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
634 #include "unlocked-io.h"
636 #ifdef INSIDE_RECURSION
637 /* Common operations on the compiled pattern. */
639 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
640 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
643 # define STORE_NUMBER(destination, number) \
645 *(destination) = (UCHAR_T)(number); \
648 # define STORE_NUMBER(destination, number) \
650 (destination)[0] = (number) & 0377; \
651 (destination)[1] = (number) >> 8; \
655 /* Same as STORE_NUMBER, except increment DESTINATION to
656 the byte after where the number is stored. Therefore, DESTINATION
657 must be an lvalue. */
658 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
660 # define STORE_NUMBER_AND_INCR(destination, number) \
662 STORE_NUMBER (destination, number); \
663 (destination) += OFFSET_ADDRESS_SIZE; \
666 /* Put into DESTINATION a number stored in two contiguous bytes starting
668 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
671 # define EXTRACT_NUMBER(destination, source) \
673 (destination) = *(source); \
676 # define EXTRACT_NUMBER(destination, source) \
678 (destination) = *(source) & 0377; \
679 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
684 static void PREFIX(extract_number) _RE_ARGS ((int *dest, UCHAR_T *source));
686 PREFIX(extract_number) (dest, source)
693 int temp = SIGN_EXTEND_CHAR (*(source + 1));
694 *dest = *source & 0377;
699 # ifndef EXTRACT_MACROS /* To debug the macros. */
700 # undef EXTRACT_NUMBER
701 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
702 # endif /* not EXTRACT_MACROS */
706 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
707 SOURCE must be an lvalue. */
709 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
711 EXTRACT_NUMBER (destination, source); \
712 (source) += OFFSET_ADDRESS_SIZE; \
716 static void PREFIX(extract_number_and_incr) _RE_ARGS ((int *destination,
719 PREFIX(extract_number_and_incr) (destination, source)
723 PREFIX(extract_number) (destination, *source);
724 *source += OFFSET_ADDRESS_SIZE;
727 # ifndef EXTRACT_MACROS
728 # undef EXTRACT_NUMBER_AND_INCR
729 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
730 PREFIX(extract_number_and_incr) (&dest, &src)
731 # endif /* not EXTRACT_MACROS */
737 /* If DEBUG is defined, Regex prints many voluminous messages about what
738 it is doing (if the variable `debug' is nonzero). If linked with the
739 main program in `iregex.c', you can enter patterns and strings
740 interactively. And if linked with the main program in `main.c' and
741 the other test files, you can run the already-written tests. */
745 # ifndef DEFINED_ONCE
747 /* We use standard I/O for debugging. */
750 /* It is useful to test things that ``must'' be true when debugging. */
755 # define DEBUG_STATEMENT(e) e
756 # define DEBUG_PRINT1(x) if (debug) printf (x)
757 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
758 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
759 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
760 # endif /* not DEFINED_ONCE */
762 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
763 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
764 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
765 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
768 /* Print the fastmap in human-readable form. */
770 # ifndef DEFINED_ONCE
772 print_fastmap (fastmap)
775 unsigned was_a_range = 0;
778 while (i < (1 << BYTEWIDTH))
784 while (i < (1 << BYTEWIDTH) && fastmap[i])
798 # endif /* not DEFINED_ONCE */
801 /* Print a compiled pattern string in human-readable form, starting at
802 the START pointer into it and ending just before the pointer END. */
805 PREFIX(print_partial_compiled_pattern) (start, end)
820 /* Loop over pattern commands. */
824 printf ("%td:\t", p - start);
826 printf ("%ld:\t", (long int) (p - start));
829 switch ((re_opcode_t) *p++)
837 printf ("/exactn/%d", mcnt);
849 printf ("/exactn_bin/%d", mcnt);
852 printf("/%lx", (long int) *p++);
856 # endif /* MBS_SUPPORT */
860 printf ("/start_memory/%d/%ld", mcnt, (long int) *p++);
865 printf ("/stop_memory/%d/%ld", mcnt, (long int) *p++);
869 printf ("/duplicate/%ld", (long int) *p++);
882 printf ("/charset [%s",
883 (re_opcode_t) *(workp - 1) == charset_not ? "^" : "");
885 length = *workp++; /* the length of char_classes */
886 for (i=0 ; i<length ; i++)
887 printf("[:%lx:]", (long int) *p++);
888 length = *workp++; /* the length of collating_symbol */
889 for (i=0 ; i<length ;)
893 PUT_CHAR((i++,*p++));
897 length = *workp++; /* the length of equivalence_class */
898 for (i=0 ; i<length ;)
902 PUT_CHAR((i++,*p++));
906 length = *workp++; /* the length of char_range */
907 for (i=0 ; i<length ; i++)
909 wchar_t range_start = *p++;
910 wchar_t range_end = *p++;
911 printf("%C-%C", range_start, range_end);
913 length = *workp++; /* the length of char */
914 for (i=0 ; i<length ; i++)
918 register int c, last = -100;
919 register int in_range = 0;
921 printf ("/charset [%s",
922 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
924 assert (p + *p < pend);
926 for (c = 0; c < 256; c++)
928 && (p[1 + (c/8)] & (1 << (c % 8))))
930 /* Are we starting a range? */
931 if (last + 1 == c && ! in_range)
936 /* Have we broken a range? */
937 else if (last + 1 != c && in_range)
967 case on_failure_jump:
968 PREFIX(extract_number_and_incr) (&mcnt, &p);
970 printf ("/on_failure_jump to %td", p + mcnt - start);
972 printf ("/on_failure_jump to %ld", (long int) (p + mcnt - start));
976 case on_failure_keep_string_jump:
977 PREFIX(extract_number_and_incr) (&mcnt, &p);
979 printf ("/on_failure_keep_string_jump to %td", p + mcnt - start);
981 printf ("/on_failure_keep_string_jump to %ld",
982 (long int) (p + mcnt - start));
986 case dummy_failure_jump:
987 PREFIX(extract_number_and_incr) (&mcnt, &p);
989 printf ("/dummy_failure_jump to %td", p + mcnt - start);
991 printf ("/dummy_failure_jump to %ld", (long int) (p + mcnt - start));
995 case push_dummy_failure:
996 printf ("/push_dummy_failure");
1000 PREFIX(extract_number_and_incr) (&mcnt, &p);
1002 printf ("/maybe_pop_jump to %td", p + mcnt - start);
1004 printf ("/maybe_pop_jump to %ld", (long int) (p + mcnt - start));
1008 case pop_failure_jump:
1009 PREFIX(extract_number_and_incr) (&mcnt, &p);
1011 printf ("/pop_failure_jump to %td", p + mcnt - start);
1013 printf ("/pop_failure_jump to %ld", (long int) (p + mcnt - start));
1018 PREFIX(extract_number_and_incr) (&mcnt, &p);
1020 printf ("/jump_past_alt to %td", p + mcnt - start);
1022 printf ("/jump_past_alt to %ld", (long int) (p + mcnt - start));
1027 PREFIX(extract_number_and_incr) (&mcnt, &p);
1029 printf ("/jump to %td", p + mcnt - start);
1031 printf ("/jump to %ld", (long int) (p + mcnt - start));
1036 PREFIX(extract_number_and_incr) (&mcnt, &p);
1038 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1040 printf ("/succeed_n to %td, %d times", p1 - start, mcnt2);
1042 printf ("/succeed_n to %ld, %d times",
1043 (long int) (p1 - start), mcnt2);
1048 PREFIX(extract_number_and_incr) (&mcnt, &p);
1050 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1051 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
1055 PREFIX(extract_number_and_incr) (&mcnt, &p);
1057 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1059 printf ("/set_number_at location %td to %d", p1 - start, mcnt2);
1061 printf ("/set_number_at location %ld to %d",
1062 (long int) (p1 - start), mcnt2);
1067 printf ("/wordbound");
1071 printf ("/notwordbound");
1075 printf ("/wordbeg");
1079 printf ("/wordend");
1084 printf ("/before_dot");
1092 printf ("/after_dot");
1096 printf ("/syntaxspec");
1098 printf ("/%d", mcnt);
1102 printf ("/notsyntaxspec");
1104 printf ("/%d", mcnt);
1109 printf ("/wordchar");
1113 printf ("/notwordchar");
1125 printf ("?%ld", (long int) *(p-1));
1132 printf ("%td:\tend of pattern.\n", p - start);
1134 printf ("%ld:\tend of pattern.\n", (long int) (p - start));
1140 PREFIX(print_compiled_pattern) (bufp)
1141 struct re_pattern_buffer *bufp;
1143 UCHAR_T *buffer = (UCHAR_T*) bufp->buffer;
1145 PREFIX(print_partial_compiled_pattern) (buffer, buffer
1146 + bufp->used / sizeof(UCHAR_T));
1147 printf ("%ld bytes used/%ld bytes allocated.\n",
1148 bufp->used, bufp->allocated);
1150 if (bufp->fastmap_accurate && bufp->fastmap)
1152 printf ("fastmap: ");
1153 print_fastmap (bufp->fastmap);
1157 printf ("re_nsub: %Zd\t", bufp->re_nsub);
1159 printf ("re_nsub: %ld\t", (long int) bufp->re_nsub);
1161 printf ("regs_alloc: %d\t", bufp->regs_allocated);
1162 printf ("can_be_null: %d\t", bufp->can_be_null);
1163 printf ("newline_anchor: %d\n", bufp->newline_anchor);
1164 printf ("no_sub: %d\t", bufp->no_sub);
1165 printf ("not_bol: %d\t", bufp->not_bol);
1166 printf ("not_eol: %d\t", bufp->not_eol);
1167 printf ("syntax: %lx\n", bufp->syntax);
1168 /* Perhaps we should print the translate table? */
1173 PREFIX(print_double_string) (where, string1, size1, string2, size2)
1174 const CHAR_T *where;
1175 const CHAR_T *string1;
1176 const CHAR_T *string2;
1188 if (FIRST_STRING_P (where))
1190 for (this_char = where - string1; this_char < size1; this_char++)
1191 PUT_CHAR (string1[this_char]);
1197 for (this_char = where - string2; this_char < size2; this_char++)
1199 PUT_CHAR (string2[this_char]);
1202 fputs ("...", stdout);
1209 # ifndef DEFINED_ONCE
1218 # else /* not DEBUG */
1220 # ifndef DEFINED_ONCE
1224 # define DEBUG_STATEMENT(e)
1225 # define DEBUG_PRINT1(x)
1226 # define DEBUG_PRINT2(x1, x2)
1227 # define DEBUG_PRINT3(x1, x2, x3)
1228 # define DEBUG_PRINT4(x1, x2, x3, x4)
1229 # endif /* not DEFINED_ONCE */
1230 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1231 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1233 # endif /* not DEBUG */
1238 /* This convert a multibyte string to a wide character string.
1239 And write their correspondances to offset_buffer(see below)
1240 and write whether each wchar_t is binary data to is_binary.
1241 This assume invalid multibyte sequences as binary data.
1242 We assume offset_buffer and is_binary is already allocated
1245 static size_t convert_mbs_to_wcs (CHAR_T *dest, const unsigned char* src,
1246 size_t len, int *offset_buffer,
1249 convert_mbs_to_wcs (dest, src, len, offset_buffer, is_binary)
1251 const unsigned char* src;
1252 size_t len; /* the length of multibyte string. */
1254 /* It hold correspondances between src(char string) and
1255 dest(wchar_t string) for optimization.
1257 dest = {'X', 'Y', 'Z'}
1258 (each "xxx", "y" and "zz" represent one multibyte character
1259 corresponding to 'X', 'Y' and 'Z'.)
1260 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1266 wchar_t *pdest = dest;
1267 const unsigned char *psrc = src;
1268 size_t wc_count = 0;
1272 size_t mb_remain = len;
1273 size_t mb_count = 0;
1275 /* Initialize the conversion state. */
1276 memset (&mbs, 0, sizeof (mbstate_t));
1278 offset_buffer[0] = 0;
1279 for( ; mb_remain > 0 ; ++wc_count, ++pdest, mb_remain -= consumed,
1282 consumed = mbrtowc (pdest, psrc, mb_remain, &mbs);
1285 /* failed to convert. maybe src contains binary data.
1286 So we consume 1 byte manualy. */
1290 is_binary[wc_count] = TRUE;
1293 is_binary[wc_count] = FALSE;
1294 /* In sjis encoding, we use yen sign as escape character in
1295 place of reverse solidus. So we convert 0x5c(yen sign in
1296 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1297 solidus in UCS2). */
1298 if (consumed == 1 && (int) *psrc == 0x5c && (int) *pdest == 0xa5)
1299 *pdest = (wchar_t) *psrc;
1301 offset_buffer[wc_count + 1] = mb_count += consumed;
1304 /* Fill remain of the buffer with sentinel. */
1305 for (i = wc_count + 1 ; i <= len ; i++)
1306 offset_buffer[i] = mb_count + 1;
1313 #else /* not INSIDE_RECURSION */
1315 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1316 also be assigned to arbitrarily: each pattern buffer stores its own
1317 syntax, so it can be changed between regex compilations. */
1318 /* This has no initializer because initialized variables in Emacs
1319 become read-only after dumping. */
1320 reg_syntax_t re_syntax_options;
1323 /* Specify the precise syntax of regexps for compilation. This provides
1324 for compatibility for various utilities which historically have
1325 different, incompatible syntaxes.
1327 The argument SYNTAX is a bit mask comprised of the various bits
1328 defined in regex.h. We return the old syntax. */
1331 re_set_syntax (syntax)
1332 reg_syntax_t syntax;
1334 reg_syntax_t ret = re_syntax_options;
1336 re_syntax_options = syntax;
1338 if (syntax & RE_DEBUG)
1340 else if (debug) /* was on but now is not */
1346 weak_alias (__re_set_syntax, re_set_syntax)
1349 /* This table gives an error message for each of the error codes listed
1350 in regex.h. Obviously the order here has to be same as there.
1351 POSIX doesn't require that we do anything for REG_NOERROR,
1352 but why not be nice? */
1354 static const char re_error_msgid[] =
1356 # define REG_NOERROR_IDX 0
1357 gettext_noop ("Success") /* REG_NOERROR */
1359 # define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success")
1360 gettext_noop ("No match") /* REG_NOMATCH */
1362 # define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match")
1363 gettext_noop ("Invalid regular expression") /* REG_BADPAT */
1365 # define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression")
1366 gettext_noop ("Invalid collation character") /* REG_ECOLLATE */
1368 # define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character")
1369 gettext_noop ("Invalid character class name") /* REG_ECTYPE */
1371 # define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name")
1372 gettext_noop ("Trailing backslash") /* REG_EESCAPE */
1374 # define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash")
1375 gettext_noop ("Invalid back reference") /* REG_ESUBREG */
1377 # define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference")
1378 gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */
1380 # define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^")
1381 gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */
1383 # define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(")
1384 gettext_noop ("Unmatched \\{") /* REG_EBRACE */
1386 # define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{")
1387 gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */
1389 # define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}")
1390 gettext_noop ("Invalid range end") /* REG_ERANGE */
1392 # define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end")
1393 gettext_noop ("Memory exhausted") /* REG_ESPACE */
1395 # define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted")
1396 gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */
1398 # define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression")
1399 gettext_noop ("Premature end of regular expression") /* REG_EEND */
1401 # define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression")
1402 gettext_noop ("Regular expression too big") /* REG_ESIZE */
1404 # define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big")
1405 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
1408 static const size_t re_error_msgid_idx[] =
1429 #endif /* INSIDE_RECURSION */
1431 #ifndef DEFINED_ONCE
1432 /* Avoiding alloca during matching, to placate r_alloc. */
1434 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1435 searching and matching functions should not call alloca. On some
1436 systems, alloca is implemented in terms of malloc, and if we're
1437 using the relocating allocator routines, then malloc could cause a
1438 relocation, which might (if the strings being searched are in the
1439 ralloc heap) shift the data out from underneath the regexp
1442 Here's another reason to avoid allocation: Emacs
1443 processes input from X in a signal handler; processing X input may
1444 call malloc; if input arrives while a matching routine is calling
1445 malloc, then we're scrod. But Emacs can't just block input while
1446 calling matching routines; then we don't notice interrupts when
1447 they come in. So, Emacs blocks input around all regexp calls
1448 except the matching calls, which it leaves unprotected, in the
1449 faith that they will not malloc. */
1451 /* Normally, this is fine. */
1452 # define MATCH_MAY_ALLOCATE
1454 /* When using GNU C, we are not REALLY using the C alloca, no matter
1455 what config.h may say. So don't take precautions for it. */
1460 /* The match routines may not allocate if (1) they would do it with malloc
1461 and (2) it's not safe for them to use malloc.
1462 Note that if REL_ALLOC is defined, matching would not use malloc for the
1463 failure stack, but we would still use it for the register vectors;
1464 so REL_ALLOC should not affect this. */
1465 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1466 # undef MATCH_MAY_ALLOCATE
1468 #endif /* not DEFINED_ONCE */
1470 #ifdef INSIDE_RECURSION
1471 /* Failure stack declarations and macros; both re_compile_fastmap and
1472 re_match_2 use a failure stack. These have to be macros because of
1473 REGEX_ALLOCATE_STACK. */
1476 /* Number of failure points for which to initially allocate space
1477 when matching. If this number is exceeded, we allocate more
1478 space, so it is not a hard limit. */
1479 # ifndef INIT_FAILURE_ALLOC
1480 # define INIT_FAILURE_ALLOC 5
1483 /* Roughly the maximum number of failure points on the stack. Would be
1484 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1485 This is a variable only so users of regex can assign to it; we never
1486 change it ourselves. */
1488 # ifdef INT_IS_16BIT
1490 # ifndef DEFINED_ONCE
1491 # if defined MATCH_MAY_ALLOCATE
1492 /* 4400 was enough to cause a crash on Alpha OSF/1,
1493 whose default stack limit is 2mb. */
1494 long int re_max_failures = 4000;
1496 long int re_max_failures = 2000;
1500 union PREFIX(fail_stack_elt)
1506 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1510 PREFIX(fail_stack_elt_t) *stack;
1511 unsigned long int size;
1512 unsigned long int avail; /* Offset of next open position. */
1513 } PREFIX(fail_stack_type);
1515 # else /* not INT_IS_16BIT */
1517 # ifndef DEFINED_ONCE
1518 # if defined MATCH_MAY_ALLOCATE
1519 /* 4400 was enough to cause a crash on Alpha OSF/1,
1520 whose default stack limit is 2mb. */
1521 int re_max_failures = 4000;
1523 int re_max_failures = 2000;
1527 union PREFIX(fail_stack_elt)
1533 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1537 PREFIX(fail_stack_elt_t) *stack;
1539 unsigned avail; /* Offset of next open position. */
1540 } PREFIX(fail_stack_type);
1542 # endif /* INT_IS_16BIT */
1544 # ifndef DEFINED_ONCE
1545 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1546 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1547 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1551 /* Define macros to initialize and free the failure stack.
1552 Do `return -2' if the alloc fails. */
1554 # ifdef MATCH_MAY_ALLOCATE
1555 # define INIT_FAIL_STACK() \
1557 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1558 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1560 if (fail_stack.stack == NULL) \
1563 fail_stack.size = INIT_FAILURE_ALLOC; \
1564 fail_stack.avail = 0; \
1567 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1569 # define INIT_FAIL_STACK() \
1571 fail_stack.avail = 0; \
1574 # define RESET_FAIL_STACK()
1578 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1580 Return 1 if succeeds, and 0 if either ran out of memory
1581 allocating space for it or it was already too large.
1583 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1585 # define DOUBLE_FAIL_STACK(fail_stack) \
1586 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1588 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1589 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1590 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1591 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1593 (fail_stack).stack == NULL \
1595 : ((fail_stack).size <<= 1, \
1599 /* Push pointer POINTER on FAIL_STACK.
1600 Return 1 if was able to do so and 0 if ran out of memory allocating
1602 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1603 ((FAIL_STACK_FULL () \
1604 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1606 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1609 /* Push a pointer value onto the failure stack.
1610 Assumes the variable `fail_stack'. Probably should only
1611 be called from within `PUSH_FAILURE_POINT'. */
1612 # define PUSH_FAILURE_POINTER(item) \
1613 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1615 /* This pushes an integer-valued item onto the failure stack.
1616 Assumes the variable `fail_stack'. Probably should only
1617 be called from within `PUSH_FAILURE_POINT'. */
1618 # define PUSH_FAILURE_INT(item) \
1619 fail_stack.stack[fail_stack.avail++].integer = (item)
1621 /* Push a fail_stack_elt_t value onto the failure stack.
1622 Assumes the variable `fail_stack'. Probably should only
1623 be called from within `PUSH_FAILURE_POINT'. */
1624 # define PUSH_FAILURE_ELT(item) \
1625 fail_stack.stack[fail_stack.avail++] = (item)
1627 /* These three POP... operations complement the three PUSH... operations.
1628 All assume that `fail_stack' is nonempty. */
1629 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1630 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1631 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1633 /* Used to omit pushing failure point id's when we're not debugging. */
1635 # define DEBUG_PUSH PUSH_FAILURE_INT
1636 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1638 # define DEBUG_PUSH(item)
1639 # define DEBUG_POP(item_addr)
1643 /* Push the information about the state we will need
1644 if we ever fail back to it.
