we hope) works properly with all combinations of
249 machines, compilers, `char' and `unsigned char' argument types.
250 (Per Bothner suggested the basic approach.) */
251 # undef SIGN_EXTEND_CHAR
253 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
254 # else /* not __STDC__ */
255 /* As in Harbison and Steele. */
256 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
260 /* How many characters in the character set. */
261 # define CHAR_SET_SIZE 256
265 extern char *re_syntax_table;
267 # else /* not SYNTAX_TABLE */
269 static char re_syntax_table[CHAR_SET_SIZE];
271 static void init_syntax_once PARAMS ((void));
281 bzero (re_syntax_table, sizeof re_syntax_table);
283 for (c = 0; c < CHAR_SET_SIZE; ++c)
285 re_syntax_table[c] = Sword;
287 re_syntax_table['_'] = Sword;
292 # endif /* not SYNTAX_TABLE */
294 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
298 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
299 use `alloca' instead of `malloc'. This is because using malloc in
300 re_search* or re_match* could cause memory leaks when C-g is used in
301 Emacs; also, malloc is slower and causes storage fragmentation. On
302 the other hand, malloc is more portable, and easier to debug.
304 Because we sometimes use alloca, some routines have to be macros,
305 not functions -- `alloca'-allocated space disappears at the end of the
306 function it is called in. */
310 # define REGEX_ALLOCATE malloc
311 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
312 # define REGEX_FREE free
314 # else /* not REGEX_MALLOC */
316 /* Emacs already defines alloca, sometimes. */
319 /* Make alloca work the best possible way. */
321 # define alloca __builtin_alloca
322 # else /* not __GNUC__ */
325 # endif /* HAVE_ALLOCA_H */
326 # endif /* not __GNUC__ */
328 # endif /* not alloca */
330 # define REGEX_ALLOCATE alloca
332 /* Assumes a `char *destination' variable. */
333 # define REGEX_REALLOCATE(source, osize, nsize) \
334 (destination = (char *) alloca (nsize), \
335 memcpy (destination, source, osize))
337 /* No need to do anything to free, after alloca. */
338 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
340 # endif /* not REGEX_MALLOC */
342 /* Define how to allocate the failure stack. */
344 # if defined REL_ALLOC && defined REGEX_MALLOC
346 # define REGEX_ALLOCATE_STACK(size) \
347 r_alloc (&failure_stack_ptr, (size))
348 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
349 r_re_alloc (&failure_stack_ptr, (nsize))
350 # define REGEX_FREE_STACK(ptr) \
351 r_alloc_free (&failure_stack_ptr)
353 # else /* not using relocating allocator */
357 # define REGEX_ALLOCATE_STACK malloc
358 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
359 # define REGEX_FREE_STACK free
361 # else /* not REGEX_MALLOC */
363 # define REGEX_ALLOCATE_STACK alloca
365 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
366 REGEX_REALLOCATE (source, osize, nsize)
367 /* No need to explicitly free anything. */
368 # define REGEX_FREE_STACK(arg)
370 # endif /* not REGEX_MALLOC */
371 # endif /* not using relocating allocator */
374 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
375 `string1' or just past its end. This works if PTR is NULL, which is
377 # define FIRST_STRING_P(ptr) \
378 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
380 /* (Re)Allocate N items of type T using malloc, or fail. */
381 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
382 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
383 # define RETALLOC_IF(addr, n, t) \
384 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
385 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
387 # define BYTEWIDTH 8 /* In bits. */
389 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
393 # define MAX(a, b) ((a) > (b) ? (a) : (b))
394 # define MIN(a, b) ((a) < (b) ? (a) : (b))
396 typedef char boolean;
400 static reg_errcode_t byte_regex_compile _RE_ARGS ((const char *pattern, size_t size,
402 struct re_pattern_buffer *bufp));
404 static int byte_re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
405 const char *string1, int size1,
406 const char *string2, int size2,
408 struct re_registers *regs,
410 static int byte_re_search_2 PARAMS ((struct re_pattern_buffer *bufp,
411 const char *string1, int size1,
412 const char *string2, int size2,
413 int startpos, int range,
414 struct re_registers *regs, int stop));
415 static int byte_re_compile_fastmap PARAMS ((struct re_pattern_buffer *bufp));
418 static reg_errcode_t wcs_regex_compile _RE_ARGS ((const char *pattern, size_t size,
420 struct re_pattern_buffer *bufp));
423 static int wcs_re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
424 const char *cstring1, int csize1,
425 const char *cstring2, int csize2,
427 struct re_registers *regs,
429 wchar_t *string1, int size1,
430 wchar_t *string2, int size2,
431 int *mbs_offset1, int *mbs_offset2));
432 static int wcs_re_search_2 PARAMS ((struct re_pattern_buffer *bufp,
433 const char *string1, int size1,
434 const char *string2, int size2,
435 int startpos, int range,
436 struct re_registers *regs, int stop));
437 static int wcs_re_compile_fastmap PARAMS ((struct re_pattern_buffer *bufp));
440 /* These are the command codes that appear in compiled regular
441 expressions. Some opcodes are followed by argument bytes. A
442 command code can specify any interpretation whatsoever for its
443 arguments. Zero bytes may appear in the compiled regular expression. */
449 /* Succeed right away--no more backtracking. */
452 /* Followed by one byte giving n, then by n literal bytes. */
456 /* Same as exactn, but contains binary data. */
460 /* Matches any (more or less) character. */
463 /* Matches any one char belonging to specified set. First
464 following byte is number of bitmap bytes. Then come bytes
465 for a bitmap saying which chars are in. Bits in each byte
466 are ordered low-bit-first. A character is in the set if its
467 bit is 1. A character too large to have a bit in the map is
468 automatically not in the set. */
469 /* ifdef MBS_SUPPORT, following element is length of character
470 classes, length of collating symbols, length of equivalence
471 classes, length of character ranges, and length of characters.
472 Next, character class element, collating symbols elements,
473 equivalence class elements, range elements, and character
475 See regex_compile function. */
478 /* Same parameters as charset, but match any character that is
479 not one of those specified. */
482 /* Start remembering the text that is matched, for storing in a
483 register. Followed by one byte with the register number, in
484 the range 0 to one less than the pattern buffer's re_nsub
485 field. Then followed by one byte with the number of groups
486 inner to this one. (This last has to be part of the
487 start_memory only because we need it in the on_failure_jump
491 /* Stop remembering the text that is matched and store it in a
492 memory register. Followed by one byte with the register
493 number, in the range 0 to one less than `re_nsub' in the
494 pattern buffer, and one byte with the number of inner groups,
495 just like `start_memory'. (We need the number of inner
496 groups here because we don't have any easy way of finding the
497 corresponding start_memory when we're at a stop_memory.) */
500 /* Match a duplicate of something remembered. Followed by one
501 byte containing the register number. */
504 /* Fail unless at beginning of line. */
507 /* Fail unless at end of line. */
510 /* Succeeds if at beginning of buffer (if emacs) or at beginning
511 of string to be matched (if not). */
514 /* Analogously, for end of buffer/string. */
517 /* Followed by two byte relative address to which to jump. */
520 /* Same as jump, but marks the end of an alternative. */
523 /* Followed by two-byte relative address of place to resume at
524 in case of failure. */
525 /* ifdef MBS_SUPPORT, the size of address is 1. */
528 /* Like on_failure_jump, but pushes a placeholder instead of the
529 current string position when executed. */
530 on_failure_keep_string_jump,
532 /* Throw away latest failure point and then jump to following
533 two-byte relative address. */
534 /* ifdef MBS_SUPPORT, the size of address is 1. */
537 /* Change to pop_failure_jump if know won't have to backtrack to
538 match; otherwise change to jump. This is used to jump
539 back to the beginning of a repeat. If what follows this jump
540 clearly won't match what the repeat does, such that we can be
541 sure that there is no use backtracking out of repetitions
542 already matched, then we change it to a pop_failure_jump.
543 Followed by two-byte address. */
544 /* ifdef MBS_SUPPORT, the size of address is 1. */
547 /* Jump to following two-byte address, and push a dummy failure
548 point. This failure point will be thrown away if an attempt
549 is made to use it for a failure. A `+' construct makes this
550 before the first repeat. Also used as an intermediary kind
551 of jump when compiling an alternative. */
552 /* ifdef MBS_SUPPORT, the size of address is 1. */
555 /* Push a dummy failure point and continue. Used at the end of
559 /* Followed by two-byte relative address and two-byte number n.
560 After matching N times, jump to the address upon failure. */
561 /* ifdef MBS_SUPPORT, the size of address is 1. */
564 /* Followed by two-byte relative address, and two-byte number n.
565 Jump to the address N times, then fail. */
566 /* ifdef MBS_SUPPORT, the size of address is 1. */
569 /* Set the following two-byte relative address to the
570 subsequent two-byte number. The address *includes* the two
572 /* ifdef MBS_SUPPORT, the size of address is 1. */
575 wordchar, /* Matches any word-constituent character. */
576 notwordchar, /* Matches any char that is not a word-constituent. */
578 wordbeg, /* Succeeds if at word beginning. */
579 wordend, /* Succeeds if at word end. */
581 wordbound, /* Succeeds if at a word boundary. */
582 notwordbound /* Succeeds if not at a word boundary. */
585 ,before_dot, /* Succeeds if before point. */
586 at_dot, /* Succeeds if at point. */
587 after_dot, /* Succeeds if after point. */
589 /* Matches any character whose syntax is specified. Followed by
590 a byte which contains a syntax code, e.g., Sword. */
593 /* Matches any character whose syntax is not that specified. */
597 #endif /* not INSIDE_RECURSION */
602 # define UCHAR_T unsigned char
603 # define COMPILED_BUFFER_VAR bufp->buffer
604 # define OFFSET_ADDRESS_SIZE 2
605 # define PREFIX(name) byte_##name
606 # define ARG_PREFIX(name) name
607 # define PUT_CHAR(c) putchar (c)
610 # define CHAR_T wchar_t
611 # define UCHAR_T wchar_t
612 # define COMPILED_BUFFER_VAR wc_buffer
613 # define OFFSET_ADDRESS_SIZE 1 /* the size which STORE_NUMBER macro use */
614 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
615 # define PREFIX(name) wcs_##name
616 # define ARG_PREFIX(name) c##name
617 /* Should we use wide stream?? */
618 # define PUT_CHAR(c) printf ("%C", c);
624 # define INSIDE_RECURSION
626 # undef INSIDE_RECURSION
629 # define INSIDE_RECURSION
631 # undef INSIDE_RECURSION
635 #ifdef INSIDE_RECURSION
636 /* Common operations on the compiled pattern. */
638 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
639 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
642 # define STORE_NUMBER(destination, number) \
644 *(destination) = (UCHAR_T)(number); \
647 # define STORE_NUMBER(destination, number) \
649 (destination)[0] = (number) & 0377; \
650 (destination)[1] = (number) >> 8; \
654 /* Same as STORE_NUMBER, except increment DESTINATION to
655 the byte after where the number is stored. Therefore, DESTINATION
656 must be an lvalue. */
657 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
659 # define STORE_NUMBER_AND_INCR(destination, number) \
661 STORE_NUMBER (destination, number); \
662 (destination) += OFFSET_ADDRESS_SIZE; \
665 /* Put into DESTINATION a number stored in two contiguous bytes starting
667 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
670 # define EXTRACT_NUMBER(destination, source) \
672 (destination) = *(source); \
675 # define EXTRACT_NUMBER(destination, source) \
677 (destination) = *(source) & 0377; \
678 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
683 static void PREFIX(extract_number) _RE_ARGS ((int *dest, UCHAR_T *source));
685 PREFIX(extract_number) (dest, source)
692 int temp = SIGN_EXTEND_CHAR (*(source + 1));
693 *dest = *source & 0377;
698 # ifndef EXTRACT_MACROS /* To debug the macros. */
699 # undef EXTRACT_NUMBER
700 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
701 # endif /* not EXTRACT_MACROS */
705 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
706 SOURCE must be an lvalue. */
708 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
710 EXTRACT_NUMBER (destination, source); \
711 (source) += OFFSET_ADDRESS_SIZE; \
715 static void PREFIX(extract_number_and_incr) _RE_ARGS ((int *destination,
718 PREFIX(extract_number_and_incr) (destination, source)
722 PREFIX(extract_number) (destination, *source);
723 *source += OFFSET_ADDRESS_SIZE;
726 # ifndef EXTRACT_MACROS
727 # undef EXTRACT_NUMBER_AND_INCR
728 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
729 PREFIX(extract_number_and_incr) (&dest, &src)
730 # endif /* not EXTRACT_MACROS */
736 /* If DEBUG is defined, Regex prints many voluminous messages about what
737 it is doing (if the variable `debug' is nonzero). If linked with the
738 main program in `iregex.c', you can enter patterns and strings
739 interactively. And if linked with the main program in `main.c' and
740 the other test files, you can run the already-written tests. */
744 # ifndef DEFINED_ONCE
746 /* We use standard I/O for debugging. */
749 /* It is useful to test things that ``must'' be true when debugging. */
754 # define DEBUG_STATEMENT(e) e
755 # define DEBUG_PRINT1(x) if (debug) printf (x)
756 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
757 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
758 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
759 # endif /* not DEFINED_ONCE */
761 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
762 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
763 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
764 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
767 /* Print the fastmap in human-readable form. */
769 # ifndef DEFINED_ONCE
771 print_fastmap (fastmap)
774 unsigned was_a_range = 0;
777 while (i < (1 << BYTEWIDTH))
783 while (i < (1 << BYTEWIDTH) && fastmap[i])
797 # endif /* not DEFINED_ONCE */
800 /* Print a compiled pattern string in human-readable form, starting at
801 the START pointer into it and ending just before the pointer END. */
804 PREFIX(print_partial_compiled_pattern) (start, end)
819 /* Loop over pattern commands. */
823 printf ("%td:\t", p - start);
825 printf ("%ld:\t", (long int) (p - start));
828 switch ((re_opcode_t) *p++)
836 printf ("/exactn/%d", mcnt);
848 printf ("/exactn_bin/%d", mcnt);
851 printf("/%lx", (long int) *p++);
855 # endif /* MBS_SUPPORT */
859 printf ("/start_memory/%d/%ld", mcnt, (long int) *p++);
864 printf ("/stop_memory/%d/%ld", mcnt, (long int) *p++);
868 printf ("/duplicate/%ld", (long int) *p++);
881 printf ("/charset [%s",
882 (re_opcode_t) *(workp - 1) == charset_not ? "^" : "");
884 length = *workp++; /* the length of char_classes */
885 for (i=0 ; i<length ; i++)
886 printf("[:%lx:]", (long int) *p++);
887 length = *workp++; /* the length of collating_symbol */
888 for (i=0 ; i<length ;)
892 PUT_CHAR((i++,*p++));
896 length = *workp++; /* the length of equivalence_class */
897 for (i=0 ; i<length ;)
901 PUT_CHAR((i++,*p++));
905 length = *workp++; /* the length of char_range */
906 for (i=0 ; i<length ; i++)
908 wchar_t range_start = *p++;
909 wchar_t range_end = *p++;
910 printf("%C-%C", range_start, range_end);
912 length = *workp++; /* the length of char */
913 for (i=0 ; i<length ; i++)
917 register int c, last = -100;
918 register int in_range = 0;
920 printf ("/charset [%s",
921 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
923 assert (p + *p < pend);
925 for (c = 0; c < 256; c++)
927 && (p[1 + (c/8)] & (1 << (c % 8))))
929 /* Are we starting a range? */
930 if (last + 1 == c && ! in_range)
935 /* Have we broken a range? */
936 else if (last + 1 != c && in_range)
966 case on_failure_jump:
967 PREFIX(extract_number_and_incr) (&mcnt, &p);
969 printf ("/on_failure_jump to %td", p + mcnt - start);
971 printf ("/on_failure_jump to %ld", (long int) (p + mcnt - start));
975 case on_failure_keep_string_jump:
976 PREFIX(extract_number_and_incr) (&mcnt, &p);
978 printf ("/on_failure_keep_string_jump to %td", p + mcnt - start);
980 printf ("/on_failure_keep_string_jump to %ld",
981 (long int) (p + mcnt - start));
985 case dummy_failure_jump:
986 PREFIX(extract_number_and_incr) (&mcnt, &p);
988 printf ("/dummy_failure_jump to %td", p + mcnt - start);
990 printf ("/dummy_failure_jump to %ld", (long int) (p + mcnt - start));
994 case push_dummy_failure:
995 printf ("/push_dummy_failure");
999 PREFIX(extract_number_and_incr) (&mcnt, &p);
1001 printf ("/maybe_pop_jump to %td", p + mcnt - start);
1003 printf ("/maybe_pop_jump to %ld", (long int) (p + mcnt - start));
1007 case pop_failure_jump:
1008 PREFIX(extract_number_and_incr) (&mcnt, &p);
1010 printf ("/pop_failure_jump to %td", p + mcnt - start);
1012 printf ("/pop_failure_jump to %ld", (long int) (p + mcnt - start));
1017 PREFIX(extract_number_and_incr) (&mcnt, &p);
1019 printf ("/jump_past_alt to %td", p + mcnt - start);
1021 printf ("/jump_past_alt to %ld", (long int) (p + mcnt - start));
1026 PREFIX(extract_number_and_incr) (&mcnt, &p);
1028 printf ("/jump to %td", p + mcnt - start);
1030 printf ("/jump to %ld", (long int) (p + mcnt - start));
1035 PREFIX(extract_number_and_incr) (&mcnt, &p);
1037 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1039 printf ("/succeed_n to %td, %d times", p1 - start, mcnt2);
1041 printf ("/succeed_n to %ld, %d times",
1042 (long int) (p1 - start), mcnt2);
1047 PREFIX(extract_number_and_incr) (&mcnt, &p);
1049 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1050 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
1054 PREFIX(extract_number_and_incr) (&mcnt, &p);
1056 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1058 printf ("/set_number_at location %td to %d", p1 - start, mcnt2);
1060 printf ("/set_number_at location %ld to %d",
1061 (long int) (p1 - start), mcnt2);
1066 printf ("/wordbound");
1070 printf ("/notwordbound");
1074 printf ("/wordbeg");
1078 printf ("/wordend");
1083 printf ("/before_dot");
1091 printf ("/after_dot");
1095 printf ("/syntaxspec");
1097 printf ("/%d", mcnt);
1101 printf ("/notsyntaxspec");
1103 printf ("/%d", mcnt);
1108 printf ("/wordchar");
1112 printf ("/notwordchar");
1124 printf ("?%ld", (long int) *(p-1));
1131 printf ("%td:\tend of pattern.\n", p - start);
1133 printf ("%ld:\tend of pattern.\n", (long int) (p - start));
1139 PREFIX(print_compiled_pattern) (bufp)
1140 struct re_pattern_buffer *bufp;
1142 UCHAR_T *buffer = (UCHAR_T*) bufp->buffer;
1144 PREFIX(print_partial_compiled_pattern) (buffer, buffer
1145 + bufp->used / sizeof(UCHAR_T));
1146 printf ("%ld bytes used/%ld bytes allocated.\n",
1147 bufp->used, bufp->allocated);
1149 if (bufp->fastmap_accurate && bufp->fastmap)
1151 printf ("fastmap: ");
1152 print_fastmap (bufp->fastmap);
1156 printf ("re_nsub: %Zd\t", bufp->re_nsub);
1158 printf ("re_nsub: %ld\t", (long int) bufp->re_nsub);
1160 printf ("regs_alloc: %d\t", bufp->regs_allocated);
1161 printf ("can_be_null: %d\t", bufp->can_be_null);
1162 printf ("newline_anchor: %d\n", bufp->newline_anchor);
1163 printf ("no_sub: %d\t", bufp->no_sub);
1164 printf ("not_bol: %d\t", bufp->not_bol);
1165 printf ("not_eol: %d\t", bufp->not_eol);
1166 printf ("syntax: %lx\n", bufp->syntax);
1167 /* Perhaps we should print the translate table? */
1172 PREFIX(print_double_string) (where, string1, size1, string2, size2)
1173 const CHAR_T *where;
1174 const CHAR_T *string1;
1175 const CHAR_T *string2;
1187 if (FIRST_STRING_P (where))
1189 for (this_char = where - string1; this_char < size1; this_char++)
1190 PUT_CHAR (string1[this_char]);
1196 for (this_char = where - string2; this_char < size2; this_char++)
1198 PUT_CHAR (string2[this_char]);
1201 fputs ("...", stdout);
1208 # ifndef DEFINED_ONCE
1217 # else /* not DEBUG */
1219 # ifndef DEFINED_ONCE
1223 # define DEBUG_STATEMENT(e)
1224 # define DEBUG_PRINT1(x)
1225 # define DEBUG_PRINT2(x1, x2)
1226 # define DEBUG_PRINT3(x1, x2, x3)
1227 # define DEBUG_PRINT4(x1, x2, x3, x4)
1228 # endif /* not DEFINED_ONCE */
1229 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1230 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1232 # endif /* not DEBUG */
1237 /* This convert a multibyte string to a wide character string.
1238 And write their correspondances to offset_buffer(see below)
1239 and write whether each wchar_t is binary data to is_binary.
1240 This assume invalid multibyte sequences as binary data.
1241 We assume offset_buffer and is_binary is already allocated
1244 static size_t convert_mbs_to_wcs (CHAR_T *dest, const unsigned char* src,
1245 size_t len, int *offset_buffer,
1248 convert_mbs_to_wcs (dest, src, len, offset_buffer, is_binary)
1250 const unsigned char* src;
1251 size_t len; /* the length of multibyte string. */
1253 /* It hold correspondances between src(char string) and
1254 dest(wchar_t string) for optimization.
1256 dest = {'X', 'Y', 'Z'}
1257 (each "xxx", "y" and "zz" represent one multibyte character
1258 corresponding to 'X', 'Y' and 'Z'.)
1259 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1265 wchar_t *pdest = dest;
1266 const unsigned char *psrc = src;
1267 size_t wc_count = 0;
1271 size_t mb_remain = len;
1272 size_t mb_count = 0;
1274 /* Initialize the conversion state. */
1275 memset (&mbs, 0, sizeof (mbstate_t));
1277 offset_buffer[0] = 0;
1278 for( ; mb_remain > 0 ; ++wc_count, ++pdest, mb_remain -= consumed,
1281 consumed = mbrtowc (pdest, psrc, mb_remain, &mbs);
1284 /* failed to convert. maybe src contains binary data.
1285 So we consume 1 byte manualy. */
1289 is_binary[wc_count] = TRUE;
1292 is_binary[wc_count] = FALSE;
1293 /* In sjis encoding, we use yen sign as escape character in
1294 place of reverse solidus. So we convert 0x5c(yen sign in
1295 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1296 solidus in UCS2). */
1297 if (consumed == 1 && (int) *psrc == 0x5c && (int) *pdest == 0xa5)
1298 *pdest = (wchar_t) *psrc;
1300 offset_buffer[wc_count + 1] = mb_count += consumed;
1303 /* Fill remain of the buffer with sentinel. */
1304 for (i = wc_count + 1 ; i <= len ; i++)
1305 offset_buffer[i] = mb_count + 1;
1312 #else /* not INSIDE_RECURSION */
1314 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1315 also be assigned to arbitrarily: each pattern buffer stores its own
1316 syntax, so it can be changed between regex compilations. */
1317 /* This has no initializer because initialized variables in Emacs
1318 become read-only after dumping. */
1319 reg_syntax_t re_syntax_options;
1322 /* Specify the precise syntax of regexps for compilation. This provides
1323 for compatibility for various utilities which historically have
1324 different, incompatible syntaxes.
