1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
6 Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
7 2002, 2005 Free Software Foundation, Inc.
8 This file is part of the GNU C Library.
10 The GNU C Library is free software; you can redistribute it and/or
11 modify it under the terms of the GNU Lesser General Public
12 License as published by the Free Software Foundation; either
13 version 2.1 of the License, or (at your option) any later version.
15 The GNU C Library is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 Lesser General Public License for more details.
20 You should have received a copy of the GNU Lesser General Public
21 License along with the GNU C Library; if not, write to the Free
22 Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
25 /* AIX requires this to be the first thing in the file. */
26 #if defined _AIX && !defined __GNUC__ && !defined REGEX_MALLOC
33 #ifndef INSIDE_RECURSION
39 /*#include <ansidecl.h>*/
41 #ifndef INSIDE_RECURSION
43 # if defined STDC_HEADERS && !defined emacs
46 /* We need this for `regex.h', and perhaps for the Emacs include files. */
47 # include <sys/types.h>
50 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
52 /* For platform which support the ISO C amendement 1 functionality we
53 support user defined character classes. */
54 # if WIDE_CHAR_SUPPORT
55 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
61 /* We have to keep the namespace clean. */
62 # define regfree(preg) __regfree (preg)
63 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
64 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
65 # define regerror(errcode, preg, errbuf, errbuf_size) \
66 __regerror(errcode, preg, errbuf, errbuf_size)
67 # define re_set_registers(bu, re, nu, st, en) \
68 __re_set_registers (bu, re, nu, st, en)
69 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
70 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
71 # define re_match(bufp, string, size, pos, regs) \
72 __re_match (bufp, string, size, pos, regs)
73 # define re_search(bufp, string, size, startpos, range, regs) \
74 __re_search (bufp, string, size, startpos, range, regs)
75 # define re_compile_pattern(pattern, length, bufp) \
76 __re_compile_pattern (pattern, length, bufp)
77 # define re_set_syntax(syntax) __re_set_syntax (syntax)
78 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
79 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
80 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
82 # define btowc __btowc
84 /* We are also using some library internals. */
85 # include <locale/localeinfo.h>
86 # include <locale/elem-hash.h>
87 # include <langinfo.h>
88 # include <locale/coll-lookup.h>
91 /* This is for other GNU distributions with internationalized messages. */
92 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
96 # define gettext(msgid) __dcgettext ("libc", msgid, LC_MESSAGES)
99 # define gettext(msgid) (msgid)
102 # ifndef gettext_noop
103 /* This define is so xgettext can find the internationalizable
105 # define gettext_noop(String) String
108 /* The `emacs' switch turns on certain matching commands
109 that make sense only in Emacs. */
116 # else /* not emacs */
118 /* If we are not linking with Emacs proper,
119 we can't use the relocating allocator
120 even if config.h says that we can. */
123 # if defined STDC_HEADERS || defined _LIBC
130 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
131 If nothing else has been done, use the method below. */
132 # ifdef INHIBIT_STRING_HEADER
133 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
134 # if !defined bzero && !defined bcopy
135 # undef INHIBIT_STRING_HEADER
140 /* This is the normal way of making sure we have a bcopy and a bzero.
141 This is used in most programs--a few other programs avoid this
142 by defining INHIBIT_STRING_HEADER. */
143 # ifndef INHIBIT_STRING_HEADER
144 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
148 # define bzero(s, n) (memset (s, '\0', n), (s))
150 # define bzero(s, n) __bzero (s, n)
154 # include <strings.h>
156 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
159 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
164 /* Define the syntax stuff for \<, \>, etc. */
166 /* This must be nonzero for the wordchar and notwordchar pattern
167 commands in re_match_2. */
172 # ifdef SWITCH_ENUM_BUG
173 # define SWITCH_ENUM_CAST(x) ((int)(x))
175 # define SWITCH_ENUM_CAST(x) (x)
178 # endif /* not emacs */
180 # if defined _LIBC || HAVE_LIMITS_H
185 # define MB_LEN_MAX 1
188 /* Get the interface, including the syntax bits. */
191 /* isalpha etc. are used for the character classes. */
194 /* Jim Meyering writes:
196 "... Some ctype macros are valid only for character codes that
197 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
198 using /bin/cc or gcc but without giving an ansi option). So, all
199 ctype uses should be through macros like ISPRINT... If
200 STDC_HEADERS is defined, then autoconf has verified that the ctype
201 macros don't need to be guarded with references to isascii. ...
202 Defining isascii to 1 should let any compiler worth its salt
203 eliminate the && through constant folding."
204 Solaris defines some of these symbols so we must undefine them first. */
207 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
208 # define ISASCII(c) 1
210 # define ISASCII(c) isascii(c)
214 # define ISBLANK(c) (ISASCII (c) && isblank (c))
216 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
219 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
221 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
225 # define ISPRINT(c) (ISASCII (c) && isprint (c))
226 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
227 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
228 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
229 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
230 # define ISLOWER(c) (ISASCII (c) && islower (c))
231 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
232 # define ISSPACE(c) (ISASCII (c) && isspace (c))
233 # define ISUPPER(c) (ISASCII (c) && isupper (c))
234 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
237 # define TOLOWER(c) _tolower(c)
239 # define TOLOWER(c) tolower(c)
243 # define NULL (void *)0
246 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
247 since ours (we hope) works properly with all combinations of
248 machines, compilers, `char' and `unsigned char' argument types.
249 (Per Bothner suggested the basic approach.) */
250 # undef SIGN_EXTEND_CHAR
252 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
253 # else /* not __STDC__ */
254 /* As in Harbison and Steele. */
255 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
259 /* How many characters in the character set. */
260 # define CHAR_SET_SIZE 256
264 extern char *re_syntax_table;
266 # else /* not SYNTAX_TABLE */
268 static char re_syntax_table[CHAR_SET_SIZE];
270 static void init_syntax_once (void);
273 init_syntax_once (void)
280 bzero (re_syntax_table, sizeof re_syntax_table);
282 for (c = 0; c < CHAR_SET_SIZE; ++c)
284 re_syntax_table[c] = Sword;
286 re_syntax_table['_'] = Sword;
291 # endif /* not SYNTAX_TABLE */
293 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
297 /* Integer type for pointers. */
298 # if !defined _LIBC && !defined HAVE_UINTPTR_T
299 typedef unsigned long int uintptr_t;
302 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
303 use `alloca' instead of `malloc'. This is because using malloc in
304 re_search* or re_match* could cause memory leaks when C-g is used in
305 Emacs; also, malloc is slower and causes storage fragmentation. On
306 the other hand, malloc is more portable, and easier to debug.
308 Because we sometimes use alloca, some routines have to be macros,
309 not functions -- `alloca'-allocated space disappears at the end of the
310 function it is called in. */
314 # define REGEX_ALLOCATE malloc
315 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
316 # define REGEX_FREE free
318 # else /* not REGEX_MALLOC */
320 /* Emacs already defines alloca, sometimes. */
323 /* Make alloca work the best possible way. */
325 # define alloca __builtin_alloca
326 # else /* not __GNUC__ */
329 # endif /* HAVE_ALLOCA_H */
330 # endif /* not __GNUC__ */
332 # endif /* not alloca */
334 # define REGEX_ALLOCATE alloca
336 /* Assumes a `char *destination' variable. */
337 # define REGEX_REALLOCATE(source, osize, nsize) \
338 (destination = (char *) alloca (nsize), \
339 memcpy (destination, source, osize))
341 /* No need to do anything to free, after alloca. */
342 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
344 # endif /* not REGEX_MALLOC */
346 /* Define how to allocate the failure stack. */
348 # if defined REL_ALLOC && defined REGEX_MALLOC
350 # define REGEX_ALLOCATE_STACK(size) \
351 r_alloc (&failure_stack_ptr, (size))
352 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
353 r_re_alloc (&failure_stack_ptr, (nsize))
354 # define REGEX_FREE_STACK(ptr) \
355 r_alloc_free (&failure_stack_ptr)
357 # else /* not using relocating allocator */
361 # define REGEX_ALLOCATE_STACK malloc
362 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
363 # define REGEX_FREE_STACK free
365 # else /* not REGEX_MALLOC */
367 # define REGEX_ALLOCATE_STACK alloca
369 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
370 REGEX_REALLOCATE (source, osize, nsize)
371 /* No need to explicitly free anything. */
372 # define REGEX_FREE_STACK(arg)
374 # endif /* not REGEX_MALLOC */
375 # endif /* not using relocating allocator */
378 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
379 `string1' or just past its end. This works if PTR is NULL, which is
381 # define FIRST_STRING_P(ptr) \
382 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
384 /* (Re)Allocate N items of type T using malloc, or fail. */
385 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
386 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
387 # define RETALLOC_IF(addr, n, t) \
388 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
389 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
391 # define BYTEWIDTH 8 /* In bits. */
393 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
397 # define MAX(a, b) ((a) > (b) ? (a) : (b))
398 # define MIN(a, b) ((a) < (b) ? (a) : (b))
400 typedef char boolean;
404 static reg_errcode_t byte_regex_compile (const char *pattern, size_t size,
406 struct re_pattern_buffer *bufp);
408 static int byte_re_match_2_internal (struct re_pattern_buffer *bufp,
409 const char *string1, int size1,
410 const char *string2, int size2,
412 struct re_registers *regs,
414 static int byte_re_search_2 (struct re_pattern_buffer *bufp,
415 const char *string1, int size1,
416 const char *string2, int size2,
417 int startpos, int range,
418 struct re_registers *regs, int stop);
419 static int byte_re_compile_fastmap (struct re_pattern_buffer *bufp);
422 static reg_errcode_t wcs_regex_compile (const char *pattern, size_t size,
424 struct re_pattern_buffer *bufp);
427 static int wcs_re_match_2_internal (struct re_pattern_buffer *bufp,
428 const char *cstring1, int csize1,
429 const char *cstring2, int csize2,
431 struct re_registers *regs,
433 wchar_t *string1, int size1,
434 wchar_t *string2, int size2,
435 int *mbs_offset1, int *mbs_offset2);
436 static int wcs_re_search_2 (struct re_pattern_buffer *bufp,
437 const char *string1, int size1,
438 const char *string2, int size2,
439 int startpos, int range,
440 struct re_registers *regs, int stop);
441 static int wcs_re_compile_fastmap (struct re_pattern_buffer *bufp);
444 /* These are the command codes that appear in compiled regular
445 expressions. Some opcodes are followed by argument bytes. A
446 command code can specify any interpretation whatsoever for its
447 arguments. Zero bytes may appear in the compiled regular expression. */
453 /* Succeed right away--no more backtracking. */
456 /* Followed by one byte giving n, then by n literal bytes. */
460 /* Same as exactn, but contains binary data. */
464 /* Matches any (more or less) character. */
467 /* Matches any one char belonging to specified set. First
468 following byte is number of bitmap bytes. Then come bytes
469 for a bitmap saying which chars are in. Bits in each byte
470 are ordered low-bit-first. A character is in the set if its
471 bit is 1. A character too large to have a bit in the map is
472 automatically not in the set. */
473 /* ifdef MBS_SUPPORT, following element is length of character
474 classes, length of collating symbols, length of equivalence
475 classes, length of character ranges, and length of characters.
476 Next, character class element, collating symbols elements,
477 equivalence class elements, range elements, and character
479 See regex_compile function. */
482 /* Same parameters as charset, but match any character that is
483 not one of those specified. */
486 /* Start remembering the text that is matched, for storing in a
487 register. Followed by one byte with the register number, in
488 the range 0 to one less than the pattern buffer's re_nsub
489 field. Then followed by one byte with the number of groups
490 inner to this one. (This last has to be part of the
491 start_memory only because we need it in the on_failure_jump
495 /* Stop remembering the text that is matched and store it in a
496 memory register. Followed by one byte with the register
497 number, in the range 0 to one less than `re_nsub' in the
498 pattern buffer, and one byte with the number of inner groups,
499 just like `start_memory'. (We need the number of inner
500 groups here because we don't have any easy way of finding the
501 corresponding start_memory when we're at a stop_memory.) */
504 /* Match a duplicate of something remembered. Followed by one
505 byte containing the register number. */
508 /* Fail unless at beginning of line. */
511 /* Fail unless at end of line. */
514 /* Succeeds if at beginning of buffer (if emacs) or at beginning
515 of string to be matched (if not). */
518 /* Analogously, for end of buffer/string. */
521 /* Followed by two byte relative address to which to jump. */
524 /* Same as jump, but marks the end of an alternative. */
527 /* Followed by two-byte relative address of place to resume at
528 in case of failure. */
529 /* ifdef MBS_SUPPORT, the size of address is 1. */
532 /* Like on_failure_jump, but pushes a placeholder instead of the
533 current string position when executed. */
534 on_failure_keep_string_jump,
536 /* Throw away latest failure point and then jump to following
537 two-byte relative address. */
538 /* ifdef MBS_SUPPORT, the size of address is 1. */
541 /* Change to pop_failure_jump if know won't have to backtrack to
542 match; otherwise change to jump. This is used to jump
543 back to the beginning of a repeat. If what follows this jump
544 clearly won't match what the repeat does, such that we can be
545 sure that there is no use backtracking out of repetitions
546 already matched, then we change it to a pop_failure_jump.
547 Followed by two-byte address. */
548 /* ifdef MBS_SUPPORT, the size of address is 1. */
551 /* Jump to following two-byte address, and push a dummy failure
552 point. This failure point will be thrown away if an attempt
553 is made to use it for a failure. A `+' construct makes this
554 before the first repeat. Also used as an intermediary kind
555 of jump when compiling an alternative. */
556 /* ifdef MBS_SUPPORT, the size of address is 1. */
559 /* Push a dummy failure point and continue. Used at the end of
563 /* Followed by two-byte relative address and two-byte number n.
564 After matching N times, jump to the address upon failure. */
565 /* ifdef MBS_SUPPORT, the size of address is 1. */
568 /* Followed by two-byte relative address, and two-byte number n.
569 Jump to the address N times, then fail. */
570 /* ifdef MBS_SUPPORT, the size of address is 1. */
573 /* Set the following two-byte relative address to the
574 subsequent two-byte number. The address *includes* the two
576 /* ifdef MBS_SUPPORT, the size of address is 1. */
579 wordchar, /* Matches any word-constituent character. */
580 notwordchar, /* Matches any char that is not a word-constituent. */
582 wordbeg, /* Succeeds if at word beginning. */
583 wordend, /* Succeeds if at word end. */
585 wordbound, /* Succeeds if at a word boundary. */
586 notwordbound /* Succeeds if not at a word boundary. */
589 ,before_dot, /* Succeeds if before point. */
590 at_dot, /* Succeeds if at point. */
591 after_dot, /* Succeeds if after point. */
593 /* Matches any character whose syntax is specified. Followed by
594 a byte which contains a syntax code, e.g., Sword. */
597 /* Matches any character whose syntax is not that specified. */
601 #endif /* not INSIDE_RECURSION */
606 # define UCHAR_T unsigned char
607 # define COMPILED_BUFFER_VAR bufp->buffer
608 # define OFFSET_ADDRESS_SIZE 2
609 # define PREFIX(name) byte_##name
610 # define ARG_PREFIX(name) name
611 # define PUT_CHAR(c) putchar (c)
614 # define CHAR_T wchar_t
615 # define UCHAR_T wchar_t
616 # define COMPILED_BUFFER_VAR wc_buffer
617 # define OFFSET_ADDRESS_SIZE 1 /* the size which STORE_NUMBER macro use */
618 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
619 # define PREFIX(name) wcs_##name
620 # define ARG_PREFIX(name) c##name
621 /* Should we use wide stream?? */
622 # define PUT_CHAR(c) printf ("%C", c);
628 # define INSIDE_RECURSION
630 # undef INSIDE_RECURSION
633 # define INSIDE_RECURSION
635 # undef INSIDE_RECURSION
639 #ifdef INSIDE_RECURSION
640 /* Common operations on the compiled pattern. */
642 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
643 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
646 # define STORE_NUMBER(destination, number) \
648 *(destination) = (UCHAR_T)(number); \
651 # define STORE_NUMBER(destination, number) \
653 (destination)[0] = (number) & 0377; \
654 (destination)[1] = (number) >> 8; \
658 /* Same as STORE_NUMBER, except increment DESTINATION to
659 the byte after where the number is stored. Therefore, DESTINATION
660 must be an lvalue. */
661 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
663 # define STORE_NUMBER_AND_INCR(destination, number) \
665 STORE_NUMBER (destination, number); \
666 (destination) += OFFSET_ADDRESS_SIZE; \
669 /* Put into DESTINATION a number stored in two contiguous bytes starting
671 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
674 # define EXTRACT_NUMBER(destination, source) \
676 (destination) = *(source); \
679 # define EXTRACT_NUMBER(destination, source) \
681 (destination) = *(source) & 0377; \
682 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
687 static void PREFIX(extract_number) (int *dest, UCHAR_T *source);
689 PREFIX(extract_number) (int *dest, UCHAR_T *source)
694 int temp = SIGN_EXTEND_CHAR (*(source + 1));
695 *dest = *source & 0377;
700 # ifndef EXTRACT_MACROS /* To debug the macros. */
701 # undef EXTRACT_NUMBER
702 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
703 # endif /* not EXTRACT_MACROS */
707 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
708 SOURCE must be an lvalue. */
710 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
712 EXTRACT_NUMBER (destination, source); \
713 (source) += OFFSET_ADDRESS_SIZE; \
717 static void PREFIX(extract_number_and_incr) (int *destination,
720 PREFIX(extract_number_and_incr) (int *destination, UCHAR_T **source)
722 PREFIX(extract_number) (destination, *source);
723 *source += OFFSET_ADDRESS_SIZE;
726 # ifndef EXTRACT_MACROS
727 # undef EXTRACT_NUMBER_AND_INCR
728 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
729 PREFIX(extract_number_and_incr) (&dest, &src)
730 # endif /* not EXTRACT_MACROS */
736 /* If DEBUG is defined, Regex prints many voluminous messages about what
737 it is doing (if the variable `debug' is nonzero). If linked with the
738 main program in `iregex.c', you can enter patterns and strings
739 interactively. And if linked with the main program in `main.c' and
740 the other test files, you can run the already-written tests. */
744 # ifndef DEFINED_ONCE
746 /* We use standard I/O for debugging. */
749 /* It is useful to test things that ``must'' be true when debugging. */
754 # define DEBUG_STATEMENT(e) e
755 # define DEBUG_PRINT1(x) if (debug) printf (x)
756 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
757 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
758 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
759 # endif /* not DEFINED_ONCE */
761 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
762 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
763 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
764 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
767 /* Print the fastmap in human-readable form. */
769 # ifndef DEFINED_ONCE
771 print_fastmap (char *fastmap)
773 unsigned was_a_range = 0;
776 while (i < (1 << BYTEWIDTH))
782 while (i < (1 << BYTEWIDTH) && fastmap[i])
796 # endif /* not DEFINED_ONCE */
799 /* Print a compiled pattern string in human-readable form, starting at
800 the START pointer into it and ending just before the pointer END. */
803 PREFIX(print_partial_compiled_pattern) (UCHAR_T *start, UCHAR_T *end)
816 /* Loop over pattern commands. */
820 printf ("%td:\t", p - start);
822 printf ("%ld:\t", (long int) (p - start));
825 switch ((re_opcode_t) *p++)
833 printf ("/exactn/%d", mcnt);
845 printf ("/exactn_bin/%d", mcnt);
848 printf("/%lx", (long int) *p++);
852 # endif /* MBS_SUPPORT */
856 printf ("/start_memory/%d/%ld", mcnt, (long int) *p++);
861 printf ("/stop_memory/%d/%ld", mcnt, (long int) *p++);
865 printf ("/duplicate/%ld", (long int) *p++);
878 printf ("/charset [%s",
879 (re_opcode_t) *(workp - 1) == charset_not ? "^" : "");
881 length = *workp++; /* the length of char_classes */
882 for (i=0 ; i<length ; i++)
883 printf("[:%lx:]", (long int) *p++);
884 length = *workp++; /* the length of collating_symbol */
885 for (i=0 ; i<length ;)
889 PUT_CHAR((i++,*p++));
893 length = *workp++; /* the length of equivalence_class */
894 for (i=0 ; i<length ;)
898 PUT_CHAR((i++,*p++));
902 length = *workp++; /* the length of char_range */
903 for (i=0 ; i<length ; i++)
905 wchar_t range_start = *p++;
906 wchar_t range_end = *p++;
907 printf("%C-%C", range_start, range_end);
909 length = *workp++; /* the length of char */
910 for (i=0 ; i<length ; i++)
914 register int c, last = -100;
915 register int in_range = 0;
917 printf ("/charset [%s",
918 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
920 assert (p + *p < pend);
922 for (c = 0; c < 256; c++)
924 && (p[1 + (c/8)] & (1 << (c % 8))))
926 /* Are we starting a range? */
927 if (last + 1 == c && ! in_range)
932 /* Have we broken a range? */
933 else if (last + 1 != c && in_range)
963 case on_failure_jump:
964 PREFIX(extract_number_and_incr) (&mcnt, &p);
966 printf ("/on_failure_jump to %td", p + mcnt - start);
968 printf ("/on_failure_jump to %ld", (long int) (p + mcnt - start));
972 case on_failure_keep_string_jump:
973 PREFIX(extract_number_and_incr) (&mcnt, &p);
975 printf ("/on_failure_keep_string_jump to %td", p + mcnt - start);
977 printf ("/on_failure_keep_string_jump to %ld",
978 (long int) (p + mcnt - start));
982 case dummy_failure_jump:
983 PREFIX(extract_number_and_incr) (&mcnt, &p);
985 printf ("/dummy_failure_jump to %td", p + mcnt - start);
987 printf ("/dummy_failure_jump to %ld", (long int) (p + mcnt - start));
991 case push_dummy_failure:
992 printf ("/push_dummy_failure");
996 PREFIX(extract_number_and_incr) (&mcnt, &p);
998 printf ("/maybe_pop_jump to %td", p + mcnt - start);
1000 printf ("/maybe_pop_jump to %ld", (long int) (p + mcnt - start));
1004 case pop_failure_jump:
1005 PREFIX(extract_number_and_incr) (&mcnt, &p);
1007 printf ("/pop_failure_jump to %td", p + mcnt - start);
1009 printf ("/pop_failure_jump to %ld", (long int) (p + mcnt - start));
1014 PREFIX(extract_number_and_incr) (&mcnt, &p);
1016 printf ("/jump_past_alt to %td", p + mcnt - start);
1018 printf ("/jump_past_alt to %ld", (long int) (p + mcnt - start));
1023 PREFIX(extract_number_and_incr) (&mcnt, &p);
1025 printf ("/jump to %td", p + mcnt - start);
1027 printf ("/jump to %ld", (long int) (p + mcnt - start));
1032 PREFIX(extract_number_and_incr) (&mcnt, &p);
1034 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1036 printf ("/succeed_n to %td, %d times", p1 - start, mcnt2);
1038 printf ("/succeed_n to %ld, %d times",
1039 (long int) (p1 - start), mcnt2);
1044 PREFIX(extract_number_and_incr) (&mcnt, &p);
1046 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1047 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
1051 PREFIX(extract_number_and_incr) (&mcnt, &p);
1053 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1055 printf ("/set_number_at location %td to %d", p1 - start, mcnt2);
1057 printf ("/set_number_at location %ld to %d",
1058 (long int) (p1 - start), mcnt2);
1063 printf ("/wordbound");
1067 printf ("/notwordbound");
1071 printf ("/wordbeg");
1075 printf ("/wordend");
1080 printf ("/before_dot");
1088 printf ("/after_dot");
1092 printf ("/syntaxspec");
1094 printf ("/%d", mcnt);
1098 printf ("/notsyntaxspec");
1100 printf ("/%d", mcnt);
1105 printf ("/wordchar");
1109 printf ("/notwordchar");
1121 printf ("?%ld", (long int) *(p-1));
1128 printf ("%td:\tend of pattern.\n", p - start);
1130 printf ("%ld:\tend of pattern.\n", (long int) (p - start));
1136 PREFIX(print_compiled_pattern) (struct re_pattern_buffer *bufp)
1138 UCHAR_T *buffer = (UCHAR_T*) bufp->buffer;
1140 PREFIX(print_partial_compiled_pattern) (buffer, buffer
1141 + bufp->used / sizeof(UCHAR_T));
1142 printf ("%ld bytes used/%ld bytes allocated.\n",
1143 bufp->used, bufp->allocated);
1145 if (bufp->fastmap_accurate && bufp->fastmap)
1147 printf ("fastmap: ");
1148 print_fastmap (bufp->fastmap);
1152 printf ("re_nsub: %Zd\t", bufp->re_nsub);
1154 printf ("re_nsub: %ld\t", (long int) bufp->re_nsub);
1156 printf ("regs_alloc: %d\t", bufp->regs_allocated);
1157 printf ("can_be_null: %d\t", bufp->can_be_null);
1158 printf ("newline_anchor: %d\n", bufp->newline_anchor);
1159 printf ("no_sub: %d\t", bufp->no_sub);
1160 printf ("not_bol: %d\t", bufp->not_bol);
1161 printf ("not_eol: %d\t", bufp->not_eol);
1162 printf ("syntax: %lx\n", bufp->syntax);
1163 /* Perhaps we should print the translate table? */
1168 PREFIX(print_double_string) (const CHAR_T *where, const CHAR_T *string1,
1169 int size1, const CHAR_T *string2, int size2)
1179 if (FIRST_STRING_P (where))
1181 for (this_char = where - string1; this_char < size1; this_char++)
1182 PUT_CHAR (string1[this_char]);
1188 for (this_char = where - string2; this_char < size2; this_char++)
1190 PUT_CHAR (string2[this_char]);
1193 fputs ("...", stdout);
1200 # ifndef DEFINED_ONCE
1208 # else /* not DEBUG */
1210 # ifndef DEFINED_ONCE
1214 # define DEBUG_STATEMENT(e)
1215 # define DEBUG_PRINT1(x)
1216 # define DEBUG_PRINT2(x1, x2)
1217 # define DEBUG_PRINT3(x1, x2, x3)
1218 # define DEBUG_PRINT4(x1, x2, x3, x4)
1219 # endif /* not DEFINED_ONCE */
1220 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1221 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1223 # endif /* not DEBUG */
1228 /* This convert a multibyte string to a wide character string.
