1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993-1999, 2000, 2001, 2002 Free Software Foundation, Inc.
6 This file is part of the GNU C Library.
8 The GNU C Library is free software; you can redistribute it and/or
9 modify it under the terms of the GNU Lesser General Public
10 License as published by the Free Software Foundation; either
11 version 2.1 of the License, or (at your option) any later version.
13 The GNU C Library is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 Lesser General Public License for more details.
18 You should have received a copy of the GNU Lesser General Public
19 License along with the GNU C Library; if not, write to the Free
20 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
23 /* This file has been modified for usage in libiberty. It includes "xregex.h"
24 instead of <regex.h>. The "xregex.h" header file renames all external
25 routines with an "x" prefix so they do not collide with the native regex
26 routines or with other components regex routines. */
27 /* AIX requires this to be the first thing in the file. */
28 #if defined _AIX && !defined __GNUC__ && !defined REGEX_MALLOC
40 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
41 # define PARAMS(args) args
43 # define PARAMS(args) ()
45 #endif /* Not PARAMS. */
47 #ifndef INSIDE_RECURSION
49 # if defined STDC_HEADERS && !defined emacs
52 /* We need this for `regex.h', and perhaps for the Emacs include files. */
53 # include <sys/types.h>
56 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
58 /* For platform which support the ISO C amendement 1 functionality we
59 support user defined character classes. */
60 # if defined _LIBC || WIDE_CHAR_SUPPORT
61 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
67 /* We have to keep the namespace clean. */
68 # define regfree(preg) __regfree (preg)
69 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
70 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
71 # define regerror(errcode, preg, errbuf, errbuf_size) \
72 __regerror(errcode, preg, errbuf, errbuf_size)
73 # define re_set_registers(bu, re, nu, st, en) \
74 __re_set_registers (bu, re, nu, st, en)
75 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
76 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
77 # define re_match(bufp, string, size, pos, regs) \
78 __re_match (bufp, string, size, pos, regs)
79 # define re_search(bufp, string, size, startpos, range, regs) \
80 __re_search (bufp, string, size, startpos, range, regs)
81 # define re_compile_pattern(pattern, length, bufp) \
82 __re_compile_pattern (pattern, length, bufp)
83 # define re_set_syntax(syntax) __re_set_syntax (syntax)
84 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
85 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
86 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
88 # define btowc __btowc
90 /* We are also using some library internals. */
91 # include <locale/localeinfo.h>
92 # include <locale/elem-hash.h>
93 # include <langinfo.h>
94 # include <locale/coll-lookup.h>
97 /* This is for other GNU distributions with internationalized messages. */
98 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
102 # define gettext(msgid) __dcgettext ("libc", msgid, LC_MESSAGES)
105 # define gettext(msgid) (msgid)
108 # ifndef gettext_noop
109 /* This define is so xgettext can find the internationalizable
111 # define gettext_noop(String) String
114 /* The `emacs' switch turns on certain matching commands
115 that make sense only in Emacs. */
122 # else /* not emacs */
124 /* If we are not linking with Emacs proper,
125 we can't use the relocating allocator
126 even if config.h says that we can. */
129 # if defined STDC_HEADERS || defined _LIBC
136 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
137 If nothing else has been done, use the method below. */
138 # ifdef INHIBIT_STRING_HEADER
139 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
140 # if !defined bzero && !defined bcopy
141 # undef INHIBIT_STRING_HEADER
146 /* This is the normal way of making sure we have a bcopy and a bzero.
147 This is used in most programs--a few other programs avoid this
148 by defining INHIBIT_STRING_HEADER. */
149 # ifndef INHIBIT_STRING_HEADER
150 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
154 # define bzero(s, n) (memset (s, '\0', n), (s))
156 # define bzero(s, n) __bzero (s, n)
160 # include <strings.h>
162 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
165 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
170 /* Define the syntax stuff for \<, \>, etc. */
172 /* This must be nonzero for the wordchar and notwordchar pattern
173 commands in re_match_2. */
178 # ifdef SWITCH_ENUM_BUG
179 # define SWITCH_ENUM_CAST(x) ((int)(x))
181 # define SWITCH_ENUM_CAST(x) (x)
184 # endif /* not emacs */
186 # if defined _LIBC || HAVE_LIMITS_H
191 # define MB_LEN_MAX 1
194 /* Get the interface, including the syntax bits. */
195 # include "xregex.h" /* change for libiberty */
197 /* isalpha etc. are used for the character classes. */
200 /* Jim Meyering writes:
202 "... Some ctype macros are valid only for character codes that
203 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
204 using /bin/cc or gcc but without giving an ansi option). So, all
205 ctype uses should be through macros like ISPRINT... If
206 STDC_HEADERS is defined, then autoconf has verified that the ctype
207 macros don't need to be guarded with references to isascii. ...
208 Defining isascii to 1 should let any compiler worth its salt
209 eliminate the && through constant folding."
210 Solaris defines some of these symbols so we must undefine them first. */
213 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
214 # define ISASCII(c) 1
216 # define ISASCII(c) isascii(c)
220 # define ISBLANK(c) (ISASCII (c) && isblank (c))
222 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
225 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
227 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
231 # define ISPRINT(c) (ISASCII (c) && isprint (c))
232 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
233 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
234 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
235 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
236 # define ISLOWER(c) (ISASCII (c) && islower (c))
237 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
238 # define ISSPACE(c) (ISASCII (c) && isspace (c))
239 # define ISUPPER(c) (ISASCII (c) && isupper (c))
240 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
243 # define TOLOWER(c) _tolower(c)
245 # define TOLOWER(c) tolower(c)
249 # define NULL (void *)0
252 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
253 since ours (we hope) works properly with all combinations of
254 machines, compilers, `char' and `unsigned char' argument types.
255 (Per Bothner suggested the basic approach.) */
256 # undef SIGN_EXTEND_CHAR
258 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
259 # else /* not __STDC__ */
260 /* As in Harbison and Steele. */
261 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
265 /* How many characters in the character set. */
266 # define CHAR_SET_SIZE 256
270 extern char *re_syntax_table;
272 # else /* not SYNTAX_TABLE */
274 static char re_syntax_table[CHAR_SET_SIZE];
276 static void init_syntax_once PARAMS ((void));
286 bzero (re_syntax_table, sizeof re_syntax_table);
288 for (c = 0; c < CHAR_SET_SIZE; ++c)
290 re_syntax_table[c] = Sword;
292 re_syntax_table['_'] = Sword;
297 # endif /* not SYNTAX_TABLE */
299 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
303 /* Integer type for pointers. */
304 # if !defined _LIBC && !defined HAVE_UINTPTR_T
305 typedef unsigned long int uintptr_t;
308 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
309 use `alloca' instead of `malloc'. This is because using malloc in
310 re_search* or re_match* could cause memory leaks when C-g is used in
311 Emacs; also, malloc is slower and causes storage fragmentation. On
312 the other hand, malloc is more portable, and easier to debug.
314 Because we sometimes use alloca, some routines have to be macros,
315 not functions -- `alloca'-allocated space disappears at the end of the
316 function it is called in. */
320 # define REGEX_ALLOCATE malloc
321 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
322 # define REGEX_FREE free
324 # else /* not REGEX_MALLOC */
326 /* Emacs already defines alloca, sometimes. */
329 /* Make alloca work the best possible way. */
331 # define alloca __builtin_alloca
332 # else /* not __GNUC__ */
335 # endif /* HAVE_ALLOCA_H */
336 # endif /* not __GNUC__ */
338 # endif /* not alloca */
340 # define REGEX_ALLOCATE alloca
342 /* Assumes a `char *destination' variable. */
343 # define REGEX_REALLOCATE(source, osize, nsize) \
344 (destination = (char *) alloca (nsize), \
345 memcpy (destination, source, osize))
347 /* No need to do anything to free, after alloca. */
348 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
350 # endif /* not REGEX_MALLOC */
352 /* Define how to allocate the failure stack. */
354 # if defined REL_ALLOC && defined REGEX_MALLOC
356 # define REGEX_ALLOCATE_STACK(size) \
357 r_alloc (&failure_stack_ptr, (size))
358 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
359 r_re_alloc (&failure_stack_ptr, (nsize))
360 # define REGEX_FREE_STACK(ptr) \
361 r_alloc_free (&failure_stack_ptr)
363 # else /* not using relocating allocator */
367 # define REGEX_ALLOCATE_STACK malloc
368 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
369 # define REGEX_FREE_STACK free
371 # else /* not REGEX_MALLOC */
373 # define REGEX_ALLOCATE_STACK alloca
375 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
376 REGEX_REALLOCATE (source, osize, nsize)
377 /* No need to explicitly free anything. */
378 # define REGEX_FREE_STACK(arg)
380 # endif /* not REGEX_MALLOC */
381 # endif /* not using relocating allocator */
384 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
385 `string1' or just past its end. This works if PTR is NULL, which is
387 # define FIRST_STRING_P(ptr) \
388 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
390 /* (Re)Allocate N items of type T using malloc, or fail. */
391 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
392 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
393 # define RETALLOC_IF(addr, n, t) \
394 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
395 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
397 # define BYTEWIDTH 8 /* In bits. */
399 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
403 # define MAX(a, b) ((a) > (b) ? (a) : (b))
404 # define MIN(a, b) ((a) < (b) ? (a) : (b))
406 typedef char boolean;
410 static reg_errcode_t byte_regex_compile _RE_ARGS ((const char *pattern, size_t size,
412 struct re_pattern_buffer *bufp));
414 static int byte_re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
415 const char *string1, int size1,
416 const char *string2, int size2,
418 struct re_registers *regs,
420 static int byte_re_search_2 PARAMS ((struct re_pattern_buffer *bufp,
421 const char *string1, int size1,
422 const char *string2, int size2,
423 int startpos, int range,
424 struct re_registers *regs, int stop));
425 static int byte_re_compile_fastmap PARAMS ((struct re_pattern_buffer *bufp));
428 static reg_errcode_t wcs_regex_compile _RE_ARGS ((const char *pattern, size_t size,
430 struct re_pattern_buffer *bufp));
433 static int wcs_re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
434 const char *cstring1, int csize1,
435 const char *cstring2, int csize2,
437 struct re_registers *regs,
439 wchar_t *string1, int size1,
440 wchar_t *string2, int size2,
441 int *mbs_offset1, int *mbs_offset2));
442 static int wcs_re_search_2 PARAMS ((struct re_pattern_buffer *bufp,
443 const char *string1, int size1,
444 const char *string2, int size2,
445 int startpos, int range,
446 struct re_registers *regs, int stop));
447 static int wcs_re_compile_fastmap PARAMS ((struct re_pattern_buffer *bufp));
450 /* These are the command codes that appear in compiled regular
451 expressions. Some opcodes are followed by argument bytes. A
452 command code can specify any interpretation whatsoever for its
453 arguments. Zero bytes may appear in the compiled regular expression. */
459 /* Succeed right away--no more backtracking. */
462 /* Followed by one byte giving n, then by n literal bytes. */
466 /* Same as exactn, but contains binary data. */
470 /* Matches any (more or less) character. */
473 /* Matches any one char belonging to specified set. First
474 following byte is number of bitmap bytes. Then come bytes
475 for a bitmap saying which chars are in. Bits in each byte
476 are ordered low-bit-first. A character is in the set if its
477 bit is 1. A character too large to have a bit in the map is
478 automatically not in the set. */
479 /* ifdef MBS_SUPPORT, following element is length of character
480 classes, length of collating symbols, length of equivalence
481 classes, length of character ranges, and length of characters.