1646 Requires variables fail_stack, regstart, regend, reg_info, and
1647 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1650 Does `return FAILURE_CODE' if runs out of memory. */
1652 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1654 char *destination; \
1655 /* Must be int, so when we don't save any registers, the arithmetic \
1656 of 0 + -1 isn't done as unsigned. */ \
1657 /* Can't be int, since there is not a shred of a guarantee that int \
1658 is wide enough to hold a value of something to which pointer can \
1660 active_reg_t this_reg; \
1662 DEBUG_STATEMENT (failure_id++); \
1663 DEBUG_STATEMENT (nfailure_points_pushed++); \
1664 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1665 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1666 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1668 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1669 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1671 /* Ensure we have enough space allocated for what we will push. */ \
1672 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1674 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1675 return failure_code; \
1677 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1678 (fail_stack).size); \
1679 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1682 /* Push the info, starting with the registers. */ \
1683 DEBUG_PRINT1 ("\n"); \
1686 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1689 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1690 DEBUG_STATEMENT (num_regs_pushed++); \
1692 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1693 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1695 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1696 PUSH_FAILURE_POINTER (regend[this_reg]); \
1698 DEBUG_PRINT2 (" info: %p\n ", \
1699 reg_info[this_reg].word.pointer); \
1700 DEBUG_PRINT2 (" match_null=%d", \
1701 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1702 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1703 DEBUG_PRINT2 (" matched_something=%d", \
1704 MATCHED_SOMETHING (reg_info[this_reg])); \
1705 DEBUG_PRINT2 (" ever_matched=%d", \
1706 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1707 DEBUG_PRINT1 ("\n"); \
1708 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1711 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1712 PUSH_FAILURE_INT (lowest_active_reg); \
1714 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1715 PUSH_FAILURE_INT (highest_active_reg); \
1717 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1718 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1719 PUSH_FAILURE_POINTER (pattern_place); \
1721 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1722 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1724 DEBUG_PRINT1 ("'\n"); \
1725 PUSH_FAILURE_POINTER (string_place); \
1727 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1728 DEBUG_PUSH (failure_id); \
1731 # ifndef DEFINED_ONCE
1732 /* This is the number of items that are pushed and popped on the stack
1733 for each register. */
1734 # define NUM_REG_ITEMS 3
1736 /* Individual items aside from the registers. */
1738 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1740 # define NUM_NONREG_ITEMS 4
1743 /* We push at most this many items on the stack. */
1744 /* We used to use (num_regs - 1), which is the number of registers
1745 this regexp will save; but that was changed to 5
1746 to avoid stack overflow for a regexp with lots of parens. */
1747 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1749 /* We actually push this many items. */
1750 # define NUM_FAILURE_ITEMS \
1752 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1756 /* How many items can still be added to the stack without overflowing it. */
1757 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1758 # endif /* not DEFINED_ONCE */
1761 /* Pops what PUSH_FAIL_STACK pushes.
1763 We restore into the parameters, all of which should be lvalues:
1764 STR -- the saved data position.
1765 PAT -- the saved pattern position.
1766 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1767 REGSTART, REGEND -- arrays of string positions.
1768 REG_INFO -- array of information about each subexpression.
1770 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1771 `pend', `string1', `size1', `string2', and `size2'. */
1772 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1774 DEBUG_STATEMENT (unsigned failure_id;) \
1775 active_reg_t this_reg; \
1776 const UCHAR_T *string_temp; \
1778 assert (!FAIL_STACK_EMPTY ()); \
1780 /* Remove failure points and point to how many regs pushed. */ \
1781 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1782 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1783 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1785 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1787 DEBUG_POP (&failure_id); \
1788 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1790 /* If the saved string location is NULL, it came from an \
1791 on_failure_keep_string_jump opcode, and we want to throw away the \
1792 saved NULL, thus retaining our current position in the string. */ \
1793 string_temp = POP_FAILURE_POINTER (); \
1794 if (string_temp != NULL) \
1795 str = (const CHAR_T *) string_temp; \
1797 DEBUG_PRINT2 (" Popping string %p: `", str); \
1798 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1799 DEBUG_PRINT1 ("'\n"); \
1801 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1802 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1803 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1805 /* Restore register info. */ \
1806 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1807 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1809 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1810 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1813 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1815 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1817 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1818 DEBUG_PRINT2 (" info: %p\n", \
1819 reg_info[this_reg].word.pointer); \
1821 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1822 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1824 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1825 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1829 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1831 reg_info[this_reg].word.integer = 0; \
1832 regend[this_reg] = 0; \
1833 regstart[this_reg] = 0; \
1835 highest_active_reg = high_reg; \
1838 set_regs_matched_done = 0; \
1839 DEBUG_STATEMENT (nfailure_points_popped++); \
1840 } /* POP_FAILURE_POINT */
1842 /* Structure for per-register (a.k.a. per-group) information.
1843 Other register information, such as the
1844 starting and ending positions (which are addresses), and the list of
1845 inner groups (which is a bits list) are maintained in separate
1848 We are making a (strictly speaking) nonportable assumption here: that
1849 the compiler will pack our bit fields into something that fits into
1850 the type of `word', i.e., is something that fits into one item on the
1854 /* Declarations and macros for re_match_2. */
1858 PREFIX(fail_stack_elt_t) word;
1861 /* This field is one if this group can match the empty string,
1862 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1863 # define MATCH_NULL_UNSET_VALUE 3
1864 unsigned match_null_string_p : 2;
1865 unsigned is_active : 1;
1866 unsigned matched_something : 1;
1867 unsigned ever_matched_something : 1;
1869 } PREFIX(register_info_type);
1871 # ifndef DEFINED_ONCE
1872 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1873 # define IS_ACTIVE(R) ((R).bits.is_active)
1874 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1875 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1878 /* Call this when have matched a real character; it sets `matched' flags
1879 for the subexpressions which we are currently inside. Also records
1880 that those subexprs have matched. */
1881 # define SET_REGS_MATCHED() \
1884 if (!set_regs_matched_done) \
1887 set_regs_matched_done = 1; \
1888 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1890 MATCHED_SOMETHING (reg_info[r]) \
1891 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1897 # endif /* not DEFINED_ONCE */
1899 /* Registers are set to a sentinel when they haven't yet matched. */
1900 static CHAR_T PREFIX(reg_unset_dummy);
1901 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1902 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1904 /* Subroutine declarations and macros for regex_compile. */
1905 static void PREFIX(store_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc, int arg));
1906 static void PREFIX(store_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1907 int arg1, int arg2));
1908 static void PREFIX(insert_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1909 int arg, UCHAR_T *end));
1910 static void PREFIX(insert_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1911 int arg1, int arg2, UCHAR_T *end));
1912 static boolean PREFIX(at_begline_loc_p) _RE_ARGS ((const CHAR_T *pattern,
1914 reg_syntax_t syntax));
1915 static boolean PREFIX(at_endline_loc_p) _RE_ARGS ((const CHAR_T *p,
1917 reg_syntax_t syntax));
1919 static reg_errcode_t wcs_compile_range _RE_ARGS ((CHAR_T range_start,
1920 const CHAR_T **p_ptr,
1923 reg_syntax_t syntax,
1926 static void insert_space _RE_ARGS ((int num, CHAR_T *loc, CHAR_T *end));
1928 static reg_errcode_t byte_compile_range _RE_ARGS ((unsigned int range_start,
1932 reg_syntax_t syntax,
1936 /* Fetch the next character in the uncompiled pattern---translating it
1937 if necessary. Also cast from a signed character in the constant
1938 string passed to us by the user to an unsigned char that we can use
1939 as an array index (in, e.g., `translate'). */
1940 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1941 because it is impossible to allocate 4GB array for some encodings
1942 which have 4 byte character_set like UCS4. */
1945 # define PATFETCH(c) \
1946 do {if (p == pend) return REG_EEND; \
1947 c = (UCHAR_T) *p++; \
1948 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1951 # define PATFETCH(c) \
1952 do {if (p == pend) return REG_EEND; \
1953 c = (unsigned char) *p++; \
1954 if (translate) c = (unsigned char) translate[c]; \
1959 /* Fetch the next character in the uncompiled pattern, with no
1961 # define PATFETCH_RAW(c) \
1962 do {if (p == pend) return REG_EEND; \
1963 c = (UCHAR_T) *p++; \
1966 /* Go backwards one character in the pattern. */
1967 # define PATUNFETCH p--
1970 /* If `translate' is non-null, return translate[D], else just D. We
1971 cast the subscript to translate because some data is declared as
1972 `char *', to avoid warnings when a string constant is passed. But
1973 when we use a character as a subscript we must make it unsigned. */
1974 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1975 because it is impossible to allocate 4GB array for some encodings
1976 which have 4 byte character_set like UCS4. */
1980 # define TRANSLATE(d) \
1981 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1982 ? (char) translate[(unsigned char) (d)] : (d))
1984 # define TRANSLATE(d) \
1985 (translate ? (char) translate[(unsigned char) (d)] : (d))
1990 /* Macros for outputting the compiled pattern into `buffer'. */
1992 /* If the buffer isn't allocated when it comes in, use this. */
1993 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1995 /* Make sure we have at least N more bytes of space in buffer. */
1997 # define GET_BUFFER_SPACE(n) \
1998 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1999 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
2002 # define GET_BUFFER_SPACE(n) \
2003 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
2007 /* Make sure we have one more byte of buffer space and then add C to it. */
2008 # define BUF_PUSH(c) \
2010 GET_BUFFER_SPACE (1); \
2011 *b++ = (UCHAR_T) (c); \
2015 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
2016 # define BUF_PUSH_2(c1, c2) \
2018 GET_BUFFER_SPACE (2); \
2019 *b++ = (UCHAR_T) (c1); \
2020 *b++ = (UCHAR_T) (c2); \
2024 /* As with BUF_PUSH_2, except for three bytes. */
2025 # define BUF_PUSH_3(c1, c2, c3) \
2027 GET_BUFFER_SPACE (3); \
2028 *b++ = (UCHAR_T) (c1); \
2029 *b++ = (UCHAR_T) (c2); \
2030 *b++ = (UCHAR_T) (c3); \
2033 /* Store a jump with opcode OP at LOC to location TO. We store a
2034 relative address offset by the three bytes the jump itself occupies. */
2035 # define STORE_JUMP(op, loc, to) \
2036 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
2038 /* Likewise, for a two-argument jump. */
2039 # define STORE_JUMP2(op, loc, to, arg) \
2040 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
2042 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
2043 # define INSERT_JUMP(op, loc, to) \
2044 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
2046 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
2047 # define INSERT_JUMP2(op, loc, to, arg) \
2048 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
2051 /* This is not an arbitrary limit: the arguments which represent offsets
2052 into the pattern are two bytes long. So if 2^16 bytes turns out to
2053 be too small, many things would have to change. */
2054 /* Any other compiler which, like MSC, has allocation limit below 2^16
2055 bytes will have to use approach similar to what was done below for
2056 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2057 reallocating to 0 bytes. Such thing is not going to work too well.
2058 You have been warned!! */
2059 # ifndef DEFINED_ONCE
2060 # if defined _MSC_VER && !defined WIN32
2061 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2062 The REALLOC define eliminates a flurry of conversion warnings,
2063 but is not required. */
2064 # define MAX_BUF_SIZE 65500L
2065 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2067 # define MAX_BUF_SIZE (1L << 16)
2068 # define REALLOC(p,s) realloc ((p), (s))
2071 /* Extend the buffer by twice its current size via realloc and
2072 reset the pointers that pointed into the old block to point to the
2073 correct places in the new one. If extending the buffer results in it
2074 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2075 # if __BOUNDED_POINTERS__
2076 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2077 # define MOVE_BUFFER_POINTER(P) \
2078 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2079 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2082 SET_HIGH_BOUND (b); \
2083 SET_HIGH_BOUND (begalt); \
2084 if (fixup_alt_jump) \
2085 SET_HIGH_BOUND (fixup_alt_jump); \
2087 SET_HIGH_BOUND (laststart); \
2088 if (pending_exact) \
2089 SET_HIGH_BOUND (pending_exact); \
2092 # define MOVE_BUFFER_POINTER(P) (P) += incr
2093 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2095 # endif /* not DEFINED_ONCE */
2098 # define EXTEND_BUFFER() \
2100 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2102 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2104 bufp->allocated <<= 1; \
2105 if (bufp->allocated > MAX_BUF_SIZE) \
2106 bufp->allocated = MAX_BUF_SIZE; \
2107 /* How many characters the new buffer can have? */ \
2108 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2109 if (wchar_count == 0) wchar_count = 1; \
2110 /* Truncate the buffer to CHAR_T align. */ \
2111 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2112 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2113 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2114 if (COMPILED_BUFFER_VAR == NULL) \
2115 return REG_ESPACE; \
2116 /* If the buffer moved, move all the pointers into it. */ \
2117 if (old_buffer != COMPILED_BUFFER_VAR) \
2119 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2120 MOVE_BUFFER_POINTER (b); \
2121 MOVE_BUFFER_POINTER (begalt); \
2122 if (fixup_alt_jump) \
2123 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2125 MOVE_BUFFER_POINTER (laststart); \
2126 if (pending_exact) \
2127 MOVE_BUFFER_POINTER (pending_exact); \
2129 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2132 # define EXTEND_BUFFER() \
2134 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2135 if (bufp->allocated == MAX_BUF_SIZE) \
2137 bufp->allocated <<= 1; \
2138 if (bufp->allocated > MAX_BUF_SIZE) \
2139 bufp->allocated = MAX_BUF_SIZE; \
2140 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2142 if (COMPILED_BUFFER_VAR == NULL) \
2143 return REG_ESPACE; \
2144 /* If the buffer moved, move all the pointers into it. */ \
2145 if (old_buffer != COMPILED_BUFFER_VAR) \
2147 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2148 MOVE_BUFFER_POINTER (b); \
2149 MOVE_BUFFER_POINTER (begalt); \
2150 if (fixup_alt_jump) \
2151 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2153 MOVE_BUFFER_POINTER (laststart); \
2154 if (pending_exact) \
2155 MOVE_BUFFER_POINTER (pending_exact); \
2157 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2161 # ifndef DEFINED_ONCE
2162 /* Since we have one byte reserved for the register number argument to
2163 {start,stop}_memory, the maximum number of groups we can report
2164 things about is what fits in that byte. */
2165 # define MAX_REGNUM 255
2167 /* But patterns can have more than `MAX_REGNUM' registers. We just
2168 ignore the excess. */
2169 typedef unsigned regnum_t;
2172 /* Macros for the compile stack. */
2174 /* Since offsets can go either forwards or backwards, this type needs to
2175 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2176 /* int may be not enough when sizeof(int) == 2. */
2177 typedef long pattern_offset_t;
2181 pattern_offset_t begalt_offset;
2182 pattern_offset_t fixup_alt_jump;
2183 pattern_offset_t inner_group_offset;
2184 pattern_offset_t laststart_offset;
2186 } compile_stack_elt_t;
2191 compile_stack_elt_t *stack;
2193 unsigned avail; /* Offset of next open position. */
2194 } compile_stack_type;
2197 # define INIT_COMPILE_STACK_SIZE 32
2199 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2200 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2202 /* The next available element. */
2203 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2205 # endif /* not DEFINED_ONCE */
2207 /* Set the bit for character C in a list. */
2208 # ifndef DEFINED_ONCE
2209 # define SET_LIST_BIT(c) \
2210 (b[((unsigned char) (c)) / BYTEWIDTH] \
2211 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2212 # endif /* DEFINED_ONCE */
2214 /* Get the next unsigned number in the uncompiled pattern. */
2215 # define GET_UNSIGNED_NUMBER(num) \
2220 if (c < '0' || c > '9') \
2222 if (num <= RE_DUP_MAX) \
2226 num = num * 10 + c - '0'; \
2231 # ifndef DEFINED_ONCE
2232 # if defined _LIBC || WIDE_CHAR_SUPPORT
2233 /* The GNU C library provides support for user-defined character classes
2234 and the functions from ISO C amendement 1. */
2235 # ifdef CHARCLASS_NAME_MAX
2236 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2238 /* This shouldn't happen but some implementation might still have this
2239 problem. Use a reasonable default value. */
2240 # define CHAR_CLASS_MAX_LENGTH 256
2244 # define IS_CHAR_CLASS(string) __wctype (string)
2246 # define IS_CHAR_CLASS(string) wctype (string)
2249 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
2251 # define IS_CHAR_CLASS(string) \
2252 (STREQ (string, "alpha") || STREQ (string, "upper") \
2253 || STREQ (string, "lower") || STREQ (string, "digit") \
2254 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
2255 || STREQ (string, "space") || STREQ (string, "print") \
2256 || STREQ (string, "punct") || STREQ (string, "graph") \
2257 || STREQ (string, "cntrl") || STREQ (string, "blank"))
2259 # endif /* DEFINED_ONCE */
2261 # ifndef MATCH_MAY_ALLOCATE
2263 /* If we cannot allocate large objects within re_match_2_internal,
2264 we make the fail stack and register vectors global.
2265 The fail stack, we grow to the maximum size when a regexp
2267 The register vectors, we adjust in size each time we
2268 compile a regexp, according to the number of registers it needs. */
2270 static PREFIX(fail_stack_type) fail_stack;
2272 /* Size with which the following vectors are currently allocated.
2273 That is so we can make them bigger as needed,
2274 but never make them smaller. */
2275 # ifdef DEFINED_ONCE
2276 static int regs_allocated_size;
2278 static const char ** regstart, ** regend;
2279 static const char ** old_regstart, ** old_regend;
2280 static const char **best_regstart, **best_regend;
2281 static const char **reg_dummy;
2282 # endif /* DEFINED_ONCE */
2284 static PREFIX(register_info_type) *PREFIX(reg_info);
2285 static PREFIX(register_info_type) *PREFIX(reg_info_dummy);
2287 /* Make the register vectors big enough for NUM_REGS registers,
2288 but don't make them smaller. */
2291 PREFIX(regex_grow_registers) (num_regs)
2294 if (num_regs > regs_allocated_size)
2296 RETALLOC_IF (regstart, num_regs, const char *);
2297 RETALLOC_IF (regend, num_regs, const char *);
2298 RETALLOC_IF (old_regstart, num_regs, const char *);
2299 RETALLOC_IF (old_regend, num_regs, const char *);
2300 RETALLOC_IF (best_regstart, num_regs, const char *);
2301 RETALLOC_IF (best_regend, num_regs, const char *);
2302 RETALLOC_IF (PREFIX(reg_info), num_regs, PREFIX(register_info_type));
2303 RETALLOC_IF (reg_dummy, num_regs, const char *);
2304 RETALLOC_IF (PREFIX(reg_info_dummy), num_regs, PREFIX(register_info_type));
2306 regs_allocated_size = num_regs;
2310 # endif /* not MATCH_MAY_ALLOCATE */
2312 # ifndef DEFINED_ONCE
2313 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
2316 # endif /* not DEFINED_ONCE */
2318 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2319 Returns one of error codes defined in `regex.h', or zero for success.
2321 Assumes the `allocated' (and perhaps `buffer') and `translate'
2322 fields are set in BUFP on entry.
2324 If it succeeds, results are put in BUFP (if it returns an error, the
2325 contents of BUFP are undefined):
2326 `buffer' is the compiled pattern;
2327 `syntax' is set to SYNTAX;
2328 `used' is set to the length of the compiled pattern;
2329 `fastmap_accurate' is zero;
2330 `re_nsub' is the number of subexpressions in PATTERN;
2331 `not_bol' and `not_eol' are zero;
2333 The `fastmap' and `newline_anchor' fields are neither
2334 examined nor set. */
2336 /* Return, freeing storage we allocated. */
2338 # define FREE_STACK_RETURN(value) \
2339 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2341 # define FREE_STACK_RETURN(value) \
2342 return (free (compile_stack.stack), value)
2345 static reg_errcode_t
2346 PREFIX(regex_compile) (ARG_PREFIX(pattern), ARG_PREFIX(size), syntax, bufp)
2347 const char *ARG_PREFIX(pattern);
2348 size_t ARG_PREFIX(size);
2349 reg_syntax_t syntax;
2350 struct re_pattern_buffer *bufp;
2352 /* We fetch characters from PATTERN here. Even though PATTERN is
2353 `char *' (i.e., signed), we declare these variables as unsigned, so
2354 they can be reliably used as array indices. */
2355 register UCHAR_T c, c1;
2358 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2359 CHAR_T *pattern, *COMPILED_BUFFER_VAR;
2361 /* offset buffer for optimization. See convert_mbs_to_wc. */
2362 int *mbs_offset = NULL;
2363 /* It hold whether each wchar_t is binary data or not. */
2364 char *is_binary = NULL;
2365 /* A flag whether exactn is handling binary data or not. */
2366 char is_exactn_bin = FALSE;
2369 /* A random temporary spot in PATTERN. */
2372 /* Points to the end of the buffer, where we should append. */
2373 register UCHAR_T *b;
2375 /* Keeps track of unclosed groups. */
2376 compile_stack_type compile_stack;
2378 /* Points to the current (ending) position in the pattern. */
2383 const CHAR_T *p = pattern;
2384 const CHAR_T *pend = pattern + size;
2387 /* How to translate the characters in the pattern. */
2388 RE_TRANSLATE_TYPE translate = bufp->translate;
2390 /* Address of the count-byte of the most recently inserted `exactn'
2391 command. This makes it possible to tell if a new exact-match
2392 character can be added to that command or if the character requires
2393 a new `exactn' command. */
2394 UCHAR_T *pending_exact = 0;
2396 /* Address of start of the most recently finished expression.