1326 The argument SYNTAX is a bit mask comprised of the various bits
1327 defined in regex.h. We return the old syntax. */
1330 re_set_syntax (syntax)
1331 reg_syntax_t syntax;
1333 reg_syntax_t ret = re_syntax_options;
1335 re_syntax_options = syntax;
1337 if (syntax & RE_DEBUG)
1339 else if (debug) /* was on but now is not */
1345 weak_alias (__re_set_syntax, re_set_syntax)
1348 /* This table gives an error message for each of the error codes listed
1349 in regex.h. Obviously the order here has to be same as there.
1350 POSIX doesn't require that we do anything for REG_NOERROR,
1351 but why not be nice? */
1353 static const char re_error_msgid[] =
1355 # define REG_NOERROR_IDX 0
1356 gettext_noop ("Success") /* REG_NOERROR */
1358 # define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success")
1359 gettext_noop ("No match") /* REG_NOMATCH */
1361 # define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match")
1362 gettext_noop ("Invalid regular expression") /* REG_BADPAT */
1364 # define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression")
1365 gettext_noop ("Invalid collation character") /* REG_ECOLLATE */
1367 # define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character")
1368 gettext_noop ("Invalid character class name") /* REG_ECTYPE */
1370 # define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name")
1371 gettext_noop ("Trailing backslash") /* REG_EESCAPE */
1373 # define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash")
1374 gettext_noop ("Invalid back reference") /* REG_ESUBREG */
1376 # define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference")
1377 gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */
1379 # define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^")
1380 gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */
1382 # define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(")
1383 gettext_noop ("Unmatched \\{") /* REG_EBRACE */
1385 # define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{")
1386 gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */
1388 # define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}")
1389 gettext_noop ("Invalid range end") /* REG_ERANGE */
1391 # define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end")
1392 gettext_noop ("Memory exhausted") /* REG_ESPACE */
1394 # define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted")
1395 gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */
1397 # define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression")
1398 gettext_noop ("Premature end of regular expression") /* REG_EEND */
1400 # define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression")
1401 gettext_noop ("Regular expression too big") /* REG_ESIZE */
1403 # define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big")
1404 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
1407 static const size_t re_error_msgid_idx[] =
1428 #endif /* INSIDE_RECURSION */
1430 #ifndef DEFINED_ONCE
1431 /* Avoiding alloca during matching, to placate r_alloc. */
1433 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1434 searching and matching functions should not call alloca. On some
1435 systems, alloca is implemented in terms of malloc, and if we're
1436 using the relocating allocator routines, then malloc could cause a
1437 relocation, which might (if the strings being searched are in the
1438 ralloc heap) shift the data out from underneath the regexp
1441 Here's another reason to avoid allocation: Emacs
1442 processes input from X in a signal handler; processing X input may
1443 call malloc; if input arrives while a matching routine is calling
1444 malloc, then we're scrod. But Emacs can't just block input while
1445 calling matching routines; then we don't notice interrupts when
1446 they come in. So, Emacs blocks input around all regexp calls
1447 except the matching calls, which it leaves unprotected, in the
1448 faith that they will not malloc. */
1450 /* Normally, this is fine. */
1451 # define MATCH_MAY_ALLOCATE
1453 /* When using GNU C, we are not REALLY using the C alloca, no matter
1454 what config.h may say. So don't take precautions for it. */
1459 /* The match routines may not allocate if (1) they would do it with malloc
1460 and (2) it's not safe for them to use malloc.
1461 Note that if REL_ALLOC is defined, matching would not use malloc for the
1462 failure stack, but we would still use it for the register vectors;
1463 so REL_ALLOC should not affect this. */
1464 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1465 # undef MATCH_MAY_ALLOCATE
1467 #endif /* not DEFINED_ONCE */
1469 #ifdef INSIDE_RECURSION
1470 /* Failure stack declarations and macros; both re_compile_fastmap and
1471 re_match_2 use a failure stack. These have to be macros because of
1472 REGEX_ALLOCATE_STACK. */
1475 /* Number of failure points for which to initially allocate space
1476 when matching. If this number is exceeded, we allocate more
1477 space, so it is not a hard limit. */
1478 # ifndef INIT_FAILURE_ALLOC
1479 # define INIT_FAILURE_ALLOC 5
1482 /* Roughly the maximum number of failure points on the stack. Would be
1483 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1484 This is a variable only so users of regex can assign to it; we never
1485 change it ourselves. */
1487 # ifdef INT_IS_16BIT
1489 # ifndef DEFINED_ONCE
1490 # if defined MATCH_MAY_ALLOCATE
1491 /* 4400 was enough to cause a crash on Alpha OSF/1,
1492 whose default stack limit is 2mb. */
1493 long int re_max_failures = 4000;
1495 long int re_max_failures = 2000;
1499 union PREFIX(fail_stack_elt)
1505 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1509 PREFIX(fail_stack_elt_t) *stack;
1510 unsigned long int size;
1511 unsigned long int avail; /* Offset of next open position. */
1512 } PREFIX(fail_stack_type);
1514 # else /* not INT_IS_16BIT */
1516 # ifndef DEFINED_ONCE
1517 # if defined MATCH_MAY_ALLOCATE
1518 /* 4400 was enough to cause a crash on Alpha OSF/1,
1519 whose default stack limit is 2mb. */
1520 int re_max_failures = 4000;
1522 int re_max_failures = 2000;
1526 union PREFIX(fail_stack_elt)
1532 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1536 PREFIX(fail_stack_elt_t) *stack;
1538 unsigned avail; /* Offset of next open position. */
1539 } PREFIX(fail_stack_type);
1541 # endif /* INT_IS_16BIT */
1543 # ifndef DEFINED_ONCE
1544 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1545 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1546 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1550 /* Define macros to initialize and free the failure stack.
1551 Do `return -2' if the alloc fails. */
1553 # ifdef MATCH_MAY_ALLOCATE
1554 # define INIT_FAIL_STACK() \
1556 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1557 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1559 if (fail_stack.stack == NULL) \
1562 fail_stack.size = INIT_FAILURE_ALLOC; \
1563 fail_stack.avail = 0; \
1566 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1568 # define INIT_FAIL_STACK() \
1570 fail_stack.avail = 0; \
1573 # define RESET_FAIL_STACK()
1577 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1579 Return 1 if succeeds, and 0 if either ran out of memory
1580 allocating space for it or it was already too large.
1582 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1584 # define DOUBLE_FAIL_STACK(fail_stack) \
1585 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1587 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1588 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1589 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1590 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1592 (fail_stack).stack == NULL \
1594 : ((fail_stack).size <<= 1, \
1598 /* Push pointer POINTER on FAIL_STACK.
1599 Return 1 if was able to do so and 0 if ran out of memory allocating
1601 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1602 ((FAIL_STACK_FULL () \
1603 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1605 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1608 /* Push a pointer value onto the failure stack.
1609 Assumes the variable `fail_stack'. Probably should only
1610 be called from within `PUSH_FAILURE_POINT'. */
1611 # define PUSH_FAILURE_POINTER(item) \
1612 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1614 /* This pushes an integer-valued item onto the failure stack.
1615 Assumes the variable `fail_stack'. Probably should only
1616 be called from within `PUSH_FAILURE_POINT'. */
1617 # define PUSH_FAILURE_INT(item) \
1618 fail_stack.stack[fail_stack.avail++].integer = (item)
1620 /* Push a fail_stack_elt_t value onto the failure stack.
1621 Assumes the variable `fail_stack'. Probably should only
1622 be called from within `PUSH_FAILURE_POINT'. */
1623 # define PUSH_FAILURE_ELT(item) \
1624 fail_stack.stack[fail_stack.avail++] = (item)
1626 /* These three POP... operations complement the three PUSH... operations.
1627 All assume that `fail_stack' is nonempty. */
1628 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1629 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1630 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1632 /* Used to omit pushing failure point id's when we're not debugging. */
1634 # define DEBUG_PUSH PUSH_FAILURE_INT
1635 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1637 # define DEBUG_PUSH(item)
1638 # define DEBUG_POP(item_addr)
1642 /* Push the information about the state we will need
1643 if we ever fail back to it.
1645 Requires variables fail_stack, regstart, regend, reg_info, and
1646 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1649 Does `return FAILURE_CODE' if runs out of memory. */
1651 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1653 char *destination; \
1654 /* Must be int, so when we don't save any registers, the arithmetic \
1655 of 0 + -1 isn't done as unsigned. */ \
1656 /* Can't be int, since there is not a shred of a guarantee that int \
1657 is wide enough to hold a value of something to which pointer can \
1659 active_reg_t this_reg; \
1661 DEBUG_STATEMENT (failure_id++); \
1662 DEBUG_STATEMENT (nfailure_points_pushed++); \
1663 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1664 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1665 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1667 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1668 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1670 /* Ensure we have enough space allocated for what we will push. */ \
1671 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1673 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1674 return failure_code; \
1676 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1677 (fail_stack).size); \
1678 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1681 /* Push the info, starting with the registers. */ \
1682 DEBUG_PRINT1 ("\n"); \
1685 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1688 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1689 DEBUG_STATEMENT (num_regs_pushed++); \
1691 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1692 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1694 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1695 PUSH_FAILURE_POINTER (regend[this_reg]); \
1697 DEBUG_PRINT2 (" info: %p\n ", \
1698 reg_info[this_reg].word.pointer); \
1699 DEBUG_PRINT2 (" match_null=%d", \
1700 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1701 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1702 DEBUG_PRINT2 (" matched_something=%d", \
1703 MATCHED_SOMETHING (reg_info[this_reg])); \
1704 DEBUG_PRINT2 (" ever_matched=%d", \
1705 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1706 DEBUG_PRINT1 ("\n"); \
1707 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1710 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1711 PUSH_FAILURE_INT (lowest_active_reg); \
1713 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1714 PUSH_FAILURE_INT (highest_active_reg); \
1716 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1717 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1718 PUSH_FAILURE_POINTER (pattern_place); \
1720 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1721 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1723 DEBUG_PRINT1 ("'\n"); \
1724 PUSH_FAILURE_POINTER (string_place); \
1726 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1727 DEBUG_PUSH (failure_id); \
1730 # ifndef DEFINED_ONCE
1731 /* This is the number of items that are pushed and popped on the stack
1732 for each register. */
1733 # define NUM_REG_ITEMS 3
1735 /* Individual items aside from the registers. */
1737 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1739 # define NUM_NONREG_ITEMS 4
1742 /* We push at most this many items on the stack. */
1743 /* We used to use (num_regs - 1), which is the number of registers
1744 this regexp will save; but that was changed to 5
1745 to avoid stack overflow for a regexp with lots of parens. */
1746 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1748 /* We actually push this many items. */
1749 # define NUM_FAILURE_ITEMS \
1751 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1755 /* How many items can still be added to the stack without overflowing it. */
1756 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1757 # endif /* not DEFINED_ONCE */
1760 /* Pops what PUSH_FAIL_STACK pushes.
1762 We restore into the parameters, all of which should be lvalues:
1763 STR -- the saved data position.
1764 PAT -- the saved pattern position.
1765 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1766 REGSTART, REGEND -- arrays of string positions.
1767 REG_INFO -- array of information about each subexpression.
1769 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1770 `pend', `string1', `size1', `string2', and `size2'. */
1771 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1773 DEBUG_STATEMENT (unsigned failure_id;) \
1774 active_reg_t this_reg; \
1775 const UCHAR_T *string_temp; \
1777 assert (!FAIL_STACK_EMPTY ()); \
1779 /* Remove failure points and point to how many regs pushed. */ \
1780 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1781 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1782 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1784 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1786 DEBUG_POP (&failure_id); \
1787 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1789 /* If the saved string location is NULL, it came from an \
1790 on_failure_keep_string_jump opcode, and we want to throw away the \
1791 saved NULL, thus retaining our current position in the string. */ \
1792 string_temp = POP_FAILURE_POINTER (); \
1793 if (string_temp != NULL) \
1794 str = (const CHAR_T *) string_temp; \
1796 DEBUG_PRINT2 (" Popping string %p: `", str); \
1797 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1798 DEBUG_PRINT1 ("'\n"); \
1800 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1801 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1802 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1804 /* Restore register info. */ \
1805 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1806 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1808 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1809 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1812 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1814 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1816 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1817 DEBUG_PRINT2 (" info: %p\n", \
1818 reg_info[this_reg].word.pointer); \
1820 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1821 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1823 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1824 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1828 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1830 reg_info[this_reg].word.integer = 0; \
1831 regend[this_reg] = 0; \
1832 regstart[this_reg] = 0; \
1834 highest_active_reg = high_reg; \
1837 set_regs_matched_done = 0; \
1838 DEBUG_STATEMENT (nfailure_points_popped++); \
1839 } /* POP_FAILURE_POINT */
1841 /* Structure for per-register (a.k.a. per-group) information.
1842 Other register information, such as the
1843 starting and ending positions (which are addresses), and the list of
1844 inner groups (which is a bits list) are maintained in separate
1847 We are making a (strictly speaking) nonportable assumption here: that
1848 the compiler will pack our bit fields into something that fits into
1849 the type of `word', i.e., is something that fits into one item on the
1853 /* Declarations and macros for re_match_2. */
1857 PREFIX(fail_stack_elt_t) word;
1860 /* This field is one if this group can match the empty string,
1861 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1862 # define MATCH_NULL_UNSET_VALUE 3
1863 unsigned match_null_string_p : 2;
1864 unsigned is_active : 1;
1865 unsigned matched_something : 1;
1866 unsigned ever_matched_something : 1;
1868 } PREFIX(register_info_type);
1870 # ifndef DEFINED_ONCE
1871 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1872 # define IS_ACTIVE(R) ((R).bits.is_active)
1873 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1874 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1877 /* Call this when have matched a real character; it sets `matched' flags
1878 for the subexpressions which we are currently inside. Also records
1879 that those subexprs have matched. */
1880 # define SET_REGS_MATCHED() \
1883 if (!set_regs_matched_done) \
1886 set_regs_matched_done = 1; \
1887 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1889 MATCHED_SOMETHING (reg_info[r]) \
1890 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1896 # endif /* not DEFINED_ONCE */
1898 /* Registers are set to a sentinel when they haven't yet matched. */
1899 static CHAR_T PREFIX(reg_unset_dummy);
1900 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1901 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1903 /* Subroutine declarations and macros for regex_compile. */
1904 static void PREFIX(store_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc, int arg));
1905 static void PREFIX(store_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1906 int arg1, int arg2));
1907 static void PREFIX(insert_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1908 int arg, UCHAR_T *end));
1909 static void PREFIX(insert_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1910 int arg1, int arg2, UCHAR_T *end));
1911 static boolean PREFIX(at_begline_loc_p) _RE_ARGS ((const CHAR_T *pattern,
1913 reg_syntax_t syntax));
1914 static boolean PREFIX(at_endline_loc_p) _RE_ARGS ((const CHAR_T *p,
1916 reg_syntax_t syntax));
1918 static reg_errcode_t wcs_compile_range _RE_ARGS ((CHAR_T range_start,
1919 const CHAR_T **p_ptr,
1922 reg_syntax_t syntax,
1925 static void insert_space _RE_ARGS ((int num, CHAR_T *loc, CHAR_T *end));
1927 static reg_errcode_t byte_compile_range _RE_ARGS ((unsigned int range_start,
1931 reg_syntax_t syntax,
1935 /* Fetch the next character in the uncompiled pattern---translating it
1936 if necessary. Also cast from a signed character in the constant
1937 string passed to us by the user to an unsigned char that we can use
1938 as an array index (in, e.g., `translate'). */
1939 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1940 because it is impossible to allocate 4GB array for some encodings
1941 which have 4 byte character_set like UCS4. */
1944 # define PATFETCH(c) \
1945 do {if (p == pend) return REG_EEND; \
1946 c = (UCHAR_T) *p++; \
1947 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1950 # define PATFETCH(c) \
1951 do {if (p == pend) return REG_EEND; \
1952 c = (unsigned char) *p++; \
1953 if (translate) c = (unsigned char) translate[c]; \
1958 /* Fetch the next character in the uncompiled pattern, with no
1960 # define PATFETCH_RAW(c) \
1961 do {if (p == pend) return REG_EEND; \
1962 c = (UCHAR_T) *p++; \
1965 /* Go backwards one character in the pattern. */
1966 # define PATUNFETCH p--
1969 /* If `translate' is non-null, return translate[D], else just D. We
1970 cast the subscript to translate because some data is declared as
1971 `char *', to avoid warnings when a string constant is passed. But
1972 when we use a character as a subscript we must make it unsigned. */
1973 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1974 because it is impossible to allocate 4GB array for some encodings
1975 which have 4 byte character_set like UCS4. */
1979 # define TRANSLATE(d) \
1980 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1981 ? (char) translate[(unsigned char) (d)] : (d))
1983 # define TRANSLATE(d) \
1984 (translate ? (char) translate[(unsigned char) (d)] : (d))
1989 /* Macros for outputting the compiled pattern into `buffer'. */
1991 /* If the buffer isn't allocated when it comes in, use this. */
1992 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1994 /* Make sure we have at least N more bytes of space in buffer. */
1996 # define GET_BUFFER_SPACE(n) \
1997 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1998 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
2001 # define GET_BUFFER_SPACE(n) \
2002 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
2006 /* Make sure we have one more byte of buffer space and then add C to it. */
2007 # define BUF_PUSH(c) \
2009 GET_BUFFER_SPACE (1); \
2010 *b++ = (UCHAR_T) (c); \
2014 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
2015 # define BUF_PUSH_2(c1, c2) \
2017 GET_BUFFER_SPACE (2); \
2018 *b++ = (UCHAR_T) (c1); \
2019 *b++ = (UCHAR_T) (c2); \
2023 /* As with BUF_PUSH_2, except for three bytes. */
2024 # define BUF_PUSH_3(c1, c2, c3) \
2026 GET_BUFFER_SPACE (3); \
2027 *b++ = (UCHAR_T) (c1); \
2028 *b++ = (UCHAR_T) (c2); \
2029 *b++ = (UCHAR_T) (c3); \
2032 /* Store a jump with opcode OP at LOC to location TO. We store a
2033 relative address offset by the three bytes the jump itself occupies. */
2034 # define STORE_JUMP(op, loc, to) \
2035 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
2037 /* Likewise, for a two-argument jump. */
2038 # define STORE_JUMP2(op, loc, to, arg) \
2039 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
2041 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
2042 # define INSERT_JUMP(op, loc, to) \
2043 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
2045 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
2046 # define INSERT_JUMP2(op, loc, to, arg) \
2047 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
2050 /* This is not an arbitrary limit: the arguments which represent offsets
2051 into the pattern are two bytes long. So if 2^16 bytes turns out to
2052 be too small, many things would have to change. */
2053 /* Any other compiler which, like MSC, has allocation limit below 2^16
2054 bytes will have to use approach similar to what was done below for
2055 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2056 reallocating to 0 bytes. Such thing is not going to work too well.
2057 You have been warned!! */
2058 # ifndef DEFINED_ONCE
2059 # if defined _MSC_VER && !defined WIN32
2060 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2061 The REALLOC define eliminates a flurry of conversion warnings,
2062 but is not required. */
2063 # define MAX_BUF_SIZE 65500L
2064 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2066 # define MAX_BUF_SIZE (1L << 16)
2067 # define REALLOC(p,s) realloc ((p), (s))
2070 /* Extend the buffer by twice its current size via realloc and
2071 reset the pointers that pointed into the old block to point to the
2072 correct places in the new one. If extending the buffer results in it
2073 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2074 # if __BOUNDED_POINTERS__
2075 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2076 # define MOVE_BUFFER_POINTER(P) \
2077 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2078 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2081 SET_HIGH_BOUND (b); \
2082 SET_HIGH_BOUND (begalt); \
2083 if (fixup_alt_jump) \
2084 SET_HIGH_BOUND (fixup_alt_jump); \
2086 SET_HIGH_BOUND (laststart); \
2087 if (pending_exact) \
2088 SET_HIGH_BOUND (pending_exact); \
2091 # define MOVE_BUFFER_POINTER(P) (P) += incr
2092 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2094 # endif /* not DEFINED_ONCE */
2097 # define EXTEND_BUFFER() \
2099 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2101 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2103 bufp->allocated <<= 1; \
2104 if (bufp->allocated > MAX_BUF_SIZE) \
2105 bufp->allocated = MAX_BUF_SIZE; \
2106 /* How many characters the new buffer can have? */ \
2107 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2108 if (wchar_count == 0) wchar_count = 1; \
2109 /* Truncate the buffer to CHAR_T align. */ \
2110 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2111 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2112 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2113 if (COMPILED_BUFFER_VAR == NULL) \
2114 return REG_ESPACE; \
2115 /* If the buffer moved, move all the pointers into it. */ \
2116 if (old_buffer != COMPILED_BUFFER_VAR) \
2118 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2119 MOVE_BUFFER_POINTER (b); \
2120 MOVE_BUFFER_POINTER (begalt); \
2121 if (fixup_alt_jump) \
2122 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2124 MOVE_BUFFER_POINTER (laststart); \
2125 if (pending_exact) \
2126 MOVE_BUFFER_POINTER (pending_exact); \
2128 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2131 # define EXTEND_BUFFER() \
2133 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2134 if (bufp->allocated == MAX_BUF_SIZE) \
2136 bufp->allocated <<= 1; \
2137 if (bufp->allocated > MAX_BUF_SIZE) \
2138 bufp->allocated = MAX_BUF_SIZE; \
2139 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2141 if (COMPILED_BUFFER_VAR == NULL) \
2142 return REG_ESPACE; \
2143 /* If the buffer moved, move all the pointers into it. */ \
2144 if (old_buffer != COMPILED_BUFFER_VAR) \
2146 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2147 MOVE_BUFFER_POINTER (b); \
2148 MOVE_BUFFER_POINTER (begalt); \
2149 if (fixup_alt_jump) \
2150 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2152 MOVE_BUFFER_POINTER (laststart); \
2153 if (pending_exact) \
2154 MOVE_BUFFER_POINTER (pending_exact); \
2156 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2160 # ifndef DEFINED_ONCE
2161 /* Since we have one byte reserved for the register number argument to
2162 {start,stop}_memory, the maximum number of groups we can report
2163 things about is what fits in that byte. */
2164 # define MAX_REGNUM 255
2166 /* But patterns can have more than `MAX_REGNUM' registers. We just
2167 ignore the excess. */
2168 typedef unsigned regnum_t;
2171 /* Macros for the compile stack. */
2173 /* Since offsets can go either forwards or backwards, this type needs to
2174 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2175 /* int may be not enough when sizeof(int) == 2. */
2176 typedef long pattern_offset_t;
2180 pattern_offset_t begalt_offset;
2181 pattern_offset_t fixup_alt_jump;
2182 pattern_offset_t inner_group_offset;
2183 pattern_offset_t laststart_offset;
2185 } compile_stack_elt_t;
2190 compile_stack_elt_t *stack;
2192 unsigned avail; /* Offset of next open position. */
2193 } compile_stack_type;
2196 # define INIT_COMPILE_STACK_SIZE 32
2198 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2199 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2201 /* The next available element. */
2202 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2204 # endif /* not DEFINED_ONCE */
2206 /* Set the bit for character C in a list. */
2207 # ifndef DEFINED_ONCE
2208 # define SET_LIST_BIT(c) \
2209 (b[((unsigned char) (c)) / BYTEWIDTH] \
2210 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2211 # endif /* DEFINED_ONCE */
2213 /* Get the next unsigned number in the uncompiled pattern. */
2214 # define GET_UNSIGNED_NUMBER(num) \
2219 if (c < '0' || c > '9') \
2221 if (num <= RE_DUP_MAX) \
2225 num = num * 10 + c - '0'; \
2230 # ifndef DEFINED_ONCE
2231 # if defined _LIBC || WIDE_CHAR_SUPPORT
2232 /* The GNU C library provides support for user-defined character classes
2233 and the functions from ISO C amendement 1. */
2234 # ifdef CHARCLASS_NAME_MAX
2235 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2237 /* This shouldn't happen but some implementation might still have this
2238 problem. Use a reasonable default value. */
2239 # define CHAR_CLASS_MAX_LENGTH 256
2243 # define IS_CHAR_CLASS(string) __wctype (string)
2245 # define IS_CHAR_CLASS(string) wctype (string)
2248 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
2250 # define IS_CHAR_CLASS(string) \
2251 (STREQ (string, "alpha") || STREQ (string, "upper") \
2252 || STREQ (string, "lower") || STREQ (string, "digit") \
2253 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
2254 || STREQ (string, "space") || STREQ (string, "print") \
2255 || STREQ (string, "punct") || STREQ (string, "graph") \
2256 || STREQ (string, "cntrl") || STREQ (string, "blank"))
2258 # endif /* DEFINED_ONCE */
2260 # ifndef MATCH_MAY_ALLOCATE
2262 /* If we cannot allocate large objects within re_match_2_internal,
2263 we make the fail stack and register vectors global.