482 Next, character class element, collating symbols elements,
483 equivalence class elements, range elements, and character
485 See regex_compile function. */
488 /* Same parameters as charset, but match any character that is
489 not one of those specified. */
492 /* Start remembering the text that is matched, for storing in a
493 register. Followed by one byte with the register number, in
494 the range 0 to one less than the pattern buffer's re_nsub
495 field. Then followed by one byte with the number of groups
496 inner to this one. (This last has to be part of the
497 start_memory only because we need it in the on_failure_jump
501 /* Stop remembering the text that is matched and store it in a
502 memory register. Followed by one byte with the register
503 number, in the range 0 to one less than `re_nsub' in the
504 pattern buffer, and one byte with the number of inner groups,
505 just like `start_memory'. (We need the number of inner
506 groups here because we don't have any easy way of finding the
507 corresponding start_memory when we're at a stop_memory.) */
510 /* Match a duplicate of something remembered. Followed by one
511 byte containing the register number. */
514 /* Fail unless at beginning of line. */
517 /* Fail unless at end of line. */
520 /* Succeeds if at beginning of buffer (if emacs) or at beginning
521 of string to be matched (if not). */
524 /* Analogously, for end of buffer/string. */
527 /* Followed by two byte relative address to which to jump. */
530 /* Same as jump, but marks the end of an alternative. */
533 /* Followed by two-byte relative address of place to resume at
534 in case of failure. */
535 /* ifdef MBS_SUPPORT, the size of address is 1. */
538 /* Like on_failure_jump, but pushes a placeholder instead of the
539 current string position when executed. */
540 on_failure_keep_string_jump,
542 /* Throw away latest failure point and then jump to following
543 two-byte relative address. */
544 /* ifdef MBS_SUPPORT, the size of address is 1. */
547 /* Change to pop_failure_jump if know won't have to backtrack to
548 match; otherwise change to jump. This is used to jump
549 back to the beginning of a repeat. If what follows this jump
550 clearly won't match what the repeat does, such that we can be
551 sure that there is no use backtracking out of repetitions
552 already matched, then we change it to a pop_failure_jump.
553 Followed by two-byte address. */
554 /* ifdef MBS_SUPPORT, the size of address is 1. */
557 /* Jump to following two-byte address, and push a dummy failure
558 point. This failure point will be thrown away if an attempt
559 is made to use it for a failure. A `+' construct makes this
560 before the first repeat. Also used as an intermediary kind
561 of jump when compiling an alternative. */
562 /* ifdef MBS_SUPPORT, the size of address is 1. */
565 /* Push a dummy failure point and continue. Used at the end of
569 /* Followed by two-byte relative address and two-byte number n.
570 After matching N times, jump to the address upon failure. */
571 /* ifdef MBS_SUPPORT, the size of address is 1. */
574 /* Followed by two-byte relative address, and two-byte number n.
575 Jump to the address N times, then fail. */
576 /* ifdef MBS_SUPPORT, the size of address is 1. */
579 /* Set the following two-byte relative address to the
580 subsequent two-byte number. The address *includes* the two
582 /* ifdef MBS_SUPPORT, the size of address is 1. */
585 wordchar, /* Matches any word-constituent character. */
586 notwordchar, /* Matches any char that is not a word-constituent. */
588 wordbeg, /* Succeeds if at word beginning. */
589 wordend, /* Succeeds if at word end. */
591 wordbound, /* Succeeds if at a word boundary. */
592 notwordbound /* Succeeds if not at a word boundary. */
595 ,before_dot, /* Succeeds if before point. */
596 at_dot, /* Succeeds if at point. */
597 after_dot, /* Succeeds if after point. */
599 /* Matches any character whose syntax is specified. Followed by
600 a byte which contains a syntax code, e.g., Sword. */
603 /* Matches any character whose syntax is not that specified. */
607 #endif /* not INSIDE_RECURSION */
612 # define UCHAR_T unsigned char
613 # define COMPILED_BUFFER_VAR bufp->buffer
614 # define OFFSET_ADDRESS_SIZE 2
615 # if defined (__STDC__) || defined (ALMOST_STDC) || defined (HAVE_STRINGIZE)
616 # define PREFIX(name) byte_##name
618 # define PREFIX(name) byte_/**/name
620 # define ARG_PREFIX(name) name
621 # define PUT_CHAR(c) putchar (c)
624 # define CHAR_T wchar_t
625 # define UCHAR_T wchar_t
626 # define COMPILED_BUFFER_VAR wc_buffer
627 # define OFFSET_ADDRESS_SIZE 1 /* the size which STORE_NUMBER macro use */
628 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
629 # if defined (__STDC__) || defined (ALMOST_STDC) || defined (HAVE_STRINGIZE)
630 # define PREFIX(name) wcs_##name
631 # define ARG_PREFIX(name) c##name
633 # define PREFIX(name) wcs_/**/name
634 # define ARG_PREFIX(name) c/**/name
636 /* Should we use wide stream?? */
637 # define PUT_CHAR(c) printf ("%C", c);
643 # define INSIDE_RECURSION
645 # undef INSIDE_RECURSION
648 # define INSIDE_RECURSION
650 # undef INSIDE_RECURSION
654 #ifdef INSIDE_RECURSION
655 /* Common operations on the compiled pattern. */
657 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
658 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
661 # define STORE_NUMBER(destination, number) \
663 *(destination) = (UCHAR_T)(number); \
666 # define STORE_NUMBER(destination, number) \
668 (destination)[0] = (number) & 0377; \
669 (destination)[1] = (number) >> 8; \
673 /* Same as STORE_NUMBER, except increment DESTINATION to
674 the byte after where the number is stored. Therefore, DESTINATION
675 must be an lvalue. */
676 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
678 # define STORE_NUMBER_AND_INCR(destination, number) \
680 STORE_NUMBER (destination, number); \
681 (destination) += OFFSET_ADDRESS_SIZE; \
684 /* Put into DESTINATION a number stored in two contiguous bytes starting
686 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
689 # define EXTRACT_NUMBER(destination, source) \
691 (destination) = *(source); \
694 # define EXTRACT_NUMBER(destination, source) \
696 (destination) = *(source) & 0377; \
697 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
702 static void PREFIX(extract_number) _RE_ARGS ((int *dest, UCHAR_T *source));
704 PREFIX(extract_number) (dest, source)
711 int temp = SIGN_EXTEND_CHAR (*(source + 1));
712 *dest = *source & 0377;
717 # ifndef EXTRACT_MACROS /* To debug the macros. */
718 # undef EXTRACT_NUMBER
719 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
720 # endif /* not EXTRACT_MACROS */
724 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
725 SOURCE must be an lvalue. */
727 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
729 EXTRACT_NUMBER (destination, source); \
730 (source) += OFFSET_ADDRESS_SIZE; \
734 static void PREFIX(extract_number_and_incr) _RE_ARGS ((int *destination,
737 PREFIX(extract_number_and_incr) (destination, source)
741 PREFIX(extract_number) (destination, *source);
742 *source += OFFSET_ADDRESS_SIZE;
745 # ifndef EXTRACT_MACROS
746 # undef EXTRACT_NUMBER_AND_INCR
747 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
748 PREFIX(extract_number_and_incr) (&dest, &src)
749 # endif /* not EXTRACT_MACROS */
755 /* If DEBUG is defined, Regex prints many voluminous messages about what
756 it is doing (if the variable `debug' is nonzero). If linked with the
757 main program in `iregex.c', you can enter patterns and strings
758 interactively. And if linked with the main program in `main.c' and
759 the other test files, you can run the already-written tests. */
763 # ifndef DEFINED_ONCE
765 /* We use standard I/O for debugging. */