1229 And write their correspondances to offset_buffer(see below)
1230 and write whether each wchar_t is binary data to is_binary.
1231 This assume invalid multibyte sequences as binary data.
1232 We assume offset_buffer and is_binary is already allocated
1235 static size_t convert_mbs_to_wcs (CHAR_T *dest, const unsigned char* src,
1236 size_t len, int *offset_buffer,
1239 convert_mbs_to_wcs (CHAR_T *dest, const unsigned char*src, size_t len,
1240 int *offset_buffer, char *is_binary)
1241 /* It hold correspondances between src(char string) and
1242 dest(wchar_t string) for optimization.
1244 dest = {'X', 'Y', 'Z'}
1245 (each "xxx", "y" and "zz" represent one multibyte character
1246 corresponding to 'X', 'Y' and 'Z'.)
1247 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1251 wchar_t *pdest = dest;
1252 const unsigned char *psrc = src;
1253 size_t wc_count = 0;
1257 size_t mb_remain = len;
1258 size_t mb_count = 0;
1260 /* Initialize the conversion state. */
1261 memset (&mbs, 0, sizeof (mbstate_t));
1263 offset_buffer[0] = 0;
1264 for( ; mb_remain > 0 ; ++wc_count, ++pdest, mb_remain -= consumed,
1268 consumed = __mbrtowc (pdest, psrc, mb_remain, &mbs);
1270 consumed = mbrtowc (pdest, psrc, mb_remain, &mbs);
1274 /* failed to convert. maybe src contains binary data.
1275 So we consume 1 byte manualy. */
1279 is_binary[wc_count] = TRUE;
1282 is_binary[wc_count] = FALSE;
1283 /* In sjis encoding, we use yen sign as escape character in
1284 place of reverse solidus. So we convert 0x5c(yen sign in
1285 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1286 solidus in UCS2). */
1287 if (consumed == 1 && (int) *psrc == 0x5c && (int) *pdest == 0xa5)
1288 *pdest = (wchar_t) *psrc;
1290 offset_buffer[wc_count + 1] = mb_count += consumed;
1293 /* Fill remain of the buffer with sentinel. */
1294 for (i = wc_count + 1 ; i <= len ; i++)
1295 offset_buffer[i] = mb_count + 1;
1302 #else /* not INSIDE_RECURSION */
1304 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1305 also be assigned to arbitrarily: each pattern buffer stores its own
1306 syntax, so it can be changed between regex compilations. */
1307 /* This has no initializer because initialized variables in Emacs
1308 become read-only after dumping. */
1309 reg_syntax_t re_syntax_options;
1312 /* Specify the precise syntax of regexps for compilation. This provides
1313 for compatibility for various utilities which historically have
1314 different, incompatible syntaxes.
1316 The argument SYNTAX is a bit mask comprised of the various bits
1317 defined in regex.h. We return the old syntax. */
1320 re_set_syntax (reg_syntax_t syntax)
1322 reg_syntax_t ret = re_syntax_options;
1324 re_syntax_options = syntax;
1326 if (syntax & RE_DEBUG)
1328 else if (debug) /* was on but now is not */
1334 weak_alias (__re_set_syntax, re_set_syntax)
1337 /* This table gives an error message for each of the error codes listed
1338 in regex.h. Obviously the order here has to be same as there.
1339 POSIX doesn't require that we do anything for REG_NOERROR,
1340 but why not be nice? */
1342 static const char *re_error_msgid[] =
1344 gettext_noop ("Success"), /* REG_NOERROR */
1345 gettext_noop ("No match"), /* REG_NOMATCH */
1346 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
1347 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
1348 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
1349 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
1350 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1351 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1352 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1353 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1354 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1355 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1356 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1357 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1358 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1359 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1360 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
1363 #endif /* INSIDE_RECURSION */
1365 #ifndef DEFINED_ONCE
1366 /* Avoiding alloca during matching, to placate r_alloc. */
1368 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1369 searching and matching functions should not call alloca. On some
1370 systems, alloca is implemented in terms of malloc, and if we're
1371 using the relocating allocator routines, then malloc could cause a
1372 relocation, which might (if the strings being searched are in the
1373 ralloc heap) shift the data out from underneath the regexp
1376 Here's another reason to avoid allocation: Emacs
1377 processes input from X in a signal handler; processing X input may
1378 call malloc; if input arrives while a matching routine is calling
1379 malloc, then we're scrod. But Emacs can't just block input while
1380 calling matching routines; then we don't notice interrupts when
1381 they come in. So, Emacs blocks input around all regexp calls
1382 except the matching calls, which it leaves unprotected, in the
1383 faith that they will not malloc. */
1385 /* Normally, this is fine. */
1386 # define MATCH_MAY_ALLOCATE
1388 /* When using GNU C, we are not REALLY using the C alloca, no matter
1389 what config.h may say. So don't take precautions for it. */
1394 /* The match routines may not allocate if (1) they would do it with malloc
1395 and (2) it's not safe for them to use malloc.
1396 Note that if REL_ALLOC is defined, matching would not use malloc for the
1397 failure stack, but we would still use it for the register vectors;
1398 so REL_ALLOC should not affect this. */
1399 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1400 # undef MATCH_MAY_ALLOCATE
1402 #endif /* not DEFINED_ONCE */
1404 #ifdef INSIDE_RECURSION
1405 /* Failure stack declarations and macros; both re_compile_fastmap and
1406 re_match_2 use a failure stack. These have to be macros because of
1407 REGEX_ALLOCATE_STACK. */
1410 /* Number of failure points for which to initially allocate space
1411 when matching. If this number is exceeded, we allocate more
1412 space, so it is not a hard limit. */
1413 # ifndef INIT_FAILURE_ALLOC
1414 # define INIT_FAILURE_ALLOC 5
1417 /* Roughly the maximum number of failure points on the stack. Would be
1418 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1419 This is a variable only so users of regex can assign to it; we never
1420 change it ourselves. */
1423 # ifndef DEFINED_ONCE
1425 # ifdef INT_IS_16BIT
1426 # define RE_M_F_TYPE long int
1428 # define RE_M_F_TYPE int
1429 # endif /* INT_IS_16BIT */
1431 # ifdef MATCH_MAY_ALLOCATE
1432 /* 4400 was enough to cause a crash on Alpha OSF/1,
1433 whose default stack limit is 2mb. */
1434 # define RE_M_F_DEFAULT 4000
1436 # define RE_M_F_DEFAULT 2000
1437 # endif /* MATCH_MAY_ALLOCATE */
1439 # include <shlib-compat.h>
1441 # if SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_3)
1442 link_warning (re_max_failures, "the 're_max_failures' variable is obsolete and will go away.")
1443 RE_M_F_TYPE re_max_failures = RE_M_F_DEFAULT;
1445 RE_M_F_TYPE re_max_failures attribute_hidden = RE_M_F_DEFAULT;
1446 # endif /* SHLIB_COMPAT */
1449 # undef RE_M_F_DEFAULT
1451 # endif /* DEFINED_ONCE */
1453 # ifdef INT_IS_16BIT
1455 union PREFIX(fail_stack_elt)
1461 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1465 PREFIX(fail_stack_elt_t) *stack;
1466 unsigned long int size;
1467 unsigned long int avail; /* Offset of next open position. */
1468 } PREFIX(fail_stack_type);
1470 # else /* not INT_IS_16BIT */
1472 union PREFIX(fail_stack_elt)
1478 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1482 PREFIX(fail_stack_elt_t) *stack;
1484 unsigned avail; /* Offset of next open position. */
1485 } PREFIX(fail_stack_type);
1487 # endif /* INT_IS_16BIT */
1489 # ifndef DEFINED_ONCE
1490 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1491 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1492 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1496 /* Define macros to initialize and free the failure stack.
1497 Do `return -2' if the alloc fails. */
1499 # ifdef MATCH_MAY_ALLOCATE
1500 # define INIT_FAIL_STACK() \
1502 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1503 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1505 if (fail_stack.stack == NULL) \
1508 fail_stack.size = INIT_FAILURE_ALLOC; \
1509 fail_stack.avail = 0; \
1512 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1514 # define INIT_FAIL_STACK() \
1516 fail_stack.avail = 0; \
1519 # define RESET_FAIL_STACK()
1523 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1525 Return 1 if succeeds, and 0 if either ran out of memory
1526 allocating space for it or it was already too large.
1528 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1530 # define DOUBLE_FAIL_STACK(fail_stack) \
1531 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1533 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1534 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1535 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1536 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1538 (fail_stack).stack == NULL \
1540 : ((fail_stack).size <<= 1, \
1544 /* Push pointer POINTER on FAIL_STACK.
1545 Return 1 if was able to do so and 0 if ran out of memory allocating
1547 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1548 ((FAIL_STACK_FULL () \
1549 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1551 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1554 /* Push a pointer value onto the failure stack.
1555 Assumes the variable `fail_stack'. Probably should only
1556 be called from within `PUSH_FAILURE_POINT'. */
1557 # define PUSH_FAILURE_POINTER(item) \
1558 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1560 /* This pushes an integer-valued item onto the failure stack.
1561 Assumes the variable `fail_stack'. Probably should only
1562 be called from within `PUSH_FAILURE_POINT'. */
1563 # define PUSH_FAILURE_INT(item) \
1564 fail_stack.stack[fail_stack.avail++].integer = (item)
1566 /* Push a fail_stack_elt_t value onto the failure stack.
1567 Assumes the variable `fail_stack'. Probably should only
1568 be called from within `PUSH_FAILURE_POINT'. */
1569 # define PUSH_FAILURE_ELT(item) \
1570 fail_stack.stack[fail_stack.avail++] = (item)
1572 /* These three POP... operations complement the three PUSH... operations.
1573 All assume that `fail_stack' is nonempty. */
1574 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1575 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1576 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1578 /* Used to omit pushing failure point id's when we're not debugging. */
1580 # define DEBUG_PUSH PUSH_FAILURE_INT
1581 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1583 # define DEBUG_PUSH(item)
1584 # define DEBUG_POP(item_addr)
1588 /* Push the information about the state we will need
1589 if we ever fail back to it.
1591 Requires variables fail_stack, regstart, regend, reg_info, and
1592 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1595 Does `return FAILURE_CODE' if runs out of memory. */
1597 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1599 char *destination; \
1600 /* Must be int, so when we don't save any registers, the arithmetic \
1601 of 0 + -1 isn't done as unsigned. */ \
1602 /* Can't be int, since there is not a shred of a guarantee that int \
1603 is wide enough to hold a value of something to which pointer can \
1605 active_reg_t this_reg; \
1607 DEBUG_STATEMENT (failure_id++); \
1608 DEBUG_STATEMENT (nfailure_points_pushed++); \
1609 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1610 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1611 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1613 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1614 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1616 /* Ensure we have enough space allocated for what we will push. */ \
1617 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1619 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1620 return failure_code; \
1622 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1623 (fail_stack).size); \
1624 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1627 /* Push the info, starting with the registers. */ \
1628 DEBUG_PRINT1 ("\n"); \
1631 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1634 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1635 DEBUG_STATEMENT (num_regs_pushed++); \
1637 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1638 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1640 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1641 PUSH_FAILURE_POINTER (regend[this_reg]); \
1643 DEBUG_PRINT2 (" info: %p\n ", \
1644 reg_info[this_reg].word.pointer); \
1645 DEBUG_PRINT2 (" match_null=%d", \
1646 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1647 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1648 DEBUG_PRINT2 (" matched_something=%d", \
1649 MATCHED_SOMETHING (reg_info[this_reg])); \
1650 DEBUG_PRINT2 (" ever_matched=%d", \
1651 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1652 DEBUG_PRINT1 ("\n"); \
1653 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1656 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1657 PUSH_FAILURE_INT (lowest_active_reg); \
1659 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1660 PUSH_FAILURE_INT (highest_active_reg); \
1662 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1663 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1664 PUSH_FAILURE_POINTER (pattern_place); \
1666 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1667 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1669 DEBUG_PRINT1 ("'\n"); \
1670 PUSH_FAILURE_POINTER (string_place); \
1672 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1673 DEBUG_PUSH (failure_id); \
1676 # ifndef DEFINED_ONCE
1677 /* This is the number of items that are pushed and popped on the stack
1678 for each register. */
1679 # define NUM_REG_ITEMS 3
1681 /* Individual items aside from the registers. */
1683 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1685 # define NUM_NONREG_ITEMS 4
1688 /* We push at most this many items on the stack. */
1689 /* We used to use (num_regs - 1), which is the number of registers
1690 this regexp will save; but that was changed to 5
1691 to avoid stack overflow for a regexp with lots of parens. */
1692 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1694 /* We actually push this many items. */
1695 # define NUM_FAILURE_ITEMS \
1697 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1701 /* How many items can still be added to the stack without overflowing it. */
1702 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1703 # endif /* not DEFINED_ONCE */
1706 /* Pops what PUSH_FAIL_STACK pushes.
1708 We restore into the parameters, all of which should be lvalues:
1709 STR -- the saved data position.
1710 PAT -- the saved pattern position.
1711 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1712 REGSTART, REGEND -- arrays of string positions.
1713 REG_INFO -- array of information about each subexpression.
1715 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1716 `pend', `string1', `size1', `string2', and `size2'. */
1717 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1719 DEBUG_STATEMENT (unsigned failure_id;) \
1720 active_reg_t this_reg; \
1721 const UCHAR_T *string_temp; \
1723 assert (!FAIL_STACK_EMPTY ()); \
1725 /* Remove failure points and point to how many regs pushed. */ \
1726 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1727 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1728 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1730 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1732 DEBUG_POP (&failure_id); \
1733 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1735 /* If the saved string location is NULL, it came from an \
1736 on_failure_keep_string_jump opcode, and we want to throw away the \
1737 saved NULL, thus retaining our current position in the string. */ \
1738 string_temp = POP_FAILURE_POINTER (); \
1739 if (string_temp != NULL) \
1740 str = (const CHAR_T *) string_temp; \
1742 DEBUG_PRINT2 (" Popping string %p: `", str); \
1743 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1744 DEBUG_PRINT1 ("'\n"); \
1746 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1747 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1748 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1750 /* Restore register info. */ \
1751 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1752 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1754 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1755 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1758 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1760 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1762 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1763 DEBUG_PRINT2 (" info: %p\n", \
1764 reg_info[this_reg].word.pointer); \
1766 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1767 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1769 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1770 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1774 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1776 reg_info[this_reg].word.integer = 0; \
1777 regend[this_reg] = 0; \
1778 regstart[this_reg] = 0; \
1780 highest_active_reg = high_reg; \
1783 set_regs_matched_done = 0; \
1784 DEBUG_STATEMENT (nfailure_points_popped++); \
1785 } /* POP_FAILURE_POINT */
1787 /* Structure for per-register (a.k.a. per-group) information.
1788 Other register information, such as the
1789 starting and ending positions (which are addresses), and the list of
1790 inner groups (which is a bits list) are maintained in separate
1793 We are making a (strictly speaking) nonportable assumption here: that
1794 the compiler will pack our bit fields into something that fits into
1795 the type of `word', i.e., is something that fits into one item on the
1799 /* Declarations and macros for re_match_2. */
1803 PREFIX(fail_stack_elt_t) word;
1806 /* This field is one if this group can match the empty string,
1807 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1808 # define MATCH_NULL_UNSET_VALUE 3
1809 unsigned match_null_string_p : 2;
1810 unsigned is_active : 1;
1811 unsigned matched_something : 1;
1812 unsigned ever_matched_something : 1;
1814 } PREFIX(register_info_type);
1816 # ifndef DEFINED_ONCE
1817 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1818 # define IS_ACTIVE(R) ((R).bits.is_active)
1819 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1820 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1823 /* Call this when have matched a real character; it sets `matched' flags
1824 for the subexpressions which we are currently inside. Also records
1825 that those subexprs have matched. */
1826 # define SET_REGS_MATCHED() \
1829 if (!set_regs_matched_done) \
1832 set_regs_matched_done = 1; \
1833 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1835 MATCHED_SOMETHING (reg_info[r]) \
1836 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1842 # endif /* not DEFINED_ONCE */
1844 /* Registers are set to a sentinel when they haven't yet matched. */
1845 static CHAR_T PREFIX(reg_unset_dummy);
1846 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1847 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1849 /* Subroutine declarations and macros for regex_compile. */
1850 static void PREFIX(store_op1) (re_opcode_t op, UCHAR_T *loc, int arg);
1851 static void PREFIX(store_op2) (re_opcode_t op, UCHAR_T *loc,
1852 int arg1, int arg2);
1853 static void PREFIX(insert_op1) (re_opcode_t op, UCHAR_T *loc,
1854 int arg, UCHAR_T *end);
1855 static void PREFIX(insert_op2) (re_opcode_t op, UCHAR_T *loc,
1856 int arg1, int arg2, UCHAR_T *end);
1857 static boolean PREFIX(at_begline_loc_p) (const CHAR_T *pattern,
1859 reg_syntax_t syntax);
1860 static boolean PREFIX(at_endline_loc_p) (const CHAR_T *p,
1862 reg_syntax_t syntax);
1864 static reg_errcode_t wcs_compile_range (CHAR_T range_start,
1865 const CHAR_T **p_ptr,
1868 reg_syntax_t syntax,
1871 static void insert_space (int num, CHAR_T *loc, CHAR_T *end);
1873 static reg_errcode_t byte_compile_range (unsigned int range_start,
1876 RE_TRANSLATE_TYPE translate,
1877 reg_syntax_t syntax,
1881 /* Fetch the next character in the uncompiled pattern---translating it
1882 if necessary. Also cast from a signed character in the constant
1883 string passed to us by the user to an unsigned char that we can use
1884 as an array index (in, e.g., `translate'). */
1885 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1886 because it is impossible to allocate 4GB array for some encodings
1887 which have 4 byte character_set like UCS4. */
1890 # define PATFETCH(c) \
1891 do {if (p == pend) return REG_EEND; \
1892 c = (UCHAR_T) *p++; \
1893 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1896 # define PATFETCH(c) \
1897 do {if (p == pend) return REG_EEND; \
1898 c = (unsigned char) *p++; \
1899 if (translate) c = (unsigned char) translate[c]; \
1904 /* Fetch the next character in the uncompiled pattern, with no
1906 # define PATFETCH_RAW(c) \
1907 do {if (p == pend) return REG_EEND; \
1908 c = (UCHAR_T) *p++; \
1911 /* Go backwards one character in the pattern. */
1912 # define PATUNFETCH p--
1915 /* If `translate' is non-null, return translate[D], else just D. We
1916 cast the subscript to translate because some data is declared as
1917 `char *', to avoid warnings when a string constant is passed. But
1918 when we use a character as a subscript we must make it unsigned. */
1919 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1920 because it is impossible to allocate 4GB array for some encodings
1921 which have 4 byte character_set like UCS4. */
1925 # define TRANSLATE(d) \
1926 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1927 ? (char) translate[(unsigned char) (d)] : (d))
1929 # define TRANSLATE(d) \
1930 (translate ? (char) translate[(unsigned char) (d)] : (char) (d))
1935 /* Macros for outputting the compiled pattern into `buffer'. */
1937 /* If the buffer isn't allocated when it comes in, use this. */
1938 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1940 /* Make sure we have at least N more bytes of space in buffer. */
1942 # define GET_BUFFER_SPACE(n) \
1943 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1944 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1947 # define GET_BUFFER_SPACE(n) \
1948 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1952 /* Make sure we have one more byte of buffer space and then add C to it. */
1953 # define BUF_PUSH(c) \
1955 GET_BUFFER_SPACE (1); \
1956 *b++ = (UCHAR_T) (c); \
1960 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1961 # define BUF_PUSH_2(c1, c2) \
1963 GET_BUFFER_SPACE (2); \
1964 *b++ = (UCHAR_T) (c1); \
1965 *b++ = (UCHAR_T) (c2); \
1969 /* As with BUF_PUSH_2, except for three bytes. */
1970 # define BUF_PUSH_3(c1, c2, c3) \
1972 GET_BUFFER_SPACE (3); \
1973 *b++ = (UCHAR_T) (c1); \
1974 *b++ = (UCHAR_T) (c2); \
1975 *b++ = (UCHAR_T) (c3); \
1978 /* Store a jump with opcode OP at LOC to location TO. We store a
1979 relative address offset by the three bytes the jump itself occupies. */
1980 # define STORE_JUMP(op, loc, to) \
1981 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
1983 /* Likewise, for a two-argument jump. */
1984 # define STORE_JUMP2(op, loc, to, arg) \
1985 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
1987 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1988 # define INSERT_JUMP(op, loc, to) \
1989 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
1991 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1992 # define INSERT_JUMP2(op, loc, to, arg) \
1993 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
1996 /* This is not an arbitrary limit: the arguments which represent offsets
1997 into the pattern are two bytes long. So if 2^16 bytes turns out to
1998 be too small, many things would have to change. */
1999 /* Any other compiler which, like MSC, has allocation limit below 2^16
2000 bytes will have to use approach similar to what was done below for
2001 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2002 reallocating to 0 bytes. Such thing is not going to work too well.