2264 The fail stack, we grow to the maximum size when a regexp
2266 The register vectors, we adjust in size each time we
2267 compile a regexp, according to the number of registers it needs. */
2269 static PREFIX(fail_stack_type) fail_stack;
2271 /* Size with which the following vectors are currently allocated.
2272 That is so we can make them bigger as needed,
2273 but never make them smaller. */
2274 # ifdef DEFINED_ONCE
2275 static int regs_allocated_size;
2277 static const char ** regstart, ** regend;
2278 static const char ** old_regstart, ** old_regend;
2279 static const char **best_regstart, **best_regend;
2280 static const char **reg_dummy;
2281 # endif /* DEFINED_ONCE */
2283 static PREFIX(register_info_type) *PREFIX(reg_info);
2284 static PREFIX(register_info_type) *PREFIX(reg_info_dummy);
2286 /* Make the register vectors big enough for NUM_REGS registers,
2287 but don't make them smaller. */
2290 PREFIX(regex_grow_registers) (num_regs)
2293 if (num_regs > regs_allocated_size)
2295 RETALLOC_IF (regstart, num_regs, const char *);
2296 RETALLOC_IF (regend, num_regs, const char *);
2297 RETALLOC_IF (old_regstart, num_regs, const char *);
2298 RETALLOC_IF (old_regend, num_regs, const char *);
2299 RETALLOC_IF (best_regstart, num_regs, const char *);
2300 RETALLOC_IF (best_regend, num_regs, const char *);
2301 RETALLOC_IF (PREFIX(reg_info), num_regs, PREFIX(register_info_type));
2302 RETALLOC_IF (reg_dummy, num_regs, const char *);
2303 RETALLOC_IF (PREFIX(reg_info_dummy), num_regs, PREFIX(register_info_type));
2305 regs_allocated_size = num_regs;
2309 # endif /* not MATCH_MAY_ALLOCATE */
2311 # ifndef DEFINED_ONCE
2312 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
2315 # endif /* not DEFINED_ONCE */
2317 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2318 Returns one of error codes defined in `regex.h', or zero for success.
2320 Assumes the `allocated' (and perhaps `buffer') and `translate'
2321 fields are set in BUFP on entry.
2323 If it succeeds, results are put in BUFP (if it returns an error, the
2324 contents of BUFP are undefined):
2325 `buffer' is the compiled pattern;
2326 `syntax' is set to SYNTAX;
2327 `used' is set to the length of the compiled pattern;
2328 `fastmap_accurate' is zero;
2329 `re_nsub' is the number of subexpressions in PATTERN;
2330 `not_bol' and `not_eol' are zero;
2332 The `fastmap' and `newline_anchor' fields are neither
2333 examined nor set. */
2335 /* Return, freeing storage we allocated. */
2337 # define FREE_STACK_RETURN(value) \
2338 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2340 # define FREE_STACK_RETURN(value) \
2341 return (free (compile_stack.stack), value)
2344 static reg_errcode_t
2345 PREFIX(regex_compile) (ARG_PREFIX(pattern), ARG_PREFIX(size), syntax, bufp)
2346 const char *ARG_PREFIX(pattern);
2347 size_t ARG_PREFIX(size);
2348 reg_syntax_t syntax;
2349 struct re_pattern_buffer *bufp;
2351 /* We fetch characters from PATTERN here. Even though PATTERN is
2352 `char *' (i.e., signed), we declare these variables as unsigned, so
2353 they can be reliably used as array indices. */
2354 register UCHAR_T c, c1;
2357 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2358 CHAR_T *pattern, *COMPILED_BUFFER_VAR;
2360 /* offset buffer for optimization. See convert_mbs_to_wc. */
2361 int *mbs_offset = NULL;
2362 /* It hold whether each wchar_t is binary data or not. */
2363 char *is_binary = NULL;
2364 /* A flag whether exactn is handling binary data or not. */
2365 char is_exactn_bin = FALSE;
2368 /* A random temporary spot in PATTERN. */
2371 /* Points to the end of the buffer, where we should append. */
2372 register UCHAR_T *b;
2374 /* Keeps track of unclosed groups. */
2375 compile_stack_type compile_stack;
2377 /* Points to the current (ending) position in the pattern. */
2382 const CHAR_T *p = pattern;
2383 const CHAR_T *pend = pattern + size;
2386 /* How to translate the characters in the pattern. */
2387 RE_TRANSLATE_TYPE translate = bufp->translate;
2389 /* Address of the count-byte of the most recently inserted `exactn'
2390 command. This makes it possible to tell if a new exact-match
2391 character can be added to that command or if the character requires
2392 a new `exactn' command. */
2393 UCHAR_T *pending_exact = 0;
2395 /* Address of start of the most recently finished expression.
2396 This tells, e.g., postfix * where to find the start of its
2397 operand. Reset at the beginning of groups and alternatives. */
2398 UCHAR_T *laststart = 0;
2400 /* Address of beginning of regexp, or inside of last group. */
2403 /* Address of the place where a forward jump should go to the end of
2404 the containing expression. Each alternative of an `or' -- except the
2405 last -- ends with a forward jump of this sort. */
2406 UCHAR_T *fixup_alt_jump = 0;
2408 /* Counts open-groups as they are encountered. Remembered for the
2409 matching close-group on the compile stack, so the same register
2410 number is put in the stop_memory as the start_memory. */
2411 regnum_t regnum = 0;
2414 /* Initialize the wchar_t PATTERN and offset_buffer. */
2415 p = pend = pattern = TALLOC(csize + 1, CHAR_T);
2416 mbs_offset = TALLOC(csize + 1, int);
2417 is_binary = TALLOC(csize + 1, char);
2418 if (pattern == NULL || mbs_offset == NULL || is_binary == NULL)
2425 pattern[csize] = L'\0'; /* sentinel */
2426 size = convert_mbs_to_wcs(pattern, cpattern, csize, mbs_offset, is_binary);
2438 DEBUG_PRINT1 ("\nCompiling pattern: ");
2441 unsigned debug_count;
2443 for (debug_count = 0; debug_count < size; debug_count++)
2444 PUT_CHAR (pattern[debug_count]);
2449 /* Initialize the compile stack. */
2450 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2451 if (compile_stack.stack == NULL)
2461 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2462 compile_stack.avail = 0;
2464 /* Initialize the pattern buffer. */
2465 bufp->syntax = syntax;
2466 bufp->fastmap_accurate = 0;
2467 bufp->not_bol = bufp->not_eol = 0;
2469 /* Set `used' to zero, so that if we return an error, the pattern
2470 printer (for debugging) will think there's no pattern. We reset it
2474 /* Always count groups, whether or not bufp->no_sub is set. */
2477 #if !defined emacs && !defined SYNTAX_TABLE
2478 /* Initialize the syntax table. */
2479 init_syntax_once ();
2482 if (bufp->allocated == 0)
2485 { /* If zero allocated, but buffer is non-null, try to realloc
2486 enough space. This loses if buffer's address is bogus, but
2487 that is the user's responsibility. */
2489 /* Free bufp->buffer and allocate an array for wchar_t pattern
2492 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE/sizeof(UCHAR_T),
2495 RETALLOC (COMPILED_BUFFER_VAR, INIT_BUF_SIZE, UCHAR_T);
2499 { /* Caller did not allocate a buffer. Do it for them. */
2500 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE / sizeof(UCHAR_T),
2504 if (!COMPILED_BUFFER_VAR) FREE_STACK_RETURN (REG_ESPACE);
2506 bufp->buffer = (char*)COMPILED_BUFFER_VAR;
2508 bufp->allocated = INIT_BUF_SIZE;
2512 COMPILED_BUFFER_VAR = (UCHAR_T*) bufp->buffer;
2515 begalt = b = COMPILED_BUFFER_VAR;
2517 /* Loop through the uncompiled pattern until we're at the end. */
2526 if ( /* If at start of pattern, it's an operator. */
2528 /* If context independent, it's an operator. */
2529 || syntax & RE_CONTEXT_INDEP_ANCHORS
2530 /* Otherwise, depends on what's come before. */
2531 || PREFIX(at_begline_loc_p) (pattern, p, syntax))
2541 if ( /* If at end of pattern, it's an operator. */
2543 /* If context independent, it's an operator. */
2544 || syntax & RE_CONTEXT_INDEP_ANCHORS
2545 /* Otherwise, depends on what's next. */
2546 || PREFIX(at_endline_loc_p) (p, pend, syntax))
2556 if ((syntax & RE_BK_PLUS_QM)
2557 || (syntax & RE_LIMITED_OPS))
2561 /* If there is no previous pattern... */
2564 if (syntax & RE_CONTEXT_INVALID_OPS)
2565 FREE_STACK_RETURN (REG_BADRPT);
2566 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2571 /* Are we optimizing this jump? */
2572 boolean keep_string_p = false;
2574 /* 1 means zero (many) matches is allowed. */
2575 char zero_times_ok = 0, many_times_ok = 0;
2577 /* If there is a sequence of repetition chars, collapse it
2578 down to just one (the right one). We can't combine
2579 interval operators with these because of, e.g., `a{2}*',
2580 which should only match an even number of `a's. */
2584 zero_times_ok |= c != '+';
2585 many_times_ok |= c != '?';
2593 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2596 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2598 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2601 if (!(c1 == '+' || c1 == '?'))
2616 /* If we get here, we found another repeat character. */
2619 /* Star, etc. applied to an empty pattern is equivalent
2620 to an empty pattern. */
2624 /* Now we know whether or not zero matches is allowed
2625 and also whether or not two or more matches is allowed. */
2627 { /* More than one repetition is allowed, so put in at the
2628 end a backward relative jump from `b' to before the next
2629 jump we're going to put in below (which jumps from
2630 laststart to after this jump).
2632 But if we are at the `*' in the exact sequence `.*\n',
2633 insert an unconditional jump backwards to the .,
2634 instead of the beginning of the loop. This way we only
2635 push a failure point once, instead of every time
2636 through the loop. */
2637 assert (p - 1 > pattern);
2639 /* Allocate the space for the jump. */
2640 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2642 /* We know we are not at the first character of the pattern,
2643 because laststart was nonzero. And we've already
2644 incremented `p', by the way, to be the character after
2645 the `*'. Do we have to do something analogous here
2646 for null bytes, because of RE_DOT_NOT_NULL? */
2647 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2649 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2650 && !(syntax & RE_DOT_NEWLINE))
2651 { /* We have .*\n. */
2652 STORE_JUMP (jump, b, laststart);
2653 keep_string_p = true;
2656 /* Anything else. */
2657 STORE_JUMP (maybe_pop_jump, b, laststart -
2658 (1 + OFFSET_ADDRESS_SIZE));
2660 /* We've added more stuff to the buffer. */
2661 b += 1 + OFFSET_ADDRESS_SIZE;
2664 /* On failure, jump from laststart to b + 3, which will be the
2665 end of the buffer after this jump is inserted. */
2666 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2668 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2669 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2671 laststart, b + 1 + OFFSET_ADDRESS_SIZE);
2673 b += 1 + OFFSET_ADDRESS_SIZE;
2677 /* At least one repetition is required, so insert a
2678 `dummy_failure_jump' before the initial
2679 `on_failure_jump' instruction of the loop. This
2680 effects a skip over that instruction the first time
2681 we hit that loop. */
2682 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2683 INSERT_JUMP (dummy_failure_jump, laststart, laststart +
2684 2 + 2 * OFFSET_ADDRESS_SIZE);
2685 b += 1 + OFFSET_ADDRESS_SIZE;
2699 boolean had_char_class = false;
2701 CHAR_T range_start = 0xffffffff;
2703 unsigned int range_start = 0xffffffff;
2705 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2708 /* We assume a charset(_not) structure as a wchar_t array.
2709 charset[0] = (re_opcode_t) charset(_not)
2710 charset[1] = l (= length of char_classes)
2711 charset[2] = m (= length of collating_symbols)
2712 charset[3] = n (= length of equivalence_classes)
2713 charset[4] = o (= length of char_ranges)
2714 charset[5] = p (= length of chars)
2716 charset[6] = char_class (wctype_t)
2717 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2719 charset[l+5] = char_class (wctype_t)
2721 charset[l+6] = collating_symbol (wchar_t)
2723 charset[l+m+5] = collating_symbol (wchar_t)
2724 ifdef _LIBC we use the index if
2725 _NL_COLLATE_SYMB_EXTRAMB instead of
2728 charset[l+m+6] = equivalence_classes (wchar_t)
2730 charset[l+m+n+5] = equivalence_classes (wchar_t)
2731 ifdef _LIBC we use the index in
2732 _NL_COLLATE_WEIGHT instead of
2735 charset[l+m+n+6] = range_start
2736 charset[l+m+n+7] = range_end
2738 charset[l+m+n+2o+4] = range_start
2739 charset[l+m+n+2o+5] = range_end
2740 ifdef _LIBC we use the value looked up
2741 in _NL_COLLATE_COLLSEQ instead of
2744 charset[l+m+n+2o+6] = char
2746 charset[l+m+n+2o+p+5] = char
2750 /* We need at least 6 spaces: the opcode, the length of
2751 char_classes, the length of collating_symbols, the length of
2752 equivalence_classes, the length of char_ranges, the length of
2754 GET_BUFFER_SPACE (6);
2756 /* Save b as laststart. And We use laststart as the pointer
2757 to the first element of the charset here.
2758 In other words, laststart[i] indicates charset[i]. */
2761 /* We test `*p == '^' twice, instead of using an if
2762 statement, so we only need one BUF_PUSH. */
2763 BUF_PUSH (*p == '^' ? charset_not : charset);
2767 /* Push the length of char_classes, the length of
2768 collating_symbols, the length of equivalence_classes, the
2769 length of char_ranges and the length of chars. */
2770 BUF_PUSH_3 (0, 0, 0);
2773 /* Remember the first position in the bracket expression. */
2776 /* charset_not matches newline according to a syntax bit. */
2777 if ((re_opcode_t) b[-6] == charset_not
2778 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2781 laststart[5]++; /* Update the length of characters */
2784 /* Read in characters and ranges, setting map bits. */
2787 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2791 /* \ might escape characters inside [...] and [^...]. */
2792 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2794 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2798 laststart[5]++; /* Update the length of chars */
2803 /* Could be the end of the bracket expression. If it's
2804 not (i.e., when the bracket expression is `[]' so
2805 far), the ']' character bit gets set way below. */
2806 if (c == ']' && p != p1 + 1)
2809 /* Look ahead to see if it's a range when the last thing
2810 was a character class. */
2811 if (had_char_class && c == '-' && *p != ']')
2812 FREE_STACK_RETURN (REG_ERANGE);
2814 /* Look ahead to see if it's a range when the last thing
2815 was a character: if this is a hyphen not at the
2816 beginning or the end of a list, then it's the range
2819 && !(p - 2 >= pattern && p[-2] == '[')
2820 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2824 /* Allocate the space for range_start and range_end. */
2825 GET_BUFFER_SPACE (2);
2826 /* Update the pointer to indicate end of buffer. */
2828 ret = wcs_compile_range (range_start, &p, pend, translate,
2829 syntax, b, laststart);
2830 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2831 range_start = 0xffffffff;
2833 else if (p[0] == '-' && p[1] != ']')
2834 { /* This handles ranges made up of characters only. */
2837 /* Move past the `-'. */
2839 /* Allocate the space for range_start and range_end. */
2840 GET_BUFFER_SPACE (2);
2841 /* Update the pointer to indicate end of buffer. */
2843 ret = wcs_compile_range (c, &p, pend, translate, syntax, b,
2845 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2846 range_start = 0xffffffff;
2849 /* See if we're at the beginning of a possible character
2851 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2852 { /* Leave room for the null. */
2853 char str[CHAR_CLASS_MAX_LENGTH + 1];
2858 /* If pattern is `[[:'. */
2859 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2864 if ((c == ':' && *p == ']') || p == pend)
2866 if (c1 < CHAR_CLASS_MAX_LENGTH)
2869 /* This is in any case an invalid class name. */
2874 /* If isn't a word bracketed by `[:' and `:]':
2875 undo the ending character, the letters, and leave
2876 the leading `:' and `[' (but store them as character). */
2877 if (c == ':' && *p == ']')
2882 /* Query the character class as wctype_t. */
2883 wt = IS_CHAR_CLASS (str);
2885 FREE_STACK_RETURN (REG_ECTYPE);
2887 /* Throw away the ] at the end of the character
2891 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2893 /* Allocate the space for character class. */
2894 GET_BUFFER_SPACE(CHAR_CLASS_SIZE);
2895 /* Update the pointer to indicate end of buffer. */
2896 b += CHAR_CLASS_SIZE;
2897 /* Move data which follow character classes
2898 not to violate the data. */
2899 insert_space(CHAR_CLASS_SIZE,
2900 laststart + 6 + laststart[1],
2902 alignedp = ((uintptr_t)(laststart + 6 + laststart[1])
2903 + __alignof__(wctype_t) - 1)
2904 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
2905 /* Store the character class. */
2906 *((wctype_t*)alignedp) = wt;
2907 /* Update length of char_classes */
2908 laststart[1] += CHAR_CLASS_SIZE;
2910 had_char_class = true;
2919 laststart[5] += 2; /* Update the length of characters */
2921 had_char_class = false;
2924 else if (syntax & RE_CHAR_CLASSES && c == '[' && (*p == '='
2927 CHAR_T str[128]; /* Should be large enough. */
2928 CHAR_T delim = *p; /* '=' or '.' */
2931 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
2936 /* If pattern is `[[=' or '[[.'. */
2937 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2942 if ((c == delim && *p == ']') || p == pend)
2944 if (c1 < sizeof (str) - 1)
2947 /* This is in any case an invalid class name. */
2952 if (c == delim && *p == ']' && str[0] != '\0')
2954 unsigned int i, offset;
2955 /* If we have no collation data we use the default
2956 collation in which each character is in a class
2957 by itself. It also means that ASCII is the
2958 character set and therefore we cannot have character
2959 with more than one byte in the multibyte
2962 /* If not defined _LIBC, we push the name and
2963 `\0' for the sake of matching performance. */
2964 int datasize = c1 + 1;
2972 FREE_STACK_RETURN (REG_ECOLLATE);
2977 const int32_t *table;
2978 const int32_t *weights;
2979 const int32_t *extra;
2980 const int32_t *indirect;
2983 /* This #include defines a local function! */
2984 # include <locale/weightwc.h>
2988 /* We push the index for equivalence class. */
2991 table = (const int32_t *)
2992 _NL_CURRENT (LC_COLLATE,
2993 _NL_COLLATE_TABLEWC);
2994 weights = (const int32_t *)
2995 _NL_CURRENT (LC_COLLATE,
2996 _NL_COLLATE_WEIGHTWC);
2997 extra = (const int32_t *)
2998 _NL_CURRENT (LC_COLLATE,
2999 _NL_COLLATE_EXTRAWC);
3000 indirect = (const int32_t *)
3001 _NL_CURRENT (LC_COLLATE,
3002 _NL_COLLATE_INDIRECTWC);
3004 idx = findidx ((const wint_t**)&cp);
3005 if (idx == 0 || cp < (wint_t*) str + c1)
3006 /* This is no valid character. */
3007 FREE_STACK_RETURN (REG_ECOLLATE);
3009 str[0] = (wchar_t)idx;
3011 else /* delim == '.' */
3013 /* We push collation sequence value
3014 for collating symbol. */
3016 const int32_t *symb_table;
3017 const unsigned char *extra;
3024 /* We have to convert the name to a single-byte
3025 string. This is possible since the names
3026 consist of ASCII characters and the internal
3027 representation is UCS4. */
3028 for (i = 0; i < c1; ++i)
3029 char_str[i] = str[i];
3032 _NL_CURRENT_WORD (LC_COLLATE,
3033 _NL_COLLATE_SYMB_HASH_SIZEMB);
3034 symb_table = (const int32_t *)
3035 _NL_CURRENT (LC_COLLATE,
3036 _NL_COLLATE_SYMB_TABLEMB);
3037 extra = (const unsigned char *)
3038 _NL_CURRENT (LC_COLLATE,
3039 _NL_COLLATE_SYMB_EXTRAMB);
3041 /* Locate the character in the hashing table. */
3042 hash = elem_hash (char_str, c1);
3045 elem = hash % table_size;
3046 second = hash % (table_size - 2);
3047 while (symb_table[2 * elem] != 0)
3049 /* First compare the hashing value. */
3050 if (symb_table[2 * elem] == hash
3051 && c1 == extra[symb_table[2 * elem + 1]]
3052 && memcmp (char_str,
3053 &extra[symb_table[2 * elem + 1]
3056 /* Yep, this is the entry. */
3057 idx = symb_table[2 * elem + 1];
3058 idx += 1 + extra[idx];
3066 if (symb_table[2 * elem] != 0)
3068 /* Compute the index of the byte sequence
3070 idx += 1 + extra[idx];
3071 /* Adjust for the alignment. */
3072 idx = (idx + 3) & ~3;
3074 str[0] = (wchar_t) idx + 4;
3076 else if (symb_table[2 * elem] == 0 && c1 == 1)
3078 /* No valid character. Match it as a
3079 single byte character. */
3080 had_char_class = false;
3082 /* Update the length of characters */
3084 range_start = str[0];
3086 /* Throw away the ] at the end of the
3087 collating symbol. */
3089 /* exit from the switch block. */
3093 FREE_STACK_RETURN (REG_ECOLLATE);
3098 /* Throw away the ] at the end of the equivalence
3099 class (or collating symbol). */
3102 /* Allocate the space for the equivalence class
3103 (or collating symbol) (and '\0' if needed). */
3104 GET_BUFFER_SPACE(datasize);
3105 /* Update the pointer to indicate end of buffer. */
3109 { /* equivalence class */
3110 /* Calculate the offset of char_ranges,
3111 which is next to equivalence_classes. */
3112 offset = laststart[1] + laststart[2]
3115 insert_space(datasize, laststart + offset, b - 1);
3117 /* Write the equivalence_class and \0. */
3118 for (i = 0 ; i < datasize ; i++)
3119 laststart[offset + i] = str[i];
3121 /* Update the length of equivalence_classes. */
3122 laststart[3] += datasize;
3123 had_char_class = true;
3125 else /* delim == '.' */
3126 { /* collating symbol */
3127 /* Calculate the offset of the equivalence_classes,
3128 which is next to collating_symbols. */
3129 offset = laststart[1] + laststart[2] + 6;
3130 /* Insert space and write the collationg_symbol
3132 insert_space(datasize, laststart + offset, b-1);
3133 for (i = 0 ; i < datasize ; i++)
3134 laststart[offset + i] = str[i];
3136 /* In re_match_2_internal if range_start < -1, we
3137 assume -range_start is the offset of the
3138 collating symbol which is specified as
3139 the character of the range start. So we assign
3140 -(laststart[1] + laststart[2] + 6) to
3142 range_start = -(laststart[1] + laststart[2] + 6);
3143 /* Update the length of collating_symbol. */
3144 laststart[2] += datasize;
3145 had_char_class = false;
3155 laststart[5] += 2; /* Update the length of characters */
3156 range_start = delim;
3157 had_char_class = false;
3162 had_char_class = false;
3164 laststart[5]++; /* Update the length of characters */
3170 /* Ensure that we have enough space to push a charset: the
3171 opcode, the length count, and the bitset; 34 bytes in all. */
3172 GET_BUFFER_SPACE (34);
3176 /* We test `*p == '^' twice, instead of using an if
3177 statement, so we only need one BUF_PUSH. */
3178 BUF_PUSH (*p == '^' ? charset_not : charset);
3182 /* Remember the first position in the bracket expression. */
3185 /* Push the number of bytes in the bitmap. */
3186 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
3188 /* Clear the whole map. */
3189 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
3191 /* charset_not matches newline according to a syntax bit. */
3192 if ((re_opcode_t) b[-2] == charset_not
3193 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
3194 SET_LIST_BIT ('\n');
3196 /* Read in characters and ranges, setting map bits. */
3199 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3203 /* \ might escape characters inside [...] and [^...]. */
3204 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
3206 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3214 /* Could be the end of the bracket expression. If it's
3215 not (i.e., when the bracket expression is `[]' so
3216 far), the ']' character bit gets set way below. */
3217 if (c == ']' && p != p1 + 1)
3220 /* Look ahead to see if it's a range when the last thing
3221 was a character class. */
3222 if (had_char_class && c == '-' && *p != ']')
3223 FREE_STACK_RETURN (REG_ERANGE);
3225 /* Look ahead to see if it's a range when the last thing
3226 was a character: if this is a hyphen not at the
3227 beginning or the end of a list, then it's the range
3230 && !(p - 2 >= pattern && p[-2] == '[')
3231 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
3235 = byte_compile_range (range_start, &p, pend, translate,
3237 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3238 range_start = 0xffffffff;
3241 else if (p[0] == '-' && p[1] != ']')
3242 { /* This handles ranges made up of characters only. */
3245 /* Move past the `-'. */
3248 ret = byte_compile_range (c, &p, pend, translate, syntax, b);
3249 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3250 range_start = 0xffffffff;
3253 /* See if we're at the beginning of a possible character
3256 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
3257 { /* Leave room for the null. */
3258 char str[CHAR_CLASS_MAX_LENGTH + 1];
3263 /* If pattern is `[[:'. */
3264 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3269 if ((c == ':' && *p == ']') || p == pend)
3271 if (c1 < CHAR_CLASS_MAX_LENGTH)
3274 /* This is in any case an invalid class name. */
3279 /* If isn't a word bracketed by `[:' and `:]':
3280 undo the ending character, the letters, and leave
3281 the leading `:' and `[' (but set bits for them). */
3282 if (c == ':' && *p == ']')
3284 # if defined _LIBC || WIDE_CHAR_SUPPORT
3285 boolean is_lower = STREQ (str, "lower");
3286 boolean is_upper = STREQ (str, "upper");
3290 wt = IS_CHAR_CLASS (str);
3292 FREE_STACK_RETURN (REG_ECTYPE);
3294 /* Throw away the ] at the end of the character
3298 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3300 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
3303 if (__iswctype (__btowc (ch), wt))
3306 if (iswctype (btowc (ch), wt))
3310 if (translate && (is_upper || is_lower)
3311 && (ISUPPER (ch) || ISLOWER (ch)))
3315 had_char_class = true;
3318 boolean is_alnum = STREQ (str, "alnum");
3319 boolean is_alpha = STREQ (str, "alpha");
3320 boolean is_blank = STREQ (str, "blank");
3321 boolean is_cntrl = STREQ (str, "cntrl");
3322 boolean is_digit = STREQ (str, "digit");
3323 boolean is_graph = STREQ (str, "graph");
3324 boolean is_lower = STREQ (str, "lower");
3325 boolean is_print = STREQ (str, "print");
3326 boolean is_punct = STREQ (str, "punct");
3327 boolean is_space = STREQ (str, "space");
3328 boolean is_upper = STREQ (str, "upper");
3329 boolean is_xdigit = STREQ (str, "xdigit");
3331 if (!IS_CHAR_CLASS (str))
3332 FREE_STACK_RETURN (REG_ECTYPE);
3334 /* Throw away the ] at the end of the character
3338 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3340 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
3342 /* This was split into 3 if's to
3343 avoid an arbitrary limit in some compiler. */
3344 if ( (is_alnum && ISALNUM (ch))
3345 || (is_alpha && ISALPHA (ch))
3346 || (is_blank && ISBLANK (ch))
3347 || (is_cntrl && ISCNTRL (ch)))
3349 if ( (is_digit && ISDIGIT (ch))
3350 || (is_graph && ISGRAPH (ch))
3351 || (is_lower && ISLOWER (ch))
3352 || (is_print && ISPRINT (ch)))
3354 if ( (is_punct && ISPUNCT (ch))
3355 || (is_space && ISSPACE (ch))
3356 || (is_upper && ISUPPER (ch))
3357 || (is_xdigit && ISXDIGIT (ch)))
3359 if ( translate && (is_upper || is_lower)
3360 && (ISUPPER (ch) || ISLOWER (ch)))
3363 had_char_class = true;
3364 # endif /* libc || wctype.h */
3374 had_char_class = false;
3377 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '=')
3379 unsigned char str[MB_LEN_MAX + 1];
3382 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3388 /* If pattern is `[[='. */
3389 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3394 if ((c == '=' && *p == ']') || p == pend)
3396 if (c1 < MB_LEN_MAX)
3399 /* This is in any case an invalid class name. */
3404 if (c == '=' && *p == ']' && str[0] != '\0')
3406 /* If we have no collation data we use the default
3407 collation in which each character is in a class
3408 by itself. It also means that ASCII is the
3409 character set and therefore we cannot have character
3410 with more than one byte in the multibyte
3417 FREE_STACK_RETURN (REG_ECOLLATE);
3419 /* Throw away the ] at the end of the equivalence
3423 /* Set the bit for the character. */
3424 SET_LIST_BIT (str[0]);
3429 /* Try to match the byte sequence in `str' against
3430 those known to the collate implementation.