2397 This tells, e.g., postfix * where to find the start of its
2398 operand. Reset at the beginning of groups and alternatives. */
2399 UCHAR_T *laststart = 0;
2401 /* Address of beginning of regexp, or inside of last group. */
2404 /* Address of the place where a forward jump should go to the end of
2405 the containing expression. Each alternative of an `or' -- except the
2406 last -- ends with a forward jump of this sort. */
2407 UCHAR_T *fixup_alt_jump = 0;
2409 /* Counts open-groups as they are encountered. Remembered for the
2410 matching close-group on the compile stack, so the same register
2411 number is put in the stop_memory as the start_memory. */
2412 regnum_t regnum = 0;
2415 /* Initialize the wchar_t PATTERN and offset_buffer. */
2416 p = pend = pattern = TALLOC(csize + 1, CHAR_T);
2417 mbs_offset = TALLOC(csize + 1, int);
2418 is_binary = TALLOC(csize + 1, char);
2419 if (pattern == NULL || mbs_offset == NULL || is_binary == NULL)
2426 pattern[csize] = L'\0'; /* sentinel */
2427 size = convert_mbs_to_wcs(pattern, cpattern, csize, mbs_offset, is_binary);
2439 DEBUG_PRINT1 ("\nCompiling pattern: ");
2442 unsigned debug_count;
2444 for (debug_count = 0; debug_count < size; debug_count++)
2445 PUT_CHAR (pattern[debug_count]);
2450 /* Initialize the compile stack. */
2451 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2452 if (compile_stack.stack == NULL)
2462 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2463 compile_stack.avail = 0;
2465 /* Initialize the pattern buffer. */
2466 bufp->syntax = syntax;
2467 bufp->fastmap_accurate = 0;
2468 bufp->not_bol = bufp->not_eol = 0;
2470 /* Set `used' to zero, so that if we return an error, the pattern
2471 printer (for debugging) will think there's no pattern. We reset it
2475 /* Always count groups, whether or not bufp->no_sub is set. */
2478 #if !defined emacs && !defined SYNTAX_TABLE
2479 /* Initialize the syntax table. */
2480 init_syntax_once ();
2483 if (bufp->allocated == 0)
2486 { /* If zero allocated, but buffer is non-null, try to realloc
2487 enough space. This loses if buffer's address is bogus, but
2488 that is the user's responsibility. */
2490 /* Free bufp->buffer and allocate an array for wchar_t pattern
2493 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE/sizeof(UCHAR_T),
2496 RETALLOC (COMPILED_BUFFER_VAR, INIT_BUF_SIZE, UCHAR_T);
2500 { /* Caller did not allocate a buffer. Do it for them. */
2501 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE / sizeof(UCHAR_T),
2505 if (!COMPILED_BUFFER_VAR) FREE_STACK_RETURN (REG_ESPACE);
2507 bufp->buffer = (char*)COMPILED_BUFFER_VAR;
2509 bufp->allocated = INIT_BUF_SIZE;
2513 COMPILED_BUFFER_VAR = (UCHAR_T*) bufp->buffer;
2516 begalt = b = COMPILED_BUFFER_VAR;
2518 /* Loop through the uncompiled pattern until we're at the end. */
2527 if ( /* If at start of pattern, it's an operator. */
2529 /* If context independent, it's an operator. */
2530 || syntax & RE_CONTEXT_INDEP_ANCHORS
2531 /* Otherwise, depends on what's come before. */
2532 || PREFIX(at_begline_loc_p) (pattern, p, syntax))
2542 if ( /* If at end of pattern, it's an operator. */
2544 /* If context independent, it's an operator. */
2545 || syntax & RE_CONTEXT_INDEP_ANCHORS
2546 /* Otherwise, depends on what's next. */
2547 || PREFIX(at_endline_loc_p) (p, pend, syntax))
2557 if ((syntax & RE_BK_PLUS_QM)
2558 || (syntax & RE_LIMITED_OPS))
2562 /* If there is no previous pattern... */
2565 if (syntax & RE_CONTEXT_INVALID_OPS)
2566 FREE_STACK_RETURN (REG_BADRPT);
2567 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2572 /* Are we optimizing this jump? */
2573 boolean keep_string_p = false;
2575 /* 1 means zero (many) matches is allowed. */
2576 char zero_times_ok = 0, many_times_ok = 0;
2578 /* If there is a sequence of repetition chars, collapse it
2579 down to just one (the right one). We can't combine
2580 interval operators with these because of, e.g., `a{2}*',
2581 which should only match an even number of `a's. */
2585 zero_times_ok |= c != '+';
2586 many_times_ok |= c != '?';
2594 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2597 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2599 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2602 if (!(c1 == '+' || c1 == '?'))
2617 /* If we get here, we found another repeat character. */
2620 /* Star, etc. applied to an empty pattern is equivalent
2621 to an empty pattern. */
2625 /* Now we know whether or not zero matches is allowed
2626 and also whether or not two or more matches is allowed. */
2628 { /* More than one repetition is allowed, so put in at the
2629 end a backward relative jump from `b' to before the next
2630 jump we're going to put in below (which jumps from
2631 laststart to after this jump).
2633 But if we are at the `*' in the exact sequence `.*\n',
2634 insert an unconditional jump backwards to the .,
2635 instead of the beginning of the loop. This way we only
2636 push a failure point once, instead of every time
2637 through the loop. */
2638 assert (p - 1 > pattern);
2640 /* Allocate the space for the jump. */
2641 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2643 /* We know we are not at the first character of the pattern,
2644 because laststart was nonzero. And we've already
2645 incremented `p', by the way, to be the character after
2646 the `*'. Do we have to do something analogous here
2647 for null bytes, because of RE_DOT_NOT_NULL? */
2648 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2650 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2651 && !(syntax & RE_DOT_NEWLINE))
2652 { /* We have .*\n. */
2653 STORE_JUMP (jump, b, laststart);
2654 keep_string_p = true;
2657 /* Anything else. */
2658 STORE_JUMP (maybe_pop_jump, b, laststart -
2659 (1 + OFFSET_ADDRESS_SIZE));
2661 /* We've added more stuff to the buffer. */
2662 b += 1 + OFFSET_ADDRESS_SIZE;
2665 /* On failure, jump from laststart to b + 3, which will be the
2666 end of the buffer after this jump is inserted. */
2667 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2669 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2670 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2672 laststart, b + 1 + OFFSET_ADDRESS_SIZE);
2674 b += 1 + OFFSET_ADDRESS_SIZE;
2678 /* At least one repetition is required, so insert a
2679 `dummy_failure_jump' before the initial
2680 `on_failure_jump' instruction of the loop. This
2681 effects a skip over that instruction the first time
2682 we hit that loop. */
2683 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2684 INSERT_JUMP (dummy_failure_jump, laststart, laststart +
2685 2 + 2 * OFFSET_ADDRESS_SIZE);
2686 b += 1 + OFFSET_ADDRESS_SIZE;
2700 boolean had_char_class = false;
2702 CHAR_T range_start = 0xffffffff;
2704 unsigned int range_start = 0xffffffff;
2706 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2709 /* We assume a charset(_not) structure as a wchar_t array.
2710 charset[0] = (re_opcode_t) charset(_not)
2711 charset[1] = l (= length of char_classes)
2712 charset[2] = m (= length of collating_symbols)
2713 charset[3] = n (= length of equivalence_classes)
2714 charset[4] = o (= length of char_ranges)
2715 charset[5] = p (= length of chars)
2717 charset[6] = char_class (wctype_t)
2718 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2720 charset[l+5] = char_class (wctype_t)
2722 charset[l+6] = collating_symbol (wchar_t)
2724 charset[l+m+5] = collating_symbol (wchar_t)
2725 ifdef _LIBC we use the index if
2726 _NL_COLLATE_SYMB_EXTRAMB instead of
2729 charset[l+m+6] = equivalence_classes (wchar_t)
2731 charset[l+m+n+5] = equivalence_classes (wchar_t)
2732 ifdef _LIBC we use the index in
2733 _NL_COLLATE_WEIGHT instead of
2736 charset[l+m+n+6] = range_start
2737 charset[l+m+n+7] = range_end
2739 charset[l+m+n+2o+4] = range_start
2740 charset[l+m+n+2o+5] = range_end
2741 ifdef _LIBC we use the value looked up
2742 in _NL_COLLATE_COLLSEQ instead of
2745 charset[l+m+n+2o+6] = char
2747 charset[l+m+n+2o+p+5] = char
2751 /* We need at least 6 spaces: the opcode, the length of
2752 char_classes, the length of collating_symbols, the length of
2753 equivalence_classes, the length of char_ranges, the length of
2755 GET_BUFFER_SPACE (6);
2757 /* Save b as laststart. And We use laststart as the pointer
2758 to the first element of the charset here.
2759 In other words, laststart[i] indicates charset[i]. */
2762 /* We test `*p == '^' twice, instead of using an if
2763 statement, so we only need one BUF_PUSH. */
2764 BUF_PUSH (*p == '^' ? charset_not : charset);
2768 /* Push the length of char_classes, the length of
2769 collating_symbols, the length of equivalence_classes, the
2770 length of char_ranges and the length of chars. */
2771 BUF_PUSH_3 (0, 0, 0);
2774 /* Remember the first position in the bracket expression. */
2777 /* charset_not matches newline according to a syntax bit. */
2778 if ((re_opcode_t) b[-6] == charset_not
2779 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2782 laststart[5]++; /* Update the length of characters */
2785 /* Read in characters and ranges, setting map bits. */
2788 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2792 /* \ might escape characters inside [...] and [^...]. */
2793 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2795 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2799 laststart[5]++; /* Update the length of chars */
2804 /* Could be the end of the bracket expression. If it's
2805 not (i.e., when the bracket expression is `[]' so
2806 far), the ']' character bit gets set way below. */
2807 if (c == ']' && p != p1 + 1)
2810 /* Look ahead to see if it's a range when the last thing
2811 was a character class. */
2812 if (had_char_class && c == '-' && *p != ']')
2813 FREE_STACK_RETURN (REG_ERANGE);
2815 /* Look ahead to see if it's a range when the last thing
2816 was a character: if this is a hyphen not at the
2817 beginning or the end of a list, then it's the range
2820 && !(p - 2 >= pattern && p[-2] == '[')
2821 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2825 /* Allocate the space for range_start and range_end. */
2826 GET_BUFFER_SPACE (2);
2827 /* Update the pointer to indicate end of buffer. */
2829 ret = wcs_compile_range (range_start, &p, pend, translate,
2830 syntax, b, laststart);
2831 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2832 range_start = 0xffffffff;
2834 else if (p[0] == '-' && p[1] != ']')
2835 { /* This handles ranges made up of characters only. */
2838 /* Move past the `-'. */
2840 /* Allocate the space for range_start and range_end. */
2841 GET_BUFFER_SPACE (2);
2842 /* Update the pointer to indicate end of buffer. */
2844 ret = wcs_compile_range (c, &p, pend, translate, syntax, b,
2846 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2847 range_start = 0xffffffff;
2850 /* See if we're at the beginning of a possible character
2852 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2853 { /* Leave room for the null. */
2854 char str[CHAR_CLASS_MAX_LENGTH + 1];
2859 /* If pattern is `[[:'. */
2860 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2865 if ((c == ':' && *p == ']') || p == pend)
2867 if (c1 < CHAR_CLASS_MAX_LENGTH)
2870 /* This is in any case an invalid class name. */
2875 /* If isn't a word bracketed by `[:' and `:]':
2876 undo the ending character, the letters, and leave
2877 the leading `:' and `[' (but store them as character). */
2878 if (c == ':' && *p == ']')
2883 /* Query the character class as wctype_t. */
2884 wt = IS_CHAR_CLASS (str);
2886 FREE_STACK_RETURN (REG_ECTYPE);
2888 /* Throw away the ] at the end of the character
2892 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2894 /* Allocate the space for character class. */
2895 GET_BUFFER_SPACE(CHAR_CLASS_SIZE);
2896 /* Update the pointer to indicate end of buffer. */
2897 b += CHAR_CLASS_SIZE;
2898 /* Move data which follow character classes
2899 not to violate the data. */
2900 insert_space(CHAR_CLASS_SIZE,
2901 laststart + 6 + laststart[1],
2903 alignedp = ((uintptr_t)(laststart + 6 + laststart[1])
2904 + __alignof__(wctype_t) - 1)
2905 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
2906 /* Store the character class. */
2907 *((wctype_t*)alignedp) = wt;
2908 /* Update length of char_classes */
2909 laststart[1] += CHAR_CLASS_SIZE;
2911 had_char_class = true;
2920 laststart[5] += 2; /* Update the length of characters */
2922 had_char_class = false;
2925 else if (syntax & RE_CHAR_CLASSES && c == '[' && (*p == '='
2928 CHAR_T str[128]; /* Should be large enough. */
2929 CHAR_T delim = *p; /* '=' or '.' */
2932 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
2937 /* If pattern is `[[=' or '[[.'. */
2938 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2943 if ((c == delim && *p == ']') || p == pend)
2945 if (c1 < sizeof (str) - 1)
2948 /* This is in any case an invalid class name. */
2953 if (c == delim && *p == ']' && str[0] != '\0')
2955 unsigned int i, offset;
2956 /* If we have no collation data we use the default
2957 collation in which each character is in a class
2958 by itself. It also means that ASCII is the
2959 character set and therefore we cannot have character
2960 with more than one byte in the multibyte
2963 /* If not defined _LIBC, we push the name and
2964 `\0' for the sake of matching performance. */
2965 int datasize = c1 + 1;
2973 FREE_STACK_RETURN (REG_ECOLLATE);
2978 const int32_t *table;
2979 const int32_t *weights;
2980 const int32_t *extra;
2981 const int32_t *indirect;
2984 /* This #include defines a local function! */
2985 # include <locale/weightwc.h>
2989 /* We push the index for equivalence class. */
2992 table = (const int32_t *)
2993 _NL_CURRENT (LC_COLLATE,
2994 _NL_COLLATE_TABLEWC);
2995 weights = (const int32_t *)
2996 _NL_CURRENT (LC_COLLATE,
2997 _NL_COLLATE_WEIGHTWC);
2998 extra = (const int32_t *)
2999 _NL_CURRENT (LC_COLLATE,
3000 _NL_COLLATE_EXTRAWC);
3001 indirect = (const int32_t *)
3002 _NL_CURRENT (LC_COLLATE,
3003 _NL_COLLATE_INDIRECTWC);
3005 idx = findidx ((const wint_t**)&cp);
3006 if (idx == 0 || cp < (wint_t*) str + c1)
3007 /* This is no valid character. */
3008 FREE_STACK_RETURN (REG_ECOLLATE);
3010 str[0] = (wchar_t)idx;
3012 else /* delim == '.' */
3014 /* We push collation sequence value
3015 for collating symbol. */
3017 const int32_t *symb_table;
3018 const unsigned char *extra;
3025 /* We have to convert the name to a single-byte
3026 string. This is possible since the names
3027 consist of ASCII characters and the internal
3028 representation is UCS4. */
3029 for (i = 0; i < c1; ++i)
3030 char_str[i] = str[i];
3033 _NL_CURRENT_WORD (LC_COLLATE,
3034 _NL_COLLATE_SYMB_HASH_SIZEMB);
3035 symb_table = (const int32_t *)
3036 _NL_CURRENT (LC_COLLATE,
3037 _NL_COLLATE_SYMB_TABLEMB);
3038 extra = (const unsigned char *)
3039 _NL_CURRENT (LC_COLLATE,
3040 _NL_COLLATE_SYMB_EXTRAMB);
3042 /* Locate the character in the hashing table. */
3043 hash = elem_hash (char_str, c1);
3046 elem = hash % table_size;
3047 second = hash % (table_size - 2);
3048 while (symb_table[2 * elem] != 0)
3050 /* First compare the hashing value. */
3051 if (symb_table[2 * elem] == hash
3052 && c1 == extra[symb_table[2 * elem + 1]]
3053 && memcmp (char_str,
3054 &extra[symb_table[2 * elem + 1]
3057 /* Yep, this is the entry. */
3058 idx = symb_table[2 * elem + 1];
3059 idx += 1 + extra[idx];
3067 if (symb_table[2 * elem] != 0)
3069 /* Compute the index of the byte sequence
3071 idx += 1 + extra[idx];
3072 /* Adjust for the alignment. */
3073 idx = (idx + 3) & ~3;
3075 str[0] = (wchar_t) idx + 4;
3077 else if (symb_table[2 * elem] == 0 && c1 == 1)
3079 /* No valid character. Match it as a
3080 single byte character. */
3081 had_char_class = false;
3083 /* Update the length of characters */
3085 range_start = str[0];
3087 /* Throw away the ] at the end of the
3088 collating symbol. */
3090 /* exit from the switch block. */
3094 FREE_STACK_RETURN (REG_ECOLLATE);
3099 /* Throw away the ] at the end of the equivalence
3100 class (or collating symbol). */
3103 /* Allocate the space for the equivalence class
3104 (or collating symbol) (and '\0' if needed). */
3105 GET_BUFFER_SPACE(datasize);
3106 /* Update the pointer to indicate end of buffer. */
3110 { /* equivalence class */
3111 /* Calculate the offset of char_ranges,
3112 which is next to equivalence_classes. */
3113 offset = laststart[1] + laststart[2]
3116 insert_space(datasize, laststart + offset, b - 1);
3118 /* Write the equivalence_class and \0. */
3119 for (i = 0 ; i < datasize ; i++)
3120 laststart[offset + i] = str[i];
3122 /* Update the length of equivalence_classes. */
3123 laststart[3] += datasize;
3124 had_char_class = true;
3126 else /* delim == '.' */
3127 { /* collating symbol */
3128 /* Calculate the offset of the equivalence_classes,
3129 which is next to collating_symbols. */
3130 offset = laststart[1] + laststart[2] + 6;
3131 /* Insert space and write the collationg_symbol
3133 insert_space(datasize, laststart + offset, b-1);
3134 for (i = 0 ; i < datasize ; i++)
3135 laststart[offset + i] = str[i];
3137 /* In re_match_2_internal if range_start < -1, we
3138 assume -range_start is the offset of the
3139 collating symbol which is specified as
3140 the character of the range start. So we assign
3141 -(laststart[1] + laststart[2] + 6) to
3143 range_start = -(laststart[1] + laststart[2] + 6);
3144 /* Update the length of collating_symbol. */
3145 laststart[2] += datasize;
3146 had_char_class = false;
3156 laststart[5] += 2; /* Update the length of characters */
3157 range_start = delim;
3158 had_char_class = false;
3163 had_char_class = false;
3165 laststart[5]++; /* Update the length of characters */
3171 /* Ensure that we have enough space to push a charset: the
3172 opcode, the length count, and the bitset; 34 bytes in all. */
3173 GET_BUFFER_SPACE (34);
3177 /* We test `*p == '^' twice, instead of using an if
3178 statement, so we only need one BUF_PUSH. */
3179 BUF_PUSH (*p == '^' ? charset_not : charset);
3183 /* Remember the first position in the bracket expression. */
3186 /* Push the number of bytes in the bitmap. */
3187 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
3189 /* Clear the whole map. */
3190 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
3192 /* charset_not matches newline according to a syntax bit. */
3193 if ((re_opcode_t) b[-2] == charset_not
3194 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
3195 SET_LIST_BIT ('\n');
3197 /* Read in characters and ranges, setting map bits. */
3200 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3204 /* \ might escape characters inside [...] and [^...]. */
3205 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
3207 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3215 /* Could be the end of the bracket expression. If it's
3216 not (i.e., when the bracket expression is `[]' so
3217 far), the ']' character bit gets set way below. */
3218 if (c == ']' && p != p1 + 1)
3221 /* Look ahead to see if it's a range when the last thing
3222 was a character class. */
3223 if (had_char_class && c == '-' && *p != ']')
3224 FREE_STACK_RETURN (REG_ERANGE);
3226 /* Look ahead to see if it's a range when the last thing
3227 was a character: if this is a hyphen not at the
3228 beginning or the end of a list, then it's the range
3231 && !(p - 2 >= pattern && p[-2] == '[')
3232 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
3236 = byte_compile_range (range_start, &p, pend, translate,
3238 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3239 range_start = 0xffffffff;
3242 else if (p[0] == '-' && p[1] != ']')
3243 { /* This handles ranges made up of characters only. */
3246 /* Move past the `-'. */
3249 ret = byte_compile_range (c, &p, pend, translate, syntax, b);
3250 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3251 range_start = 0xffffffff;
3254 /* See if we're at the beginning of a possible character
3257 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
3258 { /* Leave room for the null. */
3259 char str[CHAR_CLASS_MAX_LENGTH + 1];
3264 /* If pattern is `[[:'. */
3265 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3270 if ((c == ':' && *p == ']') || p == pend)
3272 if (c1 < CHAR_CLASS_MAX_LENGTH)
3275 /* This is in any case an invalid class name. */
3280 /* If isn't a word bracketed by `[:' and `:]':
3281 undo the ending character, the letters, and leave
3282 the leading `:' and `[' (but set bits for them). */
3283 if (c == ':' && *p == ']')
3285 # if defined _LIBC || WIDE_CHAR_SUPPORT
3286 boolean is_lower = STREQ (str, "lower");
3287 boolean is_upper = STREQ (str, "upper");
3291 wt = IS_CHAR_CLASS (str);
3293 FREE_STACK_RETURN (REG_ECTYPE);
3295 /* Throw away the ] at the end of the character
3299 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3301 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
3304 if (__iswctype (__btowc (ch), wt))
3307 if (iswctype (btowc (ch), wt))
3311 if (translate && (is_upper || is_lower)
3312 && (ISUPPER (ch) || ISLOWER (ch)))
3316 had_char_class = true;
3319 boolean is_alnum = STREQ (str, "alnum");
3320 boolean is_alpha = STREQ (str, "alpha");
3321 boolean is_blank = STREQ (str, "blank");
3322 boolean is_cntrl = STREQ (str, "cntrl");
3323 boolean is_digit = STREQ (str, "digit");
3324 boolean is_graph = STREQ (str, "graph");
3325 boolean is_lower = STREQ (str, "lower");
3326 boolean is_print = STREQ (str, "print");
3327 boolean is_punct = STREQ (str, "punct");
3328 boolean is_space = STREQ (str, "space");
3329 boolean is_upper = STREQ (str, "upper");
3330 boolean is_xdigit = STREQ (str, "xdigit");
3332 if (!IS_CHAR_CLASS (str))
3333 FREE_STACK_RETURN (REG_ECTYPE);
3335 /* Throw away the ] at the end of the character
3339 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3341 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
3343 /* This was split into 3 if's to
3344 avoid an arbitrary limit in some compiler. */
3345 if ( (is_alnum && ISALNUM (ch))
3346 || (is_alpha && ISALPHA (ch))
3347 || (is_blank && ISBLANK (ch))
3348 || (is_cntrl && ISCNTRL (ch)))
3350 if ( (is_digit && ISDIGIT (ch))
3351 || (is_graph && ISGRAPH (ch))
3352 || (is_lower && ISLOWER (ch))
3353 || (is_print && ISPRINT (ch)))
3355 if ( (is_punct && ISPUNCT (ch))
3356 || (is_space && ISSPACE (ch))
3357 || (is_upper && ISUPPER (ch))
3358 || (is_xdigit && ISXDIGIT (ch)))
3360 if ( translate && (is_upper || is_lower)
3361 && (ISUPPER (ch) || ISLOWER (ch)))
3364 had_char_class = true;
3365 # endif /* libc || wctype.h */
3375 had_char_class = false;
3378 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '=')
3380 unsigned char str[MB_LEN_MAX + 1];
3383 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3389 /* If pattern is `[[='. */
3390 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3395 if ((c == '=' && *p == ']') || p == pend)
3397 if (c1 < MB_LEN_MAX)
3400 /* This is in any case an invalid class name. */
3405 if (c == '=' && *p == ']' && str[0] != '\0')
3407 /* If we have no collation data we use the default
3408 collation in which each character is in a class
3409 by itself. It also means that ASCII is the
3410 character set and therefore we cannot have character
3411 with more than one byte in the multibyte
3418 FREE_STACK_RETURN (REG_ECOLLATE);
3420 /* Throw away the ] at the end of the equivalence
3424 /* Set the bit for the character. */
3425 SET_LIST_BIT (str[0]);
3430 /* Try to match the byte sequence in `str' against
3431 those known to the collate implementation.