2003 You have been warned!! */
2004 # ifndef DEFINED_ONCE
2005 # if defined _MSC_VER && !defined WIN32
2006 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2007 The REALLOC define eliminates a flurry of conversion warnings,
2008 but is not required. */
2009 # define MAX_BUF_SIZE 65500L
2010 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2012 # define MAX_BUF_SIZE (1L << 16)
2013 # define REALLOC(p,s) realloc ((p), (s))
2016 /* Extend the buffer by twice its current size via realloc and
2017 reset the pointers that pointed into the old block to point to the
2018 correct places in the new one. If extending the buffer results in it
2019 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2020 # if __BOUNDED_POINTERS__
2021 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2022 # define MOVE_BUFFER_POINTER(P) \
2023 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2024 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2027 SET_HIGH_BOUND (b); \
2028 SET_HIGH_BOUND (begalt); \
2029 if (fixup_alt_jump) \
2030 SET_HIGH_BOUND (fixup_alt_jump); \
2032 SET_HIGH_BOUND (laststart); \
2033 if (pending_exact) \
2034 SET_HIGH_BOUND (pending_exact); \
2037 # define MOVE_BUFFER_POINTER(P) (P) += incr
2038 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2040 # endif /* not DEFINED_ONCE */
2043 # define EXTEND_BUFFER() \
2045 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2047 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2049 bufp->allocated <<= 1; \
2050 if (bufp->allocated > MAX_BUF_SIZE) \
2051 bufp->allocated = MAX_BUF_SIZE; \
2052 /* How many characters the new buffer can have? */ \
2053 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2054 if (wchar_count == 0) wchar_count = 1; \
2055 /* Truncate the buffer to CHAR_T align. */ \
2056 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2057 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2058 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2059 if (COMPILED_BUFFER_VAR == NULL) \
2060 return REG_ESPACE; \
2061 /* If the buffer moved, move all the pointers into it. */ \
2062 if (old_buffer != COMPILED_BUFFER_VAR) \
2064 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2065 MOVE_BUFFER_POINTER (b); \
2066 MOVE_BUFFER_POINTER (begalt); \
2067 if (fixup_alt_jump) \
2068 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2070 MOVE_BUFFER_POINTER (laststart); \
2071 if (pending_exact) \
2072 MOVE_BUFFER_POINTER (pending_exact); \
2074 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2077 # define EXTEND_BUFFER() \
2079 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2080 if (bufp->allocated == MAX_BUF_SIZE) \
2082 bufp->allocated <<= 1; \
2083 if (bufp->allocated > MAX_BUF_SIZE) \
2084 bufp->allocated = MAX_BUF_SIZE; \
2085 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2087 if (COMPILED_BUFFER_VAR == NULL) \
2088 return REG_ESPACE; \
2089 /* If the buffer moved, move all the pointers into it. */ \
2090 if (old_buffer != COMPILED_BUFFER_VAR) \
2092 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2093 MOVE_BUFFER_POINTER (b); \
2094 MOVE_BUFFER_POINTER (begalt); \
2095 if (fixup_alt_jump) \
2096 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2098 MOVE_BUFFER_POINTER (laststart); \
2099 if (pending_exact) \
2100 MOVE_BUFFER_POINTER (pending_exact); \
2102 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2106 # ifndef DEFINED_ONCE
2107 /* Since we have one byte reserved for the register number argument to
2108 {start,stop}_memory, the maximum number of groups we can report
2109 things about is what fits in that byte. */
2110 # define MAX_REGNUM 255
2112 /* But patterns can have more than `MAX_REGNUM' registers. We just
2113 ignore the excess. */
2114 typedef unsigned regnum_t;
2117 /* Macros for the compile stack. */
2119 /* Since offsets can go either forwards or backwards, this type needs to
2120 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2121 /* int may be not enough when sizeof(int) == 2. */
2122 typedef long pattern_offset_t;
2126 pattern_offset_t begalt_offset;
2127 pattern_offset_t fixup_alt_jump;
2128 pattern_offset_t inner_group_offset;
2129 pattern_offset_t laststart_offset;
2131 } compile_stack_elt_t;
2136 compile_stack_elt_t *stack;
2138 unsigned avail; /* Offset of next open position. */
2139 } compile_stack_type;
2142 # define INIT_COMPILE_STACK_SIZE 32
2144 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2145 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2147 /* The next available element. */
2148 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2150 # endif /* not DEFINED_ONCE */
2152 /* Set the bit for character C in a list. */
2153 # ifndef DEFINED_ONCE
2154 # define SET_LIST_BIT(c) \
2155 (b[((unsigned char) (c)) / BYTEWIDTH] \
2156 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2157 # endif /* DEFINED_ONCE */
2159 /* Get the next unsigned number in the uncompiled pattern. */
2160 # define GET_UNSIGNED_NUMBER(num) \
2165 if (c < '0' || c > '9') \
2167 if (num <= RE_DUP_MAX) \
2171 num = num * 10 + c - '0'; \
2176 # ifndef DEFINED_ONCE
2177 # if WIDE_CHAR_SUPPORT
2178 /* The GNU C library provides support for user-defined character classes
2179 and the functions from ISO C amendement 1. */
2180 # ifdef CHARCLASS_NAME_MAX
2181 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2183 /* This shouldn't happen but some implementation might still have this
2184 problem. Use a reasonable default value. */
2185 # define CHAR_CLASS_MAX_LENGTH 256
2189 # define IS_CHAR_CLASS(string) __wctype (string)
2191 # define IS_CHAR_CLASS(string) wctype (string)
2194 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
2196 # define IS_CHAR_CLASS(string) \
2197 (STREQ (string, "alpha") || STREQ (string, "upper") \
2198 || STREQ (string, "lower") || STREQ (string, "digit") \
2199 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
2200 || STREQ (string, "space") || STREQ (string, "print") \
2201 || STREQ (string, "punct") || STREQ (string, "graph") \
2202 || STREQ (string, "cntrl") || STREQ (string, "blank"))
2204 # endif /* DEFINED_ONCE */
2206 # ifndef MATCH_MAY_ALLOCATE
2208 /* If we cannot allocate large objects within re_match_2_internal,
2209 we make the fail stack and register vectors global.
2210 The fail stack, we grow to the maximum size when a regexp
2212 The register vectors, we adjust in size each time we
2213 compile a regexp, according to the number of registers it needs. */
2215 static PREFIX(fail_stack_type) fail_stack;
2217 /* Size with which the following vectors are currently allocated.
2218 That is so we can make them bigger as needed,
2219 but never make them smaller. */
2220 # ifdef DEFINED_ONCE
2221 static int regs_allocated_size;
2223 static const char ** regstart, ** regend;
2224 static const char ** old_regstart, ** old_regend;
2225 static const char **best_regstart, **best_regend;
2226 static const char **reg_dummy;
2227 # endif /* DEFINED_ONCE */
2229 static PREFIX(register_info_type) *PREFIX(reg_info);
2230 static PREFIX(register_info_type) *PREFIX(reg_info_dummy);
2232 /* Make the register vectors big enough for NUM_REGS registers,
2233 but don't make them smaller. */
2236 PREFIX(regex_grow_registers) (int num_regs)
2238 if (num_regs > regs_allocated_size)
2240 RETALLOC_IF (regstart, num_regs, const char *);
2241 RETALLOC_IF (regend, num_regs, const char *);
2242 RETALLOC_IF (old_regstart, num_regs, const char *);
2243 RETALLOC_IF (old_regend, num_regs, const char *);
2244 RETALLOC_IF (best_regstart, num_regs, const char *);
2245 RETALLOC_IF (best_regend, num_regs, const char *);
2246 RETALLOC_IF (PREFIX(reg_info), num_regs, PREFIX(register_info_type));
2247 RETALLOC_IF (reg_dummy, num_regs, const char *);
2248 RETALLOC_IF (PREFIX(reg_info_dummy), num_regs, PREFIX(register_info_type));
2250 regs_allocated_size = num_regs;
2254 # endif /* not MATCH_MAY_ALLOCATE */
2256 # ifndef DEFINED_ONCE
2257 static boolean group_in_compile_stack (compile_stack_type compile_stack,
2259 # endif /* not DEFINED_ONCE */
2261 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2262 Returns one of error codes defined in `regex.h', or zero for success.
2264 Assumes the `allocated' (and perhaps `buffer') and `translate'
2265 fields are set in BUFP on entry.
2267 If it succeeds, results are put in BUFP (if it returns an error, the
2268 contents of BUFP are undefined):
2269 `buffer' is the compiled pattern;
2270 `syntax' is set to SYNTAX;
2271 `used' is set to the length of the compiled pattern;
2272 `fastmap_accurate' is zero;
2273 `re_nsub' is the number of subexpressions in PATTERN;
2274 `not_bol' and `not_eol' are zero;
2276 The `fastmap' and `newline_anchor' fields are neither
2277 examined nor set. */
2279 /* Return, freeing storage we allocated. */
2281 # define FREE_STACK_RETURN(value) \
2282 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2284 # define FREE_STACK_RETURN(value) \
2285 return (free (compile_stack.stack), value)
2288 static reg_errcode_t
2289 PREFIX(regex_compile) (const char *ARG_PREFIX(pattern),
2290 size_t ARG_PREFIX(size), reg_syntax_t syntax,
2291 struct re_pattern_buffer *bufp)
2293 /* We fetch characters from PATTERN here. Even though PATTERN is
2294 `char *' (i.e., signed), we declare these variables as unsigned, so
2295 they can be reliably used as array indices. */
2296 register UCHAR_T c, c1;
2299 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2300 CHAR_T *pattern, *COMPILED_BUFFER_VAR;
2302 /* offset buffer for optimization. See convert_mbs_to_wc. */
2303 int *mbs_offset = NULL;
2304 /* It hold whether each wchar_t is binary data or not. */
2305 char *is_binary = NULL;
2306 /* A flag whether exactn is handling binary data or not. */
2307 char is_exactn_bin = FALSE;
2310 /* A random temporary spot in PATTERN. */
2313 /* Points to the end of the buffer, where we should append. */
2314 register UCHAR_T *b;
2316 /* Keeps track of unclosed groups. */
2317 compile_stack_type compile_stack;
2319 /* Points to the current (ending) position in the pattern. */
2324 const CHAR_T *p = pattern;
2325 const CHAR_T *pend = pattern + size;
2328 /* How to translate the characters in the pattern. */
2329 RE_TRANSLATE_TYPE translate = bufp->translate;
2331 /* Address of the count-byte of the most recently inserted `exactn'
2332 command. This makes it possible to tell if a new exact-match
2333 character can be added to that command or if the character requires
2334 a new `exactn' command. */
2335 UCHAR_T *pending_exact = 0;
2337 /* Address of start of the most recently finished expression.
2338 This tells, e.g., postfix * where to find the start of its
2339 operand. Reset at the beginning of groups and alternatives. */
2340 UCHAR_T *laststart = 0;
2342 /* Address of beginning of regexp, or inside of last group. */
2345 /* Address of the place where a forward jump should go to the end of
2346 the containing expression. Each alternative of an `or' -- except the
2347 last -- ends with a forward jump of this sort. */
2348 UCHAR_T *fixup_alt_jump = 0;
2350 /* Counts open-groups as they are encountered. Remembered for the
2351 matching close-group on the compile stack, so the same register
2352 number is put in the stop_memory as the start_memory. */
2353 regnum_t regnum = 0;
2356 /* Initialize the wchar_t PATTERN and offset_buffer. */
2357 p = pend = pattern = TALLOC(csize + 1, CHAR_T);
2358 mbs_offset = TALLOC(csize + 1, int);
2359 is_binary = TALLOC(csize + 1, char);
2360 if (pattern == NULL || mbs_offset == NULL || is_binary == NULL)
2367 pattern[csize] = L'\0'; /* sentinel */
2368 size = convert_mbs_to_wcs(pattern, cpattern, csize, mbs_offset, is_binary);
2380 DEBUG_PRINT1 ("\nCompiling pattern: ");
2383 unsigned debug_count;
2385 for (debug_count = 0; debug_count < size; debug_count++)
2386 PUT_CHAR (pattern[debug_count]);
2391 /* Initialize the compile stack. */
2392 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2393 if (compile_stack.stack == NULL)
2403 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2404 compile_stack.avail = 0;
2406 /* Initialize the pattern buffer. */
2407 bufp->syntax = syntax;
2408 bufp->fastmap_accurate = 0;
2409 bufp->not_bol = bufp->not_eol = 0;
2411 /* Set `used' to zero, so that if we return an error, the pattern
2412 printer (for debugging) will think there's no pattern. We reset it
2416 /* Always count groups, whether or not bufp->no_sub is set. */
2419 #if !defined emacs && !defined SYNTAX_TABLE
2420 /* Initialize the syntax table. */
2421 init_syntax_once ();
2424 if (bufp->allocated == 0)
2427 { /* If zero allocated, but buffer is non-null, try to realloc
2428 enough space. This loses if buffer's address is bogus, but
2429 that is the user's responsibility. */
2431 /* Free bufp->buffer and allocate an array for wchar_t pattern
2434 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE/sizeof(UCHAR_T),
2437 RETALLOC (COMPILED_BUFFER_VAR, INIT_BUF_SIZE, UCHAR_T);
2441 { /* Caller did not allocate a buffer. Do it for them. */
2442 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE / sizeof(UCHAR_T),
2446 if (!COMPILED_BUFFER_VAR) FREE_STACK_RETURN (REG_ESPACE);
2448 bufp->buffer = (char*)COMPILED_BUFFER_VAR;
2450 bufp->allocated = INIT_BUF_SIZE;
2454 COMPILED_BUFFER_VAR = (UCHAR_T*) bufp->buffer;
2457 begalt = b = COMPILED_BUFFER_VAR;
2459 /* Loop through the uncompiled pattern until we're at the end. */
2468 if ( /* If at start of pattern, it's an operator. */
2470 /* If context independent, it's an operator. */
2471 || syntax & RE_CONTEXT_INDEP_ANCHORS
2472 /* Otherwise, depends on what's come before. */
2473 || PREFIX(at_begline_loc_p) (pattern, p, syntax))
2483 if ( /* If at end of pattern, it's an operator. */
2485 /* If context independent, it's an operator. */
2486 || syntax & RE_CONTEXT_INDEP_ANCHORS
2487 /* Otherwise, depends on what's next. */
2488 || PREFIX(at_endline_loc_p) (p, pend, syntax))
2498 if ((syntax & RE_BK_PLUS_QM)
2499 || (syntax & RE_LIMITED_OPS))
2503 /* If there is no previous pattern... */
2506 if (syntax & RE_CONTEXT_INVALID_OPS)
2507 FREE_STACK_RETURN (REG_BADRPT);
2508 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2513 /* Are we optimizing this jump? */
2514 boolean keep_string_p = false;
2516 /* 1 means zero (many) matches is allowed. */
2517 char zero_times_ok = 0, many_times_ok = 0;
2519 /* If there is a sequence of repetition chars, collapse it
2520 down to just one (the right one). We can't combine
2521 interval operators with these because of, e.g., `a{2}*',
2522 which should only match an even number of `a's. */
2526 zero_times_ok |= c != '+';
2527 many_times_ok |= c != '?';
2535 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2538 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2540 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2543 if (!(c1 == '+' || c1 == '?'))
2558 /* If we get here, we found another repeat character. */
2561 /* Star, etc. applied to an empty pattern is equivalent
2562 to an empty pattern. */
2566 /* Now we know whether or not zero matches is allowed
2567 and also whether or not two or more matches is allowed. */
2569 { /* More than one repetition is allowed, so put in at the
2570 end a backward relative jump from `b' to before the next
2571 jump we're going to put in below (which jumps from
2572 laststart to after this jump).
2574 But if we are at the `*' in the exact sequence `.*\n',
2575 insert an unconditional jump backwards to the .,
2576 instead of the beginning of the loop. This way we only
2577 push a failure point once, instead of every time
2578 through the loop. */
2579 assert (p - 1 > pattern);
2581 /* Allocate the space for the jump. */
2582 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2584 /* We know we are not at the first character of the pattern,
2585 because laststart was nonzero. And we've already
2586 incremented `p', by the way, to be the character after
2587 the `*'. Do we have to do something analogous here
2588 for null bytes, because of RE_DOT_NOT_NULL? */
2589 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2591 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2592 && !(syntax & RE_DOT_NEWLINE))
2593 { /* We have .*\n. */
2594 STORE_JUMP (jump, b, laststart);
2595 keep_string_p = true;
2598 /* Anything else. */
2599 STORE_JUMP (maybe_pop_jump, b, laststart -
2600 (1 + OFFSET_ADDRESS_SIZE));
2602 /* We've added more stuff to the buffer. */
2603 b += 1 + OFFSET_ADDRESS_SIZE;
2606 /* On failure, jump from laststart to b + 3, which will be the
2607 end of the buffer after this jump is inserted. */
2608 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2610 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2611 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2613 laststart, b + 1 + OFFSET_ADDRESS_SIZE);
2615 b += 1 + OFFSET_ADDRESS_SIZE;
2619 /* At least one repetition is required, so insert a
2620 `dummy_failure_jump' before the initial
2621 `on_failure_jump' instruction of the loop. This
2622 effects a skip over that instruction the first time
2623 we hit that loop. */
2624 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2625 INSERT_JUMP (dummy_failure_jump, laststart, laststart +
2626 2 + 2 * OFFSET_ADDRESS_SIZE);
2627 b += 1 + OFFSET_ADDRESS_SIZE;
2641 boolean had_char_class = false;
2643 CHAR_T range_start = 0xffffffff;
2645 unsigned int range_start = 0xffffffff;
2647 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2650 /* We assume a charset(_not) structure as a wchar_t array.
2651 charset[0] = (re_opcode_t) charset(_not)
2652 charset[1] = l (= length of char_classes)
2653 charset[2] = m (= length of collating_symbols)
2654 charset[3] = n (= length of equivalence_classes)
2655 charset[4] = o (= length of char_ranges)
2656 charset[5] = p (= length of chars)
2658 charset[6] = char_class (wctype_t)
2659 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2661 charset[l+5] = char_class (wctype_t)
2663 charset[l+6] = collating_symbol (wchar_t)
2665 charset[l+m+5] = collating_symbol (wchar_t)
2666 ifdef _LIBC we use the index if
2667 _NL_COLLATE_SYMB_EXTRAMB instead of
2670 charset[l+m+6] = equivalence_classes (wchar_t)
2672 charset[l+m+n+5] = equivalence_classes (wchar_t)
2673 ifdef _LIBC we use the index in
2674 _NL_COLLATE_WEIGHT instead of
2677 charset[l+m+n+6] = range_start
2678 charset[l+m+n+7] = range_end
2680 charset[l+m+n+2o+4] = range_start
2681 charset[l+m+n+2o+5] = range_end
2682 ifdef _LIBC we use the value looked up
2683 in _NL_COLLATE_COLLSEQ instead of
2686 charset[l+m+n+2o+6] = char
2688 charset[l+m+n+2o+p+5] = char
2692 /* We need at least 6 spaces: the opcode, the length of
2693 char_classes, the length of collating_symbols, the length of
2694 equivalence_classes, the length of char_ranges, the length of
2696 GET_BUFFER_SPACE (6);
2698 /* Save b as laststart. And We use laststart as the pointer
2699 to the first element of the charset here.