3431 First find out whether the bytes in `str' are
3432 actually from exactly one character. */
3433 const int32_t *table;
3434 const unsigned char *weights;
3435 const unsigned char *extra;
3436 const int32_t *indirect;
3438 const unsigned char *cp = str;
3441 /* This #include defines a local function! */
3442 # include <locale/weight.h>
3444 table = (const int32_t *)
3445 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
3446 weights = (const unsigned char *)
3447 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
3448 extra = (const unsigned char *)
3449 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
3450 indirect = (const int32_t *)
3451 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
3453 idx = findidx (&cp);
3454 if (idx == 0 || cp < str + c1)
3455 /* This is no valid character. */
3456 FREE_STACK_RETURN (REG_ECOLLATE);
3458 /* Throw away the ] at the end of the equivalence
3462 /* Now we have to go throught the whole table
3463 and find all characters which have the same
3466 XXX Note that this is not entirely correct.
3467 we would have to match multibyte sequences
3468 but this is not possible with the current
3470 for (ch = 1; ch < 256; ++ch)
3471 /* XXX This test would have to be changed if we
3472 would allow matching multibyte sequences. */
3475 int32_t idx2 = table[ch];
3476 size_t len = weights[idx2];
3478 /* Test whether the lenghts match. */
3479 if (weights[idx] == len)
3481 /* They do. New compare the bytes of
3486 && (weights[idx + 1 + cnt]
3487 == weights[idx2 + 1 + cnt]))
3491 /* They match. Mark the character as
3498 had_char_class = true;
3508 had_char_class = false;
3511 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '.')
3513 unsigned char str[128]; /* Should be large enough. */
3516 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3522 /* If pattern is `[[.'. */
3523 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3528 if ((c == '.' && *p == ']') || p == pend)
3530 if (c1 < sizeof (str))
3533 /* This is in any case an invalid class name. */
3538 if (c == '.' && *p == ']' && str[0] != '\0')
3540 /* If we have no collation data we use the default
3541 collation in which each character is the name
3542 for its own class which contains only the one
3543 character. It also means that ASCII is the
3544 character set and therefore we cannot have character
3545 with more than one byte in the multibyte
3552 FREE_STACK_RETURN (REG_ECOLLATE);
3554 /* Throw away the ] at the end of the equivalence
3558 /* Set the bit for the character. */
3559 SET_LIST_BIT (str[0]);
3560 range_start = ((const unsigned char *) str)[0];
3565 /* Try to match the byte sequence in `str' against
3566 those known to the collate implementation.
3567 First find out whether the bytes in `str' are
3568 actually from exactly one character. */
3570 const int32_t *symb_table;
3571 const unsigned char *extra;
3578 _NL_CURRENT_WORD (LC_COLLATE,
3579 _NL_COLLATE_SYMB_HASH_SIZEMB);
3580 symb_table = (const int32_t *)
3581 _NL_CURRENT (LC_COLLATE,
3582 _NL_COLLATE_SYMB_TABLEMB);
3583 extra = (const unsigned char *)
3584 _NL_CURRENT (LC_COLLATE,
3585 _NL_COLLATE_SYMB_EXTRAMB);
3587 /* Locate the character in the hashing table. */
3588 hash = elem_hash (str, c1);
3591 elem = hash % table_size;
3592 second = hash % (table_size - 2);
3593 while (symb_table[2 * elem] != 0)
3595 /* First compare the hashing value. */
3596 if (symb_table[2 * elem] == hash
3597 && c1 == extra[symb_table[2 * elem + 1]]
3599 &extra[symb_table[2 * elem + 1]
3603 /* Yep, this is the entry. */
3604 idx = symb_table[2 * elem + 1];
3605 idx += 1 + extra[idx];
3613 if (symb_table[2 * elem] == 0)
3614 /* This is no valid character. */
3615 FREE_STACK_RETURN (REG_ECOLLATE);
3617 /* Throw away the ] at the end of the equivalence
3621 /* Now add the multibyte character(s) we found
3624 XXX Note that this is not entirely correct.
3625 we would have to match multibyte sequences
3626 but this is not possible with the current
3627 implementation. Also, we have to match
3628 collating symbols, which expand to more than
3629 one file, as a whole and not allow the
3630 individual bytes. */
3633 range_start = extra[idx];
3636 SET_LIST_BIT (extra[idx]);
3641 had_char_class = false;
3651 had_char_class = false;
3656 had_char_class = false;
3662 /* Discard any (non)matching list bytes that are all 0 at the
3663 end of the map. Decrease the map-length byte too. */
3664 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
3673 if (syntax & RE_NO_BK_PARENS)
3680 if (syntax & RE_NO_BK_PARENS)
3687 if (syntax & RE_NEWLINE_ALT)
3694 if (syntax & RE_NO_BK_VBAR)
3701 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
3702 goto handle_interval;
3708 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3710 /* Do not translate the character after the \, so that we can
3711 distinguish, e.g., \B from \b, even if we normally would
3712 translate, e.g., B to b. */
3718 if (syntax & RE_NO_BK_PARENS)
3719 goto normal_backslash;
3725 if (COMPILE_STACK_FULL)
3727 RETALLOC (compile_stack.stack, compile_stack.size << 1,
3728 compile_stack_elt_t);
3729 if (compile_stack.stack == NULL) return REG_ESPACE;
3731 compile_stack.size <<= 1;
3734 /* These are the values to restore when we hit end of this
3735 group. They are all relative offsets, so that if the
3736 whole pattern moves because of realloc, they will still
3738 COMPILE_STACK_TOP.begalt_offset = begalt - COMPILED_BUFFER_VAR;
3739 COMPILE_STACK_TOP.fixup_alt_jump
3740 = fixup_alt_jump ? fixup_alt_jump - COMPILED_BUFFER_VAR + 1 : 0;
3741 COMPILE_STACK_TOP.laststart_offset = b - COMPILED_BUFFER_VAR;
3742 COMPILE_STACK_TOP.regnum = regnum;
3744 /* We will eventually replace the 0 with the number of
3745 groups inner to this one. But do not push a
3746 start_memory for groups beyond the last one we can
3747 represent in the compiled pattern. */
3748 if (regnum <= MAX_REGNUM)
3750 COMPILE_STACK_TOP.inner_group_offset = b
3751 - COMPILED_BUFFER_VAR + 2;
3752 BUF_PUSH_3 (start_memory, regnum, 0);
3755 compile_stack.avail++;
3760 /* If we've reached MAX_REGNUM groups, then this open
3761 won't actually generate any code, so we'll have to
3762 clear pending_exact explicitly. */
3768 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
3770 if (COMPILE_STACK_EMPTY)
3772 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3773 goto normal_backslash;
3775 FREE_STACK_RETURN (REG_ERPAREN);
3780 { /* Push a dummy failure point at the end of the
3781 alternative for a possible future
3782 `pop_failure_jump' to pop. See comments at
3783 `push_dummy_failure' in `re_match_2'. */
3784 BUF_PUSH (push_dummy_failure);
3786 /* We allocated space for this jump when we assigned
3787 to `fixup_alt_jump', in the `handle_alt' case below. */
3788 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
3791 /* See similar code for backslashed left paren above. */
3792 if (COMPILE_STACK_EMPTY)
3794 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3797 FREE_STACK_RETURN (REG_ERPAREN);
3800 /* Since we just checked for an empty stack above, this
3801 ``can't happen''. */
3802 assert (compile_stack.avail != 0);
3804 /* We don't just want to restore into `regnum', because
3805 later groups should continue to be numbered higher,
3806 as in `(ab)c(de)' -- the second group is #2. */
3807 regnum_t this_group_regnum;
3809 compile_stack.avail--;
3810 begalt = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.begalt_offset;
3812 = COMPILE_STACK_TOP.fixup_alt_jump
3813 ? COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.fixup_alt_jump - 1
3815 laststart = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.laststart_offset;
3816 this_group_regnum = COMPILE_STACK_TOP.regnum;
3817 /* If we've reached MAX_REGNUM groups, then this open
3818 won't actually generate any code, so we'll have to
3819 clear pending_exact explicitly. */
3822 /* We're at the end of the group, so now we know how many
3823 groups were inside this one. */
3824 if (this_group_regnum <= MAX_REGNUM)
3826 UCHAR_T *inner_group_loc
3827 = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.inner_group_offset;
3829 *inner_group_loc = regnum - this_group_regnum;
3830 BUF_PUSH_3 (stop_memory, this_group_regnum,
3831 regnum - this_group_regnum);
3837 case '|': /* `\|'. */
3838 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
3839 goto normal_backslash;
3841 if (syntax & RE_LIMITED_OPS)
3844 /* Insert before the previous alternative a jump which
3845 jumps to this alternative if the former fails. */
3846 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3847 INSERT_JUMP (on_failure_jump, begalt,
3848 b + 2 + 2 * OFFSET_ADDRESS_SIZE);
3850 b += 1 + OFFSET_ADDRESS_SIZE;
3852 /* The alternative before this one has a jump after it
3853 which gets executed if it gets matched. Adjust that
3854 jump so it will jump to this alternative's analogous
3855 jump (put in below, which in turn will jump to the next
3856 (if any) alternative's such jump, etc.). The last such
3857 jump jumps to the correct final destination. A picture:
3863 If we are at `b', then fixup_alt_jump right now points to a
3864 three-byte space after `a'. We'll put in the jump, set
3865 fixup_alt_jump to right after `b', and leave behind three
3866 bytes which we'll fill in when we get to after `c'. */
3869 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
3871 /* Mark and leave space for a jump after this alternative,
3872 to be filled in later either by next alternative or
3873 when know we're at the end of a series of alternatives. */
3875 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3876 b += 1 + OFFSET_ADDRESS_SIZE;
3884 /* If \{ is a literal. */
3885 if (!(syntax & RE_INTERVALS)
3886 /* If we're at `\{' and it's not the open-interval
3888 || (syntax & RE_NO_BK_BRACES))
3889 goto normal_backslash;
3893 /* If got here, then the syntax allows intervals. */
3895 /* At least (most) this many matches must be made. */
3896 int lower_bound = -1, upper_bound = -1;
3898 /* Place in the uncompiled pattern (i.e., just after
3899 the '{') to go back to if the interval is invalid. */
3900 const CHAR_T *beg_interval = p;
3903 goto invalid_interval;
3905 GET_UNSIGNED_NUMBER (lower_bound);
3909 GET_UNSIGNED_NUMBER (upper_bound);
3910 if (upper_bound < 0)
3911 upper_bound = RE_DUP_MAX;
3914 /* Interval such as `{1}' => match exactly once. */
3915 upper_bound = lower_bound;
3917 if (! (0 <= lower_bound && lower_bound <= upper_bound))
3918 goto invalid_interval;
3920 if (!(syntax & RE_NO_BK_BRACES))
3922 if (c != '\\' || p == pend)
3923 goto invalid_interval;
3928 goto invalid_interval;
3930 /* If it's invalid to have no preceding re. */
3933 if (syntax & RE_CONTEXT_INVALID_OPS
3934 && !(syntax & RE_INVALID_INTERVAL_ORD))
3935 FREE_STACK_RETURN (REG_BADRPT);
3936 else if (syntax & RE_CONTEXT_INDEP_OPS)
3939 goto unfetch_interval;
3942 /* We just parsed a valid interval. */
3944 if (RE_DUP_MAX < upper_bound)
3945 FREE_STACK_RETURN (REG_BADBR);
3947 /* If the upper bound is zero, don't want to succeed at
3948 all; jump from `laststart' to `b + 3', which will be
3949 the end of the buffer after we insert the jump. */
3950 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3951 instead of 'b + 3'. */
3952 if (upper_bound == 0)
3954 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3955 INSERT_JUMP (jump, laststart, b + 1
3956 + OFFSET_ADDRESS_SIZE);
3957 b += 1 + OFFSET_ADDRESS_SIZE;
3960 /* Otherwise, we have a nontrivial interval. When
3961 we're all done, the pattern will look like:
3962 set_number_at <jump count> <upper bound>
3963 set_number_at <succeed_n count> <lower bound>
3964 succeed_n <after jump addr> <succeed_n count>
3966 jump_n <succeed_n addr> <jump count>
3967 (The upper bound and `jump_n' are omitted if
3968 `upper_bound' is 1, though.) */
3970 { /* If the upper bound is > 1, we need to insert
3971 more at the end of the loop. */
3972 unsigned nbytes = 2 + 4 * OFFSET_ADDRESS_SIZE +
3973 (upper_bound > 1) * (2 + 4 * OFFSET_ADDRESS_SIZE);
3975 GET_BUFFER_SPACE (nbytes);
3977 /* Initialize lower bound of the `succeed_n', even
3978 though it will be set during matching by its
3979 attendant `set_number_at' (inserted next),
3980 because `re_compile_fastmap' needs to know.
3981 Jump to the `jump_n' we might insert below. */
3982 INSERT_JUMP2 (succeed_n, laststart,
3983 b + 1 + 2 * OFFSET_ADDRESS_SIZE
3984 + (upper_bound > 1) * (1 + 2 * OFFSET_ADDRESS_SIZE)
3986 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3988 /* Code to initialize the lower bound. Insert
3989 before the `succeed_n'. The `5' is the last two
3990 bytes of this `set_number_at', plus 3 bytes of
3991 the following `succeed_n'. */
3992 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3993 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3994 of the following `succeed_n'. */
3995 PREFIX(insert_op2) (set_number_at, laststart, 1
3996 + 2 * OFFSET_ADDRESS_SIZE, lower_bound, b);
3997 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3999 if (upper_bound > 1)
4000 { /* More than one repetition is allowed, so
4001 append a backward jump to the `succeed_n'
4002 that starts this interval.
4004 When we've reached this during matching,
4005 we'll have matched the interval once, so
4006 jump back only `upper_bound - 1' times. */
4007 STORE_JUMP2 (jump_n, b, laststart
4008 + 2 * OFFSET_ADDRESS_SIZE + 1,
4010 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4012 /* The location we want to set is the second
4013 parameter of the `jump_n'; that is `b-2' as
4014 an absolute address. `laststart' will be
4015 the `set_number_at' we're about to insert;
4016 `laststart+3' the number to set, the source
4017 for the relative address. But we are
4018 inserting into the middle of the pattern --
4019 so everything is getting moved up by 5.
4020 Conclusion: (b - 2) - (laststart + 3) + 5,
4021 i.e., b - laststart.
4023 We insert this at the beginning of the loop
4024 so that if we fail during matching, we'll
4025 reinitialize the bounds. */
4026 PREFIX(insert_op2) (set_number_at, laststart,
4028 upper_bound - 1, b);
4029 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4036 if (!(syntax & RE_INVALID_INTERVAL_ORD))
4037 FREE_STACK_RETURN (p == pend ? REG_EBRACE : REG_BADBR);
4039 /* Match the characters as literals. */
4042 if (syntax & RE_NO_BK_BRACES)
4045 goto normal_backslash;
4049 /* There is no way to specify the before_dot and after_dot
4050 operators. rms says this is ok. --karl */
4058 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
4064 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
4070 if (syntax & RE_NO_GNU_OPS)
4073 BUF_PUSH (wordchar);
4078 if (syntax & RE_NO_GNU_OPS)
4081 BUF_PUSH (notwordchar);
4086 if (syntax & RE_NO_GNU_OPS)
4092 if (syntax & RE_NO_GNU_OPS)
4098 if (syntax & RE_NO_GNU_OPS)
4100 BUF_PUSH (wordbound);
4104 if (syntax & RE_NO_GNU_OPS)
4106 BUF_PUSH (notwordbound);
4110 if (syntax & RE_NO_GNU_OPS)
4116 if (syntax & RE_NO_GNU_OPS)
4121 case '1': case '2': case '3': case '4': case '5':
4122 case '6': case '7': case '8': case '9':
4123 if (syntax & RE_NO_BK_REFS)
4129 FREE_STACK_RETURN (REG_ESUBREG);
4131 /* Can't back reference to a subexpression if inside of it. */
4132 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
4136 BUF_PUSH_2 (duplicate, c1);
4142 if (syntax & RE_BK_PLUS_QM)
4145 goto normal_backslash;
4149 /* You might think it would be useful for \ to mean
4150 not to translate; but if we don't translate it
4151 it will never match anything. */
4159 /* Expects the character in `c'. */
4161 /* If no exactn currently being built. */
4164 /* If last exactn handle binary(or character) and
4165 new exactn handle character(or binary). */
4166 || is_exactn_bin != is_binary[p - 1 - pattern]
4169 /* If last exactn not at current position. */
4170 || pending_exact + *pending_exact + 1 != b
4172 /* We have only one byte following the exactn for the count. */
4173 || *pending_exact == (1 << BYTEWIDTH) - 1
4175 /* If followed by a repetition operator. */
4176 || *p == '*' || *p == '^'
4177 || ((syntax & RE_BK_PLUS_QM)
4178 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
4179 : (*p == '+' || *p == '?'))