768 /* It is useful to test things that ``must'' be true when debugging. */
773 # define DEBUG_STATEMENT(e) e
774 # define DEBUG_PRINT1(x) if (debug) printf (x)
775 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
776 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
777 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
778 # endif /* not DEFINED_ONCE */
780 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
781 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
782 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
783 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
786 /* Print the fastmap in human-readable form. */
788 # ifndef DEFINED_ONCE
790 print_fastmap (fastmap)
793 unsigned was_a_range = 0;
796 while (i < (1 << BYTEWIDTH))
802 while (i < (1 << BYTEWIDTH) && fastmap[i])
816 # endif /* not DEFINED_ONCE */
819 /* Print a compiled pattern string in human-readable form, starting at
820 the START pointer into it and ending just before the pointer END. */
823 PREFIX(print_partial_compiled_pattern) (start, end)
838 /* Loop over pattern commands. */
842 printf ("%td:\t", p - start);
844 printf ("%ld:\t", (long int) (p - start));
847 switch ((re_opcode_t) *p++)
855 printf ("/exactn/%d", mcnt);
867 printf ("/exactn_bin/%d", mcnt);
870 printf("/%lx", (long int) *p++);
874 # endif /* MBS_SUPPORT */
878 printf ("/start_memory/%d/%ld", mcnt, (long int) *p++);
883 printf ("/stop_memory/%d/%ld", mcnt, (long int) *p++);
887 printf ("/duplicate/%ld", (long int) *p++);
900 printf ("/charset [%s",
901 (re_opcode_t) *(workp - 1) == charset_not ? "^" : "");
903 length = *workp++; /* the length of char_classes */
904 for (i=0 ; i<length ; i++)
905 printf("[:%lx:]", (long int) *p++);
906 length = *workp++; /* the length of collating_symbol */
907 for (i=0 ; i<length ;)
911 PUT_CHAR((i++,*p++));
915 length = *workp++; /* the length of equivalence_class */
916 for (i=0 ; i<length ;)
920 PUT_CHAR((i++,*p++));
924 length = *workp++; /* the length of char_range */
925 for (i=0 ; i<length ; i++)
927 wchar_t range_start = *p++;
928 wchar_t range_end = *p++;
929 printf("%C-%C", range_start, range_end);
931 length = *workp++; /* the length of char */
932 for (i=0 ; i<length ; i++)
936 register int c, last = -100;
937 register int in_range = 0;
939 printf ("/charset [%s",
940 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
942 assert (p + *p < pend);
944 for (c = 0; c < 256; c++)
946 && (p[1 + (c/8)] & (1 << (c % 8))))
948 /* Are we starting a range? */
949 if (last + 1 == c && ! in_range)
954 /* Have we broken a range? */
955 else if (last + 1 != c && in_range)
985 case on_failure_jump:
986 PREFIX(extract_number_and_incr) (&mcnt, &p);
988 printf ("/on_failure_jump to %td", p + mcnt - start);
990 printf ("/on_failure_jump to %ld", (long int) (p + mcnt - start));
994 case on_failure_keep_string_jump:
995 PREFIX(extract_number_and_incr) (&mcnt, &p);
997 printf ("/on_failure_keep_string_jump to %td", p + mcnt - start);
999 printf ("/on_failure_keep_string_jump to %ld",
1000 (long int) (p + mcnt - start));
1004 case dummy_failure_jump:
1005 PREFIX(extract_number_and_incr) (&mcnt, &p);
1007 printf ("/dummy_failure_jump to %td", p + mcnt - start);
1009 printf ("/dummy_failure_jump to %ld", (long int) (p + mcnt - start));
1013 case push_dummy_failure:
1014 printf ("/push_dummy_failure");
1017 case maybe_pop_jump:
1018 PREFIX(extract_number_and_incr) (&mcnt, &p);
1020 printf ("/maybe_pop_jump to %td", p + mcnt - start);
1022 printf ("/maybe_pop_jump to %ld", (long int) (p + mcnt - start));
1026 case pop_failure_jump:
1027 PREFIX(extract_number_and_incr) (&mcnt, &p);
1029 printf ("/pop_failure_jump to %td", p + mcnt - start);
1031 printf ("/pop_failure_jump to %ld", (long int) (p + mcnt - start));
1036 PREFIX(extract_number_and_incr) (&mcnt, &p);
1038 printf ("/jump_past_alt to %td", p + mcnt - start);
1040 printf ("/jump_past_alt to %ld", (long int) (p + mcnt - start));
1045 PREFIX(extract_number_and_incr) (&mcnt, &p);
1047 printf ("/jump to %td", p + mcnt - start);
1049 printf ("/jump to %ld", (long int) (p + mcnt - start));
1054 PREFIX(extract_number_and_incr) (&mcnt, &p);
1056 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1058 printf ("/succeed_n to %td, %d times", p1 - start, mcnt2);
1060 printf ("/succeed_n to %ld, %d times",
1061 (long int) (p1 - start), mcnt2);
1066 PREFIX(extract_number_and_incr) (&mcnt, &p);
1068 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1069 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
1073 PREFIX(extract_number_and_incr) (&mcnt, &p);
1075 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1077 printf ("/set_number_at location %td to %d", p1 - start, mcnt2);
1079 printf ("/set_number_at location %ld to %d",
1080 (long int) (p1 - start), mcnt2);
1085 printf ("/wordbound");
1089 printf ("/notwordbound");
1093 printf ("/wordbeg");
1097 printf ("/wordend");
1102 printf ("/before_dot");
1110 printf ("/after_dot");
1114 printf ("/syntaxspec");
1116 printf ("/%d", mcnt);
1120 printf ("/notsyntaxspec");
1122 printf ("/%d", mcnt);
1127 printf ("/wordchar");
1131 printf ("/notwordchar");
1143 printf ("?%ld", (long int) *(p-1));
1150 printf ("%td:\tend of pattern.\n", p - start);
1152 printf ("%ld:\tend of pattern.\n", (long int) (p - start));
1158 PREFIX(print_compiled_pattern) (bufp)
1159 struct re_pattern_buffer *bufp;
1161 UCHAR_T *buffer = (UCHAR_T*) bufp->buffer;
1163 PREFIX(print_partial_compiled_pattern) (buffer, buffer
1164 + bufp->used / sizeof(UCHAR_T));
1165 printf ("%ld bytes used/%ld bytes allocated.\n",
1166 bufp->used, bufp->allocated);
1168 if (bufp->fastmap_accurate && bufp->fastmap)
1170 printf ("fastmap: ");
1171 print_fastmap (bufp->fastmap);
1175 printf ("re_nsub: %Zd\t", bufp->re_nsub);
1177 printf ("re_nsub: %ld\t", (long int) bufp->re_nsub);
1179 printf ("regs_alloc: %d\t", bufp->regs_allocated);
1180 printf ("can_be_null: %d\t", bufp->can_be_null);
1181 printf ("newline_anchor: %d\n", bufp->newline_anchor);
1182 printf ("no_sub: %d\t", bufp->no_sub);
1183 printf ("not_bol: %d\t", bufp->not_bol);
1184 printf ("not_eol: %d\t", bufp->not_eol);
1185 printf ("syntax: %lx\n", bufp->syntax);
1186 /* Perhaps we should print the translate table? */
1191 PREFIX(print_double_string) (where, string1, size1, string2, size2)
1192 const CHAR_T *where;
1193 const CHAR_T *string1;
1194 const CHAR_T *string2;
1206 if (FIRST_STRING_P (where))
1208 for (this_char = where - string1; this_char < size1; this_char++)
1209 PUT_CHAR (string1[this_char]);
1215 for (this_char = where - string2; this_char < size2; this_char++)
1217 PUT_CHAR (string2[this_char]);
1220 fputs ("...", stdout);
1227 # ifndef DEFINED_ONCE
1236 # else /* not DEBUG */
1238 # ifndef DEFINED_ONCE
1242 # define DEBUG_STATEMENT(e)
1243 # define DEBUG_PRINT1(x)
1244 # define DEBUG_PRINT2(x1, x2)
1245 # define DEBUG_PRINT3(x1, x2, x3)
1246 # define DEBUG_PRINT4(x1, x2, x3, x4)
1247 # endif /* not DEFINED_ONCE */
1248 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1249 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1251 # endif /* not DEBUG */
1256 /* This convert a multibyte string to a wide character string.
1257 And write their correspondances to offset_buffer(see below)
1258 and write whether each wchar_t is binary data to is_binary.
1259 This assume invalid multibyte sequences as binary data.
1260 We assume offset_buffer and is_binary is already allocated
1263 static size_t convert_mbs_to_wcs (CHAR_T *dest, const unsigned char* src,
1264 size_t len, int *offset_buffer,
1267 convert_mbs_to_wcs (dest, src, len, offset_buffer, is_binary)
1269 const unsigned char* src;
1270 size_t len; /* the length of multibyte string. */
1272 /* It hold correspondances between src(char string) and
1273 dest(wchar_t string) for optimization.