2700 In other words, laststart[i] indicates charset[i]. */
2703 /* We test `*p == '^' twice, instead of using an if
2704 statement, so we only need one BUF_PUSH. */
2705 BUF_PUSH (*p == '^' ? charset_not : charset);
2709 /* Push the length of char_classes, the length of
2710 collating_symbols, the length of equivalence_classes, the
2711 length of char_ranges and the length of chars. */
2712 BUF_PUSH_3 (0, 0, 0);
2715 /* Remember the first position in the bracket expression. */
2718 /* charset_not matches newline according to a syntax bit. */
2719 if ((re_opcode_t) b[-6] == charset_not
2720 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2723 laststart[5]++; /* Update the length of characters */
2726 /* Read in characters and ranges, setting map bits. */
2729 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2733 /* \ might escape characters inside [...] and [^...]. */
2734 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2736 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2740 laststart[5]++; /* Update the length of chars */
2745 /* Could be the end of the bracket expression. If it's
2746 not (i.e., when the bracket expression is `[]' so
2747 far), the ']' character bit gets set way below. */
2748 if (c == ']' && p != p1 + 1)
2751 /* Look ahead to see if it's a range when the last thing
2752 was a character class. */
2753 if (had_char_class && c == '-' && *p != ']')
2754 FREE_STACK_RETURN (REG_ERANGE);
2756 /* Look ahead to see if it's a range when the last thing
2757 was a character: if this is a hyphen not at the
2758 beginning or the end of a list, then it's the range
2761 && !(p - 2 >= pattern && p[-2] == '[')
2762 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2766 /* Allocate the space for range_start and range_end. */
2767 GET_BUFFER_SPACE (2);
2768 /* Update the pointer to indicate end of buffer. */
2770 ret = wcs_compile_range (range_start, &p, pend, translate,
2771 syntax, b, laststart);
2772 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2773 range_start = 0xffffffff;
2775 else if (p[0] == '-' && p[1] != ']')
2776 { /* This handles ranges made up of characters only. */
2779 /* Move past the `-'. */
2781 /* Allocate the space for range_start and range_end. */
2782 GET_BUFFER_SPACE (2);
2783 /* Update the pointer to indicate end of buffer. */
2785 ret = wcs_compile_range (c, &p, pend, translate, syntax, b,
2787 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2788 range_start = 0xffffffff;
2791 /* See if we're at the beginning of a possible character
2793 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2794 { /* Leave room for the null. */
2795 char str[CHAR_CLASS_MAX_LENGTH + 1];
2800 /* If pattern is `[[:'. */
2801 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2806 if ((c == ':' && *p == ']') || p == pend)
2808 if (c1 < CHAR_CLASS_MAX_LENGTH)
2811 /* This is in any case an invalid class name. */
2816 /* If isn't a word bracketed by `[:' and `:]':
2817 undo the ending character, the letters, and leave
2818 the leading `:' and `[' (but store them as character). */
2819 if (c == ':' && *p == ']')
2824 /* Query the character class as wctype_t. */
2825 wt = IS_CHAR_CLASS (str);
2827 FREE_STACK_RETURN (REG_ECTYPE);
2829 /* Throw away the ] at the end of the character
2833 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2835 /* Allocate the space for character class. */
2836 GET_BUFFER_SPACE(CHAR_CLASS_SIZE);
2837 /* Update the pointer to indicate end of buffer. */
2838 b += CHAR_CLASS_SIZE;
2839 /* Move data which follow character classes
2840 not to violate the data. */
2841 insert_space(CHAR_CLASS_SIZE,
2842 laststart + 6 + laststart[1],
2844 alignedp = ((uintptr_t)(laststart + 6 + laststart[1])
2845 + __alignof__(wctype_t) - 1)
2846 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
2847 /* Store the character class. */
2848 *((wctype_t*)alignedp) = wt;
2849 /* Update length of char_classes */
2850 laststart[1] += CHAR_CLASS_SIZE;
2852 had_char_class = true;
2861 laststart[5] += 2; /* Update the length of characters */
2863 had_char_class = false;
2866 else if (syntax & RE_CHAR_CLASSES && c == '[' && (*p == '='
2869 CHAR_T str[128]; /* Should be large enough. */
2870 CHAR_T delim = *p; /* '=' or '.' */
2873 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
2878 /* If pattern is `[[=' or '[[.'. */
2879 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2884 if ((c == delim && *p == ']') || p == pend)
2886 if (c1 < sizeof (str) - 1)
2889 /* This is in any case an invalid class name. */
2894 if (c == delim && *p == ']' && str[0] != '\0')
2896 unsigned int i, offset;
2897 /* If we have no collation data we use the default
2898 collation in which each character is in a class
2899 by itself. It also means that ASCII is the
2900 character set and therefore we cannot have character
2901 with more than one byte in the multibyte
2904 /* If not defined _LIBC, we push the name and
2905 `\0' for the sake of matching performance. */
2906 int datasize = c1 + 1;
2914 FREE_STACK_RETURN (REG_ECOLLATE);
2919 const int32_t *table;
2920 const int32_t *weights;
2921 const int32_t *extra;
2922 const int32_t *indirect;
2925 /* This #include defines a local function! */
2926 # include <locale/weightwc.h>
2930 /* We push the index for equivalence class. */
2933 table = (const int32_t *)
2934 _NL_CURRENT (LC_COLLATE,
2935 _NL_COLLATE_TABLEWC);
2936 weights = (const int32_t *)
2937 _NL_CURRENT (LC_COLLATE,
2938 _NL_COLLATE_WEIGHTWC);
2939 extra = (const int32_t *)
2940 _NL_CURRENT (LC_COLLATE,
2941 _NL_COLLATE_EXTRAWC);
2942 indirect = (const int32_t *)
2943 _NL_CURRENT (LC_COLLATE,
2944 _NL_COLLATE_INDIRECTWC);
2946 idx = findidx ((const wint_t**)&cp);
2947 if (idx == 0 || cp < (wint_t*) str + c1)
2948 /* This is no valid character. */
2949 FREE_STACK_RETURN (REG_ECOLLATE);
2951 str[0] = (wchar_t)idx;
2953 else /* delim == '.' */
2955 /* We push collation sequence value
2956 for collating symbol. */
2958 const int32_t *symb_table;
2959 const unsigned char *extra;
2966 /* We have to convert the name to a single-byte
2967 string. This is possible since the names
2968 consist of ASCII characters and the internal
2969 representation is UCS4. */
2970 for (i = 0; i < c1; ++i)
2971 char_str[i] = str[i];
2974 _NL_CURRENT_WORD (LC_COLLATE,
2975 _NL_COLLATE_SYMB_HASH_SIZEMB);
2976 symb_table = (const int32_t *)
2977 _NL_CURRENT (LC_COLLATE,
2978 _NL_COLLATE_SYMB_TABLEMB);
2979 extra = (const unsigned char *)
2980 _NL_CURRENT (LC_COLLATE,
2981 _NL_COLLATE_SYMB_EXTRAMB);
2983 /* Locate the character in the hashing table. */
2984 hash = elem_hash (char_str, c1);
2987 elem = hash % table_size;
2988 second = hash % (table_size - 2);
2989 while (symb_table[2 * elem] != 0)
2991 /* First compare the hashing value. */
2992 if (symb_table[2 * elem] == hash
2993 && c1 == extra[symb_table[2 * elem + 1]]
2994 && memcmp (char_str,
2995 &extra[symb_table[2 * elem + 1]
2998 /* Yep, this is the entry. */
2999 idx = symb_table[2 * elem + 1];
3000 idx += 1 + extra[idx];
3008 if (symb_table[2 * elem] != 0)
3010 /* Compute the index of the byte sequence
3012 idx += 1 + extra[idx];
3013 /* Adjust for the alignment. */
3014 idx = (idx + 3) & ~3;
3016 str[0] = (wchar_t) idx + 4;
3018 else if (symb_table[2 * elem] == 0 && c1 == 1)
3020 /* No valid character. Match it as a
3021 single byte character. */
3022 had_char_class = false;
3024 /* Update the length of characters */
3026 range_start = str[0];
3028 /* Throw away the ] at the end of the
3029 collating symbol. */
3031 /* exit from the switch block. */
3035 FREE_STACK_RETURN (REG_ECOLLATE);
3040 /* Throw away the ] at the end of the equivalence
3041 class (or collating symbol). */
3044 /* Allocate the space for the equivalence class
3045 (or collating symbol) (and '\0' if needed). */
3046 GET_BUFFER_SPACE(datasize);
3047 /* Update the pointer to indicate end of buffer. */
3051 { /* equivalence class */
3052 /* Calculate the offset of char_ranges,
3053 which is next to equivalence_classes. */
3054 offset = laststart[1] + laststart[2]
3057 insert_space(datasize, laststart + offset, b - 1);
3059 /* Write the equivalence_class and \0. */
3060 for (i = 0 ; i < datasize ; i++)
3061 laststart[offset + i] = str[i];
3063 /* Update the length of equivalence_classes. */
3064 laststart[3] += datasize;
3065 had_char_class = true;
3067 else /* delim == '.' */
3068 { /* collating symbol */
3069 /* Calculate the offset of the equivalence_classes,
3070 which is next to collating_symbols. */
3071 offset = laststart[1] + laststart[2] + 6;
3072 /* Insert space and write the collationg_symbol
3074 insert_space(datasize, laststart + offset, b-1);
3075 for (i = 0 ; i < datasize ; i++)
3076 laststart[offset + i] = str[i];
3078 /* In re_match_2_internal if range_start < -1, we
3079 assume -range_start is the offset of the
3080 collating symbol which is specified as
3081 the character of the range start. So we assign
3082 -(laststart[1] + laststart[2] + 6) to
3084 range_start = -(laststart[1] + laststart[2] + 6);
3085 /* Update the length of collating_symbol. */
3086 laststart[2] += datasize;
3087 had_char_class = false;
3097 laststart[5] += 2; /* Update the length of characters */
3098 range_start = delim;
3099 had_char_class = false;
3104 had_char_class = false;
3106 laststart[5]++; /* Update the length of characters */
3112 /* Ensure that we have enough space to push a charset: the
3113 opcode, the length count, and the bitset; 34 bytes in all. */
3114 GET_BUFFER_SPACE (34);
3118 /* We test `*p == '^' twice, instead of using an if
3119 statement, so we only need one BUF_PUSH. */
3120 BUF_PUSH (*p == '^' ? charset_not : charset);
3124 /* Remember the first position in the bracket expression. */
3127 /* Push the number of bytes in the bitmap. */
3128 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
3130 /* Clear the whole map. */
3131 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
3133 /* charset_not matches newline according to a syntax bit. */
3134 if ((re_opcode_t) b[-2] == charset_not
3135 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
3136 SET_LIST_BIT ('\n');
3138 /* Read in characters and ranges, setting map bits. */
3141 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3145 /* \ might escape characters inside [...] and [^...]. */
3146 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
3148 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3156 /* Could be the end of the bracket expression. If it's
3157 not (i.e., when the bracket expression is `[]' so
3158 far), the ']' character bit gets set way below. */
3159 if (c == ']' && p != p1 + 1)
3162 /* Look ahead to see if it's a range when the last thing
3163 was a character class. */
3164 if (had_char_class && c == '-' && *p != ']')
3165 FREE_STACK_RETURN (REG_ERANGE);
3167 /* Look ahead to see if it's a range when the last thing
3168 was a character: if this is a hyphen not at the
3169 beginning or the end of a list, then it's the range
3172 && !(p - 2 >= pattern && p[-2] == '[')
3173 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
3177 = byte_compile_range (range_start, &p, pend, translate,
3179 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3180 range_start = 0xffffffff;
3183 else if (p[0] == '-' && p[1] != ']')
3184 { /* This handles ranges made up of characters only. */
3187 /* Move past the `-'. */
3190 ret = byte_compile_range (c, &p, pend, translate, syntax, b);
3191 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3192 range_start = 0xffffffff;
3195 /* See if we're at the beginning of a possible character
3198 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
3199 { /* Leave room for the null. */
3200 char str[CHAR_CLASS_MAX_LENGTH + 1];
3205 /* If pattern is `[[:'. */
3206 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3211 if ((c == ':' && *p == ']') || p == pend)
3213 if (((int) c1) < CHAR_CLASS_MAX_LENGTH)
3216 /* This is in any case an invalid class name. */
3221 /* If isn't a word bracketed by `[:' and `:]':
3222 undo the ending character, the letters, and leave
3223 the leading `:' and `[' (but set bits for them). */
3224 if (c == ':' && *p == ']')
3226 # if WIDE_CHAR_SUPPORT
3227 boolean is_lower = STREQ (str, "lower");
3228 boolean is_upper = STREQ (str, "upper");
3232 wt = IS_CHAR_CLASS (str);
3234 FREE_STACK_RETURN (REG_ECTYPE);
3236 /* Throw away the ] at the end of the character
3240 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3242 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
3245 if (__iswctype (__btowc (ch), wt))
3248 if (iswctype (btowc (ch), wt))
3252 if (translate && (is_upper || is_lower)
3253 && (ISUPPER (ch) || ISLOWER (ch)))
3257 had_char_class = true;
3260 boolean is_alnum = STREQ (str, "alnum");
3261 boolean is_alpha = STREQ (str, "alpha");
3262 boolean is_blank = STREQ (str, "blank");
3263 boolean is_cntrl = STREQ (str, "cntrl");
3264 boolean is_digit = STREQ (str, "digit");
3265 boolean is_graph = STREQ (str, "graph");
3266 boolean is_lower = STREQ (str, "lower");
3267 boolean is_print = STREQ (str, "print");
3268 boolean is_punct = STREQ (str, "punct");
3269 boolean is_space = STREQ (str, "space");
3270 boolean is_upper = STREQ (str, "upper");
3271 boolean is_xdigit = STREQ (str, "xdigit");
3273 if (!IS_CHAR_CLASS (str))
3274 FREE_STACK_RETURN (REG_ECTYPE);
3276 /* Throw away the ] at the end of the character
3280 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3282 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
3284 /* This was split into 3 if's to
3285 avoid an arbitrary limit in some compiler. */
3286 if ( (is_alnum && ISALNUM (ch))
3287 || (is_alpha && ISALPHA (ch))
3288 || (is_blank && ISBLANK (ch))
3289 || (is_cntrl && ISCNTRL (ch)))
3291 if ( (is_digit && ISDIGIT (ch))
3292 || (is_graph && ISGRAPH (ch))
3293 || (is_lower && ISLOWER (ch))
3294 || (is_print && ISPRINT (ch)))
3296 if ( (is_punct && ISPUNCT (ch))
3297 || (is_space && ISSPACE (ch))
3298 || (is_upper && ISUPPER (ch))
3299 || (is_xdigit && ISXDIGIT (ch)))
3301 if ( translate && (is_upper || is_lower)
3302 && (ISUPPER (ch) || ISLOWER (ch)))
3305 had_char_class = true;
3306 # endif /* libc || wctype.h */
3316 had_char_class = false;
3319 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '=')
3321 unsigned char str[MB_LEN_MAX + 1];
3324 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3330 /* If pattern is `[[='. */
3331 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3336 if ((c == '=' && *p == ']') || p == pend)
3338 if (c1 < MB_LEN_MAX)
3341 /* This is in any case an invalid class name. */
3346 if (c == '=' && *p == ']' && str[0] != '\0')
3348 /* If we have no collation data we use the default
3349 collation in which each character is in a class
3350 by itself. It also means that ASCII is the
3351 character set and therefore we cannot have character
3352 with more than one byte in the multibyte
3359 FREE_STACK_RETURN (REG_ECOLLATE);
3361 /* Throw away the ] at the end of the equivalence
3365 /* Set the bit for the character. */
3366 SET_LIST_BIT (str[0]);
3371 /* Try to match the byte sequence in `str' against
3372 those known to the collate implementation.
3373 First find out whether the bytes in `str' are
3374 actually from exactly one character. */
3375 const int32_t *table;
3376 const unsigned char *weights;
3377 const unsigned char *extra;
3378 const int32_t *indirect;
3380 const unsigned char *cp = str;
3383 /* This #include defines a local function! */
3384 # include <locale/weight.h>
3386 table = (const int32_t *)
3387 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
3388 weights = (const unsigned char *)
3389 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
3390 extra = (const unsigned char *)
3391 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
3392 indirect = (const int32_t *)
3393 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
3395 idx = findidx (&cp);
3396 if (idx == 0 || cp < str + c1)
3397 /* This is no valid character. */
3398 FREE_STACK_RETURN (REG_ECOLLATE);
3400 /* Throw away the ] at the end of the equivalence
3404 /* Now we have to go throught the whole table
3405 and find all characters which have the same
3408 XXX Note that this is not entirely correct.
3409 we would have to match multibyte sequences
3410 but this is not possible with the current
3412 for (ch = 1; ch < 256; ++ch)
3413 /* XXX This test would have to be changed if we
3414 would allow matching multibyte sequences. */
3417 int32_t idx2 = table[ch];
3418 size_t len = weights[idx2];
3420 /* Test whether the lenghts match. */
3421 if (weights[idx] == len)
3423 /* They do. New compare the bytes of
3428 && (weights[idx + 1 + cnt]
3429 == weights[idx2 + 1 + cnt]))
3433 /* They match. Mark the character as
3440 had_char_class = true;
3450 had_char_class = false;
3453 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '.')
3455 unsigned char str[128]; /* Should be large enough. */
3458 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3464 /* If pattern is `[[.'. */
3465 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3470 if ((c == '.' && *p == ']') || p == pend)
3472 if (c1 < sizeof (str))
3475 /* This is in any case an invalid class name. */
3480 if (c == '.' && *p == ']' && str[0] != '\0')
3482 /* If we have no collation data we use the default
3483 collation in which each character is the name
3484 for its own class which contains only the one
3485 character. It also means that ASCII is the
3486 character set and therefore we cannot have character
3487 with more than one byte in the multibyte
3494 FREE_STACK_RETURN (REG_ECOLLATE);
3496 /* Throw away the ] at the end of the equivalence
3500 /* Set the bit for the character. */
3501 SET_LIST_BIT (str[0]);
3502 range_start = ((const unsigned char *) str)[0];
3507 /* Try to match the byte sequence in `str' against
3508 those known to the collate implementation.
3509 First find out whether the bytes in `str' are
3510 actually from exactly one character. */
3512 const int32_t *symb_table;
3513 const unsigned char *extra;
3520 _NL_CURRENT_WORD (LC_COLLATE,
3521 _NL_COLLATE_SYMB_HASH_SIZEMB);
3522 symb_table = (const int32_t *)
3523 _NL_CURRENT (LC_COLLATE,
3524 _NL_COLLATE_SYMB_TABLEMB);
3525 extra = (const unsigned char *)
3526 _NL_CURRENT (LC_COLLATE,
3527 _NL_COLLATE_SYMB_EXTRAMB);
3529 /* Locate the character in the hashing table. */
3530 hash = elem_hash ((const char *) str, c1);
3533 elem = hash % table_size;
3534 second = hash % (table_size - 2);
3535 while (symb_table[2 * elem] != 0)
3537 /* First compare the hashing value. */
3538 if (symb_table[2 * elem] == hash
3539 && c1 == extra[symb_table[2 * elem + 1]]
3541 &extra[symb_table[2 * elem + 1]
3545 /* Yep, this is the entry. */
3546 idx = symb_table[2 * elem + 1];
3547 idx += 1 + extra[idx];
3555 if (symb_table[2 * elem] == 0)
3556 /* This is no valid character. */
3557 FREE_STACK_RETURN (REG_ECOLLATE);
3559 /* Throw away the ] at the end of the equivalence
3563 /* Now add the multibyte character(s) we found
3566 XXX Note that this is not entirely correct.
3567 we would have to match multibyte sequences
3568 but this is not possible with the current
3569 implementation. Also, we have to match
3570 collating symbols, which expand to more than
3571 one file, as a whole and not allow the
3572 individual bytes. */
3575 range_start = extra[idx];
3578 SET_LIST_BIT (extra[idx]);
3583 had_char_class = false;
3593 had_char_class = false;
3598 had_char_class = false;
3604 /* Discard any (non)matching list bytes that are all 0 at the
3605 end of the map. Decrease the map-length byte too. */
3606 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
3615 if (syntax & RE_NO_BK_PARENS)
3622 if (syntax & RE_NO_BK_PARENS)
3629 if (syntax & RE_NEWLINE_ALT)
3636 if (syntax & RE_NO_BK_VBAR)
3643 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
3644 goto handle_interval;
3650 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3652 /* Do not translate the character after the \, so that we can
3653 distinguish, e.g., \B from \b, even if we normally would
3654 translate, e.g., B to b. */
3660 if (syntax & RE_NO_BK_PARENS)
3661 goto normal_backslash;
3667 if (COMPILE_STACK_FULL)
3669 RETALLOC (compile_stack.stack, compile_stack.size << 1,
3670 compile_stack_elt_t);
3671 if (compile_stack.stack == NULL) return REG_ESPACE;
3673 compile_stack.size <<= 1;
3676 /* These are the values to restore when we hit end of this
3677 group. They are all relative offsets, so that if the
3678 whole pattern moves because of realloc, they will still
3680 COMPILE_STACK_TOP.begalt_offset = begalt - COMPILED_BUFFER_VAR;
3681 COMPILE_STACK_TOP.fixup_alt_jump
3682 = fixup_alt_jump ? fixup_alt_jump - COMPILED_BUFFER_VAR + 1 : 0;
3683 COMPILE_STACK_TOP.laststart_offset = b - COMPILED_BUFFER_VAR;
3684 COMPILE_STACK_TOP.regnum = regnum;
3686 /* We will eventually replace the 0 with the number of
3687 groups inner to this one. But do not push a
3688 start_memory for groups beyond the last one we can
3689 represent in the compiled pattern. */
3690 if (regnum <= MAX_REGNUM)
3692 COMPILE_STACK_TOP.inner_group_offset = b
3693 - COMPILED_BUFFER_VAR + 2;
3694 BUF_PUSH_3 (start_memory, regnum, 0);
3697 compile_stack.avail++;
3702 /* If we've reached MAX_REGNUM groups, then this open
3703 won't actually generate any code, so we'll have to
3704 clear pending_exact explicitly. */
3710 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
3712 if (COMPILE_STACK_EMPTY)
3714 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3715 goto normal_backslash;
3717 FREE_STACK_RETURN (REG_ERPAREN);
3722 { /* Push a dummy failure point at the end of the
3723 alternative for a possible future
3724 `pop_failure_jump' to pop. See comments at
3725 `push_dummy_failure' in `re_match_2'. */
3726 BUF_PUSH (push_dummy_failure);
3728 /* We allocated space for this jump when we assigned
3729 to `fixup_alt_jump', in the `handle_alt' case below. */
3730 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
3733 /* See similar code for backslashed left paren above. */
3734 if (COMPILE_STACK_EMPTY)
3736 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3739 FREE_STACK_RETURN (REG_ERPAREN);
3742 /* Since we just checked for an empty stack above, this
3743 ``can't happen''. */
3744 assert (compile_stack.avail != 0);
3746 /* We don't just want to restore into `regnum', because
3747 later groups should continue to be numbered higher,
3748 as in `(ab)c(de)' -- the second group is #2. */
3749 regnum_t this_group_regnum;
3751 compile_stack.avail--;
3752 begalt = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.begalt_offset;
3754 = COMPILE_STACK_TOP.fixup_alt_jump
3755 ? COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.fixup_alt_jump - 1
3757 laststart = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.laststart_offset;
3758 this_group_regnum = COMPILE_STACK_TOP.regnum;
3759 /* If we've reached MAX_REGNUM groups, then this open
3760 won't actually generate any code, so we'll have to
3761 clear pending_exact explicitly. */
3764 /* We're at the end of the group, so now we know how many
3765 groups were inside this one. */
3766 if (this_group_regnum <= MAX_REGNUM)
3768 UCHAR_T *inner_group_loc
3769 = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.inner_group_offset;
3771 *inner_group_loc = regnum - this_group_regnum;
3772 BUF_PUSH_3 (stop_memory, this_group_regnum,
3773 regnum - this_group_regnum);
3779 case '|': /* `\|'. */
3780 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
3781 goto normal_backslash;
3783 if (syntax & RE_LIMITED_OPS)
3786 /* Insert before the previous alternative a jump which
3787 jumps to this alternative if the former fails. */
3788 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3789 INSERT_JUMP (on_failure_jump, begalt,
3790 b + 2 + 2 * OFFSET_ADDRESS_SIZE);
3792 b += 1 + OFFSET_ADDRESS_SIZE;
3794 /* The alternative before this one has a jump after it
3795 which gets executed if it gets matched. Adjust that
3796 jump so it will jump to this alternative's analogous
3797 jump (put in below, which in turn will jump to the next
3798 (if any) alternative's such jump, etc.). The last such
3799 jump jumps to the correct final destination. A picture:
3805 If we are at `b', then fixup_alt_jump right now points to a
3806 three-byte space after `a'. We'll put in the jump, set
3807 fixup_alt_jump to right after `b', and leave behind three
3808 bytes which we'll fill in when we get to after `c'. */
3811 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
3813 /* Mark and leave space for a jump after this alternative,
3814 to be filled in later either by next alternative or
3815 when know we're at the end of a series of alternatives. */
3817 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3818 b += 1 + OFFSET_ADDRESS_SIZE;
3826 /* If \{ is a literal. */
3827 if (!(syntax & RE_INTERVALS)
3828 /* If we're at `\{' and it's not the open-interval
3830 || (syntax & RE_NO_BK_BRACES))
3831 goto normal_backslash;
3835 /* If got here, then the syntax allows intervals. */
3837 /* At least (most) this many matches must be made. */
3838 int lower_bound = -1, upper_bound = -1;
3840 /* Place in the uncompiled pattern (i.e., just after
3841 the '{') to go back to if the interval is invalid. */
3842 const CHAR_T *beg_interval = p;
3845 goto invalid_interval;
3847 GET_UNSIGNED_NUMBER (lower_bound);
3851 GET_UNSIGNED_NUMBER (upper_bound);
3852 if (upper_bound < 0)
3853 upper_bound = RE_DUP_MAX;
3856 /* Interval such as `{1}' => match exactly once. */
3857 upper_bound = lower_bound;
3859 if (! (0 <= lower_bound && lower_bound <= upper_bound))
3860 goto invalid_interval;
3862 if (!(syntax & RE_NO_BK_BRACES))
3864 if (c != '\\' || p == pend)
3865 goto invalid_interval;
3870 goto invalid_interval;
3872 /* If it's invalid to have no preceding re. */
3875 if (syntax & RE_CONTEXT_INVALID_OPS
3876 && !(syntax & RE_INVALID_INTERVAL_ORD))
3877 FREE_STACK_RETURN (REG_BADRPT);
3878 else if (syntax & RE_CONTEXT_INDEP_OPS)
3881 goto unfetch_interval;
3884 /* We just parsed a valid interval. */
3886 if (RE_DUP_MAX < upper_bound)
3887 FREE_STACK_RETURN (REG_BADBR);
3889 /* If the upper bound is zero, don't want to succeed at
3890 all; jump from `laststart' to `b + 3', which will be
3891 the end of the buffer after we insert the jump. */
3892 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3893 instead of 'b + 3'. */
3894 if (upper_bound == 0)
3896 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3897 INSERT_JUMP (jump, laststart, b + 1
3898 + OFFSET_ADDRESS_SIZE);
3899 b += 1 + OFFSET_ADDRESS_SIZE;
3902 /* Otherwise, we have a nontrivial interval. When
3903 we're all done, the pattern will look like:
3904 set_number_at <jump count> <upper bound>
3905 set_number_at <succeed_n count> <lower bound>
3906 succeed_n <after jump addr> <succeed_n count>
3908 jump_n <succeed_n addr> <jump count>
3909 (The upper bound and `jump_n' are omitted if
3910 `upper_bound' is 1, though.) */
3912 { /* If the upper bound is > 1, we need to insert
3913 more at the end of the loop. */
3914 unsigned nbytes = 2 + 4 * OFFSET_ADDRESS_SIZE +
3915 (upper_bound > 1) * (2 + 4 * OFFSET_ADDRESS_SIZE);
3917 GET_BUFFER_SPACE (nbytes);
3919 /* Initialize lower bound of the `succeed_n', even
3920 though it will be set during matching by its
3921 attendant `set_number_at' (inserted next),
3922 because `re_compile_fastmap' needs to know.
3923 Jump to the `jump_n' we might insert below. */
3924 INSERT_JUMP2 (succeed_n, laststart,
3925 b + 1 + 2 * OFFSET_ADDRESS_SIZE
3926 + (upper_bound > 1) * (1 + 2 * OFFSET_ADDRESS_SIZE)
3928 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3930 /* Code to initialize the lower bound. Insert
3931 before the `succeed_n'. The `5' is the last two
3932 bytes of this `set_number_at', plus 3 bytes of
3933 the following `succeed_n'. */
3934 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3935 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3936 of the following `succeed_n'. */
3937 PREFIX(insert_op2) (set_number_at, laststart, 1
3938 + 2 * OFFSET_ADDRESS_SIZE, lower_bound, b);
3939 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3941 if (upper_bound > 1)
3942 { /* More than one repetition is allowed, so
3943 append a backward jump to the `succeed_n'
3944 that starts this interval.
3946 When we've reached this during matching,
3947 we'll have matched the interval once, so
3948 jump back only `upper_bound - 1' times. */
3949 STORE_JUMP2 (jump_n, b, laststart
3950 + 2 * OFFSET_ADDRESS_SIZE + 1,
3952 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3954 /* The location we want to set is the second
3955 parameter of the `jump_n'; that is `b-2' as
3956 an absolute address. `laststart' will be
3957 the `set_number_at' we're about to insert;
3958 `laststart+3' the number to set, the source
3959 for the relative address. But we are
3960 inserting into the middle of the pattern --
3961 so everything is getting moved up by 5.
3962 Conclusion: (b - 2) - (laststart + 3) + 5,
3963 i.e., b - laststart.