4180 || ((syntax & RE_INTERVALS)
4181 && ((syntax & RE_NO_BK_BRACES)
4183 : (p[0] == '\\' && p[1] == '{'))))
4185 /* Start building a new exactn. */
4190 /* Is this exactn binary data or character? */
4191 is_exactn_bin = is_binary[p - 1 - pattern];
4193 BUF_PUSH_2 (exactn_bin, 0);
4195 BUF_PUSH_2 (exactn, 0);
4197 BUF_PUSH_2 (exactn, 0);
4199 pending_exact = b - 1;
4206 } /* while p != pend */
4209 /* Through the pattern now. */
4212 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
4214 if (!COMPILE_STACK_EMPTY)
4215 FREE_STACK_RETURN (REG_EPAREN);
4217 /* If we don't want backtracking, force success
4218 the first time we reach the end of the compiled pattern. */
4219 if (syntax & RE_NO_POSIX_BACKTRACKING)
4227 free (compile_stack.stack);
4229 /* We have succeeded; set the length of the buffer. */
4231 bufp->used = (uintptr_t) b - (uintptr_t) COMPILED_BUFFER_VAR;
4233 bufp->used = b - bufp->buffer;
4239 DEBUG_PRINT1 ("\nCompiled pattern: \n");
4240 PREFIX(print_compiled_pattern) (bufp);
4244 #ifndef MATCH_MAY_ALLOCATE
4245 /* Initialize the failure stack to the largest possible stack. This
4246 isn't necessary unless we're trying to avoid calling alloca in
4247 the search and match routines. */
4249 int num_regs = bufp->re_nsub + 1;
4251 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4252 is strictly greater than re_max_failures, the largest possible stack
4253 is 2 * re_max_failures failure points. */
4254 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
4256 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
4259 if (! fail_stack.stack)
4261 = (PREFIX(fail_stack_elt_t) *) xmalloc (fail_stack.size
4262 * sizeof (PREFIX(fail_stack_elt_t)));
4265 = (PREFIX(fail_stack_elt_t) *) xrealloc (fail_stack.stack,
4267 * sizeof (PREFIX(fail_stack_elt_t))));
4268 # else /* not emacs */
4269 if (! fail_stack.stack)
4271 = (PREFIX(fail_stack_elt_t) *) malloc (fail_stack.size
4272 * sizeof (PREFIX(fail_stack_elt_t)));
4275 = (PREFIX(fail_stack_elt_t) *) realloc (fail_stack.stack,
4277 * sizeof (PREFIX(fail_stack_elt_t))));
4278 # endif /* not emacs */
4281 PREFIX(regex_grow_registers) (num_regs);
4283 #endif /* not MATCH_MAY_ALLOCATE */
4286 } /* regex_compile */
4288 /* Subroutines for `regex_compile'. */
4290 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4291 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4294 PREFIX(store_op1) (op, loc, arg)
4299 *loc = (UCHAR_T) op;
4300 STORE_NUMBER (loc + 1, arg);
4304 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4305 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4308 PREFIX(store_op2) (op, loc, arg1, arg2)
4313 *loc = (UCHAR_T) op;
4314 STORE_NUMBER (loc + 1, arg1);
4315 STORE_NUMBER (loc + 1 + OFFSET_ADDRESS_SIZE, arg2);
4319 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4320 for OP followed by two-byte integer parameter ARG. */
4321 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4324 PREFIX(insert_op1) (op, loc, arg, end)
4330 register UCHAR_T *pfrom = end;
4331 register UCHAR_T *pto = end + 1 + OFFSET_ADDRESS_SIZE;
4333 while (pfrom != loc)
4336 PREFIX(store_op1) (op, loc, arg);
4340 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4341 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4344 PREFIX(insert_op2) (op, loc, arg1, arg2, end)
4350 register UCHAR_T *pfrom = end;
4351 register UCHAR_T *pto = end + 1 + 2 * OFFSET_ADDRESS_SIZE;
4353 while (pfrom != loc)
4356 PREFIX(store_op2) (op, loc, arg1, arg2);
4360 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4361 after an alternative or a begin-subexpression. We assume there is at
4362 least one character before the ^. */
4365 PREFIX(at_begline_loc_p) (pattern, p, syntax)
4366 const CHAR_T *pattern, *p;
4367 reg_syntax_t syntax;
4369 const CHAR_T *prev = p - 2;
4370 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
4373 /* After a subexpression? */
4374 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
4375 /* After an alternative? */
4376 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
4380 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4381 at least one character after the $, i.e., `P < PEND'. */
4384 PREFIX(at_endline_loc_p) (p, pend, syntax)
4385 const CHAR_T *p, *pend;
4386 reg_syntax_t syntax;
4388 const CHAR_T *next = p;
4389 boolean next_backslash = *next == '\\';
4390 const CHAR_T *next_next = p + 1 < pend ? p + 1 : 0;
4393 /* Before a subexpression? */
4394 (syntax & RE_NO_BK_PARENS ? *next == ')'
4395 : next_backslash && next_next && *next_next == ')')
4396 /* Before an alternative? */
4397 || (syntax & RE_NO_BK_VBAR ? *next == '|'
4398 : next_backslash && next_next && *next_next == '|');
4401 #else /* not INSIDE_RECURSION */
4403 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4404 false if it's not. */
4407 group_in_compile_stack (compile_stack, regnum)
4408 compile_stack_type compile_stack;
4413 for (this_element = compile_stack.avail - 1;
4416 if (compile_stack.stack[this_element].regnum == regnum)
4421 #endif /* not INSIDE_RECURSION */
4423 #ifdef INSIDE_RECURSION
4426 /* This insert space, which size is "num", into the pattern at "loc".
4427 "end" must point the end of the allocated buffer. */
4429 insert_space (num, loc, end)
4434 register CHAR_T *pto = end;
4435 register CHAR_T *pfrom = end - num;
4437 while (pfrom >= loc)
4443 static reg_errcode_t
4444 wcs_compile_range (range_start_char, p_ptr, pend, translate, syntax, b,
4446 CHAR_T range_start_char;
4447 const CHAR_T **p_ptr, *pend;
4448 CHAR_T *char_set, *b;
4449 RE_TRANSLATE_TYPE translate;
4450 reg_syntax_t syntax;
4452 const CHAR_T *p = *p_ptr;
4453 CHAR_T range_start, range_end;
4457 uint32_t start_val, end_val;
4463 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
4466 const char *collseq = (const char *) _NL_CURRENT(LC_COLLATE,
4467 _NL_COLLATE_COLLSEQWC);
4468 const unsigned char *extra = (const unsigned char *)
4469 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
4471 if (range_start_char < -1)
4473 /* range_start is a collating symbol. */
4475 /* Retreive the index and get collation sequence value. */
4476 wextra = (int32_t*)(extra + char_set[-range_start_char]);
4477 start_val = wextra[1 + *wextra];
4480 start_val = collseq_table_lookup(collseq, TRANSLATE(range_start_char));
4482 end_val = collseq_table_lookup (collseq, TRANSLATE (p[0]));
4484 /* Report an error if the range is empty and the syntax prohibits
4486 ret = ((syntax & RE_NO_EMPTY_RANGES)
4487 && (start_val > end_val))? REG_ERANGE : REG_NOERROR;
4489 /* Insert space to the end of the char_ranges. */
4490 insert_space(2, b - char_set[5] - 2, b - 1);
4491 *(b - char_set[5] - 2) = (wchar_t)start_val;
4492 *(b - char_set[5] - 1) = (wchar_t)end_val;
4493 char_set[4]++; /* ranges_index */
4498 range_start = (range_start_char >= 0)? TRANSLATE (range_start_char):
4500 range_end = TRANSLATE (p[0]);
4501 /* Report an error if the range is empty and the syntax prohibits
4503 ret = ((syntax & RE_NO_EMPTY_RANGES)
4504 && (range_start > range_end))? REG_ERANGE : REG_NOERROR;
4506 /* Insert space to the end of the char_ranges. */
4507 insert_space(2, b - char_set[5] - 2, b - 1);
4508 *(b - char_set[5] - 2) = range_start;
4509 *(b - char_set[5] - 1) = range_end;
4510 char_set[4]++; /* ranges_index */
4512 /* Have to increment the pointer into the pattern string, so the
4513 caller isn't still at the ending character. */
4519 /* Read the ending character of a range (in a bracket expression) from the
4520 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4521 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4522 Then we set the translation of all bits between the starting and
4523 ending characters (inclusive) in the compiled pattern B.
4525 Return an error code.
4527 We use these short variable names so we can use the same macros as
4528 `regex_compile' itself. */
4530 static reg_errcode_t
4531 byte_compile_range (range_start_char, p_ptr, pend, translate, syntax, b)
4532 unsigned int range_start_char;
4533 const char **p_ptr, *pend;
4534 RE_TRANSLATE_TYPE translate;
4535 reg_syntax_t syntax;
4539 const char *p = *p_ptr;
4542 const unsigned char *collseq;
4543 unsigned int start_colseq;
4544 unsigned int end_colseq;
4552 /* Have to increment the pointer into the pattern string, so the
4553 caller isn't still at the ending character. */
4556 /* Report an error if the range is empty and the syntax prohibits this. */
4557 ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
4560 collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
4561 _NL_COLLATE_COLLSEQMB);
4563 start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
4564 end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
4565 for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
4567 unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
4569 if (start_colseq <= this_colseq && this_colseq <= end_colseq)
4571 SET_LIST_BIT (TRANSLATE (this_char));
4576 /* Here we see why `this_char' has to be larger than an `unsigned
4577 char' -- we would otherwise go into an infinite loop, since all
4578 characters <= 0xff. */
4579 range_start_char = TRANSLATE (range_start_char);
4580 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4581 and some compilers cast it to int implicitly, so following for_loop
4582 may fall to (almost) infinite loop.
4583 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4584 To avoid this, we cast p[0] to unsigned int and truncate it. */
4585 end_char = ((unsigned)TRANSLATE(p[0]) & ((1 << BYTEWIDTH) - 1));
4587 for (this_char = range_start_char; this_char <= end_char; ++this_char)
4589 SET_LIST_BIT (TRANSLATE (this_char));
4598 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4599 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4600 characters can start a string that matches the pattern. This fastmap
4601 is used by re_search to skip quickly over impossible starting points.
4603 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4604 area as BUFP->fastmap.
4606 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4609 Returns 0 if we succeed, -2 if an internal error. */
4612 /* local function for re_compile_fastmap.
4613 truncate wchar_t character to char. */
4614 static unsigned char truncate_wchar (CHAR_T c);
4616 static unsigned char
4620 unsigned char buf[MB_LEN_MAX];
4621 int retval = wctomb(buf, c);
4622 return retval > 0 ? buf[0] : (unsigned char)c;
4627 PREFIX(re_compile_fastmap) (bufp)
4628 struct re_pattern_buffer *bufp;
4631 #ifdef MATCH_MAY_ALLOCATE
4632 PREFIX(fail_stack_type) fail_stack;
4634 #ifndef REGEX_MALLOC
4638 register char *fastmap = bufp->fastmap;
4641 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4642 pattern to (char*) in regex_compile. */
4643 UCHAR_T *pattern = (UCHAR_T*)bufp->buffer;
4644 register UCHAR_T *pend = (UCHAR_T*) (bufp->buffer + bufp->used);
4646 UCHAR_T *pattern = bufp->buffer;
4647 register UCHAR_T *pend = pattern + bufp->used;
4649 UCHAR_T *p = pattern;
4652 /* This holds the pointer to the failure stack, when
4653 it is allocated relocatably. */
4654 fail_stack_elt_t *failure_stack_ptr;
4657 /* Assume that each path through the pattern can be null until
4658 proven otherwise. We set this false at the bottom of switch
4659 statement, to which we get only if a particular path doesn't
4660 match the empty string. */
4661 boolean path_can_be_null = true;
4663 /* We aren't doing a `succeed_n' to begin with. */
4664 boolean succeed_n_p = false;
4666 assert (fastmap != NULL && p != NULL);
4669 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
4670 bufp->fastmap_accurate = 1; /* It will be when we're done. */
4671 bufp->can_be_null = 0;
4675 if (p == pend || *p == succeed)
4677 /* We have reached the (effective) end of pattern. */
4678 if (!FAIL_STACK_EMPTY ())
4680 bufp->can_be_null |= path_can_be_null;
4682 /* Reset for next path. */
4683 path_can_be_null = true;
4685 p = fail_stack.stack[--fail_stack.avail].pointer;
4693 /* We should never be about to go beyond the end of the pattern. */
4696 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4699 /* I guess the idea here is to simply not bother with a fastmap
4700 if a backreference is used, since it's too hard to figure out
4701 the fastmap for the corresponding group. Setting
4702 `can_be_null' stops `re_search_2' from using the fastmap, so
4703 that is all we do. */
4705 bufp->can_be_null = 1;
4709 /* Following are the cases which match a character. These end
4714 fastmap[truncate_wchar(p[1])] = 1;
4728 /* It is hard to distinguish fastmap from (multi byte) characters
4729 which depends on current locale. */
4734 bufp->can_be_null = 1;
4738 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4739 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
4745 /* Chars beyond end of map must be allowed. */
4746 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
4749 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4750 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
4756 for (j = 0; j < (1 << BYTEWIDTH); j++)
4757 if (SYNTAX (j) == Sword)
4763 for (j = 0; j < (1 << BYTEWIDTH); j++)
4764 if (SYNTAX (j) != Sword)
4771 int fastmap_newline = fastmap['\n'];
4773 /* `.' matches anything ... */
4774 for (j = 0; j < (1 << BYTEWIDTH); j++)
4777 /* ... except perhaps newline. */
4778 if (!(bufp->syntax & RE_DOT_NEWLINE))
4779 fastmap['\n'] = fastmap_newline;
4781 /* Return if we have already set `can_be_null'; if we have,
4782 then the fastmap is irrelevant. Something's wrong here. */
4783 else if (bufp->can_be_null)
4786 /* Otherwise, have to check alternative paths. */
4793 for (j = 0; j < (1 << BYTEWIDTH); j++)
4794 if (SYNTAX (j) == (enum syntaxcode) k)
4801 for (j = 0; j < (1 << BYTEWIDTH); j++)
4802 if (SYNTAX (j) != (enum syntaxcode) k)
4807 /* All cases after this match the empty string. These end with
4827 case push_dummy_failure:
4832 case pop_failure_jump:
4833 case maybe_pop_jump:
4836 case dummy_failure_jump:
4837 EXTRACT_NUMBER_AND_INCR (j, p);
4842 /* Jump backward implies we just went through the body of a
4843 loop and matched nothing. Opcode jumped to should be
4844 `on_failure_jump' or `succeed_n'. Just treat it like an
4845 ordinary jump. For a * loop, it has pushed its failure
4846 point already; if so, discard that as redundant. */
4847 if ((re_opcode_t) *p != on_failure_jump
4848 && (re_opcode_t) *p != succeed_n)
4852 EXTRACT_NUMBER_AND_INCR (j, p);
4855 /* If what's on the stack is where we are now, pop it. */
4856 if (!FAIL_STACK_EMPTY ()
4857 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
4863 case on_failure_jump:
4864 case on_failure_keep_string_jump:
4865 handle_on_failure_jump:
4866 EXTRACT_NUMBER_AND_INCR (j, p);
4868 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4869 end of the pattern. We don't want to push such a point,
4870 since when we restore it above, entering the switch will
4871 increment `p' past the end of the pattern. We don't need
4872 to push such a point since we obviously won't find any more
4873 fastmap entries beyond `pend'. Such a pattern can match
4874 the null string, though. */
4877 if (!PUSH_PATTERN_OP (p + j, fail_stack))
4879 RESET_FAIL_STACK ();
4884 bufp->can_be_null = 1;
4888 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
4889 succeed_n_p = false;
4896 /* Get to the number of times to succeed. */
4897 p += OFFSET_ADDRESS_SIZE;
4899 /* Increment p past the n for when k != 0. */
4900 EXTRACT_NUMBER_AND_INCR (k, p);
4903 p -= 2 * OFFSET_ADDRESS_SIZE;
4904 succeed_n_p = true; /* Spaghetti code alert. */
4905 goto handle_on_failure_jump;
4911 p += 2 * OFFSET_ADDRESS_SIZE;
4922 abort (); /* We have listed all the cases. */
4925 /* Getting here means we have found the possible starting
4926 characters for one path of the pattern -- and that the empty
4927 string does not match. We need not follow this path further.
4928 Instead, look at the next alternative (remembered on the
4929 stack), or quit if no more. The test at the top of the loop
4930 does these things. */
4931 path_can_be_null = false;
4935 /* Set `can_be_null' for the last path (also the first path, if the
4936 pattern is empty). */
4937 bufp->can_be_null |= path_can_be_null;
4940 RESET_FAIL_STACK ();
4944 #else /* not INSIDE_RECURSION */
4947 re_compile_fastmap (bufp)
4948 struct re_pattern_buffer *bufp;
4951 if (MB_CUR_MAX != 1)
4952 return wcs_re_compile_fastmap(bufp);
4955 return byte_re_compile_fastmap(bufp);
4956 } /* re_compile_fastmap */
4958 weak_alias (__re_compile_fastmap, re_compile_fastmap)
4962 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4963 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4964 this memory for recording register information. STARTS and ENDS
4965 must be allocated using the malloc library routine, and must each
4966 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4968 If NUM_REGS == 0, then subsequent matches should allocate their own
4971 Unless this function is called, the first search or match using
4972 PATTERN_BUFFER will allocate its own register data, without
4973 freeing the old data. */
4976 re_set_registers (bufp, regs, num_regs, starts, ends)
4977 struct re_pattern_buffer *bufp;
4978 struct re_registers *regs;
4980 regoff_t *starts, *ends;
4984 bufp->regs_allocated = REGS_REALLOCATE;
4985 regs->num_regs = num_regs;
4986 regs->start = starts;
4991 bufp->regs_allocated = REGS_UNALLOCATED;
4993 regs->start = regs->end = (regoff_t *) 0;
4997 weak_alias (__re_set_registers, re_set_registers)
5000 /* Searching routines. */
5002 /* Like re_search_2, below, but only one string is specified, and
5003 doesn't let you say where to stop matching. */
5006 re_search (bufp, string, size, startpos, range, regs)
5007 struct re_pattern_buffer *bufp;
5009 int size, startpos, range;
5010 struct re_registers *regs;
5012 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
5016 weak_alias (__re_search, re_search)
5020 /* Using the compiled pattern in BUFP->buffer, first tries to match the
5021 virtual concatenation of STRING1 and STRING2, starting first at index
5022 STARTPOS, then at STARTPOS + 1, and so on.
5024 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
5026 RANGE is how far to scan while trying to match. RANGE = 0 means try
5027 only at STARTPOS; in general, the last start tried is STARTPOS +
5030 In REGS, return the indices of the virtual concatenation of STRING1
5031 and STRING2 that matched the entire BUFP->buffer and its contained
5034 Do not consider matching one past the index STOP in the virtual
5035 concatenation of STRING1 and STRING2.
5037 We return either the position in the strings at which the match was
5038 found, -1 if no match, or -2 if error (such as failure
5042 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
5043 struct re_pattern_buffer *bufp;
5044 const char *string1, *string2;
5048 struct re_registers *regs;
5052 if (MB_CUR_MAX != 1)
5053 return wcs_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5057 return byte_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5061 weak_alias (__re_search_2, re_search_2)
5064 #endif /* not INSIDE_RECURSION */
5066 #ifdef INSIDE_RECURSION
5068 #ifdef MATCH_MAY_ALLOCATE
5069 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
5071 # define FREE_VAR(var) if (var) free (var); var = NULL
5075 # define MAX_ALLOCA_SIZE 2000
5077 # define FREE_WCS_BUFFERS() \
5079 if (size1 > MAX_ALLOCA_SIZE) \
5081 free (wcs_string1); \
5082 free (mbs_offset1); \
5086 FREE_VAR (wcs_string1); \
5087 FREE_VAR (mbs_offset1); \
5089 if (size2 > MAX_ALLOCA_SIZE) \
5091 free (wcs_string2); \
5092 free (mbs_offset2); \
5096 FREE_VAR (wcs_string2); \
5097 FREE_VAR (mbs_offset2); \
5105 PREFIX(re_search_2) (bufp, string1, size1, string2, size2, startpos, range,
5107 struct re_pattern_buffer *bufp;
5108 const char *string1, *string2;
5112 struct re_registers *regs;
5116 register char *fastmap = bufp->fastmap;
5117 register RE_TRANSLATE_TYPE translate = bufp->translate;
5118 int total_size = size1 + size2;
5119 int endpos = startpos + range;
5121 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5122 wchar_t *wcs_string1 = NULL, *wcs_string2 = NULL;
5123 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5124 int wcs_size1 = 0, wcs_size2 = 0;
5125 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5126 int *mbs_offset1 = NULL, *mbs_offset2 = NULL;
5127 /* They hold whether each wchar_t is binary data or not. */
5128 char *is_binary = NULL;
5131 /* Check for out-of-range STARTPOS. */
5132 if (startpos < 0 || startpos > total_size)
5135 /* Fix up RANGE if it might eventually take us outside
5136 the virtual concatenation of STRING1 and STRING2.
5137 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5139 range = 0 - startpos;
5140 else if (endpos > total_size)
5141 range = total_size - startpos;
5143 /* If the search isn't to be a backwards one, don't waste time in a
5144 search for a pattern that must be anchored. */
5145 if (bufp->used > 0 && range > 0
5146 && ((re_opcode_t) bufp->buffer[0] == begbuf
5147 /* `begline' is like `begbuf' if it cannot match at newlines. */
5148 || ((re_opcode_t) bufp->buffer[0] == begline
5149 && !bufp->newline_anchor)))
5158 /* In a forward search for something that starts with \=.