3432 First find out whether the bytes in `str' are
3433 actually from exactly one character. */
3434 const int32_t *table;
3435 const unsigned char *weights;
3436 const unsigned char *extra;
3437 const int32_t *indirect;
3439 const unsigned char *cp = str;
3442 /* This #include defines a local function! */
3443 # include <locale/weight.h>
3445 table = (const int32_t *)
3446 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
3447 weights = (const unsigned char *)
3448 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
3449 extra = (const unsigned char *)
3450 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
3451 indirect = (const int32_t *)
3452 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
3454 idx = findidx (&cp);
3455 if (idx == 0 || cp < str + c1)
3456 /* This is no valid character. */
3457 FREE_STACK_RETURN (REG_ECOLLATE);
3459 /* Throw away the ] at the end of the equivalence
3463 /* Now we have to go throught the whole table
3464 and find all characters which have the same
3467 XXX Note that this is not entirely correct.
3468 we would have to match multibyte sequences
3469 but this is not possible with the current
3471 for (ch = 1; ch < 256; ++ch)
3472 /* XXX This test would have to be changed if we
3473 would allow matching multibyte sequences. */
3476 int32_t idx2 = table[ch];
3477 size_t len = weights[idx2];
3479 /* Test whether the lenghts match. */
3480 if (weights[idx] == len)
3482 /* They do. New compare the bytes of
3487 && (weights[idx + 1 + cnt]
3488 == weights[idx2 + 1 + cnt]))
3492 /* They match. Mark the character as
3499 had_char_class = true;
3509 had_char_class = false;
3512 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '.')
3514 unsigned char str[128]; /* Should be large enough. */
3517 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3523 /* If pattern is `[[.'. */
3524 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3529 if ((c == '.' && *p == ']') || p == pend)
3531 if (c1 < sizeof (str))
3534 /* This is in any case an invalid class name. */
3539 if (c == '.' && *p == ']' && str[0] != '\0')
3541 /* If we have no collation data we use the default
3542 collation in which each character is the name
3543 for its own class which contains only the one
3544 character. It also means that ASCII is the
3545 character set and therefore we cannot have character
3546 with more than one byte in the multibyte
3553 FREE_STACK_RETURN (REG_ECOLLATE);
3555 /* Throw away the ] at the end of the equivalence
3559 /* Set the bit for the character. */
3560 SET_LIST_BIT (str[0]);
3561 range_start = ((const unsigned char *) str)[0];
3566 /* Try to match the byte sequence in `str' against
3567 those known to the collate implementation.
3568 First find out whether the bytes in `str' are
3569 actually from exactly one character. */
3571 const int32_t *symb_table;
3572 const unsigned char *extra;
3579 _NL_CURRENT_WORD (LC_COLLATE,
3580 _NL_COLLATE_SYMB_HASH_SIZEMB);
3581 symb_table = (const int32_t *)
3582 _NL_CURRENT (LC_COLLATE,
3583 _NL_COLLATE_SYMB_TABLEMB);
3584 extra = (const unsigned char *)
3585 _NL_CURRENT (LC_COLLATE,
3586 _NL_COLLATE_SYMB_EXTRAMB);
3588 /* Locate the character in the hashing table. */
3589 hash = elem_hash (str, c1);
3592 elem = hash % table_size;
3593 second = hash % (table_size - 2);
3594 while (symb_table[2 * elem] != 0)
3596 /* First compare the hashing value. */
3597 if (symb_table[2 * elem] == hash
3598 && c1 == extra[symb_table[2 * elem + 1]]
3600 &extra[symb_table[2 * elem + 1]
3604 /* Yep, this is the entry. */
3605 idx = symb_table[2 * elem + 1];
3606 idx += 1 + extra[idx];
3614 if (symb_table[2 * elem] == 0)
3615 /* This is no valid character. */
3616 FREE_STACK_RETURN (REG_ECOLLATE);
3618 /* Throw away the ] at the end of the equivalence
3622 /* Now add the multibyte character(s) we found
3625 XXX Note that this is not entirely correct.
3626 we would have to match multibyte sequences
3627 but this is not possible with the current
3628 implementation. Also, we have to match
3629 collating symbols, which expand to more than
3630 one file, as a whole and not allow the
3631 individual bytes. */
3634 range_start = extra[idx];
3637 SET_LIST_BIT (extra[idx]);
3642 had_char_class = false;
3652 had_char_class = false;
3657 had_char_class = false;
3663 /* Discard any (non)matching list bytes that are all 0 at the
3664 end of the map. Decrease the map-length byte too. */
3665 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
3674 if (syntax & RE_NO_BK_PARENS)
3681 if (syntax & RE_NO_BK_PARENS)
3688 if (syntax & RE_NEWLINE_ALT)
3695 if (syntax & RE_NO_BK_VBAR)
3702 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
3703 goto handle_interval;
3709 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3711 /* Do not translate the character after the \, so that we can
3712 distinguish, e.g., \B from \b, even if we normally would
3713 translate, e.g., B to b. */
3719 if (syntax & RE_NO_BK_PARENS)
3720 goto normal_backslash;
3726 if (COMPILE_STACK_FULL)
3728 RETALLOC (compile_stack.stack, compile_stack.size << 1,
3729 compile_stack_elt_t);
3730 if (compile_stack.stack == NULL) return REG_ESPACE;
3732 compile_stack.size <<= 1;
3735 /* These are the values to restore when we hit end of this
3736 group. They are all relative offsets, so that if the
3737 whole pattern moves because of realloc, they will still
3739 COMPILE_STACK_TOP.begalt_offset = begalt - COMPILED_BUFFER_VAR;
3740 COMPILE_STACK_TOP.fixup_alt_jump
3741 = fixup_alt_jump ? fixup_alt_jump - COMPILED_BUFFER_VAR + 1 : 0;
3742 COMPILE_STACK_TOP.laststart_offset = b - COMPILED_BUFFER_VAR;
3743 COMPILE_STACK_TOP.regnum = regnum;
3745 /* We will eventually replace the 0 with the number of
3746 groups inner to this one. But do not push a
3747 start_memory for groups beyond the last one we can
3748 represent in the compiled pattern. */
3749 if (regnum <= MAX_REGNUM)
3751 COMPILE_STACK_TOP.inner_group_offset = b
3752 - COMPILED_BUFFER_VAR + 2;
3753 BUF_PUSH_3 (start_memory, regnum, 0);
3756 compile_stack.avail++;
3761 /* If we've reached MAX_REGNUM groups, then this open
3762 won't actually generate any code, so we'll have to
3763 clear pending_exact explicitly. */
3769 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
3771 if (COMPILE_STACK_EMPTY)
3773 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3774 goto normal_backslash;
3776 FREE_STACK_RETURN (REG_ERPAREN);
3781 { /* Push a dummy failure point at the end of the
3782 alternative for a possible future
3783 `pop_failure_jump' to pop. See comments at
3784 `push_dummy_failure' in `re_match_2'. */
3785 BUF_PUSH (push_dummy_failure);
3787 /* We allocated space for this jump when we assigned
3788 to `fixup_alt_jump', in the `handle_alt' case below. */
3789 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
3792 /* See similar code for backslashed left paren above. */
3793 if (COMPILE_STACK_EMPTY)
3795 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3798 FREE_STACK_RETURN (REG_ERPAREN);
3801 /* Since we just checked for an empty stack above, this
3802 ``can't happen''. */
3803 assert (compile_stack.avail != 0);
3805 /* We don't just want to restore into `regnum', because
3806 later groups should continue to be numbered higher,
3807 as in `(ab)c(de)' -- the second group is #2. */
3808 regnum_t this_group_regnum;
3810 compile_stack.avail--;
3811 begalt = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.begalt_offset;
3813 = COMPILE_STACK_TOP.fixup_alt_jump
3814 ? COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.fixup_alt_jump - 1
3816 laststart = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.laststart_offset;
3817 this_group_regnum = COMPILE_STACK_TOP.regnum;
3818 /* If we've reached MAX_REGNUM groups, then this open
3819 won't actually generate any code, so we'll have to
3820 clear pending_exact explicitly. */
3823 /* We're at the end of the group, so now we know how many
3824 groups were inside this one. */
3825 if (this_group_regnum <= MAX_REGNUM)
3827 UCHAR_T *inner_group_loc
3828 = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.inner_group_offset;
3830 *inner_group_loc = regnum - this_group_regnum;
3831 BUF_PUSH_3 (stop_memory, this_group_regnum,
3832 regnum - this_group_regnum);
3838 case '|': /* `\|'. */
3839 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
3840 goto normal_backslash;
3842 if (syntax & RE_LIMITED_OPS)
3845 /* Insert before the previous alternative a jump which
3846 jumps to this alternative if the former fails. */
3847 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3848 INSERT_JUMP (on_failure_jump, begalt,
3849 b + 2 + 2 * OFFSET_ADDRESS_SIZE);
3851 b += 1 + OFFSET_ADDRESS_SIZE;
3853 /* The alternative before this one has a jump after it
3854 which gets executed if it gets matched. Adjust that
3855 jump so it will jump to this alternative's analogous
3856 jump (put in below, which in turn will jump to the next
3857 (if any) alternative's such jump, etc.). The last such
3858 jump jumps to the correct final destination. A picture:
3864 If we are at `b', then fixup_alt_jump right now points to a
3865 three-byte space after `a'. We'll put in the jump, set
3866 fixup_alt_jump to right after `b', and leave behind three
3867 bytes which we'll fill in when we get to after `c'. */
3870 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
3872 /* Mark and leave space for a jump after this alternative,
3873 to be filled in later either by next alternative or
3874 when know we're at the end of a series of alternatives. */
3876 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3877 b += 1 + OFFSET_ADDRESS_SIZE;
3885 /* If \{ is a literal. */
3886 if (!(syntax & RE_INTERVALS)
3887 /* If we're at `\{' and it's not the open-interval
3889 || (syntax & RE_NO_BK_BRACES))
3890 goto normal_backslash;
3894 /* If got here, then the syntax allows intervals. */
3896 /* At least (most) this many matches must be made. */
3897 int lower_bound = -1, upper_bound = -1;
3899 /* Place in the uncompiled pattern (i.e., just after
3900 the '{') to go back to if the interval is invalid. */
3901 const CHAR_T *beg_interval = p;
3904 goto invalid_interval;
3906 GET_UNSIGNED_NUMBER (lower_bound);
3910 GET_UNSIGNED_NUMBER (upper_bound);
3911 if (upper_bound < 0)
3912 upper_bound = RE_DUP_MAX;
3915 /* Interval such as `{1}' => match exactly once. */
3916 upper_bound = lower_bound;
3918 if (! (0 <= lower_bound && lower_bound <= upper_bound))
3919 goto invalid_interval;
3921 if (!(syntax & RE_NO_BK_BRACES))
3923 if (c != '\\' || p == pend)
3924 goto invalid_interval;
3929 goto invalid_interval;
3931 /* If it's invalid to have no preceding re. */
3934 if (syntax & RE_CONTEXT_INVALID_OPS
3935 && !(syntax & RE_INVALID_INTERVAL_ORD))
3936 FREE_STACK_RETURN (REG_BADRPT);
3937 else if (syntax & RE_CONTEXT_INDEP_OPS)
3940 goto unfetch_interval;
3943 /* We just parsed a valid interval. */
3945 if (RE_DUP_MAX < upper_bound)
3946 FREE_STACK_RETURN (REG_BADBR);
3948 /* If the upper bound is zero, don't want to succeed at
3949 all; jump from `laststart' to `b + 3', which will be
3950 the end of the buffer after we insert the jump. */
3951 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3952 instead of 'b + 3'. */
3953 if (upper_bound == 0)
3955 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3956 INSERT_JUMP (jump, laststart, b + 1
3957 + OFFSET_ADDRESS_SIZE);
3958 b += 1 + OFFSET_ADDRESS_SIZE;
3961 /* Otherwise, we have a nontrivial interval. When
3962 we're all done, the pattern will look like:
3963 set_number_at <jump count> <upper bound>
3964 set_number_at <succeed_n count> <lower bound>
3965 succeed_n <after jump addr> <succeed_n count>
3967 jump_n <succeed_n addr> <jump count>
3968 (The upper bound and `jump_n' are omitted if
3969 `upper_bound' is 1, though.) */
3971 { /* If the upper bound is > 1, we need to insert
3972 more at the end of the loop. */
3973 unsigned nbytes = 2 + 4 * OFFSET_ADDRESS_SIZE +
3974 (upper_bound > 1) * (2 + 4 * OFFSET_ADDRESS_SIZE);
3976 GET_BUFFER_SPACE (nbytes);
3978 /* Initialize lower bound of the `succeed_n', even
3979 though it will be set during matching by its
3980 attendant `set_number_at' (inserted next),
3981 because `re_compile_fastmap' needs to know.
3982 Jump to the `jump_n' we might insert below. */
3983 INSERT_JUMP2 (succeed_n, laststart,
3984 b + 1 + 2 * OFFSET_ADDRESS_SIZE
3985 + (upper_bound > 1) * (1 + 2 * OFFSET_ADDRESS_SIZE)
3987 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3989 /* Code to initialize the lower bound. Insert
3990 before the `succeed_n'. The `5' is the last two
3991 bytes of this `set_number_at', plus 3 bytes of
3992 the following `succeed_n'. */
3993 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3994 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3995 of the following `succeed_n'. */
3996 PREFIX(insert_op2) (set_number_at, laststart, 1
3997 + 2 * OFFSET_ADDRESS_SIZE, lower_bound, b);
3998 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4000 if (upper_bound > 1)
4001 { /* More than one repetition is allowed, so
4002 append a backward jump to the `succeed_n'
4003 that starts this interval.
4005 When we've reached this during matching,
4006 we'll have matched the interval once, so
4007 jump back only `upper_bound - 1' times. */
4008 STORE_JUMP2 (jump_n, b, laststart
4009 + 2 * OFFSET_ADDRESS_SIZE + 1,
4011 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4013 /* The location we want to set is the second
4014 parameter of the `jump_n'; that is `b-2' as
4015 an absolute address. `laststart' will be
4016 the `set_number_at' we're about to insert;
4017 `laststart+3' the number to set, the source
4018 for the relative address. But we are
4019 inserting into the middle of the pattern --
4020 so everything is getting moved up by 5.
4021 Conclusion: (b - 2) - (laststart + 3) + 5,
4022 i.e., b - laststart.
4024 We insert this at the beginning of the loop
4025 so that if we fail during matching, we'll
4026 reinitialize the bounds. */
4027 PREFIX(insert_op2) (set_number_at, laststart,
4029 upper_bound - 1, b);
4030 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4037 if (!(syntax & RE_INVALID_INTERVAL_ORD))
4038 FREE_STACK_RETURN (p == pend ? REG_EBRACE : REG_BADBR);
4040 /* Match the characters as literals. */
4043 if (syntax & RE_NO_BK_BRACES)
4046 goto normal_backslash;
4050 /* There is no way to specify the before_dot and after_dot
4051 operators. rms says this is ok. --karl */
4059 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
4065 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
4071 if (syntax & RE_NO_GNU_OPS)
4074 BUF_PUSH (wordchar);
4079 if (syntax & RE_NO_GNU_OPS)
4082 BUF_PUSH (notwordchar);
4087 if (syntax & RE_NO_GNU_OPS)
4093 if (syntax & RE_NO_GNU_OPS)
4099 if (syntax & RE_NO_GNU_OPS)
4101 BUF_PUSH (wordbound);
4105 if (syntax & RE_NO_GNU_OPS)
4107 BUF_PUSH (notwordbound);
4111 if (syntax & RE_NO_GNU_OPS)
4117 if (syntax & RE_NO_GNU_OPS)
4122 case '1': case '2': case '3': case '4': case '5':
4123 case '6': case '7': case '8': case '9':
4124 if (syntax & RE_NO_BK_REFS)
4130 FREE_STACK_RETURN (REG_ESUBREG);
4132 /* Can't back reference to a subexpression if inside of it. */
4133 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
4137 BUF_PUSH_2 (duplicate, c1);
4143 if (syntax & RE_BK_PLUS_QM)
4146 goto normal_backslash;
4150 /* You might think it would be useful for \ to mean
4151 not to translate; but if we don't translate it
4152 it will never match anything. */
4160 /* Expects the character in `c'. */
4162 /* If no exactn currently being built. */
4165 /* If last exactn handle binary(or character) and
4166 new exactn handle character(or binary). */
4167 || is_exactn_bin != is_binary[p - 1 - pattern]
4170 /* If last exactn not at current position. */
4171 || pending_exact + *pending_exact + 1 != b
4173 /* We have only one byte following the exactn for the count. */
4174 || *pending_exact == (1 << BYTEWIDTH) - 1
4176 /* If followed by a repetition operator. */
4177 || *p == '*' || *p == '^'
4178 || ((syntax & RE_BK_PLUS_QM)
4179 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
4180 : (*p == '+' || *p == '?'))
4181 || ((syntax & RE_INTERVALS)
4182 && ((syntax & RE_NO_BK_BRACES)
4184 : (p[0] == '\\' && p[1] == '{'))))
4186 /* Start building a new exactn. */
4191 /* Is this exactn binary data or character? */
4192 is_exactn_bin = is_binary[p - 1 - pattern];
4194 BUF_PUSH_2 (exactn_bin, 0);
4196 BUF_PUSH_2 (exactn, 0);
4198 BUF_PUSH_2 (exactn, 0);
4200 pending_exact = b - 1;
4207 } /* while p != pend */
4210 /* Through the pattern now. */
4213 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
4215 if (!COMPILE_STACK_EMPTY)
4216 FREE_STACK_RETURN (REG_EPAREN);
4218 /* If we don't want backtracking, force success
4219 the first time we reach the end of the compiled pattern. */
4220 if (syntax & RE_NO_POSIX_BACKTRACKING)
4228 free (compile_stack.stack);
4230 /* We have succeeded; set the length of the buffer. */
4232 bufp->used = (uintptr_t) b - (uintptr_t) COMPILED_BUFFER_VAR;
4234 bufp->used = b - bufp->buffer;
4240 DEBUG_PRINT1 ("\nCompiled pattern: \n");
4241 PREFIX(print_compiled_pattern) (bufp);
4245 #ifndef MATCH_MAY_ALLOCATE
4246 /* Initialize the failure stack to the largest possible stack. This
4247 isn't necessary unless we're trying to avoid calling alloca in
4248 the search and match routines. */
4250 int num_regs = bufp->re_nsub + 1;
4252 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4253 is strictly greater than re_max_failures, the largest possible stack
4254 is 2 * re_max_failures failure points. */
4255 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
4257 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
4260 if (! fail_stack.stack)
4262 = (PREFIX(fail_stack_elt_t) *) xmalloc (fail_stack.size
4263 * sizeof (PREFIX(fail_stack_elt_t)));
4266 = (PREFIX(fail_stack_elt_t) *) xrealloc (fail_stack.stack,
4268 * sizeof (PREFIX(fail_stack_elt_t))));
4269 # else /* not emacs */
4270 if (! fail_stack.stack)
4272 = (PREFIX(fail_stack_elt_t) *) malloc (fail_stack.size
4273 * sizeof (PREFIX(fail_stack_elt_t)));
4276 = (PREFIX(fail_stack_elt_t) *) realloc (fail_stack.stack,
4278 * sizeof (PREFIX(fail_stack_elt_t))));
4279 # endif /* not emacs */
4282 PREFIX(regex_grow_registers) (num_regs);
4284 #endif /* not MATCH_MAY_ALLOCATE */
4287 } /* regex_compile */
4289 /* Subroutines for `regex_compile'. */
4291 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4292 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4295 PREFIX(store_op1) (op, loc, arg)
4300 *loc = (UCHAR_T) op;
4301 STORE_NUMBER (loc + 1, arg);
4305 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4306 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4309 PREFIX(store_op2) (op, loc, arg1, arg2)
4314 *loc = (UCHAR_T) op;
4315 STORE_NUMBER (loc + 1, arg1);
4316 STORE_NUMBER (loc + 1 + OFFSET_ADDRESS_SIZE, arg2);
4320 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4321 for OP followed by two-byte integer parameter ARG. */
4322 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4325 PREFIX(insert_op1) (op, loc, arg, end)
4331 register UCHAR_T *pfrom = end;
4332 register UCHAR_T *pto = end + 1 + OFFSET_ADDRESS_SIZE;
4334 while (pfrom != loc)
4337 PREFIX(store_op1) (op, loc, arg);
4341 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4342 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4345 PREFIX(insert_op2) (op, loc, arg1, arg2, end)
4351 register UCHAR_T *pfrom = end;
4352 register UCHAR_T *pto = end + 1 + 2 * OFFSET_ADDRESS_SIZE;
4354 while (pfrom != loc)
4357 PREFIX(store_op2) (op, loc, arg1, arg2);
4361 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4362 after an alternative or a begin-subexpression. We assume there is at
4363 least one character before the ^. */
4366 PREFIX(at_begline_loc_p) (pattern, p, syntax)
4367 const CHAR_T *pattern, *p;
4368 reg_syntax_t syntax;
4370 const CHAR_T *prev = p - 2;
4371 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
4374 /* After a subexpression? */
4375 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
4376 /* After an alternative? */
4377 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
4381 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4382 at least one character after the $, i.e., `P < PEND'. */
4385 PREFIX(at_endline_loc_p) (p, pend, syntax)
4386 const CHAR_T *p, *pend;
4387 reg_syntax_t syntax;
4389 const CHAR_T *next = p;
4390 boolean next_backslash = *next == '\\';
4391 const CHAR_T *next_next = p + 1 < pend ? p + 1 : 0;
4394 /* Before a subexpression? */
4395 (syntax & RE_NO_BK_PARENS ? *next == ')'
4396 : next_backslash && next_next && *next_next == ')')
4397 /* Before an alternative? */
4398 || (syntax & RE_NO_BK_VBAR ? *next == '|'
4399 : next_backslash && next_next && *next_next == '|');
4402 #else /* not INSIDE_RECURSION */
4404 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4405 false if it's not. */
4408 group_in_compile_stack (compile_stack, regnum)
4409 compile_stack_type compile_stack;
4414 for (this_element = compile_stack.avail - 1;
4417 if (compile_stack.stack[this_element].regnum == regnum)
4422 #endif /* not INSIDE_RECURSION */
4424 #ifdef INSIDE_RECURSION
4427 /* This insert space, which size is "num", into the pattern at "loc".