3965 We insert this at the beginning of the loop
3966 so that if we fail during matching, we'll
3967 reinitialize the bounds. */
3968 PREFIX(insert_op2) (set_number_at, laststart,
3970 upper_bound - 1, b);
3971 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3978 if (!(syntax & RE_INVALID_INTERVAL_ORD))
3979 FREE_STACK_RETURN (p == pend ? REG_EBRACE : REG_BADBR);
3981 /* Match the characters as literals. */
3984 if (syntax & RE_NO_BK_BRACES)
3987 goto normal_backslash;
3991 /* There is no way to specify the before_dot and after_dot
3992 operators. rms says this is ok. --karl */
4000 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
4006 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
4012 if (syntax & RE_NO_GNU_OPS)
4015 BUF_PUSH (wordchar);
4020 if (syntax & RE_NO_GNU_OPS)
4023 BUF_PUSH (notwordchar);
4028 if (syntax & RE_NO_GNU_OPS)
4034 if (syntax & RE_NO_GNU_OPS)
4040 if (syntax & RE_NO_GNU_OPS)
4042 BUF_PUSH (wordbound);
4046 if (syntax & RE_NO_GNU_OPS)
4048 BUF_PUSH (notwordbound);
4052 if (syntax & RE_NO_GNU_OPS)
4058 if (syntax & RE_NO_GNU_OPS)
4063 case '1': case '2': case '3': case '4': case '5':
4064 case '6': case '7': case '8': case '9':
4065 if (syntax & RE_NO_BK_REFS)
4071 FREE_STACK_RETURN (REG_ESUBREG);
4073 /* Can't back reference to a subexpression if inside of it. */
4074 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
4078 BUF_PUSH_2 (duplicate, c1);
4084 if (syntax & RE_BK_PLUS_QM)
4087 goto normal_backslash;
4091 /* You might think it would be useful for \ to mean
4092 not to translate; but if we don't translate it
4093 it will never match anything. */
4101 /* Expects the character in `c'. */
4103 /* If no exactn currently being built. */
4106 /* If last exactn handle binary(or character) and
4107 new exactn handle character(or binary). */
4108 || is_exactn_bin != is_binary[p - 1 - pattern]
4111 /* If last exactn not at current position. */
4112 || pending_exact + *pending_exact + 1 != b
4114 /* We have only one byte following the exactn for the count. */
4115 || *pending_exact == (1 << BYTEWIDTH) - 1
4117 /* If followed by a repetition operator. */
4118 || *p == '*' || *p == '^'
4119 || ((syntax & RE_BK_PLUS_QM)
4120 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
4121 : (*p == '+' || *p == '?'))
4122 || ((syntax & RE_INTERVALS)
4123 && ((syntax & RE_NO_BK_BRACES)
4125 : (p[0] == '\\' && p[1] == '{'))))
4127 /* Start building a new exactn. */
4132 /* Is this exactn binary data or character? */
4133 is_exactn_bin = is_binary[p - 1 - pattern];
4135 BUF_PUSH_2 (exactn_bin, 0);
4137 BUF_PUSH_2 (exactn, 0);
4139 BUF_PUSH_2 (exactn, 0);
4141 pending_exact = b - 1;
4148 } /* while p != pend */
4151 /* Through the pattern now. */
4154 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
4156 if (!COMPILE_STACK_EMPTY)
4157 FREE_STACK_RETURN (REG_EPAREN);
4159 /* If we don't want backtracking, force success
4160 the first time we reach the end of the compiled pattern. */
4161 if (syntax & RE_NO_POSIX_BACKTRACKING)
4169 free (compile_stack.stack);
4171 /* We have succeeded; set the length of the buffer. */
4173 bufp->used = (uintptr_t) b - (uintptr_t) COMPILED_BUFFER_VAR;
4175 bufp->used = b - bufp->buffer;
4181 DEBUG_PRINT1 ("\nCompiled pattern: \n");
4182 PREFIX(print_compiled_pattern) (bufp);
4186 #ifndef MATCH_MAY_ALLOCATE
4187 /* Initialize the failure stack to the largest possible stack. This
4188 isn't necessary unless we're trying to avoid calling alloca in
4189 the search and match routines. */
4191 int num_regs = bufp->re_nsub + 1;
4193 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4194 is strictly greater than re_max_failures, the largest possible stack
4195 is 2 * re_max_failures failure points. */
4196 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
4198 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
4201 if (! fail_stack.stack)
4203 = (PREFIX(fail_stack_elt_t) *) xmalloc (fail_stack.size
4204 * sizeof (PREFIX(fail_stack_elt_t)));
4207 = (PREFIX(fail_stack_elt_t) *) xrealloc (fail_stack.stack,
4209 * sizeof (PREFIX(fail_stack_elt_t))));
4210 # else /* not emacs */
4211 if (! fail_stack.stack)
4213 = (PREFIX(fail_stack_elt_t) *) malloc (fail_stack.size
4214 * sizeof (PREFIX(fail_stack_elt_t)));
4217 = (PREFIX(fail_stack_elt_t) *) realloc (fail_stack.stack,
4219 * sizeof (PREFIX(fail_stack_elt_t))));
4220 # endif /* not emacs */
4223 PREFIX(regex_grow_registers) (num_regs);
4225 #endif /* not MATCH_MAY_ALLOCATE */
4228 } /* regex_compile */
4230 /* Subroutines for `regex_compile'. */
4232 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4233 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4236 PREFIX(store_op1) (re_opcode_t op, UCHAR_T *loc, int arg)
4238 *loc = (UCHAR_T) op;
4239 STORE_NUMBER (loc + 1, arg);
4243 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4244 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4247 PREFIX(store_op2) (re_opcode_t op, UCHAR_T *loc, int arg1, int arg2)
4249 *loc = (UCHAR_T) op;
4250 STORE_NUMBER (loc + 1, arg1);
4251 STORE_NUMBER (loc + 1 + OFFSET_ADDRESS_SIZE, arg2);
4255 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4256 for OP followed by two-byte integer parameter ARG. */
4257 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4260 PREFIX(insert_op1) (re_opcode_t op, UCHAR_T *loc, int arg, UCHAR_T *end)
4262 register UCHAR_T *pfrom = end;
4263 register UCHAR_T *pto = end + 1 + OFFSET_ADDRESS_SIZE;
4265 while (pfrom != loc)
4268 PREFIX(store_op1) (op, loc, arg);
4272 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4273 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4276 PREFIX(insert_op2) (re_opcode_t op, UCHAR_T *loc, int arg1,
4277 int arg2, UCHAR_T *end)
4279 register UCHAR_T *pfrom = end;
4280 register UCHAR_T *pto = end + 1 + 2 * OFFSET_ADDRESS_SIZE;
4282 while (pfrom != loc)
4285 PREFIX(store_op2) (op, loc, arg1, arg2);
4289 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4290 after an alternative or a begin-subexpression. We assume there is at
4291 least one character before the ^. */
4294 PREFIX(at_begline_loc_p) (const CHAR_T *pattern, const CHAR_T *p,
4295 reg_syntax_t syntax)
4297 const CHAR_T *prev = p - 2;
4298 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
4301 /* After a subexpression? */
4302 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
4303 /* After an alternative? */
4304 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
4308 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4309 at least one character after the $, i.e., `P < PEND'. */
4312 PREFIX(at_endline_loc_p) (const CHAR_T *p, const CHAR_T *pend,
4313 reg_syntax_t syntax)
4315 const CHAR_T *next = p;
4316 boolean next_backslash = *next == '\\';
4317 const CHAR_T *next_next = p + 1 < pend ? p + 1 : 0;
4320 /* Before a subexpression? */
4321 (syntax & RE_NO_BK_PARENS ? *next == ')'
4322 : next_backslash && next_next && *next_next == ')')
4323 /* Before an alternative? */
4324 || (syntax & RE_NO_BK_VBAR ? *next == '|'
4325 : next_backslash && next_next && *next_next == '|');
4328 #else /* not INSIDE_RECURSION */
4330 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4331 false if it's not. */
4334 group_in_compile_stack (compile_stack_type compile_stack, regnum_t regnum)
4338 for (this_element = compile_stack.avail - 1;
4341 if (compile_stack.stack[this_element].regnum == regnum)
4346 #endif /* not INSIDE_RECURSION */
4348 #ifdef INSIDE_RECURSION
4351 /* This insert space, which size is "num", into the pattern at "loc".
4352 "end" must point the end of the allocated buffer. */
4354 insert_space (int num, CHAR_T *loc, CHAR_T *end)
4356 register CHAR_T *pto = end;
4357 register CHAR_T *pfrom = end - num;
4359 while (pfrom >= loc)
4365 static reg_errcode_t
4366 wcs_compile_range (CHAR_T range_start_char, const CHAR_T **p_ptr,
4367 const CHAR_T *pend, RE_TRANSLATE_TYPE translate,
4368 reg_syntax_t syntax, CHAR_T *b, CHAR_T *char_set)
4370 const CHAR_T *p = *p_ptr;
4371 CHAR_T range_start, range_end;
4375 uint32_t start_val, end_val;
4381 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
4384 const char *collseq = (const char *) _NL_CURRENT(LC_COLLATE,
4385 _NL_COLLATE_COLLSEQWC);
4386 const unsigned char *extra = (const unsigned char *)
4387 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
4389 if (range_start_char < -1)
4391 /* range_start is a collating symbol. */
4393 /* Retreive the index and get collation sequence value. */
4394 wextra = (int32_t*)(extra + char_set[-range_start_char]);
4395 start_val = wextra[1 + *wextra];
4398 start_val = collseq_table_lookup(collseq, TRANSLATE(range_start_char));
4400 end_val = collseq_table_lookup (collseq, TRANSLATE (p[0]));
4402 /* Report an error if the range is empty and the syntax prohibits
4404 ret = ((syntax & RE_NO_EMPTY_RANGES)
4405 && (start_val > end_val))? REG_ERANGE : REG_NOERROR;
4407 /* Insert space to the end of the char_ranges. */
4408 insert_space(2, b - char_set[5] - 2, b - 1);
4409 *(b - char_set[5] - 2) = (wchar_t)start_val;
4410 *(b - char_set[5] - 1) = (wchar_t)end_val;
4411 char_set[4]++; /* ranges_index */
4416 range_start = (range_start_char >= 0)? TRANSLATE (range_start_char):
4418 range_end = TRANSLATE (p[0]);
4419 /* Report an error if the range is empty and the syntax prohibits
4421 ret = ((syntax & RE_NO_EMPTY_RANGES)
4422 && (range_start > range_end))? REG_ERANGE : REG_NOERROR;
4424 /* Insert space to the end of the char_ranges. */
4425 insert_space(2, b - char_set[5] - 2, b - 1);
4426 *(b - char_set[5] - 2) = range_start;
4427 *(b - char_set[5] - 1) = range_end;
4428 char_set[4]++; /* ranges_index */
4430 /* Have to increment the pointer into the pattern string, so the
4431 caller isn't still at the ending character. */
4437 /* Read the ending character of a range (in a bracket expression) from the
4438 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4439 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4440 Then we set the translation of all bits between the starting and
4441 ending characters (inclusive) in the compiled pattern B.
4443 Return an error code.
4445 We use these short variable names so we can use the same macros as
4446 `regex_compile' itself. */
4448 static reg_errcode_t
4449 byte_compile_range (unsigned int range_start_char, const char **p_ptr,
4450 const char *pend, RE_TRANSLATE_TYPE translate,
4451 reg_syntax_t syntax, unsigned char *b)
4454 const char *p = *p_ptr;
4457 const unsigned char *collseq;
4458 unsigned int start_colseq;
4459 unsigned int end_colseq;
4467 /* Have to increment the pointer into the pattern string, so the
4468 caller isn't still at the ending character. */
4471 /* Report an error if the range is empty and the syntax prohibits this. */
4472 ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
4475 collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
4476 _NL_COLLATE_COLLSEQMB);
4478 start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
4479 end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
4480 for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
4482 unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
4484 if (start_colseq <= this_colseq && this_colseq <= end_colseq)
4486 SET_LIST_BIT (TRANSLATE (this_char));
4491 /* Here we see why `this_char' has to be larger than an `unsigned
4492 char' -- we would otherwise go into an infinite loop, since all
4493 characters <= 0xff. */
4494 range_start_char = TRANSLATE (range_start_char);
4495 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4496 and some compilers cast it to int implicitly, so following for_loop
4497 may fall to (almost) infinite loop.
4498 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4499 To avoid this, we cast p[0] to unsigned int and truncate it. */
4500 end_char = ((unsigned)TRANSLATE(p[0]) & ((1 << BYTEWIDTH) - 1));
4502 for (this_char = range_start_char; this_char <= end_char; ++this_char)
4504 SET_LIST_BIT (TRANSLATE (this_char));
4513 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4514 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4515 characters can start a string that matches the pattern. This fastmap
4516 is used by re_search to skip quickly over impossible starting points.
4518 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4519 area as BUFP->fastmap.
4521 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4524 Returns 0 if we succeed, -2 if an internal error. */
4527 /* local function for re_compile_fastmap.
4528 truncate wchar_t character to char. */
4529 static unsigned char truncate_wchar (CHAR_T c);
4531 static unsigned char
4532 truncate_wchar (CHAR_T c)
4534 unsigned char buf[MB_CUR_MAX];
4537 memset (&state, '\0', sizeof (state));
4539 retval = __wcrtomb (buf, c, &state);
4541 retval = wcrtomb (buf, c, &state);
4543 return retval > 0 ? buf[0] : (unsigned char) c;
4548 PREFIX(re_compile_fastmap) (struct re_pattern_buffer *bufp)
4551 #ifdef MATCH_MAY_ALLOCATE
4552 PREFIX(fail_stack_type) fail_stack;
4554 #ifndef REGEX_MALLOC
4558 register char *fastmap = bufp->fastmap;
4561 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4562 pattern to (char*) in regex_compile. */
4563 UCHAR_T *pattern = (UCHAR_T*)bufp->buffer;
4564 register UCHAR_T *pend = (UCHAR_T*) (bufp->buffer + bufp->used);
4566 UCHAR_T *pattern = bufp->buffer;
4567 register UCHAR_T *pend = pattern + bufp->used;
4569 UCHAR_T *p = pattern;
4572 /* This holds the pointer to the failure stack, when
4573 it is allocated relocatably. */
4574 fail_stack_elt_t *failure_stack_ptr;
4577 /* Assume that each path through the pattern can be null until
4578 proven otherwise. We set this false at the bottom of switch
4579 statement, to which we get only if a particular path doesn't
4580 match the empty string. */
4581 boolean path_can_be_null = true;
4583 /* We aren't doing a `succeed_n' to begin with. */
4584 boolean succeed_n_p = false;
4586 assert (fastmap != NULL && p != NULL);
4589 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
4590 bufp->fastmap_accurate = 1; /* It will be when we're done. */
4591 bufp->can_be_null = 0;
4595 if (p == pend || *p == (UCHAR_T) succeed)
4597 /* We have reached the (effective) end of pattern. */
4598 if (!FAIL_STACK_EMPTY ())
4600 bufp->can_be_null |= path_can_be_null;
4602 /* Reset for next path. */
4603 path_can_be_null = true;
4605 p = fail_stack.stack[--fail_stack.avail].pointer;
4613 /* We should never be about to go beyond the end of the pattern. */
4616 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4619 /* I guess the idea here is to simply not bother with a fastmap
4620 if a backreference is used, since it's too hard to figure out
4621 the fastmap for the corresponding group. Setting
4622 `can_be_null' stops `re_search_2' from using the fastmap, so
4623 that is all we do. */
4625 bufp->can_be_null = 1;
4629 /* Following are the cases which match a character. These end
4634 fastmap[truncate_wchar(p[1])] = 1;
4648 /* It is hard to distinguish fastmap from (multi byte) characters
4649 which depends on current locale. */
4654 bufp->can_be_null = 1;
4658 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4659 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
4665 /* Chars beyond end of map must be allowed. */
4666 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
4669 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4670 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
4676 for (j = 0; j < (1 << BYTEWIDTH); j++)
4677 if (SYNTAX (j) == Sword)
4683 for (j = 0; j < (1 << BYTEWIDTH); j++)
4684 if (SYNTAX (j) != Sword)
4691 int fastmap_newline = fastmap['\n'];
4693 /* `.' matches anything ... */
4694 for (j = 0; j < (1 << BYTEWIDTH); j++)
4697 /* ... except perhaps newline. */
4698 if (!(bufp->syntax & RE_DOT_NEWLINE))
4699 fastmap['\n'] = fastmap_newline;
4701 /* Return if we have already set `can_be_null'; if we have,
4702 then the fastmap is irrelevant. Something's wrong here. */
4703 else if (bufp->can_be_null)
4706 /* Otherwise, have to check alternative paths. */
4713 for (j = 0; j < (1 << BYTEWIDTH); j++)
4714 if (SYNTAX (j) == (enum syntaxcode) k)
4721 for (j = 0; j < (1 << BYTEWIDTH); j++)
4722 if (SYNTAX (j) != (enum syntaxcode) k)
4727 /* All cases after this match the empty string. These end with
4747 case push_dummy_failure:
4752 case pop_failure_jump:
4753 case maybe_pop_jump:
4756 case dummy_failure_jump:
4757 EXTRACT_NUMBER_AND_INCR (j, p);
4762 /* Jump backward implies we just went through the body of a
4763 loop and matched nothing. Opcode jumped to should be
4764 `on_failure_jump' or `succeed_n'. Just treat it like an
4765 ordinary jump. For a * loop, it has pushed its failure
4766 point already; if so, discard that as redundant. */
4767 if ((re_opcode_t) *p != on_failure_jump
4768 && (re_opcode_t) *p != succeed_n)
4772 EXTRACT_NUMBER_AND_INCR (j, p);
4775 /* If what's on the stack is where we are now, pop it. */
4776 if (!FAIL_STACK_EMPTY ()
4777 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
4783 case on_failure_jump:
4784 case on_failure_keep_string_jump:
4785 handle_on_failure_jump:
4786 EXTRACT_NUMBER_AND_INCR (j, p);
4788 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4789 end of the pattern. We don't want to push such a point,
4790 since when we restore it above, entering the switch will
4791 increment `p' past the end of the pattern. We don't need
4792 to push such a point since we obviously won't find any more
4793 fastmap entries beyond `pend'. Such a pattern can match
4794 the null string, though. */
4797 if (!PUSH_PATTERN_OP (p + j, fail_stack))
4799 RESET_FAIL_STACK ();
4804 bufp->can_be_null = 1;
4808 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
4809 succeed_n_p = false;
4816 /* Get to the number of times to succeed. */
4817 p += OFFSET_ADDRESS_SIZE;
4819 /* Increment p past the n for when k != 0. */
4820 EXTRACT_NUMBER_AND_INCR (k, p);
4823 p -= 2 * OFFSET_ADDRESS_SIZE;
4824 succeed_n_p = true; /* Spaghetti code alert. */
4825 goto handle_on_failure_jump;
4831 p += 2 * OFFSET_ADDRESS_SIZE;
4842 abort (); /* We have listed all the cases. */
4845 /* Getting here means we have found the possible starting
4846 characters for one path of the pattern -- and that the empty
4847 string does not match. We need not follow this path further.
4848 Instead, look at the next alternative (remembered on the
4849 stack), or quit if no more. The test at the top of the loop
4850 does these things. */
4851 path_can_be_null = false;
4855 /* Set `can_be_null' for the last path (also the first path, if the
4856 pattern is empty). */
4857 bufp->can_be_null |= path_can_be_null;
4860 RESET_FAIL_STACK ();
4864 #else /* not INSIDE_RECURSION */
4867 re_compile_fastmap (struct re_pattern_buffer *bufp)
4870 if (MB_CUR_MAX != 1)
4871 return wcs_re_compile_fastmap(bufp);
4874 return byte_re_compile_fastmap(bufp);
4875 } /* re_compile_fastmap */
4877 weak_alias (__re_compile_fastmap, re_compile_fastmap)
4881 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4882 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4883 this memory for recording register information. STARTS and ENDS
4884 must be allocated using the malloc library routine, and must each
4885 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4887 If NUM_REGS == 0, then subsequent matches should allocate their own
4890 Unless this function is called, the first search or match using
4891 PATTERN_BUFFER will allocate its own register data, without
4892 freeing the old data. */
4895 re_set_registers (struct re_pattern_buffer *bufp,
4896 struct re_registers *regs, unsigned num_regs,
4897 regoff_t *starts, regoff_t *ends)
4901 bufp->regs_allocated = REGS_REALLOCATE;
4902 regs->num_regs = num_regs;
4903 regs->start = starts;
4908 bufp->regs_allocated = REGS_UNALLOCATED;
4910 regs->start = regs->end = (regoff_t *) 0;
4914 weak_alias (__re_set_registers, re_set_registers)
4917 /* Searching routines. */
4919 /* Like re_search_2, below, but only one string is specified, and
4920 doesn't let you say where to stop matching. */
4923 re_search (struct re_pattern_buffer *bufp, const char *string, int size,
4924 int startpos, int range, struct re_registers *regs)
4926 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
4930 weak_alias (__re_search, re_search)
4934 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4935 virtual concatenation of STRING1 and STRING2, starting first at index
4936 STARTPOS, then at STARTPOS + 1, and so on.
4938 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
4940 RANGE is how far to scan while trying to match. RANGE = 0 means try
4941 only at STARTPOS; in general, the last start tried is STARTPOS +
4944 In REGS, return the indices of the virtual concatenation of STRING1
4945 and STRING2 that matched the entire BUFP->buffer and its contained
4948 Do not consider matching one past the index STOP in the virtual
4949 concatenation of STRING1 and STRING2.
4951 We return either the position in the strings at which the match was
4952 found, -1 if no match, or -2 if error (such as failure
4956 re_search_2 (struct re_pattern_buffer *bufp, const char *string1, int size1,
4957 const char *string2, int size2, int startpos, int range,
4958 struct re_registers *regs, int stop)
4961 if (MB_CUR_MAX != 1)
4962 return wcs_re_search_2 (bufp, string1, size1, string2, size2, startpos,
4966 return byte_re_search_2 (bufp, string1, size1, string2, size2, startpos,
4970 weak_alias (__re_search_2, re_search_2)
4973 #endif /* not INSIDE_RECURSION */
4975 #ifdef INSIDE_RECURSION
4977 #ifdef MATCH_MAY_ALLOCATE
4978 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
4980 # define FREE_VAR(var) if (var) free (var); var = NULL
4984 # define MAX_ALLOCA_SIZE 2000
4986 # define FREE_WCS_BUFFERS() \
4988 if (size1 > MAX_ALLOCA_SIZE) \
4990 free (wcs_string1); \
4991 free (mbs_offset1); \
4995 FREE_VAR (wcs_string1); \
4996 FREE_VAR (mbs_offset1); \
4998 if (size2 > MAX_ALLOCA_SIZE) \
5000 free (wcs_string2); \
5001 free (mbs_offset2); \
5005 FREE_VAR (wcs_string2); \
5006 FREE_VAR (mbs_offset2); \
5014 PREFIX(re_search_2) (struct re_pattern_buffer *bufp, const char *string1,
5015 int size1, const char *string2, int size2,
5016 int startpos, int range,
5017 struct re_registers *regs, int stop)
5020 register char *fastmap = bufp->fastmap;
5021 register RE_TRANSLATE_TYPE translate = bufp->translate;
5022 int total_size = size1 + size2;
5023 int endpos = startpos + range;
5025 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5026 wchar_t *wcs_string1 = NULL, *wcs_string2 = NULL;
5027 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5028 int wcs_size1 = 0, wcs_size2 = 0;
5029 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5030 int *mbs_offset1 = NULL, *mbs_offset2 = NULL;
5031 /* They hold whether each wchar_t is binary data or not. */
5032 char *is_binary = NULL;
5035 /* Check for out-of-range STARTPOS. */
5036 if (startpos < 0 || startpos > total_size)
5039 /* Fix up RANGE if it might eventually take us outside
5040 the virtual concatenation of STRING1 and STRING2.
5041 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5043 range = 0 - startpos;
5044 else if (endpos > total_size)
5045 range = total_size - startpos;
5047 /* If the search isn't to be a backwards one, don't waste time in a
5048 search for a pattern that must be anchored. */
5049 if (bufp->used > 0 && range > 0
5050 && ((re_opcode_t) bufp->buffer[0] == begbuf
5051 /* `begline' is like `begbuf' if it cannot match at newlines. */
5052 || ((re_opcode_t) bufp->buffer[0] == begline
5053 && !bufp->newline_anchor)))
5062 /* In a forward search for something that starts with \=.