5159 don't keep searching past point. */
5160 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
5162 range = PT - startpos;
5168 /* Update the fastmap now if not correct already. */
5169 if (fastmap && !bufp->fastmap_accurate)
5170 if (re_compile_fastmap (bufp) == -2)
5174 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5175 fill them with converted string. */
5178 if (size1 > MAX_ALLOCA_SIZE)
5180 wcs_string1 = TALLOC (size1 + 1, CHAR_T);
5181 mbs_offset1 = TALLOC (size1 + 1, int);
5182 is_binary = TALLOC (size1 + 1, char);
5186 wcs_string1 = REGEX_TALLOC (size1 + 1, CHAR_T);
5187 mbs_offset1 = REGEX_TALLOC (size1 + 1, int);
5188 is_binary = REGEX_TALLOC (size1 + 1, char);
5190 if (!wcs_string1 || !mbs_offset1 || !is_binary)
5192 if (size1 > MAX_ALLOCA_SIZE)
5200 FREE_VAR (wcs_string1);
5201 FREE_VAR (mbs_offset1);
5202 FREE_VAR (is_binary);
5206 wcs_size1 = convert_mbs_to_wcs(wcs_string1, string1, size1,
5207 mbs_offset1, is_binary);
5208 wcs_string1[wcs_size1] = L'\0'; /* for a sentinel */
5209 if (size1 > MAX_ALLOCA_SIZE)
5212 FREE_VAR (is_binary);
5216 if (size2 > MAX_ALLOCA_SIZE)
5218 wcs_string2 = TALLOC (size2 + 1, CHAR_T);
5219 mbs_offset2 = TALLOC (size2 + 1, int);
5220 is_binary = TALLOC (size2 + 1, char);
5224 wcs_string2 = REGEX_TALLOC (size2 + 1, CHAR_T);
5225 mbs_offset2 = REGEX_TALLOC (size2 + 1, int);
5226 is_binary = REGEX_TALLOC (size2 + 1, char);
5228 if (!wcs_string2 || !mbs_offset2 || !is_binary)
5230 FREE_WCS_BUFFERS ();
5231 if (size2 > MAX_ALLOCA_SIZE)
5234 FREE_VAR (is_binary);
5237 wcs_size2 = convert_mbs_to_wcs(wcs_string2, string2, size2,
5238 mbs_offset2, is_binary);
5239 wcs_string2[wcs_size2] = L'\0'; /* for a sentinel */
5240 if (size2 > MAX_ALLOCA_SIZE)
5243 FREE_VAR (is_binary);
5248 /* Loop through the string, looking for a place to start matching. */
5251 /* If a fastmap is supplied, skip quickly over characters that
5252 cannot be the start of a match. If the pattern can match the
5253 null string, however, we don't need to skip characters; we want
5254 the first null string. */
5255 if (fastmap && startpos < total_size && !bufp->can_be_null)
5257 if (range > 0) /* Searching forwards. */
5259 register const char *d;
5260 register int lim = 0;
5263 if (startpos < size1 && startpos + range >= size1)
5264 lim = range - (size1 - startpos);
5266 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
5268 /* Written out as an if-else to avoid testing `translate'
5272 && !fastmap[(unsigned char)
5273 translate[(unsigned char) *d++]])
5276 while (range > lim && !fastmap[(unsigned char) *d++])
5279 startpos += irange - range;
5281 else /* Searching backwards. */
5283 register CHAR_T c = (size1 == 0 || startpos >= size1
5284 ? string2[startpos - size1]
5285 : string1[startpos]);
5287 if (!fastmap[(unsigned char) TRANSLATE (c)])
5292 /* If can't match the null string, and that's all we have left, fail. */
5293 if (range >= 0 && startpos == total_size && fastmap
5294 && !bufp->can_be_null)
5297 FREE_WCS_BUFFERS ();
5303 val = wcs_re_match_2_internal (bufp, string1, size1, string2,
5304 size2, startpos, regs, stop,
5305 wcs_string1, wcs_size1,
5306 wcs_string2, wcs_size2,
5307 mbs_offset1, mbs_offset2);
5309 val = byte_re_match_2_internal (bufp, string1, size1, string2,
5310 size2, startpos, regs, stop);
5313 #ifndef REGEX_MALLOC
5322 FREE_WCS_BUFFERS ();
5330 FREE_WCS_BUFFERS ();
5350 FREE_WCS_BUFFERS ();
5356 /* This converts PTR, a pointer into one of the search wchar_t strings
5357 `string1' and `string2' into an multibyte string offset from the
5358 beginning of that string. We use mbs_offset to optimize.
5359 See convert_mbs_to_wcs. */
5360 # define POINTER_TO_OFFSET(ptr) \
5361 (FIRST_STRING_P (ptr) \
5362 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5363 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5366 /* This converts PTR, a pointer into one of the search strings `string1'
5367 and `string2' into an offset from the beginning of that string. */
5368 # define POINTER_TO_OFFSET(ptr) \
5369 (FIRST_STRING_P (ptr) \
5370 ? ((regoff_t) ((ptr) - string1)) \
5371 : ((regoff_t) ((ptr) - string2 + size1)))
5374 /* Macros for dealing with the split strings in re_match_2. */
5376 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5378 /* Call before fetching a character with *d. This switches over to
5379 string2 if necessary. */
5380 #define PREFETCH() \
5383 /* End of string2 => fail. */ \
5384 if (dend == end_match_2) \
5386 /* End of string1 => advance to string2. */ \
5388 dend = end_match_2; \
5391 /* Test if at very beginning or at very end of the virtual concatenation
5392 of `string1' and `string2'. If only one string, it's `string2'. */
5393 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5394 #define AT_STRINGS_END(d) ((d) == end2)
5397 /* Test if D points to a character which is word-constituent. We have
5398 two special cases to check for: if past the end of string1, look at
5399 the first character in string2; and if before the beginning of
5400 string2, look at the last character in string1. */
5402 /* Use internationalized API instead of SYNTAX. */
5403 # define WORDCHAR_P(d) \
5404 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5405 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5406 || ((d) == end1 ? *string2 \
5407 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) == L'_')
5409 # define WORDCHAR_P(d) \
5410 (SYNTAX ((d) == end1 ? *string2 \
5411 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5415 /* Disabled due to a compiler bug -- see comment at case wordbound */
5417 /* Test if the character before D and the one at D differ with respect
5418 to being word-constituent. */
5419 #define AT_WORD_BOUNDARY(d) \
5420 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5421 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5424 /* Free everything we malloc. */
5425 #ifdef MATCH_MAY_ALLOCATE
5427 # define FREE_VARIABLES() \
5429 REGEX_FREE_STACK (fail_stack.stack); \
5430 FREE_VAR (regstart); \
5431 FREE_VAR (regend); \
5432 FREE_VAR (old_regstart); \
5433 FREE_VAR (old_regend); \
5434 FREE_VAR (best_regstart); \
5435 FREE_VAR (best_regend); \
5436 FREE_VAR (reg_info); \
5437 FREE_VAR (reg_dummy); \
5438 FREE_VAR (reg_info_dummy); \
5439 if (!cant_free_wcs_buf) \
5441 FREE_VAR (string1); \
5442 FREE_VAR (string2); \
5443 FREE_VAR (mbs_offset1); \
5444 FREE_VAR (mbs_offset2); \
5448 # define FREE_VARIABLES() \
5450 REGEX_FREE_STACK (fail_stack.stack); \
5451 FREE_VAR (regstart); \
5452 FREE_VAR (regend); \
5453 FREE_VAR (old_regstart); \
5454 FREE_VAR (old_regend); \
5455 FREE_VAR (best_regstart); \
5456 FREE_VAR (best_regend); \
5457 FREE_VAR (reg_info); \
5458 FREE_VAR (reg_dummy); \
5459 FREE_VAR (reg_info_dummy); \
5464 # define FREE_VARIABLES() \
5466 if (!cant_free_wcs_buf) \
5468 FREE_VAR (string1); \
5469 FREE_VAR (string2); \
5470 FREE_VAR (mbs_offset1); \
5471 FREE_VAR (mbs_offset2); \
5475 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
5477 #endif /* not MATCH_MAY_ALLOCATE */
5479 /* These values must meet several constraints. They must not be valid
5480 register values; since we have a limit of 255 registers (because
5481 we use only one byte in the pattern for the register number), we can
5482 use numbers larger than 255. They must differ by 1, because of
5483 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5484 be larger than the value for the highest register, so we do not try
5485 to actually save any registers when none are active. */
5486 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5487 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5489 #else /* not INSIDE_RECURSION */
5490 /* Matching routines. */
5492 #ifndef emacs /* Emacs never uses this. */
5493 /* re_match is like re_match_2 except it takes only a single string. */
5496 re_match (bufp, string, size, pos, regs)
5497 struct re_pattern_buffer *bufp;
5500 struct re_registers *regs;
5504 if (MB_CUR_MAX != 1)
5505 result = wcs_re_match_2_internal (bufp, NULL, 0, string, size,
5507 NULL, 0, NULL, 0, NULL, NULL);
5510 result = byte_re_match_2_internal (bufp, NULL, 0, string, size,
5512 # ifndef REGEX_MALLOC
5520 weak_alias (__re_match, re_match)
5522 #endif /* not emacs */
5524 #endif /* not INSIDE_RECURSION */
5526 #ifdef INSIDE_RECURSION
5527 static boolean PREFIX(group_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5529 PREFIX(register_info_type) *reg_info));
5530 static boolean PREFIX(alt_match_null_string_p) _RE_ARGS ((UCHAR_T *p,
5532 PREFIX(register_info_type) *reg_info));
5533 static boolean PREFIX(common_op_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5535 PREFIX(register_info_type) *reg_info));
5536 static int PREFIX(bcmp_translate) _RE_ARGS ((const CHAR_T *s1, const CHAR_T *s2,
5537 int len, char *translate));
5538 #else /* not INSIDE_RECURSION */
5540 /* re_match_2 matches the compiled pattern in BUFP against the
5541 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5542 and SIZE2, respectively). We start matching at POS, and stop
5545 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5546 store offsets for the substring each group matched in REGS. See the
5547 documentation for exactly how many groups we fill.
5549 We return -1 if no match, -2 if an internal error (such as the
5550 failure stack overflowing). Otherwise, we return the length of the
5551 matched substring. */
5554 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
5555 struct re_pattern_buffer *bufp;
5556 const char *string1, *string2;
5559 struct re_registers *regs;
5564 if (MB_CUR_MAX != 1)
5565 result = wcs_re_match_2_internal (bufp, string1, size1, string2, size2,
5567 NULL, 0, NULL, 0, NULL, NULL);
5570 result = byte_re_match_2_internal (bufp, string1, size1, string2, size2,
5573 #ifndef REGEX_MALLOC
5581 weak_alias (__re_match_2, re_match_2)
5584 #endif /* not INSIDE_RECURSION */
5586 #ifdef INSIDE_RECURSION
5589 static int count_mbs_length PARAMS ((int *, int));
5591 /* This check the substring (from 0, to length) of the multibyte string,
5592 to which offset_buffer correspond. And count how many wchar_t_characters
5593 the substring occupy. We use offset_buffer to optimization.
5594 See convert_mbs_to_wcs. */
5597 count_mbs_length(offset_buffer, length)
5603 /* Check whether the size is valid. */
5607 if (offset_buffer == NULL)
5610 /* If there are no multibyte character, offset_buffer[i] == i.
5611 Optmize for this case. */
5612 if (offset_buffer[length] == length)
5615 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5621 int middle = (lower + upper) / 2;
5622 if (middle == lower || middle == upper)
5624 if (offset_buffer[middle] > length)
5626 else if (offset_buffer[middle] < length)
5636 /* This is a separate function so that we can force an alloca cleanup
5640 wcs_re_match_2_internal (bufp, cstring1, csize1, cstring2, csize2, pos,
5641 regs, stop, string1, size1, string2, size2,
5642 mbs_offset1, mbs_offset2)
5643 struct re_pattern_buffer *bufp;
5644 const char *cstring1, *cstring2;
5647 struct re_registers *regs;
5649 /* string1 == string2 == NULL means string1/2, size1/2 and
5650 mbs_offset1/2 need seting up in this function. */
5651 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5652 wchar_t *string1, *string2;
5653 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5655 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5656 int *mbs_offset1, *mbs_offset2;
5659 byte_re_match_2_internal (bufp, string1, size1,string2, size2, pos,
5661 struct re_pattern_buffer *bufp;
5662 const char *string1, *string2;
5665 struct re_registers *regs;
5669 /* General temporaries. */
5673 /* They hold whether each wchar_t is binary data or not. */
5674 char *is_binary = NULL;
5675 /* If true, we can't free string1/2, mbs_offset1/2. */
5676 int cant_free_wcs_buf = 1;
5679 /* Just past the end of the corresponding string. */
5680 const CHAR_T *end1, *end2;
5682 /* Pointers into string1 and string2, just past the last characters in
5683 each to consider matching. */
5684 const CHAR_T *end_match_1, *end_match_2;
5686 /* Where we are in the data, and the end of the current string. */
5687 const CHAR_T *d, *dend;
5689 /* Where we are in the pattern, and the end of the pattern. */
5691 UCHAR_T *pattern, *p;
5692 register UCHAR_T *pend;
5694 UCHAR_T *p = bufp->buffer;
5695 register UCHAR_T *pend = p + bufp->used;
5698 /* Mark the opcode just after a start_memory, so we can test for an
5699 empty subpattern when we get to the stop_memory. */
5700 UCHAR_T *just_past_start_mem = 0;
5702 /* We use this to map every character in the string. */
5703 RE_TRANSLATE_TYPE translate = bufp->translate;
5705 /* Failure point stack. Each place that can handle a failure further
5706 down the line pushes a failure point on this stack. It consists of
5707 restart, regend, and reg_info for all registers corresponding to
5708 the subexpressions we're currently inside, plus the number of such
5709 registers, and, finally, two char *'s. The first char * is where
5710 to resume scanning the pattern; the second one is where to resume
5711 scanning the strings. If the latter is zero, the failure point is
5712 a ``dummy''; if a failure happens and the failure point is a dummy,
5713 it gets discarded and the next next one is tried. */
5714 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5715 PREFIX(fail_stack_type) fail_stack;
5718 static unsigned failure_id;
5719 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
5723 /* This holds the pointer to the failure stack, when
5724 it is allocated relocatably. */
5725 fail_stack_elt_t *failure_stack_ptr;
5728 /* We fill all the registers internally, independent of what we
5729 return, for use in backreferences. The number here includes
5730 an element for register zero. */
5731 size_t num_regs = bufp->re_nsub + 1;
5733 /* The currently active registers. */
5734 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
5735 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
5737 /* Information on the contents of registers. These are pointers into
5738 the input strings; they record just what was matched (on this
5739 attempt) by a subexpression part of the pattern, that is, the
5740 regnum-th regstart pointer points to where in the pattern we began
5741 matching and the regnum-th regend points to right after where we
5742 stopped matching the regnum-th subexpression. (The zeroth register
5743 keeps track of what the whole pattern matches.) */
5744 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5745 const CHAR_T **regstart, **regend;
5748 /* If a group that's operated upon by a repetition operator fails to
5749 match anything, then the register for its start will need to be
5750 restored because it will have been set to wherever in the string we
5751 are when we last see its open-group operator. Similarly for a
5753 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5754 const CHAR_T **old_regstart, **old_regend;
5757 /* The is_active field of reg_info helps us keep track of which (possibly
5758 nested) subexpressions we are currently in. The matched_something
5759 field of reg_info[reg_num] helps us tell whether or not we have
5760 matched any of the pattern so far this time through the reg_num-th
5761 subexpression. These two fields get reset each time through any
5762 loop their register is in. */
5763 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5764 PREFIX(register_info_type) *reg_info;
5767 /* The following record the register info as found in the above
5768 variables when we find a match better than any we've seen before.
5769 This happens as we backtrack through the failure points, which in
5770 turn happens only if we have not yet matched the entire string. */
5771 unsigned best_regs_set = false;
5772 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5773 const CHAR_T **best_regstart, **best_regend;
5776 /* Logically, this is `best_regend[0]'. But we don't want to have to
5777 allocate space for that if we're not allocating space for anything
5778 else (see below). Also, we never need info about register 0 for
5779 any of the other register vectors, and it seems rather a kludge to
5780 treat `best_regend' differently than the rest. So we keep track of
5781 the end of the best match so far in a separate variable. We
5782 initialize this to NULL so that when we backtrack the first time
5783 and need to test it, it's not garbage. */
5784 const CHAR_T *match_end = NULL;
5786 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5787 int set_regs_matched_done = 0;
5789 /* Used when we pop values we don't care about. */
5790 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5791 const CHAR_T **reg_dummy;
5792 PREFIX(register_info_type) *reg_info_dummy;
5796 /* Counts the total number of registers pushed. */
5797 unsigned num_regs_pushed = 0;
5800 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
5804 #ifdef MATCH_MAY_ALLOCATE
5805 /* Do not bother to initialize all the register variables if there are
5806 no groups in the pattern, as it takes a fair amount of time. If
5807 there are groups, we include space for register 0 (the whole
5808 pattern), even though we never use it, since it simplifies the
5809 array indexing. We should fix this. */
5812 regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5813 regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5814 old_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5815 old_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5816 best_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5817 best_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5818 reg_info = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5819 reg_dummy = REGEX_TALLOC (num_regs, const CHAR_T *);
5820 reg_info_dummy = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5822 if (!(regstart && regend && old_regstart && old_regend && reg_info
5823 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
5831 /* We must initialize all our variables to NULL, so that
5832 `FREE_VARIABLES' doesn't try to free them. */
5833 regstart = regend = old_regstart = old_regend = best_regstart
5834 = best_regend = reg_dummy = NULL;
5835 reg_info = reg_info_dummy = (PREFIX(register_info_type) *) NULL;
5837 #endif /* MATCH_MAY_ALLOCATE */
5839 /* The starting position is bogus. */
5841 if (pos < 0 || pos > csize1 + csize2)
5843 if (pos < 0 || pos > size1 + size2)
5851 /* Allocate wchar_t array for string1 and string2 and
5852 fill them with converted string. */
5853 if (string1 == NULL && string2 == NULL)
5855 /* We need seting up buffers here. */
5857 /* We must free wcs buffers in this function. */
5858 cant_free_wcs_buf = 0;
5862 string1 = REGEX_TALLOC (csize1 + 1, CHAR_T);
5863 mbs_offset1 = REGEX_TALLOC (csize1 + 1, int);
5864 is_binary = REGEX_TALLOC (csize1 + 1, char);
5865 if (!string1 || !mbs_offset1 || !is_binary)
5868 FREE_VAR (mbs_offset1);
5869 FREE_VAR (is_binary);
5875 string2 = REGEX_TALLOC (csize2 + 1, CHAR_T);
5876 mbs_offset2 = REGEX_TALLOC (csize2 + 1, int);
5877 is_binary = REGEX_TALLOC (csize2 + 1, char);
5878 if (!string2 || !mbs_offset2 || !is_binary)
5881 FREE_VAR (mbs_offset1);
5883 FREE_VAR (mbs_offset2);
5884 FREE_VAR (is_binary);
5887 size2 = convert_mbs_to_wcs(string2, cstring2, csize2,
5888 mbs_offset2, is_binary);
5889 string2[size2] = L'\0'; /* for a sentinel */
5890 FREE_VAR (is_binary);
5894 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5895 pattern to (char*) in regex_compile. */
5896 p = pattern = (CHAR_T*)bufp->buffer;
5897 pend = (CHAR_T*)(bufp->buffer + bufp->used);
5901 /* Initialize subexpression text positions to -1 to mark ones that no
5902 start_memory/stop_memory has been seen for. Also initialize the
5903 register information struct. */
5904 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
5906 regstart[mcnt] = regend[mcnt]
5907 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
5909 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
5910 IS_ACTIVE (reg_info[mcnt]) = 0;
5911 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5912 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5915 /* We move `string1' into `string2' if the latter's empty -- but not if
5916 `string1' is null. */
5917 if (size2 == 0 && string1 != NULL)
5924 mbs_offset2 = mbs_offset1;
5930 end1 = string1 + size1;
5931 end2 = string2 + size2;
5933 /* Compute where to stop matching, within the two strings. */
5937 mcnt = count_mbs_length(mbs_offset1, stop);
5938 end_match_1 = string1 + mcnt;
5939 end_match_2 = string2;
5943 if (stop > csize1 + csize2)
5944 stop = csize1 + csize2;
5946 mcnt = count_mbs_length(mbs_offset2, stop-csize1);
5947 end_match_2 = string2 + mcnt;
5950 { /* count_mbs_length return error. */
5957 end_match_1 = string1 + stop;
5958 end_match_2 = string2;
5963 end_match_2 = string2 + stop - size1;
5967 /* `p' scans through the pattern as `d' scans through the data.
5968 `dend' is the end of the input string that `d' points within. `d'
5969 is advanced into the following input string whenever necessary, but
5970 this happens before fetching; therefore, at the beginning of the
5971 loop, `d' can be pointing at the end of a string, but it cannot
5974 if (size1 > 0 && pos <= csize1)
5976 mcnt = count_mbs_length(mbs_offset1, pos);
5982 mcnt = count_mbs_length(mbs_offset2, pos-csize1);
5988 { /* count_mbs_length return error. */
5993 if (size1 > 0 && pos <= size1)
6000 d = string2 + pos - size1;
6005 DEBUG_PRINT1 ("The compiled pattern is:\n");
6006 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
6007 DEBUG_PRINT1 ("The string to match is: `");
6008 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
6009 DEBUG_PRINT1 ("'\n");
6011 /* This loops over pattern commands. It exits by returning from the
6012 function if the match is complete, or it drops through if the match
6013 fails at this starting point in the input data. */
6017 DEBUG_PRINT2 ("\n%p: ", p);
6019 DEBUG_PRINT2 ("\n0x%x: ", p);
6023 { /* End of pattern means we might have succeeded. */
6024 DEBUG_PRINT1 ("end of pattern ... ");
6026 /* If we haven't matched the entire string, and we want the
6027 longest match, try backtracking. */
6028 if (d != end_match_2)
6030 /* 1 if this match ends in the same string (string1 or string2)
6031 as the best previous match. */
6032 boolean same_str_p = (FIRST_STRING_P (match_end)
6033 == MATCHING_IN_FIRST_STRING);
6034 /* 1 if this match is the best seen so far. */
6035 boolean best_match_p;
6037 /* AIX compiler got confused when this was combined
6038 with the previous declaration. */
6040 best_match_p = d > match_end;
6042 best_match_p = !MATCHING_IN_FIRST_STRING;
6044 DEBUG_PRINT1 ("backtracking.\n");
6046 if (!FAIL_STACK_EMPTY ())
6047 { /* More failure points to try. */
6049 /* If exceeds best match so far, save it. */
6050 if (!best_regs_set || best_match_p)
6052 best_regs_set = true;
6055 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
6057 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6059 best_regstart[mcnt] = regstart[mcnt];
6060 best_regend[mcnt] = regend[mcnt];
6066 /* If no failure points, don't restore garbage. And if
6067 last match is real best match, don't restore second
6069 else if (best_regs_set && !best_match_p)
6072 /* Restore best match. It may happen that `dend ==
6073 end_match_1' while the restored d is in string2.
6074 For example, the pattern `x.*y.*z' against the
6075 strings `x-' and `y-z-', if the two strings are
6076 not consecutive in memory. */
6077 DEBUG_PRINT1 ("Restoring best registers.\n");
6080 dend = ((d >= string1 && d <= end1)
6081 ? end_match_1 : end_match_2);
6083 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6085 regstart[mcnt] = best_regstart[mcnt];
6086 regend[mcnt] = best_regend[mcnt];
6089 } /* d != end_match_2 */
6092 DEBUG_PRINT1 ("Accepting match.\n");
6093 /* If caller wants register contents data back, do it. */
6094 if (regs && !bufp->no_sub)
6096 /* Have the register data arrays been allocated? */
6097 if (bufp->regs_allocated == REGS_UNALLOCATED)
6098 { /* No. So allocate them with malloc. We need one
6099 extra element beyond `num_regs' for the `-1' marker
6101 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
6102 regs->start = TALLOC (regs->num_regs, regoff_t);
6103 regs->end = TALLOC (regs->num_regs, regoff_t);
6104 if (regs->start == NULL || regs->end == NULL)
6109 bufp->regs_allocated = REGS_REALLOCATE;
6111 else if (bufp->regs_allocated == REGS_REALLOCATE)
6112 { /* Yes. If we need more elements than were already
6113 allocated, reallocate them. If we need fewer, just
6115 if (regs->num_regs < num_regs + 1)
6117 regs->num_regs = num_regs + 1;
6118 RETALLOC (regs->start, regs->num_regs, regoff_t);
6119 RETALLOC (regs->end, regs->num_regs, regoff_t);
6120 if (regs->start == NULL || regs->end == NULL)
6129 /* These braces fend off a "empty body in an else-statement"
6130 warning under GCC when assert expands to nothing. */
6131 assert (bufp->regs_allocated == REGS_FIXED);
6134 /* Convert the pointer data in `regstart' and `regend' to
6135 indices. Register zero has to be set differently,
6136 since we haven't kept track of any info for it. */
6137 if (regs->num_regs > 0)
6139 regs->start[0] = pos;
6141 if (MATCHING_IN_FIRST_STRING)
6142 regs->end[0] = mbs_offset1 != NULL ?
6143 mbs_offset1[d-string1] : 0;
6145 regs->end[0] = csize1 + (mbs_offset2 != NULL ?