4428 "end" must point the end of the allocated buffer. */
4430 insert_space (num, loc, end)
4435 register CHAR_T *pto = end;
4436 register CHAR_T *pfrom = end - num;
4438 while (pfrom >= loc)
4444 static reg_errcode_t
4445 wcs_compile_range (range_start_char, p_ptr, pend, translate, syntax, b,
4447 CHAR_T range_start_char;
4448 const CHAR_T **p_ptr, *pend;
4449 CHAR_T *char_set, *b;
4450 RE_TRANSLATE_TYPE translate;
4451 reg_syntax_t syntax;
4453 const CHAR_T *p = *p_ptr;
4454 CHAR_T range_start, range_end;
4458 uint32_t start_val, end_val;
4464 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
4467 const char *collseq = (const char *) _NL_CURRENT(LC_COLLATE,
4468 _NL_COLLATE_COLLSEQWC);
4469 const unsigned char *extra = (const unsigned char *)
4470 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
4472 if (range_start_char < -1)
4474 /* range_start is a collating symbol. */
4476 /* Retreive the index and get collation sequence value. */
4477 wextra = (int32_t*)(extra + char_set[-range_start_char]);
4478 start_val = wextra[1 + *wextra];
4481 start_val = collseq_table_lookup(collseq, TRANSLATE(range_start_char));
4483 end_val = collseq_table_lookup (collseq, TRANSLATE (p[0]));
4485 /* Report an error if the range is empty and the syntax prohibits
4487 ret = ((syntax & RE_NO_EMPTY_RANGES)
4488 && (start_val > end_val))? REG_ERANGE : REG_NOERROR;
4490 /* Insert space to the end of the char_ranges. */
4491 insert_space(2, b - char_set[5] - 2, b - 1);
4492 *(b - char_set[5] - 2) = (wchar_t)start_val;
4493 *(b - char_set[5] - 1) = (wchar_t)end_val;
4494 char_set[4]++; /* ranges_index */
4499 range_start = (range_start_char >= 0)? TRANSLATE (range_start_char):
4501 range_end = TRANSLATE (p[0]);
4502 /* Report an error if the range is empty and the syntax prohibits
4504 ret = ((syntax & RE_NO_EMPTY_RANGES)
4505 && (range_start > range_end))? REG_ERANGE : REG_NOERROR;
4507 /* Insert space to the end of the char_ranges. */
4508 insert_space(2, b - char_set[5] - 2, b - 1);
4509 *(b - char_set[5] - 2) = range_start;
4510 *(b - char_set[5] - 1) = range_end;
4511 char_set[4]++; /* ranges_index */
4513 /* Have to increment the pointer into the pattern string, so the
4514 caller isn't still at the ending character. */
4520 /* Read the ending character of a range (in a bracket expression) from the
4521 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4522 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4523 Then we set the translation of all bits between the starting and
4524 ending characters (inclusive) in the compiled pattern B.
4526 Return an error code.
4528 We use these short variable names so we can use the same macros as
4529 `regex_compile' itself. */
4531 static reg_errcode_t
4532 byte_compile_range (range_start_char, p_ptr, pend, translate, syntax, b)
4533 unsigned int range_start_char;
4534 const char **p_ptr, *pend;
4535 RE_TRANSLATE_TYPE translate;
4536 reg_syntax_t syntax;
4540 const char *p = *p_ptr;
4543 const unsigned char *collseq;
4544 unsigned int start_colseq;
4545 unsigned int end_colseq;
4553 /* Have to increment the pointer into the pattern string, so the
4554 caller isn't still at the ending character. */
4557 /* Report an error if the range is empty and the syntax prohibits this. */
4558 ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
4561 collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
4562 _NL_COLLATE_COLLSEQMB);
4564 start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
4565 end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
4566 for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
4568 unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
4570 if (start_colseq <= this_colseq && this_colseq <= end_colseq)
4572 SET_LIST_BIT (TRANSLATE (this_char));
4577 /* Here we see why `this_char' has to be larger than an `unsigned
4578 char' -- we would otherwise go into an infinite loop, since all
4579 characters <= 0xff. */
4580 range_start_char = TRANSLATE (range_start_char);
4581 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4582 and some compilers cast it to int implicitly, so following for_loop
4583 may fall to (almost) infinite loop.
4584 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4585 To avoid this, we cast p[0] to unsigned int and truncate it. */
4586 end_char = ((unsigned)TRANSLATE(p[0]) & ((1 << BYTEWIDTH) - 1));
4588 for (this_char = range_start_char; this_char <= end_char; ++this_char)
4590 SET_LIST_BIT (TRANSLATE (this_char));
4599 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4600 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4601 characters can start a string that matches the pattern. This fastmap
4602 is used by re_search to skip quickly over impossible starting points.
4604 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4605 area as BUFP->fastmap.
4607 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4610 Returns 0 if we succeed, -2 if an internal error. */
4613 /* local function for re_compile_fastmap.
4614 truncate wchar_t character to char. */
4615 static unsigned char truncate_wchar (CHAR_T c);
4617 static unsigned char
4621 unsigned char buf[MB_LEN_MAX];
4622 int retval = wctomb(buf, c);
4623 return retval > 0 ? buf[0] : (unsigned char)c;
4628 PREFIX(re_compile_fastmap) (bufp)
4629 struct re_pattern_buffer *bufp;
4632 #ifdef MATCH_MAY_ALLOCATE
4633 PREFIX(fail_stack_type) fail_stack;
4635 #ifndef REGEX_MALLOC
4639 register char *fastmap = bufp->fastmap;
4642 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4643 pattern to (char*) in regex_compile. */
4644 UCHAR_T *pattern = (UCHAR_T*)bufp->buffer;
4645 register UCHAR_T *pend = (UCHAR_T*) (bufp->buffer + bufp->used);
4647 UCHAR_T *pattern = bufp->buffer;
4648 register UCHAR_T *pend = pattern + bufp->used;
4650 UCHAR_T *p = pattern;
4653 /* This holds the pointer to the failure stack, when
4654 it is allocated relocatably. */
4655 fail_stack_elt_t *failure_stack_ptr;
4658 /* Assume that each path through the pattern can be null until
4659 proven otherwise. We set this false at the bottom of switch
4660 statement, to which we get only if a particular path doesn't
4661 match the empty string. */
4662 boolean path_can_be_null = true;
4664 /* We aren't doing a `succeed_n' to begin with. */
4665 boolean succeed_n_p = false;
4667 assert (fastmap != NULL && p != NULL);
4670 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
4671 bufp->fastmap_accurate = 1; /* It will be when we're done. */
4672 bufp->can_be_null = 0;
4676 if (p == pend || *p == succeed)
4678 /* We have reached the (effective) end of pattern. */
4679 if (!FAIL_STACK_EMPTY ())
4681 bufp->can_be_null |= path_can_be_null;
4683 /* Reset for next path. */
4684 path_can_be_null = true;
4686 p = fail_stack.stack[--fail_stack.avail].pointer;
4694 /* We should never be about to go beyond the end of the pattern. */
4697 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4700 /* I guess the idea here is to simply not bother with a fastmap
4701 if a backreference is used, since it's too hard to figure out
4702 the fastmap for the corresponding group. Setting
4703 `can_be_null' stops `re_search_2' from using the fastmap, so
4704 that is all we do. */
4706 bufp->can_be_null = 1;
4710 /* Following are the cases which match a character. These end
4715 fastmap[truncate_wchar(p[1])] = 1;
4729 /* It is hard to distinguish fastmap from (multi byte) characters
4730 which depends on current locale. */
4735 bufp->can_be_null = 1;
4739 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4740 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
4746 /* Chars beyond end of map must be allowed. */
4747 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
4750 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4751 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
4757 for (j = 0; j < (1 << BYTEWIDTH); j++)
4758 if (SYNTAX (j) == Sword)
4764 for (j = 0; j < (1 << BYTEWIDTH); j++)
4765 if (SYNTAX (j) != Sword)
4772 int fastmap_newline = fastmap['\n'];
4774 /* `.' matches anything ... */
4775 for (j = 0; j < (1 << BYTEWIDTH); j++)
4778 /* ... except perhaps newline. */
4779 if (!(bufp->syntax & RE_DOT_NEWLINE))
4780 fastmap['\n'] = fastmap_newline;
4782 /* Return if we have already set `can_be_null'; if we have,
4783 then the fastmap is irrelevant. Something's wrong here. */
4784 else if (bufp->can_be_null)
4787 /* Otherwise, have to check alternative paths. */
4794 for (j = 0; j < (1 << BYTEWIDTH); j++)
4795 if (SYNTAX (j) == (enum syntaxcode) k)
4802 for (j = 0; j < (1 << BYTEWIDTH); j++)
4803 if (SYNTAX (j) != (enum syntaxcode) k)
4808 /* All cases after this match the empty string. These end with
4828 case push_dummy_failure:
4833 case pop_failure_jump:
4834 case maybe_pop_jump:
4837 case dummy_failure_jump:
4838 EXTRACT_NUMBER_AND_INCR (j, p);
4843 /* Jump backward implies we just went through the body of a
4844 loop and matched nothing. Opcode jumped to should be
4845 `on_failure_jump' or `succeed_n'. Just treat it like an
4846 ordinary jump. For a * loop, it has pushed its failure
4847 point already; if so, discard that as redundant. */
4848 if ((re_opcode_t) *p != on_failure_jump
4849 && (re_opcode_t) *p != succeed_n)
4853 EXTRACT_NUMBER_AND_INCR (j, p);
4856 /* If what's on the stack is where we are now, pop it. */
4857 if (!FAIL_STACK_EMPTY ()
4858 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
4864 case on_failure_jump:
4865 case on_failure_keep_string_jump:
4866 handle_on_failure_jump:
4867 EXTRACT_NUMBER_AND_INCR (j, p);
4869 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4870 end of the pattern. We don't want to push such a point,
4871 since when we restore it above, entering the switch will
4872 increment `p' past the end of the pattern. We don't need
4873 to push such a point since we obviously won't find any more
4874 fastmap entries beyond `pend'. Such a pattern can match
4875 the null string, though. */
4878 if (!PUSH_PATTERN_OP (p + j, fail_stack))
4880 RESET_FAIL_STACK ();
4885 bufp->can_be_null = 1;
4889 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
4890 succeed_n_p = false;
4897 /* Get to the number of times to succeed. */
4898 p += OFFSET_ADDRESS_SIZE;
4900 /* Increment p past the n for when k != 0. */
4901 EXTRACT_NUMBER_AND_INCR (k, p);
4904 p -= 2 * OFFSET_ADDRESS_SIZE;
4905 succeed_n_p = true; /* Spaghetti code alert. */
4906 goto handle_on_failure_jump;
4912 p += 2 * OFFSET_ADDRESS_SIZE;
4923 abort (); /* We have listed all the cases. */
4926 /* Getting here means we have found the possible starting
4927 characters for one path of the pattern -- and that the empty
4928 string does not match. We need not follow this path further.
4929 Instead, look at the next alternative (remembered on the
4930 stack), or quit if no more. The test at the top of the loop
4931 does these things. */
4932 path_can_be_null = false;
4936 /* Set `can_be_null' for the last path (also the first path, if the
4937 pattern is empty). */
4938 bufp->can_be_null |= path_can_be_null;
4941 RESET_FAIL_STACK ();
4945 #else /* not INSIDE_RECURSION */
4948 re_compile_fastmap (bufp)
4949 struct re_pattern_buffer *bufp;
4952 if (MB_CUR_MAX != 1)
4953 return wcs_re_compile_fastmap(bufp);
4956 return byte_re_compile_fastmap(bufp);
4957 } /* re_compile_fastmap */
4959 weak_alias (__re_compile_fastmap, re_compile_fastmap)
4963 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4964 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4965 this memory for recording register information. STARTS and ENDS
4966 must be allocated using the malloc library routine, and must each
4967 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4969 If NUM_REGS == 0, then subsequent matches should allocate their own
4972 Unless this function is called, the first search or match using
4973 PATTERN_BUFFER will allocate its own register data, without
4974 freeing the old data. */
4977 re_set_registers (bufp, regs, num_regs, starts, ends)
4978 struct re_pattern_buffer *bufp;
4979 struct re_registers *regs;
4981 regoff_t *starts, *ends;
4985 bufp->regs_allocated = REGS_REALLOCATE;
4986 regs->num_regs = num_regs;
4987 regs->start = starts;
4992 bufp->regs_allocated = REGS_UNALLOCATED;
4994 regs->start = regs->end = (regoff_t *) 0;
4998 weak_alias (__re_set_registers, re_set_registers)
5001 /* Searching routines. */
5003 /* Like re_search_2, below, but only one string is specified, and
5004 doesn't let you say where to stop matching. */
5007 re_search (bufp, string, size, startpos, range, regs)
5008 struct re_pattern_buffer *bufp;
5010 int size, startpos, range;
5011 struct re_registers *regs;
5013 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
5017 weak_alias (__re_search, re_search)
5021 /* Using the compiled pattern in BUFP->buffer, first tries to match the
5022 virtual concatenation of STRING1 and STRING2, starting first at index
5023 STARTPOS, then at STARTPOS + 1, and so on.
5025 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
5027 RANGE is how far to scan while trying to match. RANGE = 0 means try
5028 only at STARTPOS; in general, the last start tried is STARTPOS +
5031 In REGS, return the indices of the virtual concatenation of STRING1
5032 and STRING2 that matched the entire BUFP->buffer and its contained
5035 Do not consider matching one past the index STOP in the virtual
5036 concatenation of STRING1 and STRING2.
5038 We return either the position in the strings at which the match was
5039 found, -1 if no match, or -2 if error (such as failure
5043 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
5044 struct re_pattern_buffer *bufp;
5045 const char *string1, *string2;
5049 struct re_registers *regs;
5053 if (MB_CUR_MAX != 1)
5054 return wcs_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5058 return byte_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5062 weak_alias (__re_search_2, re_search_2)
5065 #endif /* not INSIDE_RECURSION */
5067 #ifdef INSIDE_RECURSION
5069 #ifdef MATCH_MAY_ALLOCATE
5070 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
5072 # define FREE_VAR(var) if (var) free (var); var = NULL
5076 # define MAX_ALLOCA_SIZE 2000
5078 # define FREE_WCS_BUFFERS() \
5080 if (size1 > MAX_ALLOCA_SIZE) \
5082 free (wcs_string1); \
5083 free (mbs_offset1); \
5087 FREE_VAR (wcs_string1); \
5088 FREE_VAR (mbs_offset1); \
5090 if (size2 > MAX_ALLOCA_SIZE) \
5092 free (wcs_string2); \
5093 free (mbs_offset2); \
5097 FREE_VAR (wcs_string2); \
5098 FREE_VAR (mbs_offset2); \
5106 PREFIX(re_search_2) (bufp, string1, size1, string2, size2, startpos, range,
5108 struct re_pattern_buffer *bufp;
5109 const char *string1, *string2;
5113 struct re_registers *regs;
5117 register char *fastmap = bufp->fastmap;
5118 register RE_TRANSLATE_TYPE translate = bufp->translate;
5119 int total_size = size1 + size2;
5120 int endpos = startpos + range;
5122 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5123 wchar_t *wcs_string1 = NULL, *wcs_string2 = NULL;
5124 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5125 int wcs_size1 = 0, wcs_size2 = 0;
5126 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5127 int *mbs_offset1 = NULL, *mbs_offset2 = NULL;
5128 /* They hold whether each wchar_t is binary data or not. */
5129 char *is_binary = NULL;
5132 /* Check for out-of-range STARTPOS. */
5133 if (startpos < 0 || startpos > total_size)
5136 /* Fix up RANGE if it might eventually take us outside
5137 the virtual concatenation of STRING1 and STRING2.
5138 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5140 range = 0 - startpos;
5141 else if (endpos > total_size)
5142 range = total_size - startpos;
5144 /* If the search isn't to be a backwards one, don't waste time in a
5145 search for a pattern that must be anchored. */
5146 if (bufp->used > 0 && range > 0
5147 && ((re_opcode_t) bufp->buffer[0] == begbuf
5148 /* `begline' is like `begbuf' if it cannot match at newlines. */
5149 || ((re_opcode_t) bufp->buffer[0] == begline
5150 && !bufp->newline_anchor)))
5159 /* In a forward search for something that starts with \=.
5160 don't keep searching past point. */
5161 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
5163 range = PT - startpos;
5169 /* Update the fastmap now if not correct already. */
5170 if (fastmap && !bufp->fastmap_accurate)
5171 if (re_compile_fastmap (bufp) == -2)
5175 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5176 fill them with converted string. */
5179 if (size1 > MAX_ALLOCA_SIZE)
5181 wcs_string1 = TALLOC (size1 + 1, CHAR_T);
5182 mbs_offset1 = TALLOC (size1 + 1, int);
5183 is_binary = TALLOC (size1 + 1, char);
5187 wcs_string1 = REGEX_TALLOC (size1 + 1, CHAR_T);
5188 mbs_offset1 = REGEX_TALLOC (size1 + 1, int);
5189 is_binary = REGEX_TALLOC (size1 + 1, char);
5191 if (!wcs_string1 || !mbs_offset1 || !is_binary)
5193 if (size1 > MAX_ALLOCA_SIZE)
5201 FREE_VAR (wcs_string1);
5202 FREE_VAR (mbs_offset1);
5203 FREE_VAR (is_binary);
5207 wcs_size1 = convert_mbs_to_wcs(wcs_string1, string1, size1,
5208 mbs_offset1, is_binary);
5209 wcs_string1[wcs_size1] = L'\0'; /* for a sentinel */
5210 if (size1 > MAX_ALLOCA_SIZE)
5213 FREE_VAR (is_binary);
5217 if (size2 > MAX_ALLOCA_SIZE)
5219 wcs_string2 = TALLOC (size2 + 1, CHAR_T);
5220 mbs_offset2 = TALLOC (size2 + 1, int);
5221 is_binary = TALLOC (size2 + 1, char);
5225 wcs_string2 = REGEX_TALLOC (size2 + 1, CHAR_T);
5226 mbs_offset2 = REGEX_TALLOC (size2 + 1, int);
5227 is_binary = REGEX_TALLOC (size2 + 1, char);
5229 if (!wcs_string2 || !mbs_offset2 || !is_binary)
5231 FREE_WCS_BUFFERS ();
5232 if (size2 > MAX_ALLOCA_SIZE)
5235 FREE_VAR (is_binary);
5238 wcs_size2 = convert_mbs_to_wcs(wcs_string2, string2, size2,
5239 mbs_offset2, is_binary);
5240 wcs_string2[wcs_size2] = L'\0'; /* for a sentinel */
5241 if (size2 > MAX_ALLOCA_SIZE)
5244 FREE_VAR (is_binary);
5249 /* Loop through the string, looking for a place to start matching. */
5252 /* If a fastmap is supplied, skip quickly over characters that
5253 cannot be the start of a match. If the pattern can match the
5254 null string, however, we don't need to skip characters; we want
5255 the first null string. */
5256 if (fastmap && startpos < total_size && !bufp->can_be_null)
5258 if (range > 0) /* Searching forwards. */
5260 register const char *d;
5261 register int lim = 0;
5264 if (startpos < size1 && startpos + range >= size1)
5265 lim = range - (size1 - startpos);
5267 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
5269 /* Written out as an if-else to avoid testing `translate'
5273 && !fastmap[(unsigned char)
5274 translate[(unsigned char) *d++]])
5277 while (range > lim && !fastmap[(unsigned char) *d++])
5280 startpos += irange - range;
5282 else /* Searching backwards. */
5284 register CHAR_T c = (size1 == 0 || startpos >= size1
5285 ? string2[startpos - size1]
5286 : string1[startpos]);
5288 if (!fastmap[(unsigned char) TRANSLATE (c)])
5293 /* If can't match the null string, and that's all we have left, fail. */
5294 if (range >= 0 && startpos == total_size && fastmap
5295 && !bufp->can_be_null)
5298 FREE_WCS_BUFFERS ();
5304 val = wcs_re_match_2_internal (bufp, string1, size1, string2,
5305 size2, startpos, regs, stop,
5306 wcs_string1, wcs_size1,
5307 wcs_string2, wcs_size2,
5308 mbs_offset1, mbs_offset2);
5310 val = byte_re_match_2_internal (bufp, string1, size1, string2,
5311 size2, startpos, regs, stop);
5314 #ifndef REGEX_MALLOC
5323 FREE_WCS_BUFFERS ();
5331 FREE_WCS_BUFFERS ();
5351 FREE_WCS_BUFFERS ();
5357 /* This converts PTR, a pointer into one of the search wchar_t strings
5358 `string1' and `string2' into an multibyte string offset from the
5359 beginning of that string. We use mbs_offset to optimize.