5063 don't keep searching past point. */
5064 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
5066 range = PT - startpos;
5072 /* Update the fastmap now if not correct already. */
5073 if (fastmap && !bufp->fastmap_accurate)
5074 if (re_compile_fastmap (bufp) == -2)
5078 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5079 fill them with converted string. */
5082 if (size1 > MAX_ALLOCA_SIZE)
5084 wcs_string1 = TALLOC (size1 + 1, CHAR_T);
5085 mbs_offset1 = TALLOC (size1 + 1, int);
5086 is_binary = TALLOC (size1 + 1, char);
5090 wcs_string1 = REGEX_TALLOC (size1 + 1, CHAR_T);
5091 mbs_offset1 = REGEX_TALLOC (size1 + 1, int);
5092 is_binary = REGEX_TALLOC (size1 + 1, char);
5094 if (!wcs_string1 || !mbs_offset1 || !is_binary)
5096 if (size1 > MAX_ALLOCA_SIZE)
5104 FREE_VAR (wcs_string1);
5105 FREE_VAR (mbs_offset1);
5106 FREE_VAR (is_binary);
5110 wcs_size1 = convert_mbs_to_wcs(wcs_string1, string1, size1,
5111 mbs_offset1, is_binary);
5112 wcs_string1[wcs_size1] = L'\0'; /* for a sentinel */
5113 if (size1 > MAX_ALLOCA_SIZE)
5116 FREE_VAR (is_binary);
5120 if (size2 > MAX_ALLOCA_SIZE)
5122 wcs_string2 = TALLOC (size2 + 1, CHAR_T);
5123 mbs_offset2 = TALLOC (size2 + 1, int);
5124 is_binary = TALLOC (size2 + 1, char);
5128 wcs_string2 = REGEX_TALLOC (size2 + 1, CHAR_T);
5129 mbs_offset2 = REGEX_TALLOC (size2 + 1, int);
5130 is_binary = REGEX_TALLOC (size2 + 1, char);
5132 if (!wcs_string2 || !mbs_offset2 || !is_binary)
5134 FREE_WCS_BUFFERS ();
5135 if (size2 > MAX_ALLOCA_SIZE)
5138 FREE_VAR (is_binary);
5141 wcs_size2 = convert_mbs_to_wcs(wcs_string2, string2, size2,
5142 mbs_offset2, is_binary);
5143 wcs_string2[wcs_size2] = L'\0'; /* for a sentinel */
5144 if (size2 > MAX_ALLOCA_SIZE)
5147 FREE_VAR (is_binary);
5152 /* Loop through the string, looking for a place to start matching. */
5155 /* If a fastmap is supplied, skip quickly over characters that
5156 cannot be the start of a match. If the pattern can match the
5157 null string, however, we don't need to skip characters; we want
5158 the first null string. */
5159 if (fastmap && startpos < total_size && !bufp->can_be_null)
5161 if (range > 0) /* Searching forwards. */
5163 register const char *d;
5164 register int lim = 0;
5167 if (startpos < size1 && startpos + range >= size1)
5168 lim = range - (size1 - startpos);
5170 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
5172 /* Written out as an if-else to avoid testing `translate'
5176 && !fastmap[(unsigned char)
5177 translate[(unsigned char) *d++]])
5180 while (range > lim && !fastmap[(unsigned char) *d++])
5183 startpos += irange - range;
5185 else /* Searching backwards. */
5187 register CHAR_T c = (size1 == 0 || startpos >= size1
5188 ? string2[startpos - size1]
5189 : string1[startpos]);
5191 if (!fastmap[(unsigned char) TRANSLATE (c)])
5196 /* If can't match the null string, and that's all we have left, fail. */
5197 if (range >= 0 && startpos == total_size && fastmap
5198 && !bufp->can_be_null)
5201 FREE_WCS_BUFFERS ();
5207 val = wcs_re_match_2_internal (bufp, string1, size1, string2,
5208 size2, startpos, regs, stop,
5209 wcs_string1, wcs_size1,
5210 wcs_string2, wcs_size2,
5211 mbs_offset1, mbs_offset2);
5213 val = byte_re_match_2_internal (bufp, string1, size1, string2,
5214 size2, startpos, regs, stop);
5217 #ifndef REGEX_MALLOC
5226 FREE_WCS_BUFFERS ();
5234 FREE_WCS_BUFFERS ();
5254 FREE_WCS_BUFFERS ();
5260 /* This converts PTR, a pointer into one of the search wchar_t strings
5261 `string1' and `string2' into an multibyte string offset from the
5262 beginning of that string. We use mbs_offset to optimize.
5263 See convert_mbs_to_wcs. */
5264 # define POINTER_TO_OFFSET(ptr) \
5265 (FIRST_STRING_P (ptr) \
5266 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5267 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5270 /* This converts PTR, a pointer into one of the search strings `string1'
5271 and `string2' into an offset from the beginning of that string. */
5272 # define POINTER_TO_OFFSET(ptr) \
5273 (FIRST_STRING_P (ptr) \
5274 ? ((regoff_t) ((ptr) - string1)) \
5275 : ((regoff_t) ((ptr) - string2 + size1)))
5278 /* Macros for dealing with the split strings in re_match_2. */
5280 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5282 /* Call before fetching a character with *d. This switches over to
5283 string2 if necessary. */
5284 #define PREFETCH() \
5287 /* End of string2 => fail. */ \
5288 if (dend == end_match_2) \
5290 /* End of string1 => advance to string2. */ \
5292 dend = end_match_2; \
5295 /* Test if at very beginning or at very end of the virtual concatenation
5296 of `string1' and `string2'. If only one string, it's `string2'. */
5297 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5298 #define AT_STRINGS_END(d) ((d) == end2)
5301 /* Test if D points to a character which is word-constituent. We have
5302 two special cases to check for: if past the end of string1, look at
5303 the first character in string2; and if before the beginning of
5304 string2, look at the last character in string1. */
5306 /* Use internationalized API instead of SYNTAX. */
5307 # define WORDCHAR_P(d) \
5308 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5309 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5310 || ((d) == end1 ? *string2 \
5311 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) == L'_')
5313 # define WORDCHAR_P(d) \
5314 (SYNTAX ((d) == end1 ? *string2 \
5315 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5319 /* Disabled due to a compiler bug -- see comment at case wordbound */
5321 /* Test if the character before D and the one at D differ with respect
5322 to being word-constituent. */
5323 #define AT_WORD_BOUNDARY(d) \
5324 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5325 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5328 /* Free everything we malloc. */
5329 #ifdef MATCH_MAY_ALLOCATE
5331 # define FREE_VARIABLES() \
5333 REGEX_FREE_STACK (fail_stack.stack); \
5334 FREE_VAR (regstart); \
5335 FREE_VAR (regend); \
5336 FREE_VAR (old_regstart); \
5337 FREE_VAR (old_regend); \
5338 FREE_VAR (best_regstart); \
5339 FREE_VAR (best_regend); \
5340 FREE_VAR (reg_info); \
5341 FREE_VAR (reg_dummy); \
5342 FREE_VAR (reg_info_dummy); \
5343 if (!cant_free_wcs_buf) \
5345 FREE_VAR (string1); \
5346 FREE_VAR (string2); \
5347 FREE_VAR (mbs_offset1); \
5348 FREE_VAR (mbs_offset2); \
5352 # define FREE_VARIABLES() \
5354 REGEX_FREE_STACK (fail_stack.stack); \
5355 FREE_VAR (regstart); \
5356 FREE_VAR (regend); \
5357 FREE_VAR (old_regstart); \
5358 FREE_VAR (old_regend); \
5359 FREE_VAR (best_regstart); \
5360 FREE_VAR (best_regend); \
5361 FREE_VAR (reg_info); \
5362 FREE_VAR (reg_dummy); \
5363 FREE_VAR (reg_info_dummy); \
5368 # define FREE_VARIABLES() \
5370 if (!cant_free_wcs_buf) \
5372 FREE_VAR (string1); \
5373 FREE_VAR (string2); \
5374 FREE_VAR (mbs_offset1); \
5375 FREE_VAR (mbs_offset2); \
5379 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
5381 #endif /* not MATCH_MAY_ALLOCATE */
5383 /* These values must meet several constraints. They must not be valid
5384 register values; since we have a limit of 255 registers (because
5385 we use only one byte in the pattern for the register number), we can
5386 use numbers larger than 255. They must differ by 1, because of
5387 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5388 be larger than the value for the highest register, so we do not try
5389 to actually save any registers when none are active. */
5390 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5391 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5393 #else /* not INSIDE_RECURSION */
5394 /* Matching routines. */
5396 #ifndef emacs /* Emacs never uses this. */
5397 /* re_match is like re_match_2 except it takes only a single string. */
5400 re_match (struct re_pattern_buffer *bufp, const char *string,
5401 int size, int pos, struct re_registers *regs)
5405 if (MB_CUR_MAX != 1)
5406 result = wcs_re_match_2_internal (bufp, NULL, 0, string, size,
5408 NULL, 0, NULL, 0, NULL, NULL);
5411 result = byte_re_match_2_internal (bufp, NULL, 0, string, size,
5413 # ifndef REGEX_MALLOC
5421 weak_alias (__re_match, re_match)
5423 #endif /* not emacs */
5425 #endif /* not INSIDE_RECURSION */
5427 #ifdef INSIDE_RECURSION
5428 static boolean PREFIX(group_match_null_string_p) (UCHAR_T **p,
5430 PREFIX(register_info_type) *reg_info);
5431 static boolean PREFIX(alt_match_null_string_p) (UCHAR_T *p,
5433 PREFIX(register_info_type) *reg_info);
5434 static boolean PREFIX(common_op_match_null_string_p) (UCHAR_T **p,
5436 PREFIX(register_info_type) *reg_info);
5437 static int PREFIX(bcmp_translate) (const CHAR_T *s1, const CHAR_T *s2,
5439 RE_TRANSLATE_TYPE translate);
5440 #else /* not INSIDE_RECURSION */
5442 /* re_match_2 matches the compiled pattern in BUFP against the
5443 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5444 and SIZE2, respectively). We start matching at POS, and stop
5447 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5448 store offsets for the substring each group matched in REGS. See the
5449 documentation for exactly how many groups we fill.
5451 We return -1 if no match, -2 if an internal error (such as the
5452 failure stack overflowing). Otherwise, we return the length of the
5453 matched substring. */
5456 re_match_2 (struct re_pattern_buffer *bufp, const char *string1, int size1,
5457 const char *string2, int size2, int pos,
5458 struct re_registers *regs, int stop)
5462 if (MB_CUR_MAX != 1)
5463 result = wcs_re_match_2_internal (bufp, string1, size1, string2, size2,
5465 NULL, 0, NULL, 0, NULL, NULL);
5468 result = byte_re_match_2_internal (bufp, string1, size1, string2, size2,
5471 #ifndef REGEX_MALLOC
5479 weak_alias (__re_match_2, re_match_2)
5482 #endif /* not INSIDE_RECURSION */
5484 #ifdef INSIDE_RECURSION
5487 static int count_mbs_length (int *, int);
5489 /* This check the substring (from 0, to length) of the multibyte string,
5490 to which offset_buffer correspond. And count how many wchar_t_characters
5491 the substring occupy. We use offset_buffer to optimization.
5492 See convert_mbs_to_wcs. */
5495 count_mbs_length(int *offset_buffer, int length)
5499 /* Check whether the size is valid. */
5503 if (offset_buffer == NULL)
5506 /* If there are no multibyte character, offset_buffer[i] == i.
5507 Optmize for this case. */
5508 if (offset_buffer[length] == length)
5511 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5517 int middle = (lower + upper) / 2;
5518 if (middle == lower || middle == upper)
5520 if (offset_buffer[middle] > length)
5522 else if (offset_buffer[middle] < length)
5532 /* This is a separate function so that we can force an alloca cleanup
5536 wcs_re_match_2_internal (struct re_pattern_buffer *bufp,
5537 const char *cstring1, int csize1,
5538 const char *cstring2, int csize2,
5540 struct re_registers *regs,
5542 /* string1 == string2 == NULL means string1/2, size1/2 and
5543 mbs_offset1/2 need seting up in this function. */
5544 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5545 wchar_t *string1, int size1,
5546 wchar_t *string2, int size2,
5547 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5548 int *mbs_offset1, int *mbs_offset2)
5551 byte_re_match_2_internal (struct re_pattern_buffer *bufp,
5552 const char *string1, int size1,
5553 const char *string2, int size2,
5555 struct re_registers *regs, int stop)
5558 /* General temporaries. */
5562 /* They hold whether each wchar_t is binary data or not. */
5563 char *is_binary = NULL;
5564 /* If true, we can't free string1/2, mbs_offset1/2. */
5565 int cant_free_wcs_buf = 1;
5568 /* Just past the end of the corresponding string. */
5569 const CHAR_T *end1, *end2;
5571 /* Pointers into string1 and string2, just past the last characters in
5572 each to consider matching. */
5573 const CHAR_T *end_match_1, *end_match_2;
5575 /* Where we are in the data, and the end of the current string. */
5576 const CHAR_T *d, *dend;
5578 /* Where we are in the pattern, and the end of the pattern. */
5580 UCHAR_T *pattern, *p;
5581 register UCHAR_T *pend;
5583 UCHAR_T *p = bufp->buffer;
5584 register UCHAR_T *pend = p + bufp->used;
5587 /* Mark the opcode just after a start_memory, so we can test for an
5588 empty subpattern when we get to the stop_memory. */
5589 UCHAR_T *just_past_start_mem = 0;
5591 /* We use this to map every character in the string. */
5592 RE_TRANSLATE_TYPE translate = bufp->translate;
5594 /* Failure point stack. Each place that can handle a failure further
5595 down the line pushes a failure point on this stack. It consists of
5596 restart, regend, and reg_info for all registers corresponding to
5597 the subexpressions we're currently inside, plus the number of such
5598 registers, and, finally, two char *'s. The first char * is where
5599 to resume scanning the pattern; the second one is where to resume
5600 scanning the strings. If the latter is zero, the failure point is
5601 a ``dummy''; if a failure happens and the failure point is a dummy,
5602 it gets discarded and the next next one is tried. */
5603 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5604 PREFIX(fail_stack_type) fail_stack;
5607 static unsigned failure_id;
5608 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
5612 /* This holds the pointer to the failure stack, when
5613 it is allocated relocatably. */
5614 fail_stack_elt_t *failure_stack_ptr;
5617 /* We fill all the registers internally, independent of what we
5618 return, for use in backreferences. The number here includes
5619 an element for register zero. */
5620 size_t num_regs = bufp->re_nsub + 1;
5622 /* The currently active registers. */
5623 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
5624 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
5626 /* Information on the contents of registers. These are pointers into
5627 the input strings; they record just what was matched (on this
5628 attempt) by a subexpression part of the pattern, that is, the
5629 regnum-th regstart pointer points to where in the pattern we began
5630 matching and the regnum-th regend points to right after where we
5631 stopped matching the regnum-th subexpression. (The zeroth register
5632 keeps track of what the whole pattern matches.) */
5633 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5634 const CHAR_T **regstart, **regend;
5637 /* If a group that's operated upon by a repetition operator fails to
5638 match anything, then the register for its start will need to be
5639 restored because it will have been set to wherever in the string we
5640 are when we last see its open-group operator. Similarly for a
5642 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5643 const CHAR_T **old_regstart, **old_regend;
5646 /* The is_active field of reg_info helps us keep track of which (possibly
5647 nested) subexpressions we are currently in. The matched_something
5648 field of reg_info[reg_num] helps us tell whether or not we have
5649 matched any of the pattern so far this time through the reg_num-th
5650 subexpression. These two fields get reset each time through any
5651 loop their register is in. */
5652 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5653 PREFIX(register_info_type) *reg_info;
5656 /* The following record the register info as found in the above
5657 variables when we find a match better than any we've seen before.
5658 This happens as we backtrack through the failure points, which in
5659 turn happens only if we have not yet matched the entire string. */
5660 unsigned best_regs_set = false;
5661 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5662 const CHAR_T **best_regstart, **best_regend;
5665 /* Logically, this is `best_regend[0]'. But we don't want to have to
5666 allocate space for that if we're not allocating space for anything
5667 else (see below). Also, we never need info about register 0 for
5668 any of the other register vectors, and it seems rather a kludge to
5669 treat `best_regend' differently than the rest. So we keep track of
5670 the end of the best match so far in a separate variable. We
5671 initialize this to NULL so that when we backtrack the first time
5672 and need to test it, it's not garbage. */
5673 const CHAR_T *match_end = NULL;
5675 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5676 int set_regs_matched_done = 0;
5678 /* Used when we pop values we don't care about. */
5679 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5680 const CHAR_T **reg_dummy;
5681 PREFIX(register_info_type) *reg_info_dummy;
5685 /* Counts the total number of registers pushed. */
5686 unsigned num_regs_pushed = 0;
5689 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
5693 #ifdef MATCH_MAY_ALLOCATE
5694 /* Do not bother to initialize all the register variables if there are
5695 no groups in the pattern, as it takes a fair amount of time. If
5696 there are groups, we include space for register 0 (the whole
5697 pattern), even though we never use it, since it simplifies the
5698 array indexing. We should fix this. */
5701 regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5702 regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5703 old_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5704 old_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5705 best_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5706 best_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5707 reg_info = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5708 reg_dummy = REGEX_TALLOC (num_regs, const CHAR_T *);
5709 reg_info_dummy = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5711 if (!(regstart && regend && old_regstart && old_regend && reg_info
5712 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
5720 /* We must initialize all our variables to NULL, so that
5721 `FREE_VARIABLES' doesn't try to free them. */
5722 regstart = regend = old_regstart = old_regend = best_regstart
5723 = best_regend = reg_dummy = NULL;
5724 reg_info = reg_info_dummy = (PREFIX(register_info_type) *) NULL;
5726 #endif /* MATCH_MAY_ALLOCATE */
5728 /* The starting position is bogus. */
5730 if (pos < 0 || pos > csize1 + csize2)
5732 if (pos < 0 || pos > size1 + size2)
5740 /* Allocate wchar_t array for string1 and string2 and
5741 fill them with converted string. */
5742 if (string1 == NULL && string2 == NULL)
5744 /* We need seting up buffers here. */
5746 /* We must free wcs buffers in this function. */
5747 cant_free_wcs_buf = 0;
5751 string1 = REGEX_TALLOC (csize1 + 1, CHAR_T);
5752 mbs_offset1 = REGEX_TALLOC (csize1 + 1, int);
5753 is_binary = REGEX_TALLOC (csize1 + 1, char);
5754 if (!string1 || !mbs_offset1 || !is_binary)
5757 FREE_VAR (mbs_offset1);
5758 FREE_VAR (is_binary);
5764 string2 = REGEX_TALLOC (csize2 + 1, CHAR_T);
5765 mbs_offset2 = REGEX_TALLOC (csize2 + 1, int);
5766 is_binary = REGEX_TALLOC (csize2 + 1, char);
5767 if (!string2 || !mbs_offset2 || !is_binary)
5770 FREE_VAR (mbs_offset1);
5772 FREE_VAR (mbs_offset2);
5773 FREE_VAR (is_binary);
5776 size2 = convert_mbs_to_wcs(string2, cstring2, csize2,
5777 mbs_offset2, is_binary);
5778 string2[size2] = L'\0'; /* for a sentinel */
5779 FREE_VAR (is_binary);
5783 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5784 pattern to (char*) in regex_compile. */
5785 p = pattern = (CHAR_T*)bufp->buffer;
5786 pend = (CHAR_T*)(bufp->buffer + bufp->used);
5790 /* Initialize subexpression text positions to -1 to mark ones that no
5791 start_memory/stop_memory has been seen for. Also initialize the
5792 register information struct. */
5793 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
5795 regstart[mcnt] = regend[mcnt]
5796 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
5798 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
5799 IS_ACTIVE (reg_info[mcnt]) = 0;
5800 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5801 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5804 /* We move `string1' into `string2' if the latter's empty -- but not if
5805 `string1' is null. */
5806 if (size2 == 0 && string1 != NULL)
5813 mbs_offset2 = mbs_offset1;
5819 end1 = string1 + size1;
5820 end2 = string2 + size2;
5822 /* Compute where to stop matching, within the two strings. */
5826 mcnt = count_mbs_length(mbs_offset1, stop);
5827 end_match_1 = string1 + mcnt;
5828 end_match_2 = string2;
5832 if (stop > csize1 + csize2)
5833 stop = csize1 + csize2;
5835 mcnt = count_mbs_length(mbs_offset2, stop-csize1);
5836 end_match_2 = string2 + mcnt;
5839 { /* count_mbs_length return error. */
5846 end_match_1 = string1 + stop;
5847 end_match_2 = string2;
5852 end_match_2 = string2 + stop - size1;
5856 /* `p' scans through the pattern as `d' scans through the data.
5857 `dend' is the end of the input string that `d' points within. `d'
5858 is advanced into the following input string whenever necessary, but
5859 this happens before fetching; therefore, at the beginning of the
5860 loop, `d' can be pointing at the end of a string, but it cannot
5863 if (size1 > 0 && pos <= csize1)
5865 mcnt = count_mbs_length(mbs_offset1, pos);
5871 mcnt = count_mbs_length(mbs_offset2, pos-csize1);
5877 { /* count_mbs_length return error. */
5882 if (size1 > 0 && pos <= size1)
5889 d = string2 + pos - size1;
5894 DEBUG_PRINT1 ("The compiled pattern is:\n");
5895 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
5896 DEBUG_PRINT1 ("The string to match is: `");
5897 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
5898 DEBUG_PRINT1 ("'\n");
5900 /* This loops over pattern commands. It exits by returning from the
5901 function if the match is complete, or it drops through if the match
5902 fails at this starting point in the input data. */
5906 DEBUG_PRINT2 ("\n%p: ", p);
5908 DEBUG_PRINT2 ("\n0x%x: ", p);
5912 { /* End of pattern means we might have succeeded. */
5913 DEBUG_PRINT1 ("end of pattern ... ");
5915 /* If we haven't matched the entire string, and we want the
5916 longest match, try backtracking. */
5917 if (d != end_match_2)
5919 /* 1 if this match ends in the same string (string1 or string2)
5920 as the best previous match. */
5921 boolean same_str_p = (FIRST_STRING_P (match_end)
5922 == MATCHING_IN_FIRST_STRING);
5923 /* 1 if this match is the best seen so far. */
5924 boolean best_match_p;
5926 /* AIX compiler got confused when this was combined
5927 with the previous declaration. */
5929 best_match_p = d > match_end;
5931 best_match_p = !MATCHING_IN_FIRST_STRING;
5933 DEBUG_PRINT1 ("backtracking.\n");
5935 if (!FAIL_STACK_EMPTY ())
5936 { /* More failure points to try. */
5938 /* If exceeds best match so far, save it. */
5939 if (!best_regs_set || best_match_p)
5941 best_regs_set = true;
5944 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
5946 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
5948 best_regstart[mcnt] = regstart[mcnt];
5949 best_regend[mcnt] = regend[mcnt];
5955 /* If no failure points, don't restore garbage. And if
5956 last match is real best match, don't restore second
5958 else if (best_regs_set && !best_match_p)
5961 /* Restore best match. It may happen that `dend ==
5962 end_match_1' while the restored d is in string2.
5963 For example, the pattern `x.*y.*z' against the
5964 strings `x-' and `y-z-', if the two strings are
5965 not consecutive in memory. */
5966 DEBUG_PRINT1 ("Restoring best registers.\n");
5969 dend = ((d >= string1 && d <= end1)
5970 ? end_match_1 : end_match_2);
5972 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
5974 regstart[mcnt] = best_regstart[mcnt];
5975 regend[mcnt] = best_regend[mcnt];
5978 } /* d != end_match_2 */
5981 DEBUG_PRINT1 ("Accepting match.\n");
5982 /* If caller wants register contents data back, do it. */
5983 if (regs && !bufp->no_sub)
5985 /* Have the register data arrays been allocated? */
5986 if (bufp->regs_allocated == REGS_UNALLOCATED)
5987 { /* No. So allocate them with malloc. We need one
5988 extra element beyond `num_regs' for the `-1' marker
5990 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
5991 regs->start = TALLOC (regs->num_regs, regoff_t);
5992 regs->end = TALLOC (regs->num_regs, regoff_t);
5993 if (regs->start == NULL || regs->end == NULL)
5998 bufp->regs_allocated = REGS_REALLOCATE;
6000 else if (bufp->regs_allocated == REGS_REALLOCATE)
6001 { /* Yes. If we need more elements than were already
6002 allocated, reallocate them. If we need fewer, just
6004 if (regs->num_regs < num_regs + 1)
6006 regs->num_regs = num_regs + 1;
6007 RETALLOC (regs->start, regs->num_regs, regoff_t);
6008 RETALLOC (regs->end, regs->num_regs, regoff_t);
6009 if (regs->start == NULL || regs->end == NULL)
6018 /* These braces fend off a "empty body in an else-statement"
6019 warning under GCC when assert expands to nothing. */
6020 assert (bufp->regs_allocated == REGS_FIXED);
6023 /* Convert the pointer data in `regstart' and `regend' to
6024 indices. Register zero has to be set differently,
6025 since we haven't kept track of any info for it. */
6026 if (regs->num_regs > 0)
6028 regs->start[0] = pos;
6030 if (MATCHING_IN_FIRST_STRING)
6031 regs->end[0] = mbs_offset1 != NULL ?