6146 mbs_offset2[d-string2] : 0);
6148 regs->end[0] = (MATCHING_IN_FIRST_STRING
6149 ? ((regoff_t) (d - string1))
6150 : ((regoff_t) (d - string2 + size1)));
6154 /* Go through the first `min (num_regs, regs->num_regs)'
6155 registers, since that is all we initialized. */
6156 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
6159 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
6160 regs->start[mcnt] = regs->end[mcnt] = -1;
6164 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
6166 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
6170 /* If the regs structure we return has more elements than
6171 were in the pattern, set the extra elements to -1. If
6172 we (re)allocated the registers, this is the case,
6173 because we always allocate enough to have at least one
6175 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
6176 regs->start[mcnt] = regs->end[mcnt] = -1;
6177 } /* regs && !bufp->no_sub */
6179 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
6180 nfailure_points_pushed, nfailure_points_popped,
6181 nfailure_points_pushed - nfailure_points_popped);
6182 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
6185 if (MATCHING_IN_FIRST_STRING)
6186 mcnt = mbs_offset1 != NULL ? mbs_offset1[d-string1] : 0;
6188 mcnt = (mbs_offset2 != NULL ? mbs_offset2[d-string2] : 0) +
6192 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
6197 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
6203 /* Otherwise match next pattern command. */
6204 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
6206 /* Ignore these. Used to ignore the n of succeed_n's which
6207 currently have n == 0. */
6209 DEBUG_PRINT1 ("EXECUTING no_op.\n");
6213 DEBUG_PRINT1 ("EXECUTING succeed.\n");
6216 /* Match the next n pattern characters exactly. The following
6217 byte in the pattern defines n, and the n bytes after that
6218 are the characters to match. */
6224 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
6226 /* This is written out as an if-else so we don't waste time
6227 testing `translate' inside the loop. */
6236 if ((UCHAR_T) translate[(unsigned char) *d++]
6242 if (*d++ != (CHAR_T) *p++)
6246 if ((UCHAR_T) translate[(unsigned char) *d++]
6258 if (*d++ != (CHAR_T) *p++) goto fail;
6262 SET_REGS_MATCHED ();
6266 /* Match any character except possibly a newline or a null. */
6268 DEBUG_PRINT1 ("EXECUTING anychar.\n");
6272 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
6273 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
6276 SET_REGS_MATCHED ();
6277 DEBUG_PRINT2 (" Matched `%ld'.\n", (long int) *d);
6287 unsigned int i, char_class_length, coll_symbol_length,
6288 equiv_class_length, ranges_length, chars_length, length;
6289 CHAR_T *workp, *workp2, *charset_top;
6290 #define WORK_BUFFER_SIZE 128
6291 CHAR_T str_buf[WORK_BUFFER_SIZE];
6296 boolean not = (re_opcode_t) *(p - 1) == charset_not;
6298 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
6300 c = TRANSLATE (*d); /* The character to match. */
6303 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
6305 charset_top = p - 1;
6306 char_class_length = *p++;
6307 coll_symbol_length = *p++;
6308 equiv_class_length = *p++;
6309 ranges_length = *p++;
6310 chars_length = *p++;
6311 /* p points charset[6], so the address of the next instruction
6312 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6313 where l=length of char_classes, m=length of collating_symbol,
6314 n=equivalence_class, o=length of char_range,
6315 p'=length of character. */
6317 /* Update p to indicate the next instruction. */
6318 p += char_class_length + coll_symbol_length+ equiv_class_length +
6319 2*ranges_length + chars_length;
6321 /* match with char_class? */
6322 for (i = 0; i < char_class_length ; i += CHAR_CLASS_SIZE)
6325 uintptr_t alignedp = ((uintptr_t)workp
6326 + __alignof__(wctype_t) - 1)
6327 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
6328 wctype = *((wctype_t*)alignedp);
6329 workp += CHAR_CLASS_SIZE;
6330 if (iswctype((wint_t)c, wctype))
6331 goto char_set_matched;
6334 /* match with collating_symbol? */
6338 const unsigned char *extra = (const unsigned char *)
6339 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
6341 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;
6345 wextra = (int32_t*)(extra + *workp++);
6346 for (i = 0; i < *wextra; ++i)
6347 if (TRANSLATE(d[i]) != wextra[1 + i])
6352 /* Update d, however d will be incremented at
6353 char_set_matched:, we decrement d here. */
6355 goto char_set_matched;
6359 else /* (nrules == 0) */
6361 /* If we can't look up collation data, we use wcscoll
6364 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;)
6366 const CHAR_T *backup_d = d, *backup_dend = dend;
6367 length = wcslen(workp);
6369 /* If wcscoll(the collating symbol, whole string) > 0,
6370 any substring of the string never match with the
6371 collating symbol. */
6372 if (wcscoll(workp, d) > 0)
6374 workp += length + 1;
6378 /* First, we compare the collating symbol with
6379 the first character of the string.
6380 If it don't match, we add the next character to
6381 the compare buffer in turn. */
6382 for (i = 0 ; i < WORK_BUFFER_SIZE-1 ; i++, d++)
6387 if (dend == end_match_2)
6393 /* add next character to the compare buffer. */
6394 str_buf[i] = TRANSLATE(*d);
6395 str_buf[i+1] = '\0';
6397 match = wcscoll(workp, str_buf);
6399 goto char_set_matched;
6402 /* (str_buf > workp) indicate (str_buf + X > workp),
6403 because for all X (str_buf + X > str_buf).
6404 So we don't need continue this loop. */
6407 /* Otherwise(str_buf < workp),
6408 (str_buf+next_character) may equals (workp).
6409 So we continue this loop. */
6414 workp += length + 1;
6417 /* match with equivalence_class? */
6421 const CHAR_T *backup_d = d, *backup_dend = dend;
6422 /* Try to match the equivalence class against
6423 those known to the collate implementation. */
6424 const int32_t *table;
6425 const int32_t *weights;
6426 const int32_t *extra;
6427 const int32_t *indirect;
6432 /* This #include defines a local function! */
6433 # include <locale/weightwc.h>
6435 table = (const int32_t *)
6436 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEWC);
6437 weights = (const wint_t *)
6438 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTWC);
6439 extra = (const wint_t *)
6440 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAWC);
6441 indirect = (const int32_t *)
6442 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTWC);
6444 /* Write 1 collating element to str_buf, and
6448 for (i = 0 ; idx2 == 0 && i < WORK_BUFFER_SIZE - 1; i++)
6450 cp = (wint_t*)str_buf;
6453 if (dend == end_match_2)
6458 str_buf[i] = TRANSLATE(*(d+i));
6459 str_buf[i+1] = '\0'; /* sentinel */
6460 idx2 = findidx ((const wint_t**)&cp);
6463 /* Update d, however d will be incremented at
6464 char_set_matched:, we decrement d here. */
6465 d = backup_d + ((wchar_t*)cp - (wchar_t*)str_buf - 1);
6468 if (dend == end_match_2)
6477 len = weights[idx2];
6479 for (workp2 = workp + equiv_class_length ; workp < workp2 ;
6482 idx = (int32_t)*workp;
6483 /* We already checked idx != 0 in regex_compile. */
6485 if (idx2 != 0 && len == weights[idx])
6488 while (cnt < len && (weights[idx + 1 + cnt]
6489 == weights[idx2 + 1 + cnt]))
6493 goto char_set_matched;
6500 else /* (nrules == 0) */
6502 /* If we can't look up collation data, we use wcscoll
6505 for (workp2 = workp + equiv_class_length ; workp < workp2 ;)
6507 const CHAR_T *backup_d = d, *backup_dend = dend;
6508 length = wcslen(workp);
6510 /* If wcscoll(the collating symbol, whole string) > 0,
6511 any substring of the string never match with the
6512 collating symbol. */
6513 if (wcscoll(workp, d) > 0)
6515 workp += length + 1;
6519 /* First, we compare the equivalence class with
6520 the first character of the string.
6521 If it don't match, we add the next character to
6522 the compare buffer in turn. */
6523 for (i = 0 ; i < WORK_BUFFER_SIZE - 1 ; i++, d++)
6528 if (dend == end_match_2)
6534 /* add next character to the compare buffer. */
6535 str_buf[i] = TRANSLATE(*d);
6536 str_buf[i+1] = '\0';
6538 match = wcscoll(workp, str_buf);
6541 goto char_set_matched;
6544 /* (str_buf > workp) indicate (str_buf + X > workp),
6545 because for all X (str_buf + X > str_buf).
6546 So we don't need continue this loop. */
6549 /* Otherwise(str_buf < workp),
6550 (str_buf+next_character) may equals (workp).
6551 So we continue this loop. */
6556 workp += length + 1;
6560 /* match with char_range? */
6564 uint32_t collseqval;
6565 const char *collseq = (const char *)
6566 _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQWC);
6568 collseqval = collseq_table_lookup (collseq, c);
6570 for (; workp < p - chars_length ;)
6572 uint32_t start_val, end_val;
6574 /* We already compute the collation sequence value
6575 of the characters (or collating symbols). */
6576 start_val = (uint32_t) *workp++; /* range_start */
6577 end_val = (uint32_t) *workp++; /* range_end */
6579 if (start_val <= collseqval && collseqval <= end_val)
6580 goto char_set_matched;
6586 /* We set range_start_char at str_buf[0], range_end_char
6587 at str_buf[4], and compared char at str_buf[2]. */
6592 for (; workp < p - chars_length ;)
6594 wchar_t *range_start_char, *range_end_char;
6596 /* match if (range_start_char <= c <= range_end_char). */
6598 /* If range_start(or end) < 0, we assume -range_start(end)
6599 is the offset of the collating symbol which is specified
6600 as the character of the range start(end). */
6604 range_start_char = charset_top - (*workp++);
6607 str_buf[0] = *workp++;
6608 range_start_char = str_buf;
6613 range_end_char = charset_top - (*workp++);
6616 str_buf[4] = *workp++;
6617 range_end_char = str_buf + 4;
6620 if (wcscoll(range_start_char, str_buf+2) <= 0 &&
6621 wcscoll(str_buf+2, range_end_char) <= 0)
6623 goto char_set_matched;
6627 /* match with char? */
6628 for (; workp < p ; workp++)
6630 goto char_set_matched;
6637 /* Cast to `unsigned' instead of `unsigned char' in case the
6638 bit list is a full 32 bytes long. */
6639 if (c < (unsigned) (*p * BYTEWIDTH)
6640 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
6645 if (!not) goto fail;
6646 #undef WORK_BUFFER_SIZE
6648 SET_REGS_MATCHED ();
6654 /* The beginning of a group is represented by start_memory.
6655 The arguments are the register number in the next byte, and the
6656 number of groups inner to this one in the next. The text
6657 matched within the group is recorded (in the internal
6658 registers data structure) under the register number. */
6660 DEBUG_PRINT3 ("EXECUTING start_memory %ld (%ld):\n",
6661 (long int) *p, (long int) p[1]);
6663 /* Find out if this group can match the empty string. */
6664 p1 = p; /* To send to group_match_null_string_p. */
6666 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
6667 REG_MATCH_NULL_STRING_P (reg_info[*p])
6668 = PREFIX(group_match_null_string_p) (&p1, pend, reg_info);
6670 /* Save the position in the string where we were the last time
6671 we were at this open-group operator in case the group is
6672 operated upon by a repetition operator, e.g., with `(a*)*b'
6673 against `ab'; then we want to ignore where we are now in
6674 the string in case this attempt to match fails. */
6675 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6676 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
6678 DEBUG_PRINT2 (" old_regstart: %d\n",
6679 POINTER_TO_OFFSET (old_regstart[*p]));
6682 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
6684 IS_ACTIVE (reg_info[*p]) = 1;
6685 MATCHED_SOMETHING (reg_info[*p]) = 0;
6687 /* Clear this whenever we change the register activity status. */
6688 set_regs_matched_done = 0;
6690 /* This is the new highest active register. */
6691 highest_active_reg = *p;
6693 /* If nothing was active before, this is the new lowest active
6695 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
6696 lowest_active_reg = *p;
6698 /* Move past the register number and inner group count. */
6700 just_past_start_mem = p;
6705 /* The stop_memory opcode represents the end of a group. Its
6706 arguments are the same as start_memory's: the register
6707 number, and the number of inner groups. */
6709 DEBUG_PRINT3 ("EXECUTING stop_memory %ld (%ld):\n",
6710 (long int) *p, (long int) p[1]);
6712 /* We need to save the string position the last time we were at
6713 this close-group operator in case the group is operated
6714 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6715 against `aba'; then we want to ignore where we are now in
6716 the string in case this attempt to match fails. */
6717 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6718 ? REG_UNSET (regend[*p]) ? d : regend[*p]
6720 DEBUG_PRINT2 (" old_regend: %d\n",
6721 POINTER_TO_OFFSET (old_regend[*p]));
6724 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
6726 /* This register isn't active anymore. */
6727 IS_ACTIVE (reg_info[*p]) = 0;
6729 /* Clear this whenever we change the register activity status. */
6730 set_regs_matched_done = 0;
6732 /* If this was the only register active, nothing is active
6734 if (lowest_active_reg == highest_active_reg)
6736 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6737 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6740 { /* We must scan for the new highest active register, since
6741 it isn't necessarily one less than now: consider
6742 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6743 new highest active register is 1. */
6745 while (r > 0 && !IS_ACTIVE (reg_info[r]))
6748 /* If we end up at register zero, that means that we saved
6749 the registers as the result of an `on_failure_jump', not
6750 a `start_memory', and we jumped to past the innermost
6751 `stop_memory'. For example, in ((.)*) we save
6752 registers 1 and 2 as a result of the *, but when we pop
6753 back to the second ), we are at the stop_memory 1.
6754 Thus, nothing is active. */
6757 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6758 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6761 highest_active_reg = r;
6764 /* If just failed to match something this time around with a
6765 group that's operated on by a repetition operator, try to
6766 force exit from the ``loop'', and restore the register
6767 information for this group that we had before trying this
6769 if ((!MATCHED_SOMETHING (reg_info[*p])
6770 || just_past_start_mem == p - 1)
6773 boolean is_a_jump_n = false;
6777 switch ((re_opcode_t) *p1++)
6781 case pop_failure_jump:
6782 case maybe_pop_jump:
6784 case dummy_failure_jump:
6785 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6787 p1 += OFFSET_ADDRESS_SIZE;
6795 /* If the next operation is a jump backwards in the pattern
6796 to an on_failure_jump right before the start_memory
6797 corresponding to this stop_memory, exit from the loop
6798 by forcing a failure after pushing on the stack the
6799 on_failure_jump's jump in the pattern, and d. */
6800 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
6801 && (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == start_memory
6802 && p1[2+OFFSET_ADDRESS_SIZE] == *p)
6804 /* If this group ever matched anything, then restore
6805 what its registers were before trying this last
6806 failed match, e.g., with `(a*)*b' against `ab' for
6807 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6808 against `aba' for regend[3].
6810 Also restore the registers for inner groups for,
6811 e.g., `((a*)(b*))*' against `aba' (register 3 would
6812 otherwise get trashed). */
6814 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
6818 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
6820 /* Restore this and inner groups' (if any) registers. */
6821 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
6824 regstart[r] = old_regstart[r];
6826 /* xx why this test? */
6827 if (old_regend[r] >= regstart[r])
6828 regend[r] = old_regend[r];
6832 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6833 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
6839 /* Move past the register number and the inner group count. */
6844 /* \<digit> has been turned into a `duplicate' command which is
6845 followed by the numeric value of <digit> as the register number. */
6848 register const CHAR_T *d2, *dend2;
6849 int regno = *p++; /* Get which register to match against. */
6850 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
6852 /* Can't back reference a group which we've never matched. */
6853 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
6856 /* Where in input to try to start matching. */
6857 d2 = regstart[regno];
6859 /* Where to stop matching; if both the place to start and
6860 the place to stop matching are in the same string, then
6861 set to the place to stop, otherwise, for now have to use
6862 the end of the first string. */
6864 dend2 = ((FIRST_STRING_P (regstart[regno])
6865 == FIRST_STRING_P (regend[regno]))
6866 ? regend[regno] : end_match_1);
6869 /* If necessary, advance to next segment in register
6873 if (dend2 == end_match_2) break;
6874 if (dend2 == regend[regno]) break;
6876 /* End of string1 => advance to string2. */
6878 dend2 = regend[regno];
6880 /* At end of register contents => success */
6881 if (d2 == dend2) break;
6883 /* If necessary, advance to next segment in data. */
6886 /* How many characters left in this segment to match. */
6889 /* Want how many consecutive characters we can match in
6890 one shot, so, if necessary, adjust the count. */
6891 if (mcnt > dend2 - d2)
6894 /* Compare that many; failure if mismatch, else move
6897 ? PREFIX(bcmp_translate) (d, d2, mcnt, translate)
6898 : memcmp (d, d2, mcnt*sizeof(UCHAR_T)))
6900 d += mcnt, d2 += mcnt;
6902 /* Do this because we've match some characters. */
6903 SET_REGS_MATCHED ();
6909 /* begline matches the empty string at the beginning of the string
6910 (unless `not_bol' is set in `bufp'), and, if
6911 `newline_anchor' is set, after newlines. */
6913 DEBUG_PRINT1 ("EXECUTING begline.\n");
6915 if (AT_STRINGS_BEG (d))
6917 if (!bufp->not_bol) break;
6919 else if (d[-1] == '\n' && bufp->newline_anchor)
6923 /* In all other cases, we fail. */
6927 /* endline is the dual of begline. */
6929 DEBUG_PRINT1 ("EXECUTING endline.\n");
6931 if (AT_STRINGS_END (d))
6933 if (!bufp->not_eol) break;
6936 /* We have to ``prefetch'' the next character. */
6937 else if ((d == end1 ? *string2 : *d) == '\n'
6938 && bufp->newline_anchor)
6945 /* Match at the very beginning of the data. */
6947 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
6948 if (AT_STRINGS_BEG (d))
6953 /* Match at the very end of the data. */
6955 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
6956 if (AT_STRINGS_END (d))
6961 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6962 pushes NULL as the value for the string on the stack. Then
6963 `pop_failure_point' will keep the current value for the
6964 string, instead of restoring it. To see why, consider
6965 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6966 then the . fails against the \n. But the next thing we want
6967 to do is match the \n against the \n; if we restored the
6968 string value, we would be back at the foo.
6970 Because this is used only in specific cases, we don't need to
6971 check all the things that `on_failure_jump' does, to make
6972 sure the right things get saved on the stack. Hence we don't
6973 share its code. The only reason to push anything on the
6974 stack at all is that otherwise we would have to change
6975 `anychar's code to do something besides goto fail in this
6976 case; that seems worse than this. */
6977 case on_failure_keep_string_jump:
6978 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
6980 EXTRACT_NUMBER_AND_INCR (mcnt, p);
6982 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
6984 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
6987 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
6991 /* Uses of on_failure_jump:
6993 Each alternative starts with an on_failure_jump that points
6994 to the beginning of the next alternative. Each alternative
6995 except the last ends with a jump that in effect jumps past
6996 the rest of the alternatives. (They really jump to the
6997 ending jump of the following alternative, because tensioning
6998 these jumps is a hassle.)
7000 Repeats start with an on_failure_jump that points past both
7001 the repetition text and either the following jump or
7002 pop_failure_jump back to this on_failure_jump. */
7003 case on_failure_jump:
7005 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
7007 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7009 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
7011 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
7014 /* If this on_failure_jump comes right before a group (i.e.,
7015 the original * applied to a group), save the information
7016 for that group and all inner ones, so that if we fail back
7017 to this point, the group's information will be correct.
7018 For example, in \(a*\)*\1, we need the preceding group,
7019 and in \(zz\(a*\)b*\)\2, we need the inner group. */
7021 /* We can't use `p' to check ahead because we push
7022 a failure point to `p + mcnt' after we do this. */
7025 /* We need to skip no_op's before we look for the
7026 start_memory in case this on_failure_jump is happening as
7027 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
7029 while (p1 < pend && (re_opcode_t) *p1 == no_op)
7032 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
7034 /* We have a new highest active register now. This will
7035 get reset at the start_memory we are about to get to,
7036 but we will have saved all the registers relevant to
7037 this repetition op, as described above. */
7038 highest_active_reg = *(p1 + 1) + *(p1 + 2);
7039 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
7040 lowest_active_reg = *(p1 + 1);
7043 DEBUG_PRINT1 (":\n");
7044 PUSH_FAILURE_POINT (p + mcnt, d, -2);
7048 /* A smart repeat ends with `maybe_pop_jump'.
7049 We change it to either `pop_failure_jump' or `jump'. */
7050 case maybe_pop_jump:
7051 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7052 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
7054 register UCHAR_T *p2 = p;
7056 /* Compare the beginning of the repeat with what in the
7057 pattern follows its end. If we can establish that there
7058 is nothing that they would both match, i.e., that we
7059 would have to backtrack because of (as in, e.g., `a*a')
7060 then we can change to pop_failure_jump, because we'll
7061 never have to backtrack.
7063 This is not true in the case of alternatives: in
7064 `(a|ab)*' we do need to backtrack to the `ab' alternative
7065 (e.g., if the string was `ab'). But instead of trying to
7066 detect that here, the alternative has put on a dummy
7067 failure point which is what we will end up popping. */
7069 /* Skip over open/close-group commands.