1275 dest = {'X', 'Y', 'Z'}
1276 (each "xxx", "y" and "zz" represent one multibyte character
1277 corresponding to 'X', 'Y' and 'Z'.)
1278 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1284 wchar_t *pdest = dest;
1285 const unsigned char *psrc = src;
1286 size_t wc_count = 0;
1290 size_t mb_remain = len;
1291 size_t mb_count = 0;
1293 /* Initialize the conversion state. */
1294 memset (&mbs, 0, sizeof (mbstate_t));
1296 offset_buffer[0] = 0;
1297 for( ; mb_remain > 0 ; ++wc_count, ++pdest, mb_remain -= consumed,
1301 consumed = __mbrtowc (pdest, psrc, mb_remain, &mbs);
1303 consumed = mbrtowc (pdest, psrc, mb_remain, &mbs);
1307 /* failed to convert. maybe src contains binary data.
1308 So we consume 1 byte manualy. */
1312 is_binary[wc_count] = TRUE;
1315 is_binary[wc_count] = FALSE;
1316 /* In sjis encoding, we use yen sign as escape character in
1317 place of reverse solidus. So we convert 0x5c(yen sign in
1318 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1319 solidus in UCS2). */
1320 if (consumed == 1 && (int) *psrc == 0x5c && (int) *pdest == 0xa5)
1321 *pdest = (wchar_t) *psrc;
1323 offset_buffer[wc_count + 1] = mb_count += consumed;
1326 /* Fill remain of the buffer with sentinel. */
1327 for (i = wc_count + 1 ; i <= len ; i++)
1328 offset_buffer[i] = mb_count + 1;
1335 #else /* not INSIDE_RECURSION */
1337 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1338 also be assigned to arbitrarily: each pattern buffer stores its own
1339 syntax, so it can be changed between regex compilations. */
1340 /* This has no initializer because initialized variables in Emacs
1341 become read-only after dumping. */
1342 reg_syntax_t re_syntax_options;
1345 /* Specify the precise syntax of regexps for compilation. This provides
1346 for compatibility for various utilities which historically have
1347 different, incompatible syntaxes.
1349 The argument SYNTAX is a bit mask comprised of the various bits
1350 defined in regex.h. We return the old syntax. */
1353 re_set_syntax (syntax)
1354 reg_syntax_t syntax;
1356 reg_syntax_t ret = re_syntax_options;
1358 re_syntax_options = syntax;
1360 if (syntax & RE_DEBUG)
1362 else if (debug) /* was on but now is not */
1368 weak_alias (__re_set_syntax, re_set_syntax)
1371 /* This table gives an error message for each of the error codes listed
1372 in regex.h. Obviously the order here has to be same as there.
1373 POSIX doesn't require that we do anything for REG_NOERROR,
1374 but why not be nice? */
1376 static const char *re_error_msgid[] =
1378 gettext_noop ("Success"), /* REG_NOERROR */
1379 gettext_noop ("No match"), /* REG_NOMATCH */
1380 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
1381 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
1382 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
1383 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
1384 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1385 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1386 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1387 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1388 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1389 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1390 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1391 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1392 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1393 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1394 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
1397 #endif /* INSIDE_RECURSION */
1399 #ifndef DEFINED_ONCE
1400 /* Avoiding alloca during matching, to placate r_alloc. */
1402 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1403 searching and matching functions should not call alloca. On some
1404 systems, alloca is implemented in terms of malloc, and if we're
1405 using the relocating allocator routines, then malloc could cause a
1406 relocation, which might (if the strings being searched are in the
1407 ralloc heap) shift the data out from underneath the regexp
1410 Here's another reason to avoid allocation: Emacs
1411 processes input from X in a signal handler; processing X input may
1412 call malloc; if input arrives while a matching routine is calling
1413 malloc, then we're scrod. But Emacs can't just block input while
1414 calling matching routines; then we don't notice interrupts when
1415 they come in. So, Emacs blocks input around all regexp calls
1416 except the matching calls, which it leaves unprotected, in the
1417 faith that they will not malloc. */
1419 /* Normally, this is fine. */
1420 # define MATCH_MAY_ALLOCATE
1422 /* When using GNU C, we are not REALLY using the C alloca, no matter
1423 what config.h may say. So don't take precautions for it. */
1428 /* The match routines may not allocate if (1) they would do it with malloc
1429 and (2) it's not safe for them to use malloc.
1430 Note that if REL_ALLOC is defined, matching would not use malloc for the
1431 failure stack, but we would still use it for the register vectors;
1432 so REL_ALLOC should not affect this. */
1433 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1434 # undef MATCH_MAY_ALLOCATE
1436 #endif /* not DEFINED_ONCE */
1438 #ifdef INSIDE_RECURSION
1439 /* Failure stack declarations and macros; both re_compile_fastmap and
1440 re_match_2 use a failure stack. These have to be macros because of
1441 REGEX_ALLOCATE_STACK. */
1444 /* Number of failure points for which to initially allocate space
1445 when matching. If this number is exceeded, we allocate more
1446 space, so it is not a hard limit. */
1447 # ifndef INIT_FAILURE_ALLOC
1448 # define INIT_FAILURE_ALLOC 5
1451 /* Roughly the maximum number of failure points on the stack. Would be
1452 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1453 This is a variable only so users of regex can assign to it; we never
1454 change it ourselves. */
1456 # ifdef INT_IS_16BIT
1458 # ifndef DEFINED_ONCE
1459 # if defined MATCH_MAY_ALLOCATE
1460 /* 4400 was enough to cause a crash on Alpha OSF/1,
1461 whose default stack limit is 2mb. */
1462 long int re_max_failures = 4000;
1464 long int re_max_failures = 2000;
1468 union PREFIX(fail_stack_elt)
1474 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1478 PREFIX(fail_stack_elt_t) *stack;
1479 unsigned long int size;
1480 unsigned long int avail; /* Offset of next open position. */
1481 } PREFIX(fail_stack_type);
1483 # else /* not INT_IS_16BIT */
1485 # ifndef DEFINED_ONCE
1486 # if defined MATCH_MAY_ALLOCATE
1487 /* 4400 was enough to cause a crash on Alpha OSF/1,
1488 whose default stack limit is 2mb. */
1489 int re_max_failures = 4000;
1491 int re_max_failures = 2000;
1495 union PREFIX(fail_stack_elt)
1501 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1505 PREFIX(fail_stack_elt_t) *stack;
1507 unsigned avail; /* Offset of next open position. */
1508 } PREFIX(fail_stack_type);
1510 # endif /* INT_IS_16BIT */
1512 # ifndef DEFINED_ONCE
1513 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1514 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1515 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1519 /* Define macros to initialize and free the failure stack.
1520 Do `return -2' if the alloc fails. */
1522 # ifdef MATCH_MAY_ALLOCATE
1523 # define INIT_FAIL_STACK() \
1525 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1526 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1528 if (fail_stack.stack == NULL) \
1531 fail_stack.size = INIT_FAILURE_ALLOC; \
1532 fail_stack.avail = 0; \
1535 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1537 # define INIT_FAIL_STACK() \
1539 fail_stack.avail = 0; \
1542 # define RESET_FAIL_STACK()
1546 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1548 Return 1 if succeeds, and 0 if either ran out of memory
1549 allocating space for it or it was already too large.
1551 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1553 # define DOUBLE_FAIL_STACK(fail_stack) \
1554 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1556 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1557 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1558 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1559 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1561 (fail_stack).stack == NULL \
1563 : ((fail_stack).size <<= 1, \
1567 /* Push pointer POINTER on FAIL_STACK.
1568 Return 1 if was able to do so and 0 if ran out of memory allocating
1570 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1571 ((FAIL_STACK_FULL () \
1572 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1574 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1577 /* Push a pointer value onto the failure stack.
1578 Assumes the variable `fail_stack'. Probably should only
1579 be called from within `PUSH_FAILURE_POINT'. */
1580 # define PUSH_FAILURE_POINTER(item) \
1581 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1583 /* This pushes an integer-valued item onto the failure stack.
1584 Assumes the variable `fail_stack'. Probably should only
1585 be called from within `PUSH_FAILURE_POINT'. */
1586 # define PUSH_FAILURE_INT(item) \
1587 fail_stack.stack[fail_stack.avail++].integer = (item)
1589 /* Push a fail_stack_elt_t value onto the failure stack.
1590 Assumes the variable `fail_stack'. Probably should only
1591 be called from within `PUSH_FAILURE_POINT'. */
1592 # define PUSH_FAILURE_ELT(item) \
1593 fail_stack.stack[fail_stack.avail++] = (item)
1595 /* These three POP... operations complement the three PUSH... operations.
1596 All assume that `fail_stack' is nonempty. */
1597 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1598 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1599 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1601 /* Used to omit pushing failure point id's when we're not debugging. */
1603 # define DEBUG_PUSH PUSH_FAILURE_INT
1604 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1606 # define DEBUG_PUSH(item)
1607 # define DEBUG_POP(item_addr)
1611 /* Push the information about the state we will need
1612 if we ever fail back to it.