6032 mbs_offset1[d-string1] : 0;
6034 regs->end[0] = csize1 + (mbs_offset2 != NULL ?
6035 mbs_offset2[d-string2] : 0);
6037 regs->end[0] = (MATCHING_IN_FIRST_STRING
6038 ? ((regoff_t) (d - string1))
6039 : ((regoff_t) (d - string2 + size1)));
6043 /* Go through the first `min (num_regs, regs->num_regs)'
6044 registers, since that is all we initialized. */
6045 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
6048 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
6049 regs->start[mcnt] = regs->end[mcnt] = -1;
6053 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
6055 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
6059 /* If the regs structure we return has more elements than
6060 were in the pattern, set the extra elements to -1. If
6061 we (re)allocated the registers, this is the case,
6062 because we always allocate enough to have at least one
6064 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
6065 regs->start[mcnt] = regs->end[mcnt] = -1;
6066 } /* regs && !bufp->no_sub */
6068 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
6069 nfailure_points_pushed, nfailure_points_popped,
6070 nfailure_points_pushed - nfailure_points_popped);
6071 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
6074 if (MATCHING_IN_FIRST_STRING)
6075 mcnt = mbs_offset1 != NULL ? mbs_offset1[d-string1] : 0;
6077 mcnt = (mbs_offset2 != NULL ? mbs_offset2[d-string2] : 0) +
6081 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
6086 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
6092 /* Otherwise match next pattern command. */
6093 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
6095 /* Ignore these. Used to ignore the n of succeed_n's which
6096 currently have n == 0. */
6098 DEBUG_PRINT1 ("EXECUTING no_op.\n");
6102 DEBUG_PRINT1 ("EXECUTING succeed.\n");
6105 /* Match the next n pattern characters exactly. The following
6106 byte in the pattern defines n, and the n bytes after that
6107 are the characters to match. */
6113 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
6115 /* This is written out as an if-else so we don't waste time
6116 testing `translate' inside the loop. */
6125 if ((UCHAR_T) translate[(unsigned char) *d++]
6131 if (*d++ != (CHAR_T) *p++)
6135 if ((UCHAR_T) translate[(unsigned char) *d++]
6147 if (*d++ != (CHAR_T) *p++) goto fail;
6151 SET_REGS_MATCHED ();
6155 /* Match any character except possibly a newline or a null. */
6157 DEBUG_PRINT1 ("EXECUTING anychar.\n");
6161 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
6162 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
6165 SET_REGS_MATCHED ();
6166 DEBUG_PRINT2 (" Matched `%ld'.\n", (long int) *d);
6176 unsigned int i, char_class_length, coll_symbol_length,
6177 equiv_class_length, ranges_length, chars_length, length;
6178 CHAR_T *workp, *workp2, *charset_top;
6179 #define WORK_BUFFER_SIZE 128
6180 CHAR_T str_buf[WORK_BUFFER_SIZE];
6185 boolean negate = (re_opcode_t) *(p - 1) == charset_not;
6187 DEBUG_PRINT2 ("EXECUTING charset%s.\n", negate ? "_not" : "");
6189 c = TRANSLATE (*d); /* The character to match. */
6192 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
6194 charset_top = p - 1;
6195 char_class_length = *p++;
6196 coll_symbol_length = *p++;
6197 equiv_class_length = *p++;
6198 ranges_length = *p++;
6199 chars_length = *p++;
6200 /* p points charset[6], so the address of the next instruction
6201 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6202 where l=length of char_classes, m=length of collating_symbol,
6203 n=equivalence_class, o=length of char_range,
6204 p'=length of character. */
6206 /* Update p to indicate the next instruction. */
6207 p += char_class_length + coll_symbol_length+ equiv_class_length +
6208 2*ranges_length + chars_length;
6210 /* match with char_class? */
6211 for (i = 0; i < char_class_length ; i += CHAR_CLASS_SIZE)
6214 uintptr_t alignedp = ((uintptr_t)workp
6215 + __alignof__(wctype_t) - 1)
6216 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
6217 wctype = *((wctype_t*)alignedp);
6218 workp += CHAR_CLASS_SIZE;
6220 if (__iswctype((wint_t)c, wctype))
6221 goto char_set_matched;
6223 if (iswctype((wint_t)c, wctype))
6224 goto char_set_matched;
6228 /* match with collating_symbol? */
6232 const unsigned char *extra = (const unsigned char *)
6233 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
6235 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;
6239 wextra = (int32_t*)(extra + *workp++);
6240 for (i = 0; i < *wextra; ++i)
6241 if (TRANSLATE(d[i]) != wextra[1 + i])
6246 /* Update d, however d will be incremented at
6247 char_set_matched:, we decrement d here. */
6249 goto char_set_matched;
6253 else /* (nrules == 0) */
6255 /* If we can't look up collation data, we use wcscoll
6258 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;)
6260 const CHAR_T *backup_d = d, *backup_dend = dend;
6262 length = __wcslen (workp);
6264 length = wcslen (workp);
6267 /* If wcscoll(the collating symbol, whole string) > 0,
6268 any substring of the string never match with the
6269 collating symbol. */
6271 if (__wcscoll (workp, d) > 0)
6273 if (wcscoll (workp, d) > 0)
6276 workp += length + 1;
6280 /* First, we compare the collating symbol with
6281 the first character of the string.
6282 If it don't match, we add the next character to
6283 the compare buffer in turn. */
6284 for (i = 0 ; i < WORK_BUFFER_SIZE-1 ; i++, d++)
6289 if (dend == end_match_2)
6295 /* add next character to the compare buffer. */
6296 str_buf[i] = TRANSLATE(*d);
6297 str_buf[i+1] = '\0';
6300 match = __wcscoll (workp, str_buf);
6302 match = wcscoll (workp, str_buf);
6305 goto char_set_matched;
6308 /* (str_buf > workp) indicate (str_buf + X > workp),
6309 because for all X (str_buf + X > str_buf).
6310 So we don't need continue this loop. */
6313 /* Otherwise(str_buf < workp),
6314 (str_buf+next_character) may equals (workp).
6315 So we continue this loop. */
6320 workp += length + 1;
6323 /* match with equivalence_class? */
6327 const CHAR_T *backup_d = d, *backup_dend = dend;
6328 /* Try to match the equivalence class against
6329 those known to the collate implementation. */
6330 const int32_t *table;
6331 const int32_t *weights;
6332 const int32_t *extra;
6333 const int32_t *indirect;
6338 /* This #include defines a local function! */
6339 # include <locale/weightwc.h>
6341 table = (const int32_t *)
6342 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEWC);
6343 weights = (const wint_t *)
6344 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTWC);
6345 extra = (const wint_t *)
6346 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAWC);
6347 indirect = (const int32_t *)
6348 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTWC);
6350 /* Write 1 collating element to str_buf, and
6354 for (i = 0 ; idx2 == 0 && i < WORK_BUFFER_SIZE - 1; i++)
6356 cp = (wint_t*)str_buf;
6359 if (dend == end_match_2)
6364 str_buf[i] = TRANSLATE(*(d+i));
6365 str_buf[i+1] = '\0'; /* sentinel */
6366 idx2 = findidx ((const wint_t**)&cp);
6369 /* Update d, however d will be incremented at
6370 char_set_matched:, we decrement d here. */
6371 d = backup_d + ((wchar_t*)cp - (wchar_t*)str_buf - 1);
6374 if (dend == end_match_2)
6383 len = weights[idx2];
6385 for (workp2 = workp + equiv_class_length ; workp < workp2 ;
6388 idx = (int32_t)*workp;
6389 /* We already checked idx != 0 in regex_compile. */
6391 if (idx2 != 0 && len == weights[idx])
6394 while (cnt < len && (weights[idx + 1 + cnt]
6395 == weights[idx2 + 1 + cnt]))
6399 goto char_set_matched;
6406 else /* (nrules == 0) */
6408 /* If we can't look up collation data, we use wcscoll
6411 for (workp2 = workp + equiv_class_length ; workp < workp2 ;)
6413 const CHAR_T *backup_d = d, *backup_dend = dend;
6415 length = __wcslen (workp);
6417 length = wcslen (workp);
6420 /* If wcscoll(the collating symbol, whole string) > 0,
6421 any substring of the string never match with the
6422 collating symbol. */
6424 if (__wcscoll (workp, d) > 0)
6426 if (wcscoll (workp, d) > 0)
6429 workp += length + 1;
6433 /* First, we compare the equivalence class with
6434 the first character of the string.
6435 If it don't match, we add the next character to
6436 the compare buffer in turn. */
6437 for (i = 0 ; i < WORK_BUFFER_SIZE - 1 ; i++, d++)
6442 if (dend == end_match_2)
6448 /* add next character to the compare buffer. */
6449 str_buf[i] = TRANSLATE(*d);
6450 str_buf[i+1] = '\0';
6453 match = __wcscoll (workp, str_buf);
6455 match = wcscoll (workp, str_buf);
6459 goto char_set_matched;
6462 /* (str_buf > workp) indicate (str_buf + X > workp),
6463 because for all X (str_buf + X > str_buf).
6464 So we don't need continue this loop. */
6467 /* Otherwise(str_buf < workp),
6468 (str_buf+next_character) may equals (workp).
6469 So we continue this loop. */
6474 workp += length + 1;
6478 /* match with char_range? */
6482 uint32_t collseqval;
6483 const char *collseq = (const char *)
6484 _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQWC);
6486 collseqval = collseq_table_lookup (collseq, c);
6488 for (; workp < p - chars_length ;)
6490 uint32_t start_val, end_val;
6492 /* We already compute the collation sequence value
6493 of the characters (or collating symbols). */
6494 start_val = (uint32_t) *workp++; /* range_start */
6495 end_val = (uint32_t) *workp++; /* range_end */
6497 if (start_val <= collseqval && collseqval <= end_val)
6498 goto char_set_matched;
6504 /* We set range_start_char at str_buf[0], range_end_char
6505 at str_buf[4], and compared char at str_buf[2]. */
6510 for (; workp < p - chars_length ;)
6512 wchar_t *range_start_char, *range_end_char;
6514 /* match if (range_start_char <= c <= range_end_char). */
6516 /* If range_start(or end) < 0, we assume -range_start(end)
6517 is the offset of the collating symbol which is specified
6518 as the character of the range start(end). */
6522 range_start_char = charset_top - (*workp++);
6525 str_buf[0] = *workp++;
6526 range_start_char = str_buf;
6531 range_end_char = charset_top - (*workp++);
6534 str_buf[4] = *workp++;
6535 range_end_char = str_buf + 4;
6539 if (__wcscoll (range_start_char, str_buf+2) <= 0
6540 && __wcscoll (str_buf+2, range_end_char) <= 0)
6542 if (wcscoll (range_start_char, str_buf+2) <= 0
6543 && wcscoll (str_buf+2, range_end_char) <= 0)
6545 goto char_set_matched;
6549 /* match with char? */
6550 for (; workp < p ; workp++)
6552 goto char_set_matched;
6557 if (negate) goto fail;
6559 /* Cast to `unsigned' instead of `unsigned char' in case the
6560 bit list is a full 32 bytes long. */
6561 if (c < (unsigned) (*p * BYTEWIDTH)
6562 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
6567 if (!negate) goto fail;
6568 #undef WORK_BUFFER_SIZE
6570 SET_REGS_MATCHED ();
6576 /* The beginning of a group is represented by start_memory.
6577 The arguments are the register number in the next byte, and the
6578 number of groups inner to this one in the next. The text
6579 matched within the group is recorded (in the internal
6580 registers data structure) under the register number. */
6582 DEBUG_PRINT3 ("EXECUTING start_memory %ld (%ld):\n",
6583 (long int) *p, (long int) p[1]);
6585 /* Find out if this group can match the empty string. */
6586 p1 = p; /* To send to group_match_null_string_p. */
6588 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
6589 REG_MATCH_NULL_STRING_P (reg_info[*p])
6590 = PREFIX(group_match_null_string_p) (&p1, pend, reg_info);
6592 /* Save the position in the string where we were the last time
6593 we were at this open-group operator in case the group is
6594 operated upon by a repetition operator, e.g., with `(a*)*b'
6595 against `ab'; then we want to ignore where we are now in
6596 the string in case this attempt to match fails. */
6597 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6598 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
6600 DEBUG_PRINT2 (" old_regstart: %d\n",
6601 POINTER_TO_OFFSET (old_regstart[*p]));
6604 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
6606 IS_ACTIVE (reg_info[*p]) = 1;
6607 MATCHED_SOMETHING (reg_info[*p]) = 0;
6609 /* Clear this whenever we change the register activity status. */
6610 set_regs_matched_done = 0;
6612 /* This is the new highest active register. */
6613 highest_active_reg = *p;
6615 /* If nothing was active before, this is the new lowest active
6617 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
6618 lowest_active_reg = *p;
6620 /* Move past the register number and inner group count. */
6622 just_past_start_mem = p;
6627 /* The stop_memory opcode represents the end of a group. Its
6628 arguments are the same as start_memory's: the register
6629 number, and the number of inner groups. */
6631 DEBUG_PRINT3 ("EXECUTING stop_memory %ld (%ld):\n",
6632 (long int) *p, (long int) p[1]);
6634 /* We need to save the string position the last time we were at
6635 this close-group operator in case the group is operated
6636 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6637 against `aba'; then we want to ignore where we are now in
6638 the string in case this attempt to match fails. */
6639 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6640 ? REG_UNSET (regend[*p]) ? d : regend[*p]
6642 DEBUG_PRINT2 (" old_regend: %d\n",
6643 POINTER_TO_OFFSET (old_regend[*p]));
6646 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
6648 /* This register isn't active anymore. */
6649 IS_ACTIVE (reg_info[*p]) = 0;
6651 /* Clear this whenever we change the register activity status. */
6652 set_regs_matched_done = 0;
6654 /* If this was the only register active, nothing is active
6656 if (lowest_active_reg == highest_active_reg)
6658 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6659 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6662 { /* We must scan for the new highest active register, since
6663 it isn't necessarily one less than now: consider
6664 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6665 new highest active register is 1. */
6667 while (r > 0 && !IS_ACTIVE (reg_info[r]))
6670 /* If we end up at register zero, that means that we saved
6671 the registers as the result of an `on_failure_jump', not
6672 a `start_memory', and we jumped to past the innermost
6673 `stop_memory'. For example, in ((.)*) we save
6674 registers 1 and 2 as a result of the *, but when we pop
6675 back to the second ), we are at the stop_memory 1.
6676 Thus, nothing is active. */
6679 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6680 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6683 highest_active_reg = r;
6686 /* If just failed to match something this time around with a
6687 group that's operated on by a repetition operator, try to
6688 force exit from the ``loop'', and restore the register
6689 information for this group that we had before trying this
6691 if ((!MATCHED_SOMETHING (reg_info[*p])
6692 || just_past_start_mem == p - 1)
6695 boolean is_a_jump_n = false;
6699 switch ((re_opcode_t) *p1++)
6703 case pop_failure_jump:
6704 case maybe_pop_jump:
6706 case dummy_failure_jump:
6707 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6709 p1 += OFFSET_ADDRESS_SIZE;
6717 /* If the next operation is a jump backwards in the pattern
6718 to an on_failure_jump right before the start_memory
6719 corresponding to this stop_memory, exit from the loop
6720 by forcing a failure after pushing on the stack the
6721 on_failure_jump's jump in the pattern, and d. */
6722 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
6723 && (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == start_memory
6724 && p1[2+OFFSET_ADDRESS_SIZE] == *p)
6726 /* If this group ever matched anything, then restore
6727 what its registers were before trying this last
6728 failed match, e.g., with `(a*)*b' against `ab' for
6729 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6730 against `aba' for regend[3].
6732 Also restore the registers for inner groups for,
6733 e.g., `((a*)(b*))*' against `aba' (register 3 would
6734 otherwise get trashed). */
6736 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
6740 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
6742 /* Restore this and inner groups' (if any) registers. */
6743 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
6746 regstart[r] = old_regstart[r];
6748 /* xx why this test? */
6749 if (old_regend[r] >= regstart[r])
6750 regend[r] = old_regend[r];
6754 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6755 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
6761 /* Move past the register number and the inner group count. */
6766 /* \<digit> has been turned into a `duplicate' command which is
6767 followed by the numeric value of <digit> as the register number. */
6770 register const CHAR_T *d2, *dend2;
6771 int regno = *p++; /* Get which register to match against. */
6772 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
6774 /* Can't back reference a group which we've never matched. */
6775 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
6778 /* Where in input to try to start matching. */
6779 d2 = regstart[regno];
6781 /* Where to stop matching; if both the place to start and
6782 the place to stop matching are in the same string, then
6783 set to the place to stop, otherwise, for now have to use
6784 the end of the first string. */
6786 dend2 = ((FIRST_STRING_P (regstart[regno])
6787 == FIRST_STRING_P (regend[regno]))
6788 ? regend[regno] : end_match_1);
6791 /* If necessary, advance to next segment in register
6795 if (dend2 == end_match_2) break;
6796 if (dend2 == regend[regno]) break;
6798 /* End of string1 => advance to string2. */
6800 dend2 = regend[regno];
6802 /* At end of register contents => success */
6803 if (d2 == dend2) break;
6805 /* If necessary, advance to next segment in data. */
6808 /* How many characters left in this segment to match. */
6811 /* Want how many consecutive characters we can match in
6812 one shot, so, if necessary, adjust the count. */
6813 if (mcnt > dend2 - d2)
6816 /* Compare that many; failure if mismatch, else move
6819 ? PREFIX(bcmp_translate) (d, d2, mcnt, translate)
6820 : memcmp (d, d2, mcnt*sizeof(UCHAR_T)))
6822 d += mcnt, d2 += mcnt;
6824 /* Do this because we've match some characters. */
6825 SET_REGS_MATCHED ();
6831 /* begline matches the empty string at the beginning of the string
6832 (unless `not_bol' is set in `bufp'), and, if
6833 `newline_anchor' is set, after newlines. */
6835 DEBUG_PRINT1 ("EXECUTING begline.\n");
6837 if (AT_STRINGS_BEG (d))
6839 if (!bufp->not_bol) break;
6841 else if (d[-1] == '\n' && bufp->newline_anchor)
6845 /* In all other cases, we fail. */
6849 /* endline is the dual of begline. */
6851 DEBUG_PRINT1 ("EXECUTING endline.\n");
6853 if (AT_STRINGS_END (d))
6855 if (!bufp->not_eol) break;
6858 /* We have to ``prefetch'' the next character. */
6859 else if ((d == end1 ? *string2 : *d) == '\n'
6860 && bufp->newline_anchor)
6867 /* Match at the very beginning of the data. */
6869 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
6870 if (AT_STRINGS_BEG (d))
6875 /* Match at the very end of the data. */
6877 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
6878 if (AT_STRINGS_END (d))
6883 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6884 pushes NULL as the value for the string on the stack. Then
6885 `pop_failure_point' will keep the current value for the
6886 string, instead of restoring it. To see why, consider
6887 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6888 then the . fails against the \n. But the next thing we want
6889 to do is match the \n against the \n; if we restored the
6890 string value, we would be back at the foo.
6892 Because this is used only in specific cases, we don't need to
6893 check all the things that `on_failure_jump' does, to make
6894 sure the right things get saved on the stack. Hence we don't
6895 share its code. The only reason to push anything on the
6896 stack at all is that otherwise we would have to change
6897 `anychar's code to do something besides goto fail in this
6898 case; that seems worse than this. */
6899 case on_failure_keep_string_jump:
6900 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
6902 EXTRACT_NUMBER_AND_INCR (mcnt, p);
6904 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
6906 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
6909 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
6913 /* Uses of on_failure_jump:
6915 Each alternative starts with an on_failure_jump that points
6916 to the beginning of the next alternative. Each alternative
6917 except the last ends with a jump that in effect jumps past
6918 the rest of the alternatives. (They really jump to the
6919 ending jump of the following alternative, because tensioning
6920 these jumps is a hassle.)
6922 Repeats start with an on_failure_jump that points past both
6923 the repetition text and either the following jump or
6924 pop_failure_jump back to this on_failure_jump. */
6925 case on_failure_jump:
6927 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
6929 EXTRACT_NUMBER_AND_INCR (mcnt, p);
6931 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
6933 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
6936 /* If this on_failure_jump comes right before a group (i.e.,
6937 the original * applied to a group), save the information
6938 for that group and all inner ones, so that if we fail back
6939 to this point, the group's information will be correct.
6940 For example, in \(a*\)*\1, we need the preceding group,
6941 and in \(zz\(a*\)b*\)\2, we need the inner group. */
6943 /* We can't use `p' to check ahead because we push
6944 a failure point to `p + mcnt' after we do this. */
6947 /* We need to skip no_op's before we look for the
6948 start_memory in case this on_failure_jump is happening as
6949 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
6951 while (p1 < pend && (re_opcode_t) *p1 == no_op)
6954 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
6956 /* We have a new highest active register now. This will
6957 get reset at the start_memory we are about to get to,
6958 but we will have saved all the registers relevant to
6959 this repetition op, as described above. */
6960 highest_active_reg = *(p1 + 1) + *(p1 + 2);
6961 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
6962 lowest_active_reg = *(p1 + 1);
6965 DEBUG_PRINT1 (":\n");
6966 PUSH_FAILURE_POINT (p + mcnt, d, -2);
6970 /* A smart repeat ends with `maybe_pop_jump'.
6971 We change it to either `pop_failure_jump' or `jump'. */
6972 case maybe_pop_jump:
6973 EXTRACT_NUMBER_AND_INCR (mcnt, p);
6974 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
6976 register UCHAR_T *p2 = p;
6978 /* Compare the beginning of the repeat with what in the
6979 pattern follows its end. If we can establish that there
6980 is nothing that they would both match, i.e., that we
6981 would have to backtrack because of (as in, e.g., `a*a')
6982 then we can change to pop_failure_jump, because we'll
6983 never have to backtrack.
6985 This is not true in the case of alternatives: in
6986 `(a|ab)*' we do need to backtrack to the `ab' alternative
6987 (e.g., if the string was `ab'). But instead of trying to
6988 detect that here, the alternative has put on a dummy
6989 failure point which is what we will end up popping. */
6991 /* Skip over open/close-group commands.