7070 If what follows this loop is a ...+ construct,
7071 look at what begins its body, since we will have to
7072 match at least one of that. */
7076 && ((re_opcode_t) *p2 == stop_memory
7077 || (re_opcode_t) *p2 == start_memory))
7079 else if (p2 + 2 + 2 * OFFSET_ADDRESS_SIZE < pend
7080 && (re_opcode_t) *p2 == dummy_failure_jump)
7081 p2 += 2 + 2 * OFFSET_ADDRESS_SIZE;
7087 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7088 to the `maybe_finalize_jump' of this case. Examine what
7091 /* If we're at the end of the pattern, we can change. */
7094 /* Consider what happens when matching ":\(.*\)"
7095 against ":/". I don't really understand this code
7097 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7100 (" End of pattern: change to `pop_failure_jump'.\n");
7103 else if ((re_opcode_t) *p2 == exactn
7105 || (re_opcode_t) *p2 == exactn_bin
7107 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
7110 = *p2 == (UCHAR_T) endline ? '\n' : p2[2];
7112 if (((re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn
7114 || (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn_bin
7116 ) && p1[3+OFFSET_ADDRESS_SIZE] != c)
7118 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7121 DEBUG_PRINT3 (" %C != %C => pop_failure_jump.\n",
7123 (wint_t) p1[3+OFFSET_ADDRESS_SIZE]);
7125 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
7127 (char) p1[3+OFFSET_ADDRESS_SIZE]);
7132 else if ((re_opcode_t) p1[3] == charset
7133 || (re_opcode_t) p1[3] == charset_not)
7135 int not = (re_opcode_t) p1[3] == charset_not;
7137 if (c < (unsigned) (p1[4] * BYTEWIDTH)
7138 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
7141 /* `not' is equal to 1 if c would match, which means
7142 that we can't change to pop_failure_jump. */
7145 p[-3] = (unsigned char) pop_failure_jump;
7146 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7149 #endif /* not WCHAR */
7152 else if ((re_opcode_t) *p2 == charset)
7154 /* We win if the first character of the loop is not part
7156 if ((re_opcode_t) p1[3] == exactn
7157 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
7158 && (p2[2 + p1[5] / BYTEWIDTH]
7159 & (1 << (p1[5] % BYTEWIDTH)))))
7161 p[-3] = (unsigned char) pop_failure_jump;
7162 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7165 else if ((re_opcode_t) p1[3] == charset_not)
7168 /* We win if the charset_not inside the loop
7169 lists every character listed in the charset after. */
7170 for (idx = 0; idx < (int) p2[1]; idx++)
7171 if (! (p2[2 + idx] == 0
7172 || (idx < (int) p1[4]
7173 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
7178 p[-3] = (unsigned char) pop_failure_jump;
7179 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7182 else if ((re_opcode_t) p1[3] == charset)
7185 /* We win if the charset inside the loop
7186 has no overlap with the one after the loop. */
7188 idx < (int) p2[1] && idx < (int) p1[4];
7190 if ((p2[2 + idx] & p1[5 + idx]) != 0)
7193 if (idx == p2[1] || idx == p1[4])
7195 p[-3] = (unsigned char) pop_failure_jump;
7196 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7200 #endif /* not WCHAR */
7202 p -= OFFSET_ADDRESS_SIZE; /* Point at relative address again. */
7203 if ((re_opcode_t) p[-1] != pop_failure_jump)
7205 p[-1] = (UCHAR_T) jump;
7206 DEBUG_PRINT1 (" Match => jump.\n");
7207 goto unconditional_jump;
7209 /* Note fall through. */
7212 /* The end of a simple repeat has a pop_failure_jump back to
7213 its matching on_failure_jump, where the latter will push a
7214 failure point. The pop_failure_jump takes off failure
7215 points put on by this pop_failure_jump's matching
7216 on_failure_jump; we got through the pattern to here from the
7217 matching on_failure_jump, so didn't fail. */
7218 case pop_failure_jump:
7220 /* We need to pass separate storage for the lowest and
7221 highest registers, even though we don't care about the
7222 actual values. Otherwise, we will restore only one
7223 register from the stack, since lowest will == highest in
7224 `pop_failure_point'. */
7225 active_reg_t dummy_low_reg, dummy_high_reg;
7226 UCHAR_T *pdummy = NULL;
7227 const CHAR_T *sdummy = NULL;
7229 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
7230 POP_FAILURE_POINT (sdummy, pdummy,
7231 dummy_low_reg, dummy_high_reg,
7232 reg_dummy, reg_dummy, reg_info_dummy);
7234 /* Note fall through. */
7238 DEBUG_PRINT2 ("\n%p: ", p);
7240 DEBUG_PRINT2 ("\n0x%x: ", p);
7242 /* Note fall through. */
7244 /* Unconditionally jump (without popping any failure points). */
7246 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
7247 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
7248 p += mcnt; /* Do the jump. */
7250 DEBUG_PRINT2 ("(to %p).\n", p);
7252 DEBUG_PRINT2 ("(to 0x%x).\n", p);
7257 /* We need this opcode so we can detect where alternatives end
7258 in `group_match_null_string_p' et al. */
7260 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
7261 goto unconditional_jump;
7264 /* Normally, the on_failure_jump pushes a failure point, which
7265 then gets popped at pop_failure_jump. We will end up at
7266 pop_failure_jump, also, and with a pattern of, say, `a+', we
7267 are skipping over the on_failure_jump, so we have to push
7268 something meaningless for pop_failure_jump to pop. */
7269 case dummy_failure_jump:
7270 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
7271 /* It doesn't matter what we push for the string here. What
7272 the code at `fail' tests is the value for the pattern. */
7273 PUSH_FAILURE_POINT (NULL, NULL, -2);
7274 goto unconditional_jump;
7277 /* At the end of an alternative, we need to push a dummy failure
7278 point in case we are followed by a `pop_failure_jump', because
7279 we don't want the failure point for the alternative to be
7280 popped. For example, matching `(a|ab)*' against `aab'
7281 requires that we match the `ab' alternative. */
7282 case push_dummy_failure:
7283 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
7284 /* See comments just above at `dummy_failure_jump' about the
7286 PUSH_FAILURE_POINT (NULL, NULL, -2);
7289 /* Have to succeed matching what follows at least n times.
7290 After that, handle like `on_failure_jump'. */
7292 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7293 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
7296 /* Originally, this is how many times we HAVE to succeed. */
7300 p += OFFSET_ADDRESS_SIZE;
7301 STORE_NUMBER_AND_INCR (p, mcnt);
7303 DEBUG_PRINT3 (" Setting %p to %d.\n", p - OFFSET_ADDRESS_SIZE
7306 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - OFFSET_ADDRESS_SIZE
7313 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n",
7314 p + OFFSET_ADDRESS_SIZE);
7316 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n",
7317 p + OFFSET_ADDRESS_SIZE);
7321 p[1] = (UCHAR_T) no_op;
7323 p[2] = (UCHAR_T) no_op;
7324 p[3] = (UCHAR_T) no_op;
7331 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7332 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
7334 /* Originally, this is how many times we CAN jump. */
7338 STORE_NUMBER (p + OFFSET_ADDRESS_SIZE, mcnt);
7341 DEBUG_PRINT3 (" Setting %p to %d.\n", p + OFFSET_ADDRESS_SIZE,
7344 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + OFFSET_ADDRESS_SIZE,
7347 goto unconditional_jump;
7349 /* If don't have to jump any more, skip over the rest of command. */
7351 p += 2 * OFFSET_ADDRESS_SIZE;
7356 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
7358 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7360 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7362 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
7364 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
7366 STORE_NUMBER (p1, mcnt);
7371 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7372 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7373 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7374 macro and introducing temporary variables works around the bug. */
7377 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7378 if (AT_WORD_BOUNDARY (d))
7383 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7384 if (AT_WORD_BOUNDARY (d))
7390 boolean prevchar, thischar;
7392 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7393 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7396 prevchar = WORDCHAR_P (d - 1);
7397 thischar = WORDCHAR_P (d);
7398 if (prevchar != thischar)
7405 boolean prevchar, thischar;
7407 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7408 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7411 prevchar = WORDCHAR_P (d - 1);
7412 thischar = WORDCHAR_P (d);
7413 if (prevchar != thischar)
7420 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
7421 if (!AT_STRINGS_END (d) && WORDCHAR_P (d)
7422 && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
7427 DEBUG_PRINT1 ("EXECUTING wordend.\n");
7428 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
7429 && (AT_STRINGS_END (d) || !WORDCHAR_P (d)))
7435 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
7436 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
7441 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
7442 if (PTR_CHAR_POS ((unsigned char *) d) != point)
7447 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
7448 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
7453 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
7458 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
7462 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7464 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
7466 SET_REGS_MATCHED ();
7470 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
7472 goto matchnotsyntax;
7475 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
7479 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7481 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
7483 SET_REGS_MATCHED ();
7486 #else /* not emacs */
7488 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
7490 if (!WORDCHAR_P (d))
7492 SET_REGS_MATCHED ();
7497 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
7501 SET_REGS_MATCHED ();
7504 #endif /* not emacs */
7509 continue; /* Successfully executed one pattern command; keep going. */
7512 /* We goto here if a matching operation fails. */
7514 if (!FAIL_STACK_EMPTY ())
7515 { /* A restart point is known. Restore to that state. */
7516 DEBUG_PRINT1 ("\nFAIL:\n");
7517 POP_FAILURE_POINT (d, p,
7518 lowest_active_reg, highest_active_reg,
7519 regstart, regend, reg_info);
7521 /* If this failure point is a dummy, try the next one. */
7525 /* If we failed to the end of the pattern, don't examine *p. */
7529 boolean is_a_jump_n = false;
7531 /* If failed to a backwards jump that's part of a repetition
7532 loop, need to pop this failure point and use the next one. */
7533 switch ((re_opcode_t) *p)
7537 case maybe_pop_jump:
7538 case pop_failure_jump:
7541 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7544 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
7546 && (re_opcode_t) *p1 == on_failure_jump))
7554 if (d >= string1 && d <= end1)
7558 break; /* Matching at this starting point really fails. */
7562 goto restore_best_regs;
7566 return -1; /* Failure to match. */
7569 /* Subroutine definitions for re_match_2. */
7572 /* We are passed P pointing to a register number after a start_memory.
7574 Return true if the pattern up to the corresponding stop_memory can
7575 match the empty string, and false otherwise.
7577 If we find the matching stop_memory, sets P to point to one past its number.
7578 Otherwise, sets P to an undefined byte less than or equal to END.
7580 We don't handle duplicates properly (yet). */
7583 PREFIX(group_match_null_string_p) (p, end, reg_info)
7585 PREFIX(register_info_type) *reg_info;
7588 /* Point to after the args to the start_memory. */
7589 UCHAR_T *p1 = *p + 2;
7593 /* Skip over opcodes that can match nothing, and return true or
7594 false, as appropriate, when we get to one that can't, or to the
7595 matching stop_memory. */
7597 switch ((re_opcode_t) *p1)
7599 /* Could be either a loop or a series of alternatives. */
7600 case on_failure_jump:
7602 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7604 /* If the next operation is not a jump backwards in the
7609 /* Go through the on_failure_jumps of the alternatives,
7610 seeing if any of the alternatives cannot match nothing.
7611 The last alternative starts with only a jump,
7612 whereas the rest start with on_failure_jump and end
7613 with a jump, e.g., here is the pattern for `a|b|c':
7615 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7616 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7619 So, we have to first go through the first (n-1)
7620 alternatives and then deal with the last one separately. */
7623 /* Deal with the first (n-1) alternatives, which start
7624 with an on_failure_jump (see above) that jumps to right
7625 past a jump_past_alt. */
7627 while ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] ==
7630 /* `mcnt' holds how many bytes long the alternative
7631 is, including the ending `jump_past_alt' and
7634 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt -
7635 (1 + OFFSET_ADDRESS_SIZE),
7639 /* Move to right after this alternative, including the
7643 /* Break if it's the beginning of an n-th alternative
7644 that doesn't begin with an on_failure_jump. */
7645 if ((re_opcode_t) *p1 != on_failure_jump)
7648 /* Still have to check that it's not an n-th
7649 alternative that starts with an on_failure_jump. */
7651 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7652 if ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] !=
7655 /* Get to the beginning of the n-th alternative. */
7656 p1 -= 1 + OFFSET_ADDRESS_SIZE;
7661 /* Deal with the last alternative: go back and get number
7662 of the `jump_past_alt' just before it. `mcnt' contains
7663 the length of the alternative. */
7664 EXTRACT_NUMBER (mcnt, p1 - OFFSET_ADDRESS_SIZE);
7666 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt, reg_info))
7669 p1 += mcnt; /* Get past the n-th alternative. */
7675 assert (p1[1] == **p);
7681 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7684 } /* while p1 < end */
7687 } /* group_match_null_string_p */
7690 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7691 It expects P to be the first byte of a single alternative and END one
7692 byte past the last. The alternative can contain groups. */
7695 PREFIX(alt_match_null_string_p) (p, end, reg_info)
7697 PREFIX(register_info_type) *reg_info;
7704 /* Skip over opcodes that can match nothing, and break when we get
7705 to one that can't. */
7707 switch ((re_opcode_t) *p1)
7710 case on_failure_jump:
7712 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7717 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7720 } /* while p1 < end */
7723 } /* alt_match_null_string_p */
7726 /* Deals with the ops common to group_match_null_string_p and
7727 alt_match_null_string_p.
7729 Sets P to one after the op and its arguments, if any. */
7732 PREFIX(common_op_match_null_string_p) (p, end, reg_info)
7734 PREFIX(register_info_type) *reg_info;
7741 switch ((re_opcode_t) *p1++)
7761 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
7762 ret = PREFIX(group_match_null_string_p) (&p1, end, reg_info);
7764 /* Have to set this here in case we're checking a group which
7765 contains a group and a back reference to it. */
7767 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
7768 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
7774 /* If this is an optimized succeed_n for zero times, make the jump. */
7776 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7784 /* Get to the number of times to succeed. */
7785 p1 += OFFSET_ADDRESS_SIZE;
7786 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7790 p1 -= 2 * OFFSET_ADDRESS_SIZE;
7791 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7799 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
7804 p1 += 2 * OFFSET_ADDRESS_SIZE;
7807 /* All other opcodes mean we cannot match the empty string. */
7813 } /* common_op_match_null_string_p */
7816 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7817 bytes; nonzero otherwise. */
7820 PREFIX(bcmp_translate) (s1, s2, len, translate)
7821 const CHAR_T *s1, *s2;
7823 RE_TRANSLATE_TYPE translate;
7825 register const UCHAR_T *p1 = (const UCHAR_T *) s1;
7826 register const UCHAR_T *p2 = (const UCHAR_T *) s2;
7830 if (((*p1<=0xff)?translate[*p1++]:*p1++)
7831 != ((*p2<=0xff)?translate[*p2++]:*p2++))
7834 if (translate[*p1++] != translate[*p2++]) return 1;
7842 #else /* not INSIDE_RECURSION */
7844 /* Entry points for GNU code. */
7846 /* re_compile_pattern is the GNU regular expression compiler: it
7847 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7848 Returns 0 if the pattern was valid, otherwise an error string.
7850 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7851 are set in BUFP on entry.
7853 We call regex_compile to do the actual compilation. */
7856 re_compile_pattern (pattern, length, bufp)
7857 const char *pattern;
7859 struct re_pattern_buffer *bufp;
7863 /* GNU code is written to assume at least RE_NREGS registers will be set
7864 (and at least one extra will be -1). */
7865 bufp->regs_allocated = REGS_UNALLOCATED;
7867 /* And GNU code determines whether or not to get register information
7868 by passing null for the REGS argument to re_match, etc., not by
7872 /* Match anchors at newline. */
7873 bufp->newline_anchor = 1;
7876 if (MB_CUR_MAX != 1)
7877 ret = wcs_regex_compile (pattern, length, re_syntax_options, bufp);
7880 ret = byte_regex_compile (pattern, length, re_syntax_options, bufp);
7884 return gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
7887 weak_alias (__re_compile_pattern, re_compile_pattern)
7890 /* Entry points compatible with 4.2 BSD regex library. We don't define
7891 them unless specifically requested. */
7893 #if defined _REGEX_RE_COMP || defined _LIBC
7895 /* BSD has one and only one pattern buffer. */
7896 static struct re_pattern_buffer re_comp_buf;
7900 /* Make these definitions weak in libc, so POSIX programs can redefine
7901 these names if they don't use our functions, and still use
7902 regcomp/regexec below without link errors. */
7912 if (!re_comp_buf.buffer)
7913 return gettext ("No previous regular expression");
7917 if (!re_comp_buf.buffer)
7919 re_comp_buf.buffer = (unsigned char *) malloc (200);
7920 if (re_comp_buf.buffer == NULL)
7921 return (char *) gettext (re_error_msgid
7922 + re_error_msgid_idx[(int) REG_ESPACE]);
7923 re_comp_buf.allocated = 200;
7925 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
7926 if (re_comp_buf.fastmap == NULL)
7927 return (char *) gettext (re_error_msgid
7928 + re_error_msgid_idx[(int) REG_ESPACE]);
7931 /* Since `re_exec' always passes NULL for the `regs' argument, we
7932 don't need to initialize the pattern buffer fields which affect it. */
7934 /* Match anchors at newlines. */
7935 re_comp_buf.newline_anchor = 1;
7938 if (MB_CUR_MAX != 1)
7939 ret = wcs_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7942 ret = byte_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7947 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7948 return (char *) gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
7959 const int len = strlen (s);
7961 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
7964 #endif /* _REGEX_RE_COMP */
7966 /* POSIX.2 functions. Don't define these for Emacs. */
7970 /* regcomp takes a regular expression as a string and compiles it.
7972 PREG is a regex_t *. We do not expect any fields to be initialized,
7973 since POSIX says we shouldn't. Thus, we set
7975 `buffer' to the compiled pattern;
7976 `used' to the length of the compiled pattern;
7977 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7978 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7979 RE_SYNTAX_POSIX_BASIC;
7980 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7981 `fastmap' to an allocated space for the fastmap;
7982 `fastmap_accurate' to zero;
7983 `re_nsub' to the number of subexpressions in PATTERN.
7985 PATTERN is the address of the pattern string.
7987 CFLAGS is a series of bits which affect compilation.
7989 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7990 use POSIX basic syntax.
7992 If REG_NEWLINE is set, then . and [^...] don't match newline.
7993 Also, regexec will try a match beginning after every newline.
7995 If REG_ICASE is set, then we considers upper- and lowercase
7996 versions of letters to be equivalent when matching.
7998 If REG_NOSUB is set, then when PREG is passed to regexec, that
7999 routine will report only success or failure, and nothing about the
8002 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
8003 the return codes and their meanings.) */
8006 regcomp (preg, pattern, cflags)
8008 const char *pattern;
8013 = (cflags & REG_EXTENDED) ?
8014 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
8016 /* regex_compile will allocate the space for the compiled pattern. */
8018 preg->allocated = 0;
8021 /* Try to allocate space for the fastmap. */
8022 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
8024 if (cflags & REG_ICASE)
8029 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
8030 * sizeof (*(RE_TRANSLATE_TYPE)0));
8031 if (preg->translate == NULL)
8032 return (int) REG_ESPACE;
8034 /* Map uppercase characters to corresponding lowercase ones. */
8035 for (i = 0; i < CHAR_SET_SIZE; i++)
8036 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
8039 preg->translate = NULL;
8041 /* If REG_NEWLINE is set, newlines are treated differently. */
8042 if (cflags & REG_NEWLINE)
8043 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
8044 syntax &= ~RE_DOT_NEWLINE;
8045 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
8046 /* It also changes the matching behavior. */
8047 preg->newline_anchor = 1;
8050 preg->newline_anchor = 0;
8052 preg->no_sub = !!(cflags & REG_NOSUB);
8054 /* POSIX says a null character in the pattern terminates it, so we
8055 can use strlen here in compiling the pattern. */
8057 if (MB_CUR_MAX != 1)
8058 ret = wcs_regex_compile (pattern, strlen (pattern), syntax, preg);
8061 ret = byte_regex_compile (pattern, strlen (pattern), syntax, preg);
8063 /* POSIX doesn't distinguish between an unmatched open-group and an
8064 unmatched close-group: both are REG_EPAREN. */
8065 if (ret == REG_ERPAREN) ret = REG_EPAREN;
8067 if (ret == REG_NOERROR && preg->fastmap)
8069 /* Compute the fastmap now, since regexec cannot modify the pattern
8071 if (re_compile_fastmap (preg) == -2)
8073 /* Some error occurred while computing the fastmap, just forget
8075 free (preg->fastmap);
8076 preg->fastmap = NULL;
8083 weak_alias (__regcomp, regcomp)
8087 /* regexec searches for a given pattern, specified by PREG, in the
8090 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8091 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8092 least NMATCH elements, and we set them to the offsets of the
8093 corresponding matched substrings.
8095 EFLAGS specifies `execution flags' which affect matching: if
8096 REG_NOTBOL is set, then ^ does not match at the beginning of the
8097 string; if REG_NOTEOL is set, then $ does not match at the end.
8099 We return 0 if we find a match and REG_NOMATCH if not. */
8102 regexec (preg, string, nmatch, pmatch, eflags)
8103 const regex_t *preg;
8106 regmatch_t pmatch[];
8110 struct re_registers regs;
8111 regex_t private_preg;
8112 int len = strlen (string);
8113 boolean want_reg_info = !preg->no_sub && nmatch > 0;
8115 private_preg = *preg;
8117 private_preg.not_bol = !!(eflags & REG_NOTBOL);
8118 private_preg.not_eol = !!(eflags & REG_NOTEOL);
8120 /* The user has told us exactly how many registers to return
8121 information about, via `nmatch'. We have to pass that on to the
8122 matching routines. */
8123 private_preg.regs_allocated = REGS_FIXED;
8127 regs.num_regs = nmatch;
8128 regs.start = TALLOC (nmatch * 2, regoff_t);
8129 if (regs.start == NULL)
8130 return (int) REG_NOMATCH;
8131 regs.end = regs.start + nmatch;
8134 /* Perform the searching operation. */
8135 ret = re_search (&private_preg, string, len,
8136 /* start: */ 0, /* range: */ len,
8137 want_reg_info ? ®s : (struct re_registers *) 0);
8139 /* Copy the register information to the POSIX structure. */
8146 for (r = 0; r < nmatch; r++)
8148 pmatch[r].rm_so = regs.start[r];
8149 pmatch[r].rm_eo = regs.end[r];
8153 /* If we needed the temporary register info, free the space now. */
8157 /* We want zero return to mean success, unlike `re_search'. */
8158 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
8161 weak_alias (__regexec, regexec)
8165 /* Returns a message corresponding to an error code, ERRCODE, returned
8166 from either regcomp or regexec. We don't use PREG here. */
8169 regerror (errcode, preg, errbuf, errbuf_size)
8171 const regex_t *preg;
8179 || errcode >= (int) (sizeof (re_error_msgid_idx)
8180 / sizeof (re_error_msgid_idx[0])))
8181 /* Only error codes returned by the rest of the code should be passed
8182 to this routine. If we are given anything else, or if other regex
8183 code generates an invalid error code, then the program has a bug.
8184 Dump core so we can fix it. */
8187 msg = gettext (re_error_msgid + re_error_msgid_idx[errcode]);
8189 msg_size = strlen (msg) + 1; /* Includes the null. */
8191 if (errbuf_size != 0)
8193 if (msg_size > errbuf_size)
8195 #if defined HAVE_MEMPCPY || defined _LIBC
8196 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
8198 memcpy (errbuf, msg, errbuf_size - 1);
8199 errbuf[errbuf_size - 1] = 0;
8203 memcpy (errbuf, msg, msg_size);
8209 weak_alias (__regerror, regerror)
8213 /* Free dynamically allocated space used by PREG. */
8219 if (preg->buffer != NULL)
8220 free (preg->buffer);
8221 preg->buffer = NULL;
8223 preg->allocated = 0;
8226 if (preg->fastmap != NULL)
8227 free (preg->fastmap);
8228 preg->fastmap = NULL;
8229 preg->fastmap_accurate = 0;
8231 if (preg->translate != NULL)
8232 free (preg->translate);
8233 preg->translate = NULL;
8236 weak_alias (__regfree, regfree)
8239 #endif /* not emacs */
8241 #endif /* not INSIDE_RECURSION */
8245 #undef STORE_NUMBER_AND_INCR
8246 #undef EXTRACT_NUMBER
8247 #undef EXTRACT_NUMBER_AND_INCR
8249 #undef DEBUG_PRINT_COMPILED_PATTERN
8250 #undef DEBUG_PRINT_DOUBLE_STRING
8252 #undef INIT_FAIL_STACK
8253 #undef RESET_FAIL_STACK
8254 #undef DOUBLE_FAIL_STACK
8255 #undef PUSH_PATTERN_OP
8256 #undef PUSH_FAILURE_POINTER
8257 #undef PUSH_FAILURE_INT
8258 #undef PUSH_FAILURE_ELT
8259 #undef POP_FAILURE_POINTER
8260 #undef POP_FAILURE_INT
8261 #undef POP_FAILURE_ELT
8264 #undef PUSH_FAILURE_POINT
8265 #undef POP_FAILURE_POINT
8267 #undef REG_UNSET_VALUE
8275 #undef INIT_BUF_SIZE
8276 #undef GET_BUFFER_SPACE
8284 #undef EXTEND_BUFFER
8285 #undef GET_UNSIGNED_NUMBER
8286 #undef FREE_STACK_RETURN
8288 # undef POINTER_TO_OFFSET
8289 # undef MATCHING_IN_FRST_STRING
8291 # undef AT_STRINGS_BEG
8292 # undef AT_STRINGS_END
8295 # undef FREE_VARIABLES
8296 # undef NO_HIGHEST_ACTIVE_REG
8297 # undef NO_LOWEST_ACTIVE_REG
8301 # undef COMPILED_BUFFER_VAR
8302 # undef OFFSET_ADDRESS_SIZE
8303 # undef CHAR_CLASS_SIZE
8310 # define DEFINED_ONCE
projects / gnulib.git / blob