1614 Requires variables fail_stack, regstart, regend, reg_info, and
1615 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1618 Does `return FAILURE_CODE' if runs out of memory. */
1620 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1622 char *destination; \
1623 /* Must be int, so when we don't save any registers, the arithmetic \
1624 of 0 + -1 isn't done as unsigned. */ \
1625 /* Can't be int, since there is not a shred of a guarantee that int \
1626 is wide enough to hold a value of something to which pointer can \
1628 active_reg_t this_reg; \
1630 DEBUG_STATEMENT (failure_id++); \
1631 DEBUG_STATEMENT (nfailure_points_pushed++); \
1632 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1633 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1634 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1636 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1637 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1639 /* Ensure we have enough space allocated for what we will push. */ \
1640 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1642 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1643 return failure_code; \
1645 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1646 (fail_stack).size); \
1647 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1650 /* Push the info, starting with the registers. */ \
1651 DEBUG_PRINT1 ("\n"); \
1654 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1657 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1658 DEBUG_STATEMENT (num_regs_pushed++); \
1660 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1661 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1663 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1664 PUSH_FAILURE_POINTER (regend[this_reg]); \
1666 DEBUG_PRINT2 (" info: %p\n ", \
1667 reg_info[this_reg].word.pointer); \
1668 DEBUG_PRINT2 (" match_null=%d", \
1669 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1670 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1671 DEBUG_PRINT2 (" matched_something=%d", \
1672 MATCHED_SOMETHING (reg_info[this_reg])); \
1673 DEBUG_PRINT2 (" ever_matched=%d", \
1674 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1675 DEBUG_PRINT1 ("\n"); \
1676 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1679 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1680 PUSH_FAILURE_INT (lowest_active_reg); \
1682 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1683 PUSH_FAILURE_INT (highest_active_reg); \
1685 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1686 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1687 PUSH_FAILURE_POINTER (pattern_place); \
1689 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1690 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1692 DEBUG_PRINT1 ("'\n"); \
1693 PUSH_FAILURE_POINTER (string_place); \
1695 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1696 DEBUG_PUSH (failure_id); \
1699 # ifndef DEFINED_ONCE
1700 /* This is the number of items that are pushed and popped on the stack
1701 for each register. */
1702 # define NUM_REG_ITEMS 3
1704 /* Individual items aside from the registers. */
1706 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1708 # define NUM_NONREG_ITEMS 4
1711 /* We push at most this many items on the stack. */
1712 /* We used to use (num_regs - 1), which is the number of registers
1713 this regexp will save; but that was changed to 5
1714 to avoid stack overflow for a regexp with lots of parens. */
1715 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1717 /* We actually push this many items. */
1718 # define NUM_FAILURE_ITEMS \
1720 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1724 /* How many items can still be added to the stack without overflowing it. */
1725 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1726 # endif /* not DEFINED_ONCE */
1729 /* Pops what PUSH_FAIL_STACK pushes.
1731 We restore into the parameters, all of which should be lvalues:
1732 STR -- the saved data position.
1733 PAT -- the saved pattern position.
1734 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1735 REGSTART, REGEND -- arrays of string positions.
1736 REG_INFO -- array of information about each subexpression.
1738 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1739 `pend', `string1', `size1', `string2', and `size2'. */
1740 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1742 DEBUG_STATEMENT (unsigned failure_id;) \
1743 active_reg_t this_reg; \
1744 const UCHAR_T *string_temp; \
1746 assert (!FAIL_STACK_EMPTY ()); \
1748 /* Remove failure points and point to how many regs pushed. */ \
1749 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1750 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1751 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1753 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1755 DEBUG_POP (&failure_id); \
1756 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1758 /* If the saved string location is NULL, it came from an \
1759 on_failure_keep_string_jump opcode, and we want to throw away the \
1760 saved NULL, thus retaining our current position in the string. */ \
1761 string_temp = POP_FAILURE_POINTER (); \
1762 if (string_temp != NULL) \
1763 str = (const CHAR_T *) string_temp; \
1765 DEBUG_PRINT2 (" Popping string %p: `", str); \
1766 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1767 DEBUG_PRINT1 ("'\n"); \
1769 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1770 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1771 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1773 /* Restore register info. */ \
1774 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1775 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1777 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1778 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1781 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1783 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1785 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1786 DEBUG_PRINT2 (" info: %p\n", \
1787 reg_info[this_reg].word.pointer); \
1789 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1790 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1792 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1793 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1797 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1799 reg_info[this_reg].word.integer = 0; \
1800 regend[this_reg] = 0; \
1801 regstart[this_reg] = 0; \
1803 highest_active_reg = high_reg; \
1806 set_regs_matched_done = 0; \
1807 DEBUG_STATEMENT (nfailure_points_popped++); \
1808 } /* POP_FAILURE_POINT */
1810 /* Structure for per-register (a.k.a. per-group) information.
1811 Other register information, such as the
1812 starting and ending positions (which are addresses), and the list of
1813 inner groups (which is a bits list) are maintained in separate
1816 We are making a (strictly speaking) nonportable assumption here: that
1817 the compiler will pack our bit fields into something that fits into
1818 the type of `word', i.e., is something that fits into one item on the
1822 /* Declarations and macros for re_match_2. */
1826 PREFIX(fail_stack_elt_t) word;
1829 /* This field is one if this group can match the empty string,
1830 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1831 # define MATCH_NULL_UNSET_VALUE 3
1832 unsigned match_null_string_p : 2;
1833 unsigned is_active : 1;
1834 unsigned matched_something : 1;
1835 unsigned ever_matched_something : 1;
1837 } PREFIX(register_info_type);
1839 # ifndef DEFINED_ONCE
1840 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1841 # define IS_ACTIVE(R) ((R).bits.is_active)
1842 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1843 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1846 /* Call this when have matched a real character; it sets `matched' flags
1847 for the subexpressions which we are currently inside. Also records
1848 that those subexprs have matched. */
1849 # define SET_REGS_MATCHED() \
1852 if (!set_regs_matched_done) \
1855 set_regs_matched_done = 1; \
1856 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1858 MATCHED_SOMETHING (reg_info[r]) \
1859 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1865 # endif /* not DEFINED_ONCE */
1867 /* Registers are set to a sentinel when they haven't yet matched. */
1868 static CHAR_T PREFIX(reg_unset_dummy);
1869 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1870 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1872 /* Subroutine declarations and macros for regex_compile. */
1873 static void PREFIX(store_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc, int arg));
1874 static void PREFIX(store_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1875 int arg1, int arg2));
1876 static void PREFIX(insert_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1877 int arg, UCHAR_T *end));
1878 static void PREFIX(insert_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1879 int arg1, int arg2, UCHAR_T *end));
1880 static boolean PREFIX(at_begline_loc_p) _RE_ARGS ((const CHAR_T *pattern,
1882 reg_syntax_t syntax));
1883 static boolean PREFIX(at_endline_loc_p) _RE_ARGS ((const CHAR_T *p,
1885 reg_syntax_t syntax));
1887 static reg_errcode_t wcs_compile_range _RE_ARGS ((CHAR_T range_start,
1888 const CHAR_T **p_ptr,
1891 reg_syntax_t syntax,
1894 static void insert_space _RE_ARGS ((int num, CHAR_T *loc, CHAR_T *end));
1896 static reg_errcode_t byte_compile_range _RE_ARGS ((unsigned int range_start,
1900 reg_syntax_t syntax,
1904 /* Fetch the next character in the uncompiled pattern---translating it
1905 if necessary. Also cast from a signed character in the constant
1906 string passed to us by the user to an unsigned char that we can use
1907 as an array index (in, e.g., `translate'). */
1908 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1909 because it is impossible to allocate 4GB array for some encodings
1910 which have 4 byte character_set like UCS4. */
1913 # define PATFETCH(c) \
1914 do {if (p == pend) return REG_EEND; \
1915 c = (UCHAR_T) *p++; \
1916 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1919 # define PATFETCH(c) \
1920 do {if (p == pend) return REG_EEND; \
1921 c = (unsigned char) *p++; \
1922 if (translate) c = (unsigned char) translate[c]; \
1927 /* Fetch the next character in the uncompiled pattern, with no
1929 # define PATFETCH_RAW(c) \
1930 do {if (p == pend) return REG_EEND; \
1931 c = (UCHAR_T) *p++; \
1934 /* Go backwards one character in the pattern. */
1935 # define PATUNFETCH p--
1938 /* If `translate' is non-null, return translate[D], else just D. We
1939 cast the subscript to translate because some data is declared as
1940 `char *', to avoid warnings when a string constant is passed. But
1941 when we use a character as a subscript we must make it unsigned. */
1942 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1943 because it is impossible to allocate 4GB array for some encodings
1944 which have 4 byte character_set like UCS4. */
1948 # define TRANSLATE(d) \
1949 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1950 ? (char) translate[(unsigned char) (d)] : (d))
1952 # define TRANSLATE(d) \
1953 (translate ? (char) translate[(unsigned char) (d)] : (d))
1958 /* Macros for outputting the compiled pattern into `buffer'. */
1960 /* If the buffer isn't allocated when it comes in, use this. */
1961 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1963 /* Make sure we have at least N more bytes of space in buffer. */
1965 # define GET_BUFFER_SPACE(n) \
1966 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1967 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1970 # define GET_BUFFER_SPACE(n) \
1971 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1975 /* Make sure we have one more byte of buffer space and then add C to it. */
1976 # define BUF_PUSH(c) \
1978 GET_BUFFER_SPACE (1); \
1979 *b++ = (UCHAR_T) (c); \
1983 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1984 # define BUF_PUSH_2(c1, c2) \
1986 GET_BUFFER_SPACE (2); \
1987 *b++ = (UCHAR_T) (c1); \
1988 *b++ = (UCHAR_T) (c2); \
1992 /* As with BUF_PUSH_2, except for three bytes. */
1993 # define BUF_PUSH_3(c1, c2, c3) \
1995 GET_BUFFER_SPACE (3); \
1996 *b++ = (UCHAR_T) (c1); \
1997 *b++ = (UCHAR_T) (c2); \
1998 *b++ = (UCHAR_T) (c3); \
2001 /* Store a jump with opcode OP at LOC to location TO. We store a
2002 relative address offset by the three bytes the jump itself occupies. */
2003 # define STORE_JUMP(op, loc, to) \
2004 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
2006 /* Likewise, for a two-argument jump. */
2007 # define STORE_JUMP2(op, loc, to, arg) \
2008 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
2010 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
2011 # define INSERT_JUMP(op, loc, to) \
2012 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
2014 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
2015 # define INSERT_JUMP2(op, loc, to, arg) \
2016 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
2019 /* This is not an arbitrary limit: the arguments which represent offsets
2020 into the pattern are two bytes long. So if 2^16 bytes turns out to
2021 be too small, many things would have to change. */
2022 /* Any other compiler which, like MSC, has allocation limit below 2^16
2023 bytes will have to use approach similar to what was done below for
2024 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2025 reallocating to 0 bytes. Such thing is not going to work too well.