5360 See convert_mbs_to_wcs. */
5361 # define POINTER_TO_OFFSET(ptr) \
5362 (FIRST_STRING_P (ptr) \
5363 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5364 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5367 /* This converts PTR, a pointer into one of the search strings `string1'
5368 and `string2' into an offset from the beginning of that string. */
5369 # define POINTER_TO_OFFSET(ptr) \
5370 (FIRST_STRING_P (ptr) \
5371 ? ((regoff_t) ((ptr) - string1)) \
5372 : ((regoff_t) ((ptr) - string2 + size1)))
5375 /* Macros for dealing with the split strings in re_match_2. */
5377 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5379 /* Call before fetching a character with *d. This switches over to
5380 string2 if necessary. */
5381 #define PREFETCH() \
5384 /* End of string2 => fail. */ \
5385 if (dend == end_match_2) \
5387 /* End of string1 => advance to string2. */ \
5389 dend = end_match_2; \
5392 /* Test if at very beginning or at very end of the virtual concatenation
5393 of `string1' and `string2'. If only one string, it's `string2'. */
5394 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5395 #define AT_STRINGS_END(d) ((d) == end2)
5398 /* Test if D points to a character which is word-constituent. We have
5399 two special cases to check for: if past the end of string1, look at
5400 the first character in string2; and if before the beginning of
5401 string2, look at the last character in string1. */
5403 /* Use internationalized API instead of SYNTAX. */
5404 # define WORDCHAR_P(d) \
5405 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5406 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5407 || ((d) == end1 ? *string2 \
5408 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) == L'_')
5410 # define WORDCHAR_P(d) \
5411 (SYNTAX ((d) == end1 ? *string2 \
5412 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5416 /* Disabled due to a compiler bug -- see comment at case wordbound */
5418 /* Test if the character before D and the one at D differ with respect
5419 to being word-constituent. */
5420 #define AT_WORD_BOUNDARY(d) \
5421 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5422 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5425 /* Free everything we malloc. */
5426 #ifdef MATCH_MAY_ALLOCATE
5428 # define FREE_VARIABLES() \
5430 REGEX_FREE_STACK (fail_stack.stack); \
5431 FREE_VAR (regstart); \
5432 FREE_VAR (regend); \
5433 FREE_VAR (old_regstart); \
5434 FREE_VAR (old_regend); \
5435 FREE_VAR (best_regstart); \
5436 FREE_VAR (best_regend); \
5437 FREE_VAR (reg_info); \
5438 FREE_VAR (reg_dummy); \
5439 FREE_VAR (reg_info_dummy); \
5440 if (!cant_free_wcs_buf) \
5442 FREE_VAR (string1); \
5443 FREE_VAR (string2); \
5444 FREE_VAR (mbs_offset1); \
5445 FREE_VAR (mbs_offset2); \
5449 # define FREE_VARIABLES() \
5451 REGEX_FREE_STACK (fail_stack.stack); \
5452 FREE_VAR (regstart); \
5453 FREE_VAR (regend); \
5454 FREE_VAR (old_regstart); \
5455 FREE_VAR (old_regend); \
5456 FREE_VAR (best_regstart); \
5457 FREE_VAR (best_regend); \
5458 FREE_VAR (reg_info); \
5459 FREE_VAR (reg_dummy); \
5460 FREE_VAR (reg_info_dummy); \
5465 # define FREE_VARIABLES() \
5467 if (!cant_free_wcs_buf) \
5469 FREE_VAR (string1); \
5470 FREE_VAR (string2); \
5471 FREE_VAR (mbs_offset1); \
5472 FREE_VAR (mbs_offset2); \
5476 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
5478 #endif /* not MATCH_MAY_ALLOCATE */
5480 /* These values must meet several constraints. They must not be valid
5481 register values; since we have a limit of 255 registers (because
5482 we use only one byte in the pattern for the register number), we can
5483 use numbers larger than 255. They must differ by 1, because of
5484 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5485 be larger than the value for the highest register, so we do not try
5486 to actually save any registers when none are active. */
5487 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5488 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5490 #else /* not INSIDE_RECURSION */
5491 /* Matching routines. */
5493 #ifndef emacs /* Emacs never uses this. */
5494 /* re_match is like re_match_2 except it takes only a single string. */
5497 re_match (bufp, string, size, pos, regs)
5498 struct re_pattern_buffer *bufp;
5501 struct re_registers *regs;
5505 if (MB_CUR_MAX != 1)
5506 result = wcs_re_match_2_internal (bufp, NULL, 0, string, size,
5508 NULL, 0, NULL, 0, NULL, NULL);
5511 result = byte_re_match_2_internal (bufp, NULL, 0, string, size,
5513 # ifndef REGEX_MALLOC
5521 weak_alias (__re_match, re_match)
5523 #endif /* not emacs */
5525 #endif /* not INSIDE_RECURSION */
5527 #ifdef INSIDE_RECURSION
5528 static boolean PREFIX(group_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5530 PREFIX(register_info_type) *reg_info));
5531 static boolean PREFIX(alt_match_null_string_p) _RE_ARGS ((UCHAR_T *p,
5533 PREFIX(register_info_type) *reg_info));
5534 static boolean PREFIX(common_op_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5536 PREFIX(register_info_type) *reg_info));
5537 static int PREFIX(bcmp_translate) _RE_ARGS ((const CHAR_T *s1, const CHAR_T *s2,
5538 int len, char *translate));
5539 #else /* not INSIDE_RECURSION */
5541 /* re_match_2 matches the compiled pattern in BUFP against the
5542 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5543 and SIZE2, respectively). We start matching at POS, and stop
5546 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5547 store offsets for the substring each group matched in REGS. See the
5548 documentation for exactly how many groups we fill.
5550 We return -1 if no match, -2 if an internal error (such as the
5551 failure stack overflowing). Otherwise, we return the length of the
5552 matched substring. */
5555 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
5556 struct re_pattern_buffer *bufp;
5557 const char *string1, *string2;
5560 struct re_registers *regs;
5565 if (MB_CUR_MAX != 1)
5566 result = wcs_re_match_2_internal (bufp, string1, size1, string2, size2,
5568 NULL, 0, NULL, 0, NULL, NULL);
5571 result = byte_re_match_2_internal (bufp, string1, size1, string2, size2,
5574 #ifndef REGEX_MALLOC
5582 weak_alias (__re_match_2, re_match_2)
5585 #endif /* not INSIDE_RECURSION */
5587 #ifdef INSIDE_RECURSION
5590 static int count_mbs_length PARAMS ((int *, int));
5592 /* This check the substring (from 0, to length) of the multibyte string,
5593 to which offset_buffer correspond. And count how many wchar_t_characters
5594 the substring occupy. We use offset_buffer to optimization.
5595 See convert_mbs_to_wcs. */
5598 count_mbs_length(offset_buffer, length)
5604 /* Check whether the size is valid. */
5608 if (offset_buffer == NULL)
5611 /* If there are no multibyte character, offset_buffer[i] == i.
5612 Optmize for this case. */
5613 if (offset_buffer[length] == length)
5616 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5622 int middle = (lower + upper) / 2;
5623 if (middle == lower || middle == upper)
5625 if (offset_buffer[middle] > length)
5627 else if (offset_buffer[middle] < length)
5637 /* This is a separate function so that we can force an alloca cleanup
5641 wcs_re_match_2_internal (bufp, cstring1, csize1, cstring2, csize2, pos,
5642 regs, stop, string1, size1, string2, size2,
5643 mbs_offset1, mbs_offset2)
5644 struct re_pattern_buffer *bufp;
5645 const char *cstring1, *cstring2;
5648 struct re_registers *regs;
5650 /* string1 == string2 == NULL means string1/2, size1/2 and
5651 mbs_offset1/2 need seting up in this function. */
5652 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5653 wchar_t *string1, *string2;
5654 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5656 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5657 int *mbs_offset1, *mbs_offset2;
5660 byte_re_match_2_internal (bufp, string1, size1,string2, size2, pos,
5662 struct re_pattern_buffer *bufp;
5663 const char *string1, *string2;
5666 struct re_registers *regs;
5670 /* General temporaries. */
5674 /* They hold whether each wchar_t is binary data or not. */
5675 char *is_binary = NULL;
5676 /* If true, we can't free string1/2, mbs_offset1/2. */
5677 int cant_free_wcs_buf = 1;
5680 /* Just past the end of the corresponding string. */
5681 const CHAR_T *end1, *end2;
5683 /* Pointers into string1 and string2, just past the last characters in
5684 each to consider matching. */
5685 const CHAR_T *end_match_1, *end_match_2;
5687 /* Where we are in the data, and the end of the current string. */
5688 const CHAR_T *d, *dend;
5690 /* Where we are in the pattern, and the end of the pattern. */
5692 UCHAR_T *pattern, *p;
5693 register UCHAR_T *pend;
5695 UCHAR_T *p = bufp->buffer;
5696 register UCHAR_T *pend = p + bufp->used;
5699 /* Mark the opcode just after a start_memory, so we can test for an
5700 empty subpattern when we get to the stop_memory. */
5701 UCHAR_T *just_past_start_mem = 0;
5703 /* We use this to map every character in the string. */
5704 RE_TRANSLATE_TYPE translate = bufp->translate;
5706 /* Failure point stack. Each place that can handle a failure further
5707 down the line pushes a failure point on this stack. It consists of
5708 restart, regend, and reg_info for all registers corresponding to
5709 the subexpressions we're currently inside, plus the number of such
5710 registers, and, finally, two char *'s. The first char * is where
5711 to resume scanning the pattern; the second one is where to resume
5712 scanning the strings. If the latter is zero, the failure point is
5713 a ``dummy''; if a failure happens and the failure point is a dummy,
5714 it gets discarded and the next next one is tried. */
5715 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5716 PREFIX(fail_stack_type) fail_stack;
5719 static unsigned failure_id;
5720 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
5724 /* This holds the pointer to the failure stack, when
5725 it is allocated relocatably. */
5726 fail_stack_elt_t *failure_stack_ptr;
5729 /* We fill all the registers internally, independent of what we
5730 return, for use in backreferences. The number here includes
5731 an element for register zero. */
5732 size_t num_regs = bufp->re_nsub + 1;
5734 /* The currently active registers. */
5735 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
5736 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
5738 /* Information on the contents of registers. These are pointers into
5739 the input strings; they record just what was matched (on this
5740 attempt) by a subexpression part of the pattern, that is, the
5741 regnum-th regstart pointer points to where in the pattern we began
5742 matching and the regnum-th regend points to right after where we
5743 stopped matching the regnum-th subexpression. (The zeroth register
5744 keeps track of what the whole pattern matches.) */
5745 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5746 const CHAR_T **regstart, **regend;
5749 /* If a group that's operated upon by a repetition operator fails to
5750 match anything, then the register for its start will need to be
5751 restored because it will have been set to wherever in the string we
5752 are when we last see its open-group operator. Similarly for a
5754 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5755 const CHAR_T **old_regstart, **old_regend;
5758 /* The is_active field of reg_info helps us keep track of which (possibly
5759 nested) subexpressions we are currently in. The matched_something
5760 field of reg_info[reg_num] helps us tell whether or not we have
5761 matched any of the pattern so far this time through the reg_num-th
5762 subexpression. These two fields get reset each time through any
5763 loop their register is in. */
5764 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5765 PREFIX(register_info_type) *reg_info;
5768 /* The following record the register info as found in the above
5769 variables when we find a match better than any we've seen before.
5770 This happens as we backtrack through the failure points, which in
5771 turn happens only if we have not yet matched the entire string. */
5772 unsigned best_regs_set = false;
5773 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5774 const CHAR_T **best_regstart, **best_regend;
5777 /* Logically, this is `best_regend[0]'. But we don't want to have to
5778 allocate space for that if we're not allocating space for anything
5779 else (see below). Also, we never need info about register 0 for
5780 any of the other register vectors, and it seems rather a kludge to
5781 treat `best_regend' differently than the rest. So we keep track of
5782 the end of the best match so far in a separate variable. We
5783 initialize this to NULL so that when we backtrack the first time
5784 and need to test it, it's not garbage. */
5785 const CHAR_T *match_end = NULL;
5787 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5788 int set_regs_matched_done = 0;
5790 /* Used when we pop values we don't care about. */
5791 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5792 const CHAR_T **reg_dummy;
5793 PREFIX(register_info_type) *reg_info_dummy;
5797 /* Counts the total number of registers pushed. */
5798 unsigned num_regs_pushed = 0;
5801 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
5805 #ifdef MATCH_MAY_ALLOCATE
5806 /* Do not bother to initialize all the register variables if there are
5807 no groups in the pattern, as it takes a fair amount of time. If
5808 there are groups, we include space for register 0 (the whole
5809 pattern), even though we never use it, since it simplifies the
5810 array indexing. We should fix this. */
5813 regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5814 regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5815 old_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5816 old_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5817 best_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5818 best_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5819 reg_info = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5820 reg_dummy = REGEX_TALLOC (num_regs, const CHAR_T *);
5821 reg_info_dummy = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5823 if (!(regstart && regend && old_regstart && old_regend && reg_info
5824 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
5832 /* We must initialize all our variables to NULL, so that
5833 `FREE_VARIABLES' doesn't try to free them. */
5834 regstart = regend = old_regstart = old_regend = best_regstart
5835 = best_regend = reg_dummy = NULL;
5836 reg_info = reg_info_dummy = (PREFIX(register_info_type) *) NULL;
5838 #endif /* MATCH_MAY_ALLOCATE */
5840 /* The starting position is bogus. */
5842 if (pos < 0 || pos > csize1 + csize2)
5844 if (pos < 0 || pos > size1 + size2)
5852 /* Allocate wchar_t array for string1 and string2 and
5853 fill them with converted string. */
5854 if (string1 == NULL && string2 == NULL)
5856 /* We need seting up buffers here. */
5858 /* We must free wcs buffers in this function. */
5859 cant_free_wcs_buf = 0;
5863 string1 = REGEX_TALLOC (csize1 + 1, CHAR_T);
5864 mbs_offset1 = REGEX_TALLOC (csize1 + 1, int);
5865 is_binary = REGEX_TALLOC (csize1 + 1, char);
5866 if (!string1 || !mbs_offset1 || !is_binary)
5869 FREE_VAR (mbs_offset1);
5870 FREE_VAR (is_binary);
5876 string2 = REGEX_TALLOC (csize2 + 1, CHAR_T);
5877 mbs_offset2 = REGEX_TALLOC (csize2 + 1, int);
5878 is_binary = REGEX_TALLOC (csize2 + 1, char);
5879 if (!string2 || !mbs_offset2 || !is_binary)
5882 FREE_VAR (mbs_offset1);
5884 FREE_VAR (mbs_offset2);
5885 FREE_VAR (is_binary);
5888 size2 = convert_mbs_to_wcs(string2, cstring2, csize2,
5889 mbs_offset2, is_binary);
5890 string2[size2] = L'\0'; /* for a sentinel */
5891 FREE_VAR (is_binary);
5895 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5896 pattern to (char*) in regex_compile. */
5897 p = pattern = (CHAR_T*)bufp->buffer;
5898 pend = (CHAR_T*)(bufp->buffer + bufp->used);
5902 /* Initialize subexpression text positions to -1 to mark ones that no
5903 start_memory/stop_memory has been seen for. Also initialize the
5904 register information struct. */
5905 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
5907 regstart[mcnt] = regend[mcnt]
5908 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
5910 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
5911 IS_ACTIVE (reg_info[mcnt]) = 0;
5912 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5913 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5916 /* We move `string1' into `string2' if the latter's empty -- but not if
5917 `string1' is null. */
5918 if (size2 == 0 && string1 != NULL)
5925 mbs_offset2 = mbs_offset1;
5931 end1 = string1 + size1;
5932 end2 = string2 + size2;
5934 /* Compute where to stop matching, within the two strings. */
5938 mcnt = count_mbs_length(mbs_offset1, stop);
5939 end_match_1 = string1 + mcnt;
5940 end_match_2 = string2;
5944 if (stop > csize1 + csize2)
5945 stop = csize1 + csize2;
5947 mcnt = count_mbs_length(mbs_offset2, stop-csize1);
5948 end_match_2 = string2 + mcnt;
5951 { /* count_mbs_length return error. */
5958 end_match_1 = string1 + stop;
5959 end_match_2 = string2;
5964 end_match_2 = string2 + stop - size1;
5968 /* `p' scans through the pattern as `d' scans through the data.
5969 `dend' is the end of the input string that `d' points within. `d'
5970 is advanced into the following input string whenever necessary, but
5971 this happens before fetching; therefore, at the beginning of the
5972 loop, `d' can be pointing at the end of a string, but it cannot
5975 if (size1 > 0 && pos <= csize1)
5977 mcnt = count_mbs_length(mbs_offset1, pos);
5983 mcnt = count_mbs_length(mbs_offset2, pos-csize1);
5989 { /* count_mbs_length return error. */
5994 if (size1 > 0 && pos <= size1)
6001 d = string2 + pos - size1;
6006 DEBUG_PRINT1 ("The compiled pattern is:\n");
6007 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
6008 DEBUG_PRINT1 ("The string to match is: `");
6009 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
6010 DEBUG_PRINT1 ("'\n");
6012 /* This loops over pattern commands. It exits by returning from the
6013 function if the match is complete, or it drops through if the match
6014 fails at this starting point in the input data. */
6018 DEBUG_PRINT2 ("\n%p: ", p);
6020 DEBUG_PRINT2 ("\n0x%x: ", p);
6024 { /* End of pattern means we might have succeeded. */
6025 DEBUG_PRINT1 ("end of pattern ... ");
6027 /* If we haven't matched the entire string, and we want the
6028 longest match, try backtracking. */
6029 if (d != end_match_2)
6031 /* 1 if this match ends in the same string (string1 or string2)
6032 as the best previous match. */
6033 boolean same_str_p = (FIRST_STRING_P (match_end)
6034 == MATCHING_IN_FIRST_STRING);
6035 /* 1 if this match is the best seen so far. */
6036 boolean best_match_p;
6038 /* AIX compiler got confused when this was combined
6039 with the previous declaration. */
6041 best_match_p = d > match_end;
6043 best_match_p = !MATCHING_IN_FIRST_STRING;
6045 DEBUG_PRINT1 ("backtracking.\n");
6047 if (!FAIL_STACK_EMPTY ())
6048 { /* More failure points to try. */
6050 /* If exceeds best match so far, save it. */
6051 if (!best_regs_set || best_match_p)
6053 best_regs_set = true;
6056 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
6058 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6060 best_regstart[mcnt] = regstart[mcnt];
6061 best_regend[mcnt] = regend[mcnt];
6067 /* If no failure points, don't restore garbage. And if
6068 last match is real best match, don't restore second
6070 else if (best_regs_set && !best_match_p)
6073 /* Restore best match. It may happen that `dend ==
6074 end_match_1' while the restored d is in string2.
6075 For example, the pattern `x.*y.*z' against the
6076 strings `x-' and `y-z-', if the two strings are
6077 not consecutive in memory. */
6078 DEBUG_PRINT1 ("Restoring best registers.\n");
6081 dend = ((d >= string1 && d <= end1)
6082 ? end_match_1 : end_match_2);
6084 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6086 regstart[mcnt] = best_regstart[mcnt];
6087 regend[mcnt] = best_regend[mcnt];
6090 } /* d != end_match_2 */
6093 DEBUG_PRINT1 ("Accepting match.\n");
6094 /* If caller wants register contents data back, do it. */
6095 if (regs && !bufp->no_sub)
6097 /* Have the register data arrays been allocated? */
6098 if (bufp->regs_allocated == REGS_UNALLOCATED)
6099 { /* No. So allocate them with malloc. We need one
6100 extra element beyond `num_regs' for the `-1' marker
6102 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
6103 regs->start = TALLOC (regs->num_regs, regoff_t);
6104 regs->end = TALLOC (regs->num_regs, regoff_t);
6105 if (regs->start == NULL || regs->end == NULL)
6110 bufp->regs_allocated = REGS_REALLOCATE;
6112 else if (bufp->regs_allocated == REGS_REALLOCATE)
6113 { /* Yes. If we need more elements than were already
6114 allocated, reallocate them. If we need fewer, just
6116 if (regs->num_regs < num_regs + 1)
6118 regs->num_regs = num_regs + 1;
6119 RETALLOC (regs->start, regs->num_regs, regoff_t);
6120 RETALLOC (regs->end, regs->num_regs, regoff_t);
6121 if (regs->start == NULL || regs->end == NULL)
6130 /* These braces fend off a "empty body in an else-statement"
6131 warning under GCC when assert expands to nothing. */
6132 assert (bufp->regs_allocated == REGS_FIXED);
6135 /* Convert the pointer data in `regstart' and `regend' to
6136 indices. Register zero has to be set differently,
6137 since we haven't kept track of any info for it. */
6138 if (regs->num_regs > 0)
6140 regs->start[0] = pos;
6142 if (MATCHING_IN_FIRST_STRING)
6143 regs->end[0] = mbs_offset1 != NULL ?
6144 mbs_offset1[d-string1] : 0;
6146 regs->end[0] = csize1 + (mbs_offset2 != NULL ?
6147 mbs_offset2[d-string2] : 0);
6149 regs->end[0] = (MATCHING_IN_FIRST_STRING
6150 ? ((regoff_t) (d - string1))
6151 : ((regoff_t) (d - string2 + size1)));
6155 /* Go through the first `min (num_regs, regs->num_regs)'
6156 registers, since that is all we initialized. */
6157 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
6160 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
6161 regs->start[mcnt] = regs->end[mcnt] = -1;
6165 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
6167 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
6171 /* If the regs structure we return has more elements than
6172 were in the pattern, set the extra elements to -1. If
6173 we (re)allocated the registers, this is the case,
6174 because we always allocate enough to have at least one
6176 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
6177 regs->start[mcnt] = regs->end[mcnt] = -1;
6178 } /* regs && !bufp->no_sub */
6180 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
6181 nfailure_points_pushed, nfailure_points_popped,
6182 nfailure_points_pushed - nfailure_points_popped);
6183 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
6186 if (MATCHING_IN_FIRST_STRING)
6187 mcnt = mbs_offset1 != NULL ? mbs_offset1[d-string1] : 0;
6189 mcnt = (mbs_offset2 != NULL ? mbs_offset2[d-string2] : 0) +
6193 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
6198 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
6204 /* Otherwise match next pattern command. */
6205 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
6207 /* Ignore these. Used to ignore the n of succeed_n's which
6208 currently have n == 0. */
6210 DEBUG_PRINT1 ("EXECUTING no_op.\n");
6214 DEBUG_PRINT1 ("EXECUTING succeed.\n");
6217 /* Match the next n pattern characters exactly. The following
6218 byte in the pattern defines n, and the n bytes after that
6219 are the characters to match. */
6225 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
6227 /* This is written out as an if-else so we don't waste time
6228 testing `translate' inside the loop. */
6237 if ((UCHAR_T) translate[(unsigned char) *d++]
6243 if (*d++ != (CHAR_T) *p++)
6247 if ((UCHAR_T) translate[(unsigned char) *d++]
6259 if (*d++ != (CHAR_T) *p++) goto fail;
6263 SET_REGS_MATCHED ();
6267 /* Match any character except possibly a newline or a null. */
6269 DEBUG_PRINT1 ("EXECUTING anychar.\n");
6273 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
6274 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
6277 SET_REGS_MATCHED ();
6278 DEBUG_PRINT2 (" Matched `%ld'.\n", (long int) *d);
6288 unsigned int i, char_class_length, coll_symbol_length,
6289 equiv_class_length, ranges_length, chars_length, length;
6290 CHAR_T *workp, *workp2, *charset_top;
6291 #define WORK_BUFFER_SIZE 128
6292 CHAR_T str_buf[WORK_BUFFER_SIZE];
6297 boolean not = (re_opcode_t) *(p - 1) == charset_not;
6299 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
6301 c = TRANSLATE (*d); /* The character to match. */
6304 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
6306 charset_top = p - 1;
6307 char_class_length = *p++;
6308 coll_symbol_length = *p++;
6309 equiv_class_length = *p++;
6310 ranges_length = *p++;
6311 chars_length = *p++;
6312 /* p points charset[6], so the address of the next instruction
6313 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6314 where l=length of char_classes, m=length of collating_symbol,
6315 n=equivalence_class, o=length of char_range,
6316 p'=length of character. */
6318 /* Update p to indicate the next instruction. */
6319 p += char_class_length + coll_symbol_length+ equiv_class_length +
6320 2*ranges_length + chars_length;
6322 /* match with char_class? */
6323 for (i = 0; i < char_class_length ; i += CHAR_CLASS_SIZE)
6326 uintptr_t alignedp = ((uintptr_t)workp
6327 + __alignof__(wctype_t) - 1)
6328 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
6329 wctype = *((wctype_t*)alignedp);
6330 workp += CHAR_CLASS_SIZE;
6331 if (iswctype((wint_t)c, wctype))
6332 goto char_set_matched;
6335 /* match with collating_symbol? */
6339 const unsigned char *extra = (const unsigned char *)
6340 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
6342 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;
6346 wextra = (int32_t*)(extra + *workp++);
6347 for (i = 0; i < *wextra; ++i)
6348 if (TRANSLATE(d[i]) != wextra[1 + i])
6353 /* Update d, however d will be incremented at
6354 char_set_matched:, we decrement d here. */
6356 goto char_set_matched;
6360 else /* (nrules == 0) */
6362 /* If we can't look up collation data, we use wcscoll
6365 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;)
6367 const CHAR_T *backup_d = d, *backup_dend = dend;
6368 length = wcslen(workp);
6370 /* If wcscoll(the collating symbol, whole string) > 0,
6371 any substring of the string never match with the
6372 collating symbol. */
6373 if (wcscoll(workp, d) > 0)
6375 workp += length + 1;
6379 /* First, we compare the collating symbol with
6380 the first character of the string.