6992 If what follows this loop is a ...+ construct,
6993 look at what begins its body, since we will have to
6994 match at least one of that. */
6998 && ((re_opcode_t) *p2 == stop_memory
6999 || (re_opcode_t) *p2 == start_memory))
7001 else if (p2 + 2 + 2 * OFFSET_ADDRESS_SIZE < pend
7002 && (re_opcode_t) *p2 == dummy_failure_jump)
7003 p2 += 2 + 2 * OFFSET_ADDRESS_SIZE;
7009 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7010 to the `maybe_finalize_jump' of this case. Examine what
7013 /* If we're at the end of the pattern, we can change. */
7016 /* Consider what happens when matching ":\(.*\)"
7017 against ":/". I don't really understand this code
7019 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7022 (" End of pattern: change to `pop_failure_jump'.\n");
7025 else if ((re_opcode_t) *p2 == exactn
7027 || (re_opcode_t) *p2 == exactn_bin
7029 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
7032 = *p2 == (UCHAR_T) endline ? '\n' : p2[2];
7034 if (((re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn
7036 || (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn_bin
7038 ) && p1[3+OFFSET_ADDRESS_SIZE] != c)
7040 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7043 DEBUG_PRINT3 (" %C != %C => pop_failure_jump.\n",
7045 (wint_t) p1[3+OFFSET_ADDRESS_SIZE]);
7047 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
7049 (char) p1[3+OFFSET_ADDRESS_SIZE]);
7054 else if ((re_opcode_t) p1[3] == charset
7055 || (re_opcode_t) p1[3] == charset_not)
7057 int negate = (re_opcode_t) p1[3] == charset_not;
7059 if (c < (unsigned) (p1[4] * BYTEWIDTH)
7060 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
7063 /* `negate' is equal to 1 if c would match, which means
7064 that we can't change to pop_failure_jump. */
7067 p[-3] = (unsigned char) pop_failure_jump;
7068 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7071 #endif /* not WCHAR */
7074 else if ((re_opcode_t) *p2 == charset)
7076 /* We win if the first character of the loop is not part
7078 if ((re_opcode_t) p1[3] == exactn
7079 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
7080 && (p2[2 + p1[5] / BYTEWIDTH]
7081 & (1 << (p1[5] % BYTEWIDTH)))))
7083 p[-3] = (unsigned char) pop_failure_jump;
7084 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7087 else if ((re_opcode_t) p1[3] == charset_not)
7090 /* We win if the charset_not inside the loop
7091 lists every character listed in the charset after. */
7092 for (idx = 0; idx < (int) p2[1]; idx++)
7093 if (! (p2[2 + idx] == 0
7094 || (idx < (int) p1[4]
7095 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
7100 p[-3] = (unsigned char) pop_failure_jump;
7101 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7104 else if ((re_opcode_t) p1[3] == charset)
7107 /* We win if the charset inside the loop
7108 has no overlap with the one after the loop. */
7110 idx < (int) p2[1] && idx < (int) p1[4];
7112 if ((p2[2 + idx] & p1[5 + idx]) != 0)
7115 if (idx == p2[1] || idx == p1[4])
7117 p[-3] = (unsigned char) pop_failure_jump;
7118 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7122 #endif /* not WCHAR */
7124 p -= OFFSET_ADDRESS_SIZE; /* Point at relative address again. */
7125 if ((re_opcode_t) p[-1] != pop_failure_jump)
7127 p[-1] = (UCHAR_T) jump;
7128 DEBUG_PRINT1 (" Match => jump.\n");
7129 goto unconditional_jump;
7131 /* Note fall through. */
7134 /* The end of a simple repeat has a pop_failure_jump back to
7135 its matching on_failure_jump, where the latter will push a
7136 failure point. The pop_failure_jump takes off failure
7137 points put on by this pop_failure_jump's matching
7138 on_failure_jump; we got through the pattern to here from the
7139 matching on_failure_jump, so didn't fail. */
7140 case pop_failure_jump:
7142 /* We need to pass separate storage for the lowest and
7143 highest registers, even though we don't care about the
7144 actual values. Otherwise, we will restore only one
7145 register from the stack, since lowest will == highest in
7146 `pop_failure_point'. */
7147 active_reg_t dummy_low_reg, dummy_high_reg;
7148 UCHAR_T *pdummy = NULL;
7149 const CHAR_T *sdummy = NULL;
7151 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
7152 POP_FAILURE_POINT (sdummy, pdummy,
7153 dummy_low_reg, dummy_high_reg,
7154 reg_dummy, reg_dummy, reg_info_dummy);
7156 /* Note fall through. */
7160 DEBUG_PRINT2 ("\n%p: ", p);
7162 DEBUG_PRINT2 ("\n0x%x: ", p);
7164 /* Note fall through. */
7166 /* Unconditionally jump (without popping any failure points). */
7168 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
7169 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
7170 p += mcnt; /* Do the jump. */
7172 DEBUG_PRINT2 ("(to %p).\n", p);
7174 DEBUG_PRINT2 ("(to 0x%x).\n", p);
7179 /* We need this opcode so we can detect where alternatives end
7180 in `group_match_null_string_p' et al. */
7182 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
7183 goto unconditional_jump;
7186 /* Normally, the on_failure_jump pushes a failure point, which
7187 then gets popped at pop_failure_jump. We will end up at
7188 pop_failure_jump, also, and with a pattern of, say, `a+', we
7189 are skipping over the on_failure_jump, so we have to push
7190 something meaningless for pop_failure_jump to pop. */
7191 case dummy_failure_jump:
7192 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
7193 /* It doesn't matter what we push for the string here. What
7194 the code at `fail' tests is the value for the pattern. */
7195 PUSH_FAILURE_POINT (NULL, NULL, -2);
7196 goto unconditional_jump;
7199 /* At the end of an alternative, we need to push a dummy failure
7200 point in case we are followed by a `pop_failure_jump', because
7201 we don't want the failure point for the alternative to be
7202 popped. For example, matching `(a|ab)*' against `aab'
7203 requires that we match the `ab' alternative. */
7204 case push_dummy_failure:
7205 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
7206 /* See comments just above at `dummy_failure_jump' about the
7208 PUSH_FAILURE_POINT (NULL, NULL, -2);
7211 /* Have to succeed matching what follows at least n times.
7212 After that, handle like `on_failure_jump'. */
7214 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7215 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
7218 /* Originally, this is how many times we HAVE to succeed. */
7222 p += OFFSET_ADDRESS_SIZE;
7223 STORE_NUMBER_AND_INCR (p, mcnt);
7225 DEBUG_PRINT3 (" Setting %p to %d.\n", p - OFFSET_ADDRESS_SIZE
7228 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - OFFSET_ADDRESS_SIZE
7235 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n",
7236 p + OFFSET_ADDRESS_SIZE);
7238 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n",
7239 p + OFFSET_ADDRESS_SIZE);
7243 p[1] = (UCHAR_T) no_op;
7245 p[2] = (UCHAR_T) no_op;
7246 p[3] = (UCHAR_T) no_op;
7253 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7254 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
7256 /* Originally, this is how many times we CAN jump. */
7260 STORE_NUMBER (p + OFFSET_ADDRESS_SIZE, mcnt);
7263 DEBUG_PRINT3 (" Setting %p to %d.\n", p + OFFSET_ADDRESS_SIZE,
7266 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + OFFSET_ADDRESS_SIZE,
7269 goto unconditional_jump;
7271 /* If don't have to jump any more, skip over the rest of command. */
7273 p += 2 * OFFSET_ADDRESS_SIZE;
7278 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
7280 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7282 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7284 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
7286 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
7288 STORE_NUMBER (p1, mcnt);
7293 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7294 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7295 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7296 macro and introducing temporary variables works around the bug. */
7299 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7300 if (AT_WORD_BOUNDARY (d))
7305 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7306 if (AT_WORD_BOUNDARY (d))
7312 boolean prevchar, thischar;
7314 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7315 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7318 prevchar = WORDCHAR_P (d - 1);
7319 thischar = WORDCHAR_P (d);
7320 if (prevchar != thischar)
7327 boolean prevchar, thischar;
7329 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7330 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7333 prevchar = WORDCHAR_P (d - 1);
7334 thischar = WORDCHAR_P (d);
7335 if (prevchar != thischar)
7342 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
7343 if (!AT_STRINGS_END (d) && WORDCHAR_P (d)
7344 && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
7349 DEBUG_PRINT1 ("EXECUTING wordend.\n");
7350 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
7351 && (AT_STRINGS_END (d) || !WORDCHAR_P (d)))
7357 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
7358 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
7363 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
7364 if (PTR_CHAR_POS ((unsigned char *) d) != point)
7369 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
7370 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
7375 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
7380 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
7384 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7386 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
7388 SET_REGS_MATCHED ();
7392 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
7394 goto matchnotsyntax;
7397 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
7401 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7403 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
7405 SET_REGS_MATCHED ();
7408 #else /* not emacs */
7410 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
7412 if (!WORDCHAR_P (d))
7414 SET_REGS_MATCHED ();
7419 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
7423 SET_REGS_MATCHED ();
7426 #endif /* not emacs */
7431 continue; /* Successfully executed one pattern command; keep going. */
7434 /* We goto here if a matching operation fails. */
7436 if (!FAIL_STACK_EMPTY ())
7437 { /* A restart point is known. Restore to that state. */
7438 DEBUG_PRINT1 ("\nFAIL:\n");
7439 POP_FAILURE_POINT (d, p,
7440 lowest_active_reg, highest_active_reg,
7441 regstart, regend, reg_info);
7443 /* If this failure point is a dummy, try the next one. */
7447 /* If we failed to the end of the pattern, don't examine *p. */
7451 boolean is_a_jump_n = false;
7453 /* If failed to a backwards jump that's part of a repetition
7454 loop, need to pop this failure point and use the next one. */
7455 switch ((re_opcode_t) *p)
7459 case maybe_pop_jump:
7460 case pop_failure_jump:
7463 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7466 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
7468 && (re_opcode_t) *p1 == on_failure_jump))
7476 if (d >= string1 && d <= end1)
7480 break; /* Matching at this starting point really fails. */
7484 goto restore_best_regs;
7488 return -1; /* Failure to match. */
7491 /* Subroutine definitions for re_match_2. */
7494 /* We are passed P pointing to a register number after a start_memory.
7496 Return true if the pattern up to the corresponding stop_memory can
7497 match the empty string, and false otherwise.
7499 If we find the matching stop_memory, sets P to point to one past its number.
7500 Otherwise, sets P to an undefined byte less than or equal to END.
7502 We don't handle duplicates properly (yet). */
7505 PREFIX(group_match_null_string_p) (UCHAR_T **p, UCHAR_T *end,
7506 PREFIX(register_info_type) *reg_info)
7509 /* Point to after the args to the start_memory. */
7510 UCHAR_T *p1 = *p + 2;
7514 /* Skip over opcodes that can match nothing, and return true or
7515 false, as appropriate, when we get to one that can't, or to the
7516 matching stop_memory. */
7518 switch ((re_opcode_t) *p1)
7520 /* Could be either a loop or a series of alternatives. */
7521 case on_failure_jump:
7523 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7525 /* If the next operation is not a jump backwards in the
7530 /* Go through the on_failure_jumps of the alternatives,
7531 seeing if any of the alternatives cannot match nothing.
7532 The last alternative starts with only a jump,
7533 whereas the rest start with on_failure_jump and end
7534 with a jump, e.g., here is the pattern for `a|b|c':
7536 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7537 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7540 So, we have to first go through the first (n-1)
7541 alternatives and then deal with the last one separately. */
7544 /* Deal with the first (n-1) alternatives, which start
7545 with an on_failure_jump (see above) that jumps to right
7546 past a jump_past_alt. */
7548 while ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] ==
7551 /* `mcnt' holds how many bytes long the alternative
7552 is, including the ending `jump_past_alt' and
7555 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt -
7556 (1 + OFFSET_ADDRESS_SIZE),
7560 /* Move to right after this alternative, including the
7564 /* Break if it's the beginning of an n-th alternative
7565 that doesn't begin with an on_failure_jump. */
7566 if ((re_opcode_t) *p1 != on_failure_jump)
7569 /* Still have to check that it's not an n-th
7570 alternative that starts with an on_failure_jump. */
7572 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7573 if ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] !=
7576 /* Get to the beginning of the n-th alternative. */
7577 p1 -= 1 + OFFSET_ADDRESS_SIZE;
7582 /* Deal with the last alternative: go back and get number
7583 of the `jump_past_alt' just before it. `mcnt' contains
7584 the length of the alternative. */
7585 EXTRACT_NUMBER (mcnt, p1 - OFFSET_ADDRESS_SIZE);
7587 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt, reg_info))
7590 p1 += mcnt; /* Get past the n-th alternative. */
7596 assert (p1[1] == **p);
7602 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7605 } /* while p1 < end */
7608 } /* group_match_null_string_p */
7611 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7612 It expects P to be the first byte of a single alternative and END one
7613 byte past the last. The alternative can contain groups. */
7616 PREFIX(alt_match_null_string_p) (UCHAR_T *p, UCHAR_T *end,
7617 PREFIX(register_info_type) *reg_info)
7624 /* Skip over opcodes that can match nothing, and break when we get
7625 to one that can't. */
7627 switch ((re_opcode_t) *p1)
7630 case on_failure_jump:
7632 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7637 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7640 } /* while p1 < end */
7643 } /* alt_match_null_string_p */
7646 /* Deals with the ops common to group_match_null_string_p and
7647 alt_match_null_string_p.
7649 Sets P to one after the op and its arguments, if any. */
7652 PREFIX(common_op_match_null_string_p) (UCHAR_T **p, UCHAR_T *end,
7653 PREFIX(register_info_type) *reg_info)
7660 switch ((re_opcode_t) *p1++)
7680 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
7681 ret = PREFIX(group_match_null_string_p) (&p1, end, reg_info);
7683 /* Have to set this here in case we're checking a group which
7684 contains a group and a back reference to it. */
7686 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
7687 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
7693 /* If this is an optimized succeed_n for zero times, make the jump. */
7695 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7703 /* Get to the number of times to succeed. */
7704 p1 += OFFSET_ADDRESS_SIZE;
7705 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7709 p1 -= 2 * OFFSET_ADDRESS_SIZE;
7710 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7718 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
7723 p1 += 2 * OFFSET_ADDRESS_SIZE;
7726 /* All other opcodes mean we cannot match the empty string. */
7732 } /* common_op_match_null_string_p */
7735 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7736 bytes; nonzero otherwise. */
7739 PREFIX(bcmp_translate) (const CHAR_T *s1, const CHAR_T *s2, register int len,
7740 RE_TRANSLATE_TYPE translate)
7742 register const UCHAR_T *p1 = (const UCHAR_T *) s1;
7743 register const UCHAR_T *p2 = (const UCHAR_T *) s2;
7747 if (((*p1<=0xff)?translate[*p1++]:*p1++)
7748 != ((*p2<=0xff)?translate[*p2++]:*p2++))
7751 if (translate[*p1++] != translate[*p2++]) return 1;
7759 #else /* not INSIDE_RECURSION */
7761 /* Entry points for GNU code. */
7763 /* re_compile_pattern is the GNU regular expression compiler: it
7764 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7765 Returns 0 if the pattern was valid, otherwise an error string.
7767 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7768 are set in BUFP on entry.
7770 We call regex_compile to do the actual compilation. */
7773 re_compile_pattern (const char *pattern, size_t length,
7774 struct re_pattern_buffer *bufp)
7778 /* GNU code is written to assume at least RE_NREGS registers will be set
7779 (and at least one extra will be -1). */
7780 bufp->regs_allocated = REGS_UNALLOCATED;
7782 /* And GNU code determines whether or not to get register information
7783 by passing null for the REGS argument to re_match, etc., not by
7787 /* Match anchors at newline. */
7788 bufp->newline_anchor = 1;
7791 if (MB_CUR_MAX != 1)
7792 ret = wcs_regex_compile (pattern, length, re_syntax_options, bufp);
7795 ret = byte_regex_compile (pattern, length, re_syntax_options, bufp);
7799 return gettext (re_error_msgid[(int) ret]);
7802 weak_alias (__re_compile_pattern, re_compile_pattern)
7805 /* Entry points compatible with 4.2 BSD regex library. We don't define
7806 them unless specifically requested. */
7808 #if defined _REGEX_RE_COMP || defined _LIBC
7810 /* BSD has one and only one pattern buffer. */
7811 static struct re_pattern_buffer re_comp_buf;
7815 /* Make these definitions weak in libc, so POSIX programs can redefine
7816 these names if they don't use our functions, and still use
7817 regcomp/regexec below without link errors. */
7820 re_comp (const char *s)
7826 if (!re_comp_buf.buffer)
7827 return (char *) gettext ("No previous regular expression");
7831 if (!re_comp_buf.buffer)
7833 re_comp_buf.buffer = (unsigned char *) malloc (200);
7834 if (re_comp_buf.buffer == NULL)
7835 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
7836 re_comp_buf.allocated = 200;
7838 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
7839 if (re_comp_buf.fastmap == NULL)
7840 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
7843 /* Since `re_exec' always passes NULL for the `regs' argument, we
7844 don't need to initialize the pattern buffer fields which affect it. */
7846 /* Match anchors at newlines. */
7847 re_comp_buf.newline_anchor = 1;
7850 if (MB_CUR_MAX != 1)
7851 ret = wcs_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7854 ret = byte_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7859 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7860 return (char *) gettext (re_error_msgid[(int) ret]);
7868 re_exec (const char *s)
7870 const int len = strlen (s);
7872 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
7875 #endif /* _REGEX_RE_COMP */
7877 /* POSIX.2 functions. Don't define these for Emacs. */
7881 /* regcomp takes a regular expression as a string and compiles it.
7883 PREG is a regex_t *. We do not expect any fields to be initialized,
7884 since POSIX says we shouldn't. Thus, we set
7886 `buffer' to the compiled pattern;
7887 `used' to the length of the compiled pattern;
7888 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7889 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7890 RE_SYNTAX_POSIX_BASIC;
7891 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7892 `fastmap' to an allocated space for the fastmap;
7893 `fastmap_accurate' to zero;
7894 `re_nsub' to the number of subexpressions in PATTERN.
7896 PATTERN is the address of the pattern string.
7898 CFLAGS is a series of bits which affect compilation.
7900 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7901 use POSIX basic syntax.
7903 If REG_NEWLINE is set, then . and [^...] don't match newline.
7904 Also, regexec will try a match beginning after every newline.
7906 If REG_ICASE is set, then we considers upper- and lowercase
7907 versions of letters to be equivalent when matching.
7909 If REG_NOSUB is set, then when PREG is passed to regexec, that
7910 routine will report only success or failure, and nothing about the
7913 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
7914 the return codes and their meanings.) */
7917 regcomp (regex_t *preg, const char *pattern, int cflags)
7921 = (cflags & REG_EXTENDED) ?
7922 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
7924 /* regex_compile will allocate the space for the compiled pattern. */
7926 preg->allocated = 0;
7929 /* Try to allocate space for the fastmap. */
7930 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
7932 if (cflags & REG_ICASE)
7937 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
7938 * sizeof (*(RE_TRANSLATE_TYPE)0));
7939 if (preg->translate == NULL)
7940 return (int) REG_ESPACE;
7942 /* Map uppercase characters to corresponding lowercase ones. */
7943 for (i = 0; i < CHAR_SET_SIZE; i++)
7944 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
7947 preg->translate = NULL;
7949 /* If REG_NEWLINE is set, newlines are treated differently. */
7950 if (cflags & REG_NEWLINE)
7951 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
7952 syntax &= ~RE_DOT_NEWLINE;
7953 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
7954 /* It also changes the matching behavior. */
7955 preg->newline_anchor = 1;
7958 preg->newline_anchor = 0;
7960 preg->no_sub = !!(cflags & REG_NOSUB);
7962 /* POSIX says a null character in the pattern terminates it, so we
7963 can use strlen here in compiling the pattern. */
7965 if (MB_CUR_MAX != 1)
7966 ret = wcs_regex_compile (pattern, strlen (pattern), syntax, preg);
7969 ret = byte_regex_compile (pattern, strlen (pattern), syntax, preg);
7971 /* POSIX doesn't distinguish between an unmatched open-group and an
7972 unmatched close-group: both are REG_EPAREN. */
7973 if (ret == REG_ERPAREN) ret = REG_EPAREN;
7975 if (ret == REG_NOERROR && preg->fastmap)
7977 /* Compute the fastmap now, since regexec cannot modify the pattern
7979 if (re_compile_fastmap (preg) == -2)
7981 /* Some error occurred while computing the fastmap, just forget
7983 free (preg->fastmap);
7984 preg->fastmap = NULL;
7991 weak_alias (__regcomp, regcomp)
7995 /* regexec searches for a given pattern, specified by PREG, in the
7998 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
7999 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8000 least NMATCH elements, and we set them to the offsets of the
8001 corresponding matched substrings.
8003 EFLAGS specifies `execution flags' which affect matching: if
8004 REG_NOTBOL is set, then ^ does not match at the beginning of the
8005 string; if REG_NOTEOL is set, then $ does not match at the end.
8007 We return 0 if we find a match and REG_NOMATCH if not. */
8010 regexec (const regex_t *preg, const char *string, size_t nmatch,
8011 regmatch_t pmatch[], int eflags)
8014 struct re_registers regs;
8015 regex_t private_preg;
8016 int len = strlen (string);
8017 boolean want_reg_info = !preg->no_sub && nmatch > 0;
8019 private_preg = *preg;
8021 private_preg.not_bol = !!(eflags & REG_NOTBOL);
8022 private_preg.not_eol = !!(eflags & REG_NOTEOL);
8024 /* The user has told us exactly how many registers to return
8025 information about, via `nmatch'. We have to pass that on to the
8026 matching routines. */
8027 private_preg.regs_allocated = REGS_FIXED;
8031 regs.num_regs = nmatch;
8032 regs.start = TALLOC (nmatch * 2, regoff_t);
8033 if (regs.start == NULL)
8034 return (int) REG_NOMATCH;
8035 regs.end = regs.start + nmatch;
8038 /* Perform the searching operation. */
8039 ret = re_search (&private_preg, string, len,
8040 /* start: */ 0, /* range: */ len,
8041 want_reg_info ? ®s : (struct re_registers *) 0);
8043 /* Copy the register information to the POSIX structure. */
8050 for (r = 0; r < nmatch; r++)
8052 pmatch[r].rm_so = regs.start[r];
8053 pmatch[r].rm_eo = regs.end[r];
8057 /* If we needed the temporary register info, free the space now. */
8061 /* We want zero return to mean success, unlike `re_search'. */
8062 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
8065 /* EGLIBC: This is handled in regexec-compat.c. */
8066 /*weak_alias (__regexec, regexec)*/
8067 #include "regexec-compat.c"
8071 /* Returns a message corresponding to an error code, ERRCODE, returned
8072 from either regcomp or regexec. We don't use PREG here. */
8075 regerror (int errcode, const regex_t *preg __attribute__ ((unused)),
8076 char *errbuf, size_t errbuf_size)
8082 || errcode >= (int) (sizeof (re_error_msgid)
8083 / sizeof (re_error_msgid[0])))
8084 /* Only error codes returned by the rest of the code should be passed
8085 to this routine. If we are given anything else, or if other regex
8086 code generates an invalid error code, then the program has a bug.
8087 Dump core so we can fix it. */
8090 msg = gettext (re_error_msgid[errcode]);
8092 msg_size = strlen (msg) + 1; /* Includes the null. */
8094 if (errbuf_size != 0)
8096 if (msg_size > errbuf_size)
8098 #if defined HAVE_MEMPCPY || defined _LIBC
8099 *((char *) mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
8101 memcpy (errbuf, msg, errbuf_size - 1);
8102 errbuf[errbuf_size - 1] = 0;
8106 memcpy (errbuf, msg, msg_size);
8112 weak_alias (__regerror, regerror)
8116 /* Free dynamically allocated space used by PREG. */
8119 regfree (regex_t *preg)
8121 if (preg->buffer != NULL)
8122 free (preg->buffer);
8123 preg->buffer = NULL;
8125 preg->allocated = 0;
8128 if (preg->fastmap != NULL)
8129 free (preg->fastmap);
8130 preg->fastmap = NULL;
8131 preg->fastmap_accurate = 0;
8133 if (preg->translate != NULL)
8134 free (preg->translate);
8135 preg->translate = NULL;
8138 weak_alias (__regfree, regfree)
8141 #endif /* not emacs */
8143 #endif /* not INSIDE_RECURSION */
8147 #undef STORE_NUMBER_AND_INCR
8148 #undef EXTRACT_NUMBER
8149 #undef EXTRACT_NUMBER_AND_INCR
8151 #undef DEBUG_PRINT_COMPILED_PATTERN
8152 #undef DEBUG_PRINT_DOUBLE_STRING
8154 #undef INIT_FAIL_STACK
8155 #undef RESET_FAIL_STACK
8156 #undef DOUBLE_FAIL_STACK
8157 #undef PUSH_PATTERN_OP
8158 #undef PUSH_FAILURE_POINTER
8159 #undef PUSH_FAILURE_INT
8160 #undef PUSH_FAILURE_ELT
8161 #undef POP_FAILURE_POINTER
8162 #undef POP_FAILURE_INT
8163 #undef POP_FAILURE_ELT
8166 #undef PUSH_FAILURE_POINT
8167 #undef POP_FAILURE_POINT
8169 #undef REG_UNSET_VALUE
8177 #undef INIT_BUF_SIZE
8178 #undef GET_BUFFER_SPACE
8186 #undef EXTEND_BUFFER
8187 #undef GET_UNSIGNED_NUMBER
8188 #undef FREE_STACK_RETURN
8190 # undef POINTER_TO_OFFSET
8191 # undef MATCHING_IN_FRST_STRING
8193 # undef AT_STRINGS_BEG
8194 # undef AT_STRINGS_END
8197 # undef FREE_VARIABLES
8198 # undef NO_HIGHEST_ACTIVE_REG
8199 # undef NO_LOWEST_ACTIVE_REG
8203 # undef COMPILED_BUFFER_VAR
8204 # undef OFFSET_ADDRESS_SIZE
8205 # undef CHAR_CLASS_SIZE
8212 # define DEFINED_ONCE
~ianmdlvl / eglibc.git / blob