2026 You have been warned!! */
2027 # ifndef DEFINED_ONCE
2028 # if defined _MSC_VER && !defined WIN32
2029 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2030 The REALLOC define eliminates a flurry of conversion warnings,
2031 but is not required. */
2032 # define MAX_BUF_SIZE 65500L
2033 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2035 # define MAX_BUF_SIZE (1L << 16)
2036 # define REALLOC(p,s) realloc ((p), (s))
2039 /* Extend the buffer by twice its current size via realloc and
2040 reset the pointers that pointed into the old block to point to the
2041 correct places in the new one. If extending the buffer results in it
2042 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2043 # if __BOUNDED_POINTERS__
2044 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2045 # define MOVE_BUFFER_POINTER(P) \
2046 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2047 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2050 SET_HIGH_BOUND (b); \
2051 SET_HIGH_BOUND (begalt); \
2052 if (fixup_alt_jump) \
2053 SET_HIGH_BOUND (fixup_alt_jump); \
2055 SET_HIGH_BOUND (laststart); \
2056 if (pending_exact) \
2057 SET_HIGH_BOUND (pending_exact); \
2060 # define MOVE_BUFFER_POINTER(P) (P) += incr
2061 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2063 # endif /* not DEFINED_ONCE */
2066 # define EXTEND_BUFFER() \
2068 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2070 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2072 bufp->allocated <<= 1; \
2073 if (bufp->allocated > MAX_BUF_SIZE) \
2074 bufp->allocated = MAX_BUF_SIZE; \
2075 /* How many characters the new buffer can have? */ \
2076 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2077 if (wchar_count == 0) wchar_count = 1; \
2078 /* Truncate the buffer to CHAR_T align. */ \
2079 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2080 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2081 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2082 if (COMPILED_BUFFER_VAR == NULL) \
2083 return REG_ESPACE; \
2084 /* If the buffer moved, move all the pointers into it. */ \
2085 if (old_buffer != COMPILED_BUFFER_VAR) \
2087 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2088 MOVE_BUFFER_POINTER (b); \
2089 MOVE_BUFFER_POINTER (begalt); \
2090 if (fixup_alt_jump) \
2091 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2093 MOVE_BUFFER_POINTER (laststart); \
2094 if (pending_exact) \
2095 MOVE_BUFFER_POINTER (pending_exact); \
2097 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2100 # define EXTEND_BUFFER() \
2102 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2103 if (bufp->allocated == MAX_BUF_SIZE) \
2105 bufp->allocated <<= 1; \
2106 if (bufp->allocated > MAX_BUF_SIZE) \
2107 bufp->allocated = MAX_BUF_SIZE; \
2108 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2110 if (COMPILED_BUFFER_VAR == NULL) \
2111 return REG_ESPACE; \
2112 /* If the buffer moved, move all the pointers into it. */ \
2113 if (old_buffer != COMPILED_BUFFER_VAR) \
2115 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2116 MOVE_BUFFER_POINTER (b); \
2117 MOVE_BUFFER_POINTER (begalt); \
2118 if (fixup_alt_jump) \
2119 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2121 MOVE_BUFFER_POINTER (laststart); \
2122 if (pending_exact) \
2123 MOVE_BUFFER_POINTER (pending_exact); \
2125 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2129 # ifndef DEFINED_ONCE
2130 /* Since we have one byte reserved for the register number argument to
2131 {start,stop}_memory, the maximum number of groups we can report
2132 things about is what fits in that byte. */
2133 # define MAX_REGNUM 255
2135 /* But patterns can have more than `MAX_REGNUM' registers. We just
2136 ignore the excess. */
2137 typedef unsigned regnum_t;
2140 /* Macros for the compile stack. */
2142 /* Since offsets can go either forwards or backwards, this type needs to
2143 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2144 /* int may be not enough when sizeof(int) == 2. */
2145 typedef long pattern_offset_t;
2149 pattern_offset_t begalt_offset;
2150 pattern_offset_t fixup_alt_jump;
2151 pattern_offset_t inner_group_offset;
2152 pattern_offset_t laststart_offset;
2154 } compile_stack_elt_t;
2159 compile_stack_elt_t *stack;
2161 unsigned avail; /* Offset of next open position. */
2162 } compile_stack_type;
2165 # define INIT_COMPILE_STACK_SIZE 32
2167 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2168 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2170 /* The next available element. */
2171 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2173 # endif /* not DEFINED_ONCE */
2175 /* Set the bit for character C in a list. */
2176 # ifndef DEFINED_ONCE
2177 # define SET_LIST_BIT(c) \
2178 (b[((unsigned char) (c)) / BYTEWIDTH] \
2179 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2180 # endif /* DEFINED_ONCE */
2182 /* Get the next unsigned number in the uncompiled pattern. */
2183 # define GET_UNSIGNED_NUMBER(num) \
2188 if (c < '0' || c > '9') \
2190 if (num <= RE_DUP_MAX) \
2194 num = num * 10 + c - '0'; \
2199 # ifndef DEFINED_ONCE
2200 # if defined _LIBC || WIDE_CHAR_SUPPORT
2201 /* The GNU C library provides support for user-defined character classes
2202 and the functions from ISO C amendement 1. */
2203 # ifdef CHARCLASS_NAME_MAX
2204 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2206 /* This shouldn't happen but some implementation might still have this
2207 problem. Use a reasonable default value. */
2208 # define CHAR_CLASS_MAX_LENGTH 256
2212 # define IS_CHAR_CLASS(string) __wctype (string)
2214 # define IS_CHAR_CLASS(string) wctype (string)
2217 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
2219 # define IS_CHAR_CLASS(string) \
2220 (STREQ (string, "alpha") || STREQ (string, "upper") \
2221 || STREQ (string, "lower") || STREQ (string, "digit") \
2222 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
2223 || STREQ (string, "space") || STREQ (string, "print") \
2224 || STREQ (string, "punct") || STREQ (string, "graph") \
2225 || STREQ (string, "cntrl") || STREQ (string, "blank"))
2227 # endif /* DEFINED_ONCE */
2229 # ifndef MATCH_MAY_ALLOCATE
2231 /* If we cannot allocate large objects within re_match_2_internal,
2232 we make the fail stack and register vectors global.
2233 The fail stack, we grow to the maximum size when a regexp
2235 The register vectors, we adjust in size each time we
2236 compile a regexp, according to the number of registers it needs. */
2238 static PREFIX(fail_stack_type) fail_stack;
2240 /* Size with which the following vectors are currently allocated.
2241 That is so we can make them bigger as needed,
2242 but never make them smaller. */
2243 # ifdef DEFINED_ONCE
2244 static int regs_allocated_size;
2246 static const char ** regstart, ** regend;
2247 static const char ** old_regstart, ** old_regend;
2248 static const char **best_regstart, **best_regend;
2249 static const char **reg_dummy;
2250 # endif /* DEFINED_ONCE */
2252 static PREFIX(register_info_type) *PREFIX(reg_info);
2253 static PREFIX(register_info_type) *PREFIX(reg_info_dummy);
2255 /* Make the register vectors big enough for NUM_REGS registers,
2256 but don't make them smaller. */
2259 PREFIX(regex_grow_registers) (num_regs)
2262 if (num_regs > regs_allocated_size)
2264 RETALLOC_IF (regstart, num_regs, const char *);
2265 RETALLOC_IF (regend, num_regs, const char *);
2266 RETALLOC_IF (old_regstart, num_regs, const char *);
2267 RETALLOC_IF (old_regend, num_regs, const char *);
2268 RETALLOC_IF (best_regstart, num_regs, const char *);
2269 RETALLOC_IF (best_regend, num_regs, const char *);
2270 RETALLOC_IF (PREFIX(reg_info), num_regs, PREFIX(register_info_type));
2271 RETALLOC_IF (reg_dummy, num_regs, const char *);
2272 RETALLOC_IF (PREFIX(reg_info_dummy), num_regs, PREFIX(register_info_type));
2274 regs_allocated_size = num_regs;
2278 # endif /* not MATCH_MAY_ALLOCATE */
2280 # ifndef DEFINED_ONCE
2281 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
2284 # endif /* not DEFINED_ONCE */
2286 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2287 Returns one of error codes defined in `regex.h', or zero for success.