6381 If it don't match, we add the next character to
6382 the compare buffer in turn. */
6383 for (i = 0 ; i < WORK_BUFFER_SIZE-1 ; i++, d++)
6388 if (dend == end_match_2)
6394 /* add next character to the compare buffer. */
6395 str_buf[i] = TRANSLATE(*d);
6396 str_buf[i+1] = '\0';
6398 match = wcscoll(workp, str_buf);
6400 goto char_set_matched;
6403 /* (str_buf > workp) indicate (str_buf + X > workp),
6404 because for all X (str_buf + X > str_buf).
6405 So we don't need continue this loop. */
6408 /* Otherwise(str_buf < workp),
6409 (str_buf+next_character) may equals (workp).
6410 So we continue this loop. */
6415 workp += length + 1;
6418 /* match with equivalence_class? */
6422 const CHAR_T *backup_d = d, *backup_dend = dend;
6423 /* Try to match the equivalence class against
6424 those known to the collate implementation. */
6425 const int32_t *table;
6426 const int32_t *weights;
6427 const int32_t *extra;
6428 const int32_t *indirect;
6433 /* This #include defines a local function! */
6434 # include <locale/weightwc.h>
6436 table = (const int32_t *)
6437 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEWC);
6438 weights = (const wint_t *)
6439 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTWC);
6440 extra = (const wint_t *)
6441 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAWC);
6442 indirect = (const int32_t *)
6443 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTWC);
6445 /* Write 1 collating element to str_buf, and
6449 for (i = 0 ; idx2 == 0 && i < WORK_BUFFER_SIZE - 1; i++)
6451 cp = (wint_t*)str_buf;
6454 if (dend == end_match_2)
6459 str_buf[i] = TRANSLATE(*(d+i));
6460 str_buf[i+1] = '\0'; /* sentinel */
6461 idx2 = findidx ((const wint_t**)&cp);
6464 /* Update d, however d will be incremented at
6465 char_set_matched:, we decrement d here. */
6466 d = backup_d + ((wchar_t*)cp - (wchar_t*)str_buf - 1);
6469 if (dend == end_match_2)
6478 len = weights[idx2];
6480 for (workp2 = workp + equiv_class_length ; workp < workp2 ;
6483 idx = (int32_t)*workp;
6484 /* We already checked idx != 0 in regex_compile. */
6486 if (idx2 != 0 && len == weights[idx])
6489 while (cnt < len && (weights[idx + 1 + cnt]
6490 == weights[idx2 + 1 + cnt]))
6494 goto char_set_matched;
6501 else /* (nrules == 0) */
6503 /* If we can't look up collation data, we use wcscoll
6506 for (workp2 = workp + equiv_class_length ; workp < workp2 ;)
6508 const CHAR_T *backup_d = d, *backup_dend = dend;
6509 length = wcslen(workp);
6511 /* If wcscoll(the collating symbol, whole string) > 0,
6512 any substring of the string never match with the
6513 collating symbol. */
6514 if (wcscoll(workp, d) > 0)
6516 workp += length + 1;
6520 /* First, we compare the equivalence class with
6521 the first character of the string.
6522 If it don't match, we add the next character to
6523 the compare buffer in turn. */
6524 for (i = 0 ; i < WORK_BUFFER_SIZE - 1 ; i++, d++)
6529 if (dend == end_match_2)
6535 /* add next character to the compare buffer. */
6536 str_buf[i] = TRANSLATE(*d);
6537 str_buf[i+1] = '\0';
6539 match = wcscoll(workp, str_buf);
6542 goto char_set_matched;
6545 /* (str_buf > workp) indicate (str_buf + X > workp),
6546 because for all X (str_buf + X > str_buf).
6547 So we don't need continue this loop. */
6550 /* Otherwise(str_buf < workp),
6551 (str_buf+next_character) may equals (workp).
6552 So we continue this loop. */
6557 workp += length + 1;
6561 /* match with char_range? */
6565 uint32_t collseqval;
6566 const char *collseq = (const char *)
6567 _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQWC);
6569 collseqval = collseq_table_lookup (collseq, c);
6571 for (; workp < p - chars_length ;)
6573 uint32_t start_val, end_val;
6575 /* We already compute the collation sequence value
6576 of the characters (or collating symbols). */
6577 start_val = (uint32_t) *workp++; /* range_start */
6578 end_val = (uint32_t) *workp++; /* range_end */
6580 if (start_val <= collseqval && collseqval <= end_val)
6581 goto char_set_matched;
6587 /* We set range_start_char at str_buf[0], range_end_char
6588 at str_buf[4], and compared char at str_buf[2]. */
6593 for (; workp < p - chars_length ;)
6595 wchar_t *range_start_char, *range_end_char;
6597 /* match if (range_start_char <= c <= range_end_char). */
6599 /* If range_start(or end) < 0, we assume -range_start(end)
6600 is the offset of the collating symbol which is specified
6601 as the character of the range start(end). */
6605 range_start_char = charset_top - (*workp++);
6608 str_buf[0] = *workp++;
6609 range_start_char = str_buf;
6614 range_end_char = charset_top - (*workp++);
6617 str_buf[4] = *workp++;
6618 range_end_char = str_buf + 4;
6621 if (wcscoll(range_start_char, str_buf+2) <= 0 &&
6622 wcscoll(str_buf+2, range_end_char) <= 0)
6624 goto char_set_matched;
6628 /* match with char? */
6629 for (; workp < p ; workp++)
6631 goto char_set_matched;
6638 /* Cast to `unsigned' instead of `unsigned char' in case the
6639 bit list is a full 32 bytes long. */
6640 if (c < (unsigned) (*p * BYTEWIDTH)
6641 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
6646 if (!not) goto fail;
6647 #undef WORK_BUFFER_SIZE
6649 SET_REGS_MATCHED ();
6655 /* The beginning of a group is represented by start_memory.
6656 The arguments are the register number in the next byte, and the
6657 number of groups inner to this one in the next. The text
6658 matched within the group is recorded (in the internal
6659 registers data structure) under the register number. */
6661 DEBUG_PRINT3 ("EXECUTING start_memory %ld (%ld):\n",
6662 (long int) *p, (long int) p[1]);
6664 /* Find out if this group can match the empty string. */
6665 p1 = p; /* To send to group_match_null_string_p. */
6667 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
6668 REG_MATCH_NULL_STRING_P (reg_info[*p])
6669 = PREFIX(group_match_null_string_p) (&p1, pend, reg_info);
6671 /* Save the position in the string where we were the last time
6672 we were at this open-group operator in case the group is
6673 operated upon by a repetition operator, e.g., with `(a*)*b'
6674 against `ab'; then we want to ignore where we are now in
6675 the string in case this attempt to match fails. */
6676 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6677 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
6679 DEBUG_PRINT2 (" old_regstart: %d\n",
6680 POINTER_TO_OFFSET (old_regstart[*p]));
6683 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
6685 IS_ACTIVE (reg_info[*p]) = 1;
6686 MATCHED_SOMETHING (reg_info[*p]) = 0;
6688 /* Clear this whenever we change the register activity status. */
6689 set_regs_matched_done = 0;
6691 /* This is the new highest active register. */
6692 highest_active_reg = *p;
6694 /* If nothing was active before, this is the new lowest active
6696 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
6697 lowest_active_reg = *p;
6699 /* Move past the register number and inner group count. */
6701 just_past_start_mem = p;
6706 /* The stop_memory opcode represents the end of a group. Its
6707 arguments are the same as start_memory's: the register
6708 number, and the number of inner groups. */
6710 DEBUG_PRINT3 ("EXECUTING stop_memory %ld (%ld):\n",
6711 (long int) *p, (long int) p[1]);
6713 /* We need to save the string position the last time we were at
6714 this close-group operator in case the group is operated
6715 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6716 against `aba'; then we want to ignore where we are now in
6717 the string in case this attempt to match fails. */
6718 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6719 ? REG_UNSET (regend[*p]) ? d : regend[*p]
6721 DEBUG_PRINT2 (" old_regend: %d\n",
6722 POINTER_TO_OFFSET (old_regend[*p]));
6725 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
6727 /* This register isn't active anymore. */
6728 IS_ACTIVE (reg_info[*p]) = 0;
6730 /* Clear this whenever we change the register activity status. */
6731 set_regs_matched_done = 0;
6733 /* If this was the only register active, nothing is active
6735 if (lowest_active_reg == highest_active_reg)
6737 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6738 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6741 { /* We must scan for the new highest active register, since
6742 it isn't necessarily one less than now: consider
6743 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6744 new highest active register is 1. */
6746 while (r > 0 && !IS_ACTIVE (reg_info[r]))
6749 /* If we end up at register zero, that means that we saved
6750 the registers as the result of an `on_failure_jump', not
6751 a `start_memory', and we jumped to past the innermost
6752 `stop_memory'. For example, in ((.)*) we save
6753 registers 1 and 2 as a result of the *, but when we pop
6754 back to the second ), we are at the stop_memory 1.
6755 Thus, nothing is active. */
6758 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6759 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6762 highest_active_reg = r;
6765 /* If just failed to match something this time around with a
6766 group that's operated on by a repetition operator, try to
6767 force exit from the ``loop'', and restore the register
6768 information for this group that we had before trying this
6770 if ((!MATCHED_SOMETHING (reg_info[*p])
6771 || just_past_start_mem == p - 1)
6774 boolean is_a_jump_n = false;
6778 switch ((re_opcode_t) *p1++)
6782 case pop_failure_jump:
6783 case maybe_pop_jump:
6785 case dummy_failure_jump:
6786 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6788 p1 += OFFSET_ADDRESS_SIZE;
6796 /* If the next operation is a jump backwards in the pattern
6797 to an on_failure_jump right before the start_memory
6798 corresponding to this stop_memory, exit from the loop
6799 by forcing a failure after pushing on the stack the
6800 on_failure_jump's jump in the pattern, and d. */
6801 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
6802 && (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == start_memory
6803 && p1[2+OFFSET_ADDRESS_SIZE] == *p)
6805 /* If this group ever matched anything, then restore
6806 what its registers were before trying this last
6807 failed match, e.g., with `(a*)*b' against `ab' for
6808 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6809 against `aba' for regend[3].
6811 Also restore the registers for inner groups for,
6812 e.g., `((a*)(b*))*' against `aba' (register 3 would
6813 otherwise get trashed). */
6815 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
6819 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
6821 /* Restore this and inner groups' (if any) registers. */
6822 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
6825 regstart[r] = old_regstart[r];
6827 /* xx why this test? */
6828 if (old_regend[r] >= regstart[r])
6829 regend[r] = old_regend[r];
6833 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6834 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
6840 /* Move past the register number and the inner group count. */
6845 /* \<digit> has been turned into a `duplicate' command which is
6846 followed by the numeric value of <digit> as the register number. */
6849 register const CHAR_T *d2, *dend2;
6850 int regno = *p++; /* Get which register to match against. */
6851 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
6853 /* Can't back reference a group which we've never matched. */
6854 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
6857 /* Where in input to try to start matching. */
6858 d2 = regstart[regno];
6860 /* Where to stop matching; if both the place to start and
6861 the place to stop matching are in the same string, then
6862 set to the place to stop, otherwise, for now have to use
6863 the end of the first string. */
6865 dend2 = ((FIRST_STRING_P (regstart[regno])
6866 == FIRST_STRING_P (regend[regno]))
6867 ? regend[regno] : end_match_1);
6870 /* If necessary, advance to next segment in register
6874 if (dend2 == end_match_2) break;
6875 if (dend2 == regend[regno]) break;
6877 /* End of string1 => advance to string2. */
6879 dend2 = regend[regno];
6881 /* At end of register contents => success */
6882 if (d2 == dend2) break;
6884 /* If necessary, advance to next segment in data. */
6887 /* How many characters left in this segment to match. */
6890 /* Want how many consecutive characters we can match in
6891 one shot, so, if necessary, adjust the count. */
6892 if (mcnt > dend2 - d2)
6895 /* Compare that many; failure if mismatch, else move
6898 ? PREFIX(bcmp_translate) (d, d2, mcnt, translate)
6899 : memcmp (d, d2, mcnt*sizeof(UCHAR_T)))
6901 d += mcnt, d2 += mcnt;
6903 /* Do this because we've match some characters. */
6904 SET_REGS_MATCHED ();
6910 /* begline matches the empty string at the beginning of the string
6911 (unless `not_bol' is set in `bufp'), and, if
6912 `newline_anchor' is set, after newlines. */
6914 DEBUG_PRINT1 ("EXECUTING begline.\n");
6916 if (AT_STRINGS_BEG (d))
6918 if (!bufp->not_bol) break;
6920 else if (d[-1] == '\n' && bufp->newline_anchor)
6924 /* In all other cases, we fail. */
6928 /* endline is the dual of begline. */
6930 DEBUG_PRINT1 ("EXECUTING endline.\n");
6932 if (AT_STRINGS_END (d))
6934 if (!bufp->not_eol) break;
6937 /* We have to ``prefetch'' the next character. */
6938 else if ((d == end1 ? *string2 : *d) == '\n'
6939 && bufp->newline_anchor)
6946 /* Match at the very beginning of the data. */
6948 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
6949 if (AT_STRINGS_BEG (d))
6954 /* Match at the very end of the data. */
6956 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
6957 if (AT_STRINGS_END (d))
6962 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6963 pushes NULL as the value for the string on the stack. Then
6964 `pop_failure_point' will keep the current value for the
6965 string, instead of restoring it. To see why, consider
6966 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6967 then the . fails against the \n. But the next thing we want
6968 to do is match the \n against the \n; if we restored the
6969 string value, we would be back at the foo.
6971 Because this is used only in specific cases, we don't need to
6972 check all the things that `on_failure_jump' does, to make
6973 sure the right things get saved on the stack. Hence we don't
6974 share its code. The only reason to push anything on the
6975 stack at all is that otherwise we would have to change
6976 `anychar's code to do something besides goto fail in this
6977 case; that seems worse than this. */
6978 case on_failure_keep_string_jump:
6979 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
6981 EXTRACT_NUMBER_AND_INCR (mcnt, p);
6983 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
6985 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
6988 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
6992 /* Uses of on_failure_jump:
6994 Each alternative starts with an on_failure_jump that points
6995 to the beginning of the next alternative. Each alternative
6996 except the last ends with a jump that in effect jumps past
6997 the rest of the alternatives. (They really jump to the
6998 ending jump of the following alternative, because tensioning
6999 these jumps is a hassle.)
7001 Repeats start with an on_failure_jump that points past both
7002 the repetition text and either the following jump or
7003 pop_failure_jump back to this on_failure_jump. */
7004 case on_failure_jump:
7006 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
7008 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7010 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
7012 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
7015 /* If this on_failure_jump comes right before a group (i.e.,
7016 the original * applied to a group), save the information
7017 for that group and all inner ones, so that if we fail back
7018 to this point, the group's information will be correct.
7019 For example, in \(a*\)*\1, we need the preceding group,
7020 and in \(zz\(a*\)b*\)\2, we need the inner group. */
7022 /* We can't use `p' to check ahead because we push
7023 a failure point to `p + mcnt' after we do this. */
7026 /* We need to skip no_op's before we look for the
7027 start_memory in case this on_failure_jump is happening as
7028 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
7030 while (p1 < pend && (re_opcode_t) *p1 == no_op)
7033 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
7035 /* We have a new highest active register now. This will
7036 get reset at the start_memory we are about to get to,
7037 but we will have saved all the registers relevant to
7038 this repetition op, as described above. */
7039 highest_active_reg = *(p1 + 1) + *(p1 + 2);
7040 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
7041 lowest_active_reg = *(p1 + 1);
7044 DEBUG_PRINT1 (":\n");
7045 PUSH_FAILURE_POINT (p + mcnt, d, -2);
7049 /* A smart repeat ends with `maybe_pop_jump'.
7050 We change it to either `pop_failure_jump' or `jump'. */
7051 case maybe_pop_jump:
7052 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7053 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
7055 register UCHAR_T *p2 = p;
7057 /* Compare the beginning of the repeat with what in the
7058 pattern follows its end. If we can establish that there
7059 is nothing that they would both match, i.e., that we
7060 would have to backtrack because of (as in, e.g., `a*a')
7061 then we can change to pop_failure_jump, because we'll
7062 never have to backtrack.
7064 This is not true in the case of alternatives: in
7065 `(a|ab)*' we do need to backtrack to the `ab' alternative
7066 (e.g., if the string was `ab'). But instead of trying to
7067 detect that here, the alternative has put on a dummy
7068 failure point which is what we will end up popping. */
7070 /* Skip over open/close-group commands.
7071 If what follows this loop is a ...+ construct,
7072 look at what begins its body, since we will have to
7073 match at least one of that. */
7077 && ((re_opcode_t) *p2 == stop_memory
7078 || (re_opcode_t) *p2 == start_memory))
7080 else if (p2 + 2 + 2 * OFFSET_ADDRESS_SIZE < pend
7081 && (re_opcode_t) *p2 == dummy_failure_jump)
7082 p2 += 2 + 2 * OFFSET_ADDRESS_SIZE;
7088 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7089 to the `maybe_finalize_jump' of this case. Examine what
7092 /* If we're at the end of the pattern, we can change. */
7095 /* Consider what happens when matching ":\(.*\)"
7096 against ":/". I don't really understand this code
7098 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7101 (" End of pattern: change to `pop_failure_jump'.\n");
7104 else if ((re_opcode_t) *p2 == exactn
7106 || (re_opcode_t) *p2 == exactn_bin
7108 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
7111 = *p2 == (UCHAR_T) endline ? '\n' : p2[2];
7113 if (((re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn
7115 || (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn_bin
7117 ) && p1[3+OFFSET_ADDRESS_SIZE] != c)
7119 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7122 DEBUG_PRINT3 (" %C != %C => pop_failure_jump.\n",
7124 (wint_t) p1[3+OFFSET_ADDRESS_SIZE]);
7126 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
7128 (char) p1[3+OFFSET_ADDRESS_SIZE]);
7133 else if ((re_opcode_t) p1[3] == charset
7134 || (re_opcode_t) p1[3] == charset_not)
7136 int not = (re_opcode_t) p1[3] == charset_not;
7138 if (c < (unsigned) (p1[4] * BYTEWIDTH)
7139 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
7142 /* `not' is equal to 1 if c would match, which means
7143 that we can't change to pop_failure_jump. */
7146 p[-3] = (unsigned char) pop_failure_jump;
7147 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7150 #endif /* not WCHAR */
7153 else if ((re_opcode_t) *p2 == charset)
7155 /* We win if the first character of the loop is not part
7157 if ((re_opcode_t) p1[3] == exactn
7158 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
7159 && (p2[2 + p1[5] / BYTEWIDTH]
7160 & (1 << (p1[5] % BYTEWIDTH)))))
7162 p[-3] = (unsigned char) pop_failure_jump;
7163 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7166 else if ((re_opcode_t) p1[3] == charset_not)
7169 /* We win if the charset_not inside the loop
7170 lists every character listed in the charset after. */
7171 for (idx = 0; idx < (int) p2[1]; idx++)
7172 if (! (p2[2 + idx] == 0
7173 || (idx < (int) p1[4]
7174 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
7179 p[-3] = (unsigned char) pop_failure_jump;
7180 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7183 else if ((re_opcode_t) p1[3] == charset)
7186 /* We win if the charset inside the loop
7187 has no overlap with the one after the loop. */
7189 idx < (int) p2[1] && idx < (int) p1[4];
7191 if ((p2[2 + idx] & p1[5 + idx]) != 0)
7194 if (idx == p2[1] || idx == p1[4])
7196 p[-3] = (unsigned char) pop_failure_jump;
7197 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7201 #endif /* not WCHAR */
7203 p -= OFFSET_ADDRESS_SIZE; /* Point at relative address again. */
7204 if ((re_opcode_t) p[-1] != pop_failure_jump)
7206 p[-1] = (UCHAR_T) jump;
7207 DEBUG_PRINT1 (" Match => jump.\n");
7208 goto unconditional_jump;
7210 /* Note fall through. */
7213 /* The end of a simple repeat has a pop_failure_jump back to
7214 its matching on_failure_jump, where the latter will push a
7215 failure point. The pop_failure_jump takes off failure
7216 points put on by this pop_failure_jump's matching
7217 on_failure_jump; we got through the pattern to here from the
7218 matching on_failure_jump, so didn't fail. */
7219 case pop_failure_jump:
7221 /* We need to pass separate storage for the lowest and
7222 highest registers, even though we don't care about the
7223 actual values. Otherwise, we will restore only one
7224 register from the stack, since lowest will == highest in
7225 `pop_failure_point'. */
7226 active_reg_t dummy_low_reg, dummy_high_reg;
7227 UCHAR_T *pdummy = NULL;
7228 const CHAR_T *sdummy = NULL;
7230 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
7231 POP_FAILURE_POINT (sdummy, pdummy,
7232 dummy_low_reg, dummy_high_reg,
7233 reg_dummy, reg_dummy, reg_info_dummy);
7235 /* Note fall through. */
7239 DEBUG_PRINT2 ("\n%p: ", p);
7241 DEBUG_PRINT2 ("\n0x%x: ", p);
7243 /* Note fall through. */
7245 /* Unconditionally jump (without popping any failure points). */
7247 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
7248 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
7249 p += mcnt; /* Do the jump. */
7251 DEBUG_PRINT2 ("(to %p).\n", p);
7253 DEBUG_PRINT2 ("(to 0x%x).\n", p);
7258 /* We need this opcode so we can detect where alternatives end
7259 in `group_match_null_string_p' et al. */
7261 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
7262 goto unconditional_jump;
7265 /* Normally, the on_failure_jump pushes a failure point, which
7266 then gets popped at pop_failure_jump. We will end up at
7267 pop_failure_jump, also, and with a pattern of, say, `a+', we
7268 are skipping over the on_failure_jump, so we have to push
7269 something meaningless for pop_failure_jump to pop. */
7270 case dummy_failure_jump:
7271 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
7272 /* It doesn't matter what we push for the string here. What
7273 the code at `fail' tests is the value for the pattern. */
7274 PUSH_FAILURE_POINT (NULL, NULL, -2);
7275 goto unconditional_jump;
7278 /* At the end of an alternative, we need to push a dummy failure
7279 point in case we are followed by a `pop_failure_jump', because
7280 we don't want the failure point for the alternative to be
7281 popped. For example, matching `(a|ab)*' against `aab'
7282 requires that we match the `ab' alternative. */
7283 case push_dummy_failure:
7284 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
7285 /* See comments just above at `dummy_failure_jump' about the
7287 PUSH_FAILURE_POINT (NULL, NULL, -2);
7290 /* Have to succeed matching what follows at least n times.