2289 Assumes the `allocated' (and perhaps `buffer') and `translate'
2290 fields are set in BUFP on entry.
2292 If it succeeds, results are put in BUFP (if it returns an error, the
2293 contents of BUFP are undefined):
2294 `buffer' is the compiled pattern;
2295 `syntax' is set to SYNTAX;
2296 `used' is set to the length of the compiled pattern;
2297 `fastmap_accurate' is zero;
2298 `re_nsub' is the number of subexpressions in PATTERN;
2299 `not_bol' and `not_eol' are zero;
2301 The `fastmap' and `newline_anchor' fields are neither
2302 examined nor set. */
2304 /* Return, freeing storage we allocated. */
2306 # define FREE_STACK_RETURN(value) \
2307 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2309 # define FREE_STACK_RETURN(value) \
2310 return (free (compile_stack.stack), value)
2313 static reg_errcode_t
2314 PREFIX(regex_compile) (ARG_PREFIX(pattern), ARG_PREFIX(size), syntax, bufp)
2315 const char *ARG_PREFIX(pattern);
2316 size_t ARG_PREFIX(size);
2317 reg_syntax_t syntax;
2318 struct re_pattern_buffer *bufp;
2320 /* We fetch characters from PATTERN here. Even though PATTERN is
2321 `char *' (i.e., signed), we declare these variables as unsigned, so
2322 they can be reliably used as array indices. */
2323 register UCHAR_T c, c1;
2326 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2327 CHAR_T *pattern, *COMPILED_BUFFER_VAR;
2329 /* offset buffer for optimization. See convert_mbs_to_wc. */
2330 int *mbs_offset = NULL;
2331 /* It hold whether each wchar_t is binary data or not. */
2332 char *is_binary = NULL;
2333 /* A flag whether exactn is handling binary data or not. */
2334 char is_exactn_bin = FALSE;
2337 /* A random temporary spot in PATTERN. */
2340 /* Points to the end of the buffer, where we should append. */
2341 register UCHAR_T *b;
2343 /* Keeps track of unclosed groups. */
2344 compile_stack_type compile_stack;
2346 /* Points to the current (ending) position in the pattern. */
2351 const CHAR_T *p = pattern;
2352 const CHAR_T *pend = pattern + size;
2355 /* How to translate the characters in the pattern. */
2356 RE_TRANSLATE_TYPE translate = bufp->translate;
2358 /* Address of the count-byte of the most recently inserted `exactn'
2359 command. This makes it possible to tell if a new exact-match
2360 character can be added to that command or if the character requires
2361 a new `exactn' command. */
2362 UCHAR_T *pending_exact = 0;
2364 /* Address of start of the most recently finished expression.
2365 This tells, e.g., postfix * where to find the start of its
2366 operand. Reset at the beginning of groups and alternatives. */
2367 UCHAR_T *laststart = 0;
2369 /* Address of beginning of regexp, or inside of last group. */
2372 /* Address of the place where a forward jump should go to the end of
2373 the containing expression. Each alternative of an `or' -- except the
2374 last -- ends with a forward jump of this sort. */
2375 UCHAR_T *fixup_alt_jump = 0;
2377 /* Counts open-groups as they are encountered. Remembered for the
2378 matching close-group on the compile stack, so the same register
2379 number is put in the stop_memory as the start_memory. */
2380 regnum_t regnum = 0;
2383 /* Initialize the wchar_t PATTERN and offset_buffer. */
2384 p = pend = pattern = TALLOC(csize + 1, CHAR_T);
2385 mbs_offset = TALLOC(csize + 1, int);
2386 is_binary = TALLOC(csize + 1, char);
2387 if (pattern == NULL || mbs_offset == NULL || is_binary == NULL)
2394 pattern[csize] = L'\0'; /* sentinel */
2395 size = convert_mbs_to_wcs(pattern, cpattern, csize, mbs_offset, is_binary);
2407 DEBUG_PRINT1 ("\nCompiling pattern: ");
2410 unsigned debug_count;
2412 for (debug_count = 0; debug_count < size; debug_count++)
2413 PUT_CHAR (pattern[debug_count]);
2418 /* Initialize the compile stack. */
2419 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2420 if (compile_stack.stack == NULL)
2430 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2431 compile_stack.avail = 0;
2433 /* Initialize the pattern buffer. */
2434 bufp->syntax = syntax;
2435 bufp->fastmap_accurate = 0;
2436 bufp->not_bol = bufp->not_eol = 0;
2438 /* Set `used' to zero, so that if we return an error, the pattern
2439 printer (for debugging) will think there's no pattern. We reset it
2443 /* Always count groups, whether or not bufp->no_sub is set. */
2446 #if !defined emacs && !defined SYNTAX_TABLE
2447 /* Initialize the syntax table. */
2448 init_syntax_once ();
2451 if (bufp->allocated == 0)
2454 { /* If zero allocated, but buffer is non-null, try to realloc
2455 enough space. This loses if buffer's address is bogus, but
2456 that is the user's responsibility. */
2458 /* Free bufp->buffer and allocate an array for wchar_t pattern
2461 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE/sizeof(UCHAR_T),
2464 RETALLOC (COMPILED_BUFFER_VAR, INIT_BUF_SIZE, UCHAR_T);
2468 { /* Caller did not allocate a buffer. Do it for them. */
2469 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE / sizeof(UCHAR_T),
2473 if (!COMPILED_BUFFER_VAR) FREE_STACK_RETURN (REG_ESPACE);
2475 bufp->buffer = (char*)COMPILED_BUFFER_VAR;
2477 bufp->allocated = INIT_BUF_SIZE;
2481 COMPILED_BUFFER_VAR = (UCHAR_T*) bufp->buffer;
2484 begalt = b = COMPILED_BUFFER_VAR;
2486 /* Loop through the uncompiled pattern until we're at the end. */
2495 if ( /* If at start of pattern, it's an operator. */
2497 /* If context independent, it's an operator. */
2498 || syntax & RE_CONTEXT_INDEP_ANCHORS
2499 /* Otherwise, depends on what's come before. */
2500 || PREFIX(at_begline_loc_p) (pattern, p, syntax))
2510 if ( /* If at end of pattern, it's an operator. */
2512 /* If context independent, it's an operator. */
2513 || syntax & RE_CONTEXT_INDEP_ANCHORS
2514 /* Otherwise, depends on what's next. */
2515 || PREFIX(at_endline_loc_p) (p, pend, syntax))
2525 if ((syntax & RE_BK_PLUS_QM)
2526 || (syntax & RE_LIMITED_OPS))
2530 /* If there is no previous pattern... */
2533 if (syntax & RE_CONTEXT_INVALID_OPS)
2534 FREE_STACK_RETURN (REG_BADRPT);
2535 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2540 /* Are we optimizing this jump? */
2541 boolean keep_string_p = false;
2543 /* 1 means zero (many) matches is allowed. */
2544 char zero_times_ok = 0, many_times_ok = 0;
2546 /* If there is a sequence of repetition chars, collapse it
2547 down to just one (the right one). We can't combine
2548 interval operators with these because of, e.g., `a{2}*',
2549 which should only match an even number of `a's. */
2553 zero_times_ok |= c != '+';
2554 many_times_ok |= c != '?';
2562 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2565 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2567 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2570 if (!(c1 == '+' || c1 == '?'))
2585 /* If we get here, we found another repeat character. */
2588 /* Star, etc. applied to an empty pattern is equivalent
2589 to an empty pattern. */
2593 /* Now we know whether or not zero matches is allowed
2594 and also whether or not two or more matches is allowed. */
2596 { /* More than one repetition is allowed, so put in at the
2597 end a backward relative jump from `b' to before the next
2598 jump we're going to put in below (which jumps from
2599 laststart to after this jump).
2601 But if we are at the `*' in the exact sequence `.*\n',
2602 insert an unconditional jump backwards to the .,
2603 instead of the beginning of the loop. This way we only
2604 push a failure point once, instead of every time
2605 through the loop. */
2606 assert (p - 1 > pattern);
2608 /* Allocate the space for the jump. */
2609 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2611 /* We know we are not at the first character of the pattern,
2612 because laststart was nonzero. And we've already
2613 incremented `p', by the way, to be the character after
2614 the `*'. Do we have to do something analogous here
2615 for null bytes, because of RE_DOT_NOT_NULL? */
2616 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2618 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2619 && !(syntax & RE_DOT_NEWLINE))
2620 { /* We have .*\n. */
2621 STORE_JUMP (jump, b, laststart);
2622 keep_string_p = true;
2625 /* Anything else. */
2626 STORE_JUMP (maybe_pop_jump, b, laststart -
2627 (1 + OFFSET_ADDRESS_SIZE));
2629 /* We've added more stuff to the buffer. */
2630 b += 1 + OFFSET_ADDRESS_SIZE;
2633 /* On failure, jump from laststart to b + 3, which will be the
2634 end of the buffer after this jump is inserted. */
2635 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2637 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2638 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2640 laststart, b + 1 + OFFSET_ADDRESS_SIZE);
2642 b += 1 + OFFSET_ADDRESS_SIZE;
2646 /* At least one repetition is required, so insert a
2647 `dummy_failure_jump' before the initial
2648 `on_failure_jump' instruction of the loop. This
2649 effects a skip over that instruction the first time
2650 we hit that loop. */
2651 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2652 INSERT_JUMP (dummy_failure_jump, laststart, laststart +
2653 2 + 2 * OFFSET_ADDRESS_SIZE);
2654 b += 1 + OFFSET_ADDRESS_SIZE;
2668 boolean had_char_class = false;
2670 CHAR_T range_start = 0xffffffff;
2672 unsigned int range_start = 0xffffffff;
2674 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2677 /* We assume a charset(_not) structure as a wchar_t array.
2678 charset[0] = (re_opcode_t) charset(_not)
2679 charset[1] = l (= length of char_classes)
2680 charset[2] = m (= length of collating_symbols)
2681 charset[3] = n (= length of equivalence_classes)
2682 charset[4] = o (= length of char_ranges)
2683 charset[5] = p (= length of chars)
2685 charset[6] = char_class (wctype_t)
2686 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2688 charset[l+5] = char_class (wctype_t)
2690 charset[l+6] = collating_symbol (wchar_t)
2692 charset[l+m+5] = collating_symbol (wchar_t)
2693 ifdef _LIBC we use the index if
2694 _NL_COLLATE_SYMB_EXTRAMB instead of
2697 charset[l+m+6] = equivalence_classes (wchar_t)
2699 charset[l+m+n+5] = equivalence_classes (wchar_t)
2700 ifdef _LIBC we use the index in
2701 _NL_COLLATE_WEIGHT instead of
2704 charset[l+m+n+6] = range_start
2705 charset[l+m+n+7] = range_end
2707 charset[l+m+n+2o+4] = range_start
2708 charset[l+m+n+2o+5] = range_end
2709 ifdef _LIBC we use the value looked up
2710 in _NL_COLLATE_COLLSEQ instead of
2713 charset[l+m+n+2o+6] = char
2715 charset[l+m+n+2o+p+5] = char
2719 /* We need at least 6 spaces: the opcode, the length of
2720 char_classes, the length of collating_symbols, the length of
2721 equivalence_classes, the length of char_ranges, the length of
2723 GET_BUFFER_SPACE (6);
2725 /* Save b as laststart. And We use laststart as the pointer
2726 to the first element of the charset here.
2727 In other words, laststart[i] indicates charset[i]. */
2730 /* We test `*p == '^' twice, instead of using an if
2731 statement, so we only need one BUF_PUSH. */
2732 BUF_PUSH (*p == '^' ? charset_not : charset);
2736 /* Push the length of char_classes, the length of
2737 collating_symbols, the length of equivalence_classes, the
2738 length of char_ranges and the length of chars. */
2739 BUF_PUSH_3 (0, 0, 0);
2742 /* Remember the first position in the bracket expression. */
2745 /* charset_not matches newline according to a syntax bit. */
2746 if ((re_opcode_t) b[-6] == charset_not
2747 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2750 laststart[5]++; /* Update the length of characters */
2753 /* Read in characters and ranges, setting map bits. */
2756 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2760 /* \ might escape characters inside [...] and [^...]. */
2761 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2763 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2767 laststart[5]++; /* Update the length of chars */
2772 /* Could be the end of the bracket expression. If it's
2773 not (i.e., when the bracket expression is `[]' so
2774 far), the ']' character bit gets set way below. */
2775 if (c == ']' && p != p1 + 1)
2778 /* Look ahead to see if it's a range when the last thing
2779 was a character class. */
2780 if (had_char_class && c == '-' && *p != ']')
2781 FREE_STACK_RETURN (REG_ERANGE);
2783 /* Look ahead to see if it's a range when the last thing
2784 was a character: if this is a hyphen not at the
2785 beginning or the end of a list, then it's the range
2788 && !(p - 2 >= pattern && p[-2] == '[')
2789 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2793 /* Allocate the space for range_start and range_end. */
2794 GET_BUFFER_SPACE (2);
2795 /* Update the pointer to indicate end of buffer. */
2797 ret = wcs_compile_range (range_start, &p, pend, translate,
2798 syntax, b, laststart);
2799 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2800 range_start = 0xffffffff;
2802 else if (p[0] == '-' && p[1] != ']')
2803 { /* This handles ranges made up of characters only. */
2806 /* Move past the `-'. */
2808 /* Allocate the space for range_start and range_end. */
2809 GET_BUFFER_SPACE (2);
2810 /* Update the pointer to indicate end of buffer. */
2812 ret = wcs_compile_range (c, &p, pend, translate, syntax, b,
2814 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2815 range_start = 0xffffffff;
2818 /* See if we're at the beginning of a possible character
2820 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2821 { /* Leave room for the null. */
2822 char str[CHAR_CLASS_MAX_LENGTH + 1];
2827 /* If pattern is `[[:'. */
2828 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2833 if ((c == ':' && *p == ']') || p == pend)
2835 if (c1 < CHAR_CLASS_MAX_LENGTH)
2838 /* This is in any case an invalid class name. */
2843 /* If isn't a word bracketed by `[:' and `:]':
2844 undo the ending character, the letters, and leave
2845 the leading `:' and `[' (but store them as character). */
2846 if (c == ':' && *p == ']')
2851 /* Query the character class as wctype_t. */
2852 wt = IS_CHAR_CLASS (str);
2854 FREE_STACK_RETURN (REG_ECTYPE);
2856 /* Throw away the ] at the end of the character
2860 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2862 /* Allocate the space for character class. */
2863 GET_BUFFER_SPACE(CHAR_CLASS_SIZE);
2864 /* Update the pointer to indicate end of buffer. */
2865 b += CHAR_CLASS_SIZE;
2866 /* Move data which follow character classes
2867 not to violate the data. */
2868 insert_space(CHAR_CLASS_SIZE,
2869 laststart + 6 + laststart[1],
2871 alignedp = ((uintptr_t)(laststart + 6 + laststart[1])
2872 + __alignof__(wctype_t) - 1)
2873 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
2874 /* Store the character class. */
2875 *((wctype_t*)alignedp) = wt;
2876 /* Update length of char_classes */
2877 laststart[1] += CHAR_CLASS_SIZE;
2879 had_char_class = true;
2888 laststart[5] += 2; /* Update the length of characters */
2890 had_char_class = false;
2893 else if (syntax & RE_CHAR_CLASSES && c == '[' && (*p == '='
2896 CHAR_T str[128]; /* Should be large enough. */
2897 CHAR_T delim = *p; /* '=' or '.' */
2900 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
2905 /* If pattern is `[[=' or '[[.'. */
2906 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2911 if ((c == delim && *p == ']') || p == pend)
2913 if (c1 < sizeof (str) - 1)
2916 /* This is in any case an invalid class name. */
2921 if (c == delim && *p == ']' && str[0] != '\0')
2923 unsigned int i, offset;
2924 /* If we have no collation data we use the default
2925 collation in which each character is in a class
2926 by itself. It also means that ASCII is the
2927 character set and therefore we cannot have character
2928 with more than one byte in the multibyte
2931 /* If not defined _LIBC, we push the name and
2932 `\0' for the sake of matching performance. */
2933 int datasize = c1 + 1;
2941 FREE_STACK_RETURN (REG_ECOLLATE);
2946 const int32_t *table;
2947 const int32_t *weights;
2948 const int32_t *extra;
2949 const int32_t *indirect;
2952 /* This #include defines a local function! */
2953 # include <locale/weightwc.h>
2957 /* We push the index for equivalence class. */
2960 table = (const int32_t *)
2961 _NL_CURRENT (LC_COLLATE,
2962 _NL_COLLATE_TABLEWC);
2963 weights = (const int32_t *)
2964 _NL_CURRENT (LC_COLLATE,
2965 _NL_COLLATE_WEIGHTWC);
2966 extra = (const int32_t *)
2967 _NL_CURRENT (LC_COLLATE,
2968 _NL_COLLATE_EXTRAWC);
2969 indirect = (const int32_t *)
2970 _NL_CURRENT (LC_COLLATE,
2971 _NL_COLLATE_INDIRECTWC);
2973 idx = findidx ((const wint_t**)&cp);
2974 if (idx == 0 || cp < (wint_t*) str + c1)
2975 /* This is no valid character. */
2976 FREE_STACK_RETURN (REG_ECOLLATE);
2978 str[0] = (wchar_t)idx;
2980 else /* delim == '.' */
2982 /* We push collation sequence value
2983 for collating symbol. */
2985 const int32_t *symb_table;
2986 const unsigned char *extra;
2993 /* We have to convert the name to a single-byte
2994 string. This is possible since the names
2995 consist of ASCII characters and the internal
2996 representation is UCS4. */
2997 for (i = 0; i < c1; ++i)
2998 char_str[i] = str[i];
3001 _NL_CURRENT_WORD (LC_COLLATE,
3002 _NL_COLLATE_SYMB_HASH_SIZEMB);
3003 symb_table = (const int32_t *)
3004 _NL_CURRENT (LC_COLLATE,
3005 _NL_COLLATE_SYMB_TABLEMB);
3006 extra = (const unsigned char *)
3007 _NL_CURRENT (LC_COLLATE,
3008 _NL_COLLATE_SYMB_EXTRAMB);
3010 /* Locate the character in the hashing table. */
3011 hash = elem_hash (char_str, c1);
3014 elem = hash % table_size;
3015 second = hash % (table_size - 2);
3016 while (symb_table[2 * elem] != 0)
3018 /* First compare the hashing value. */
3019 if (symb_table[2 * elem] == hash
3020 && c1 == extra[symb_table[2 * elem + 1]]
3021 && memcmp (char_str,
3022 &extra[symb_table[2 * elem + 1]
3025 /* Yep, this is the entry. */