7291 After that, handle like `on_failure_jump'. */
7293 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7294 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
7297 /* Originally, this is how many times we HAVE to succeed. */
7301 p += OFFSET_ADDRESS_SIZE;
7302 STORE_NUMBER_AND_INCR (p, mcnt);
7304 DEBUG_PRINT3 (" Setting %p to %d.\n", p - OFFSET_ADDRESS_SIZE
7307 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - OFFSET_ADDRESS_SIZE
7314 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n",
7315 p + OFFSET_ADDRESS_SIZE);
7317 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n",
7318 p + OFFSET_ADDRESS_SIZE);
7322 p[1] = (UCHAR_T) no_op;
7324 p[2] = (UCHAR_T) no_op;
7325 p[3] = (UCHAR_T) no_op;
7332 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7333 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
7335 /* Originally, this is how many times we CAN jump. */
7339 STORE_NUMBER (p + OFFSET_ADDRESS_SIZE, mcnt);
7342 DEBUG_PRINT3 (" Setting %p to %d.\n", p + OFFSET_ADDRESS_SIZE,
7345 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + OFFSET_ADDRESS_SIZE,
7348 goto unconditional_jump;
7350 /* If don't have to jump any more, skip over the rest of command. */
7352 p += 2 * OFFSET_ADDRESS_SIZE;
7357 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
7359 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7361 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7363 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
7365 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
7367 STORE_NUMBER (p1, mcnt);
7372 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7373 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7374 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7375 macro and introducing temporary variables works around the bug. */
7378 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7379 if (AT_WORD_BOUNDARY (d))
7384 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7385 if (AT_WORD_BOUNDARY (d))
7391 boolean prevchar, thischar;
7393 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7394 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7397 prevchar = WORDCHAR_P (d - 1);
7398 thischar = WORDCHAR_P (d);
7399 if (prevchar != thischar)
7406 boolean prevchar, thischar;
7408 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7409 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7412 prevchar = WORDCHAR_P (d - 1);
7413 thischar = WORDCHAR_P (d);
7414 if (prevchar != thischar)
7421 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
7422 if (!AT_STRINGS_END (d) && WORDCHAR_P (d)
7423 && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
7428 DEBUG_PRINT1 ("EXECUTING wordend.\n");
7429 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
7430 && (AT_STRINGS_END (d) || !WORDCHAR_P (d)))
7436 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
7437 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
7442 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
7443 if (PTR_CHAR_POS ((unsigned char *) d) != point)
7448 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
7449 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
7454 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
7459 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
7463 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7465 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
7467 SET_REGS_MATCHED ();
7471 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
7473 goto matchnotsyntax;
7476 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
7480 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7482 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
7484 SET_REGS_MATCHED ();
7487 #else /* not emacs */
7489 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
7491 if (!WORDCHAR_P (d))
7493 SET_REGS_MATCHED ();
7498 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
7502 SET_REGS_MATCHED ();
7505 #endif /* not emacs */
7510 continue; /* Successfully executed one pattern command; keep going. */
7513 /* We goto here if a matching operation fails. */
7515 if (!FAIL_STACK_EMPTY ())
7516 { /* A restart point is known. Restore to that state. */
7517 DEBUG_PRINT1 ("\nFAIL:\n");
7518 POP_FAILURE_POINT (d, p,
7519 lowest_active_reg, highest_active_reg,
7520 regstart, regend, reg_info);
7522 /* If this failure point is a dummy, try the next one. */
7526 /* If we failed to the end of the pattern, don't examine *p. */
7530 boolean is_a_jump_n = false;
7532 /* If failed to a backwards jump that's part of a repetition
7533 loop, need to pop this failure point and use the next one. */
7534 switch ((re_opcode_t) *p)
7538 case maybe_pop_jump:
7539 case pop_failure_jump:
7542 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7545 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
7547 && (re_opcode_t) *p1 == on_failure_jump))
7555 if (d >= string1 && d <= end1)
7559 break; /* Matching at this starting point really fails. */
7563 goto restore_best_regs;
7567 return -1; /* Failure to match. */
7570 /* Subroutine definitions for re_match_2. */
7573 /* We are passed P pointing to a register number after a start_memory.
7575 Return true if the pattern up to the corresponding stop_memory can
7576 match the empty string, and false otherwise.
7578 If we find the matching stop_memory, sets P to point to one past its number.
7579 Otherwise, sets P to an undefined byte less than or equal to END.
7581 We don't handle duplicates properly (yet). */
7584 PREFIX(group_match_null_string_p) (p, end, reg_info)
7586 PREFIX(register_info_type) *reg_info;
7589 /* Point to after the args to the start_memory. */
7590 UCHAR_T *p1 = *p + 2;
7594 /* Skip over opcodes that can match nothing, and return true or
7595 false, as appropriate, when we get to one that can't, or to the
7596 matching stop_memory. */
7598 switch ((re_opcode_t) *p1)
7600 /* Could be either a loop or a series of alternatives. */
7601 case on_failure_jump:
7603 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7605 /* If the next operation is not a jump backwards in the
7610 /* Go through the on_failure_jumps of the alternatives,
7611 seeing if any of the alternatives cannot match nothing.
7612 The last alternative starts with only a jump,
7613 whereas the rest start with on_failure_jump and end
7614 with a jump, e.g., here is the pattern for `a|b|c':
7616 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7617 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7620 So, we have to first go through the first (n-1)
7621 alternatives and then deal with the last one separately. */
7624 /* Deal with the first (n-1) alternatives, which start
7625 with an on_failure_jump (see above) that jumps to right
7626 past a jump_past_alt. */
7628 while ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] ==
7631 /* `mcnt' holds how many bytes long the alternative
7632 is, including the ending `jump_past_alt' and
7635 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt -
7636 (1 + OFFSET_ADDRESS_SIZE),
7640 /* Move to right after this alternative, including the
7644 /* Break if it's the beginning of an n-th alternative
7645 that doesn't begin with an on_failure_jump. */
7646 if ((re_opcode_t) *p1 != on_failure_jump)
7649 /* Still have to check that it's not an n-th
7650 alternative that starts with an on_failure_jump. */
7652 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7653 if ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] !=
7656 /* Get to the beginning of the n-th alternative. */
7657 p1 -= 1 + OFFSET_ADDRESS_SIZE;
7662 /* Deal with the last alternative: go back and get number
7663 of the `jump_past_alt' just before it. `mcnt' contains
7664 the length of the alternative. */
7665 EXTRACT_NUMBER (mcnt, p1 - OFFSET_ADDRESS_SIZE);
7667 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt, reg_info))
7670 p1 += mcnt; /* Get past the n-th alternative. */
7676 assert (p1[1] == **p);
7682 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7685 } /* while p1 < end */
7688 } /* group_match_null_string_p */
7691 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7692 It expects P to be the first byte of a single alternative and END one
7693 byte past the last. The alternative can contain groups. */
7696 PREFIX(alt_match_null_string_p) (p, end, reg_info)
7698 PREFIX(register_info_type) *reg_info;
7705 /* Skip over opcodes that can match nothing, and break when we get
7706 to one that can't. */
7708 switch ((re_opcode_t) *p1)
7711 case on_failure_jump:
7713 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7718 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7721 } /* while p1 < end */
7724 } /* alt_match_null_string_p */
7727 /* Deals with the ops common to group_match_null_string_p and
7728 alt_match_null_string_p.
7730 Sets P to one after the op and its arguments, if any. */
7733 PREFIX(common_op_match_null_string_p) (p, end, reg_info)
7735 PREFIX(register_info_type) *reg_info;
7742 switch ((re_opcode_t) *p1++)
7762 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
7763 ret = PREFIX(group_match_null_string_p) (&p1, end, reg_info);
7765 /* Have to set this here in case we're checking a group which
7766 contains a group and a back reference to it. */
7768 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
7769 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
7775 /* If this is an optimized succeed_n for zero times, make the jump. */
7777 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7785 /* Get to the number of times to succeed. */
7786 p1 += OFFSET_ADDRESS_SIZE;
7787 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7791 p1 -= 2 * OFFSET_ADDRESS_SIZE;
7792 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7800 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
7805 p1 += 2 * OFFSET_ADDRESS_SIZE;
7808 /* All other opcodes mean we cannot match the empty string. */
7814 } /* common_op_match_null_string_p */
7817 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7818 bytes; nonzero otherwise. */
7821 PREFIX(bcmp_translate) (s1, s2, len, translate)
7822 const CHAR_T *s1, *s2;
7824 RE_TRANSLATE_TYPE translate;
7826 register const UCHAR_T *p1 = (const UCHAR_T *) s1;
7827 register const UCHAR_T *p2 = (const UCHAR_T *) s2;
7831 if (((*p1<=0xff)?translate[*p1++]:*p1++)
7832 != ((*p2<=0xff)?translate[*p2++]:*p2++))
7835 if (translate[*p1++] != translate[*p2++]) return 1;
7843 #else /* not INSIDE_RECURSION */
7845 /* Entry points for GNU code. */
7847 /* re_compile_pattern is the GNU regular expression compiler: it
7848 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7849 Returns 0 if the pattern was valid, otherwise an error string.
7851 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7852 are set in BUFP on entry.
7854 We call regex_compile to do the actual compilation. */
7857 re_compile_pattern (pattern, length, bufp)
7858 const char *pattern;
7860 struct re_pattern_buffer *bufp;
7864 /* GNU code is written to assume at least RE_NREGS registers will be set
7865 (and at least one extra will be -1). */
7866 bufp->regs_allocated = REGS_UNALLOCATED;
7868 /* And GNU code determines whether or not to get register information
7869 by passing null for the REGS argument to re_match, etc., not by
7873 /* Match anchors at newline. */
7874 bufp->newline_anchor = 1;
7877 if (MB_CUR_MAX != 1)
7878 ret = wcs_regex_compile (pattern, length, re_syntax_options, bufp);
7881 ret = byte_regex_compile (pattern, length, re_syntax_options, bufp);
7885 return gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
7888 weak_alias (__re_compile_pattern, re_compile_pattern)
7891 /* Entry points compatible with 4.2 BSD regex library. We don't define
7892 them unless specifically requested. */
7894 #if defined _REGEX_RE_COMP || defined _LIBC
7896 /* BSD has one and only one pattern buffer. */
7897 static struct re_pattern_buffer re_comp_buf;
7901 /* Make these definitions weak in libc, so POSIX programs can redefine
7902 these names if they don't use our functions, and still use
7903 regcomp/regexec below without link errors. */
7913 if (!re_comp_buf.buffer)
7914 return gettext ("No previous regular expression");
7918 if (!re_comp_buf.buffer)
7920 re_comp_buf.buffer = (unsigned char *) malloc (200);
7921 if (re_comp_buf.buffer == NULL)
7922 return (char *) gettext (re_error_msgid
7923 + re_error_msgid_idx[(int) REG_ESPACE]);
7924 re_comp_buf.allocated = 200;
7926 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
7927 if (re_comp_buf.fastmap == NULL)
7928 return (char *) gettext (re_error_msgid
7929 + re_error_msgid_idx[(int) REG_ESPACE]);
7932 /* Since `re_exec' always passes NULL for the `regs' argument, we
7933 don't need to initialize the pattern buffer fields which affect it. */
7935 /* Match anchors at newlines. */
7936 re_comp_buf.newline_anchor = 1;
7939 if (MB_CUR_MAX != 1)
7940 ret = wcs_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7943 ret = byte_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7948 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7949 return (char *) gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
7960 const int len = strlen (s);
7962 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
7965 #endif /* _REGEX_RE_COMP */
7967 /* POSIX.2 functions. Don't define these for Emacs. */
7971 /* regcomp takes a regular expression as a string and compiles it.
7973 PREG is a regex_t *. We do not expect any fields to be initialized,
7974 since POSIX says we shouldn't. Thus, we set
7976 `buffer' to the compiled pattern;
7977 `used' to the length of the compiled pattern;
7978 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7979 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7980 RE_SYNTAX_POSIX_BASIC;
7981 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7982 `fastmap' to an allocated space for the fastmap;
7983 `fastmap_accurate' to zero;
7984 `re_nsub' to the number of subexpressions in PATTERN.
7986 PATTERN is the address of the pattern string.
7988 CFLAGS is a series of bits which affect compilation.
7990 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7991 use POSIX basic syntax.
7993 If REG_NEWLINE is set, then . and [^...] don't match newline.
7994 Also, regexec will try a match beginning after every newline.
7996 If REG_ICASE is set, then we considers upper- and lowercase
7997 versions of letters to be equivalent when matching.
7999 If REG_NOSUB is set, then when PREG is passed to regexec, that
8000 routine will report only success or failure, and nothing about the
8003 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
8004 the return codes and their meanings.) */
8007 regcomp (preg, pattern, cflags)
8009 const char *pattern;
8014 = (cflags & REG_EXTENDED) ?
8015 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
8017 /* regex_compile will allocate the space for the compiled pattern. */
8019 preg->allocated = 0;
8022 /* Try to allocate space for the fastmap. */
8023 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
8025 if (cflags & REG_ICASE)
8030 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
8031 * sizeof (*(RE_TRANSLATE_TYPE)0));
8032 if (preg->translate == NULL)
8033 return (int) REG_ESPACE;
8035 /* Map uppercase characters to corresponding lowercase ones. */
8036 for (i = 0; i < CHAR_SET_SIZE; i++)
8037 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
8040 preg->translate = NULL;
8042 /* If REG_NEWLINE is set, newlines are treated differently. */
8043 if (cflags & REG_NEWLINE)
8044 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
8045 syntax &= ~RE_DOT_NEWLINE;
8046 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
8047 /* It also changes the matching behavior. */
8048 preg->newline_anchor = 1;
8051 preg->newline_anchor = 0;
8053 preg->no_sub = !!(cflags & REG_NOSUB);
8055 /* POSIX says a null character in the pattern terminates it, so we
8056 can use strlen here in compiling the pattern. */
8058 if (MB_CUR_MAX != 1)
8059 ret = wcs_regex_compile (pattern, strlen (pattern), syntax, preg);
8062 ret = byte_regex_compile (pattern, strlen (pattern), syntax, preg);
8064 /* POSIX doesn't distinguish between an unmatched open-group and an
8065 unmatched close-group: both are REG_EPAREN. */
8066 if (ret == REG_ERPAREN) ret = REG_EPAREN;
8068 if (ret == REG_NOERROR && preg->fastmap)
8070 /* Compute the fastmap now, since regexec cannot modify the pattern
8072 if (re_compile_fastmap (preg) == -2)
8074 /* Some error occurred while computing the fastmap, just forget
8076 free (preg->fastmap);
8077 preg->fastmap = NULL;
8084 weak_alias (__regcomp, regcomp)
8088 /* regexec searches for a given pattern, specified by PREG, in the
8091 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8092 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8093 least NMATCH elements, and we set them to the offsets of the
8094 corresponding matched substrings.
8096 EFLAGS specifies `execution flags' which affect matching: if
8097 REG_NOTBOL is set, then ^ does not match at the beginning of the
8098 string; if REG_NOTEOL is set, then $ does not match at the end.
8100 We return 0 if we find a match and REG_NOMATCH if not. */
8103 regexec (preg, string, nmatch, pmatch, eflags)
8104 const regex_t *preg;
8107 regmatch_t pmatch[];
8111 struct re_registers regs;
8112 regex_t private_preg;
8113 int len = strlen (string);
8114 boolean want_reg_info = !preg->no_sub && nmatch > 0;
8116 private_preg = *preg;
8118 private_preg.not_bol = !!(eflags & REG_NOTBOL);
8119 private_preg.not_eol = !!(eflags & REG_NOTEOL);
8121 /* The user has told us exactly how many registers to return
8122 information about, via `nmatch'. We have to pass that on to the
8123 matching routines. */
8124 private_preg.regs_allocated = REGS_FIXED;
8128 regs.num_regs = nmatch;
8129 regs.start = TALLOC (nmatch * 2, regoff_t);
8130 if (regs.start == NULL)
8131 return (int) REG_NOMATCH;
8132 regs.end = regs.start + nmatch;
8135 /* Perform the searching operation. */
8136 ret = re_search (&private_preg, string, len,
8137 /* start: */ 0, /* range: */ len,
8138 want_reg_info ? ®s : (struct re_registers *) 0);
8140 /* Copy the register information to the POSIX structure. */
8147 for (r = 0; r < nmatch; r++)
8149 pmatch[r].rm_so = regs.start[r];
8150 pmatch[r].rm_eo = regs.end[r];
8154 /* If we needed the temporary register info, free the space now. */
8158 /* We want zero return to mean success, unlike `re_search'. */
8159 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
8162 weak_alias (__regexec, regexec)
8166 /* Returns a message corresponding to an error code, ERRCODE, returned
8167 from either regcomp or regexec. We don't use PREG here. */
8170 regerror (errcode, preg, errbuf, errbuf_size)
8172 const regex_t *preg;
8180 || errcode >= (int) (sizeof (re_error_msgid_idx)
8181 / sizeof (re_error_msgid_idx[0])))
8182 /* Only error codes returned by the rest of the code should be passed
8183 to this routine. If we are given anything else, or if other regex
8184 code generates an invalid error code, then the program has a bug.
8185 Dump core so we can fix it. */
8188 msg = gettext (re_error_msgid + re_error_msgid_idx[errcode]);
8190 msg_size = strlen (msg) + 1; /* Includes the null. */
8192 if (errbuf_size != 0)
8194 if (msg_size > errbuf_size)
8196 #if defined HAVE_MEMPCPY || defined _LIBC
8197 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
8199 memcpy (errbuf, msg, errbuf_size - 1);
8200 errbuf[errbuf_size - 1] = 0;
8204 memcpy (errbuf, msg, msg_size);
8210 weak_alias (__regerror, regerror)
8214 /* Free dynamically allocated space used by PREG. */
8220 if (preg->buffer != NULL)
8221 free (preg->buffer);
8222 preg->buffer = NULL;
8224 preg->allocated = 0;
8227 if (preg->fastmap != NULL)
8228 free (preg->fastmap);
8229 preg->fastmap = NULL;
8230 preg->fastmap_accurate = 0;
8232 if (preg->translate != NULL)
8233 free (preg->translate);
8234 preg->translate = NULL;
8237 weak_alias (__regfree, regfree)
8240 #endif /* not emacs */
8242 #endif /* not INSIDE_RECURSION */
8246 #undef STORE_NUMBER_AND_INCR
8247 #undef EXTRACT_NUMBER
8248 #undef EXTRACT_NUMBER_AND_INCR
8250 #undef DEBUG_PRINT_COMPILED_PATTERN
8251 #undef DEBUG_PRINT_DOUBLE_STRING
8253 #undef INIT_FAIL_STACK
8254 #undef RESET_FAIL_STACK
8255 #undef DOUBLE_FAIL_STACK
8256 #undef PUSH_PATTERN_OP
8257 #undef PUSH_FAILURE_POINTER
8258 #undef PUSH_FAILURE_INT
8259 #undef PUSH_FAILURE_ELT
8260 #undef POP_FAILURE_POINTER
8261 #undef POP_FAILURE_INT
8262 #undef POP_FAILURE_ELT
8265 #undef PUSH_FAILURE_POINT
8266 #undef POP_FAILURE_POINT
8268 #undef REG_UNSET_VALUE
8276 #undef INIT_BUF_SIZE
8277 #undef GET_BUFFER_SPACE
8285 #undef EXTEND_BUFFER
8286 #undef GET_UNSIGNED_NUMBER
8287 #undef FREE_STACK_RETURN
8289 # undef POINTER_TO_OFFSET
8290 # undef MATCHING_IN_FRST_STRING
8292 # undef AT_STRINGS_BEG
8293 # undef AT_STRINGS_END
8296 # undef FREE_VARIABLES
8297 # undef NO_HIGHEST_ACTIVE_REG
8298 # undef NO_LOWEST_ACTIVE_REG
8302 # undef COMPILED_BUFFER_VAR
8303 # undef OFFSET_ADDRESS_SIZE
8304 # undef CHAR_CLASS_SIZE
8311 # define DEFINED_ONCE
projects / gnulib.git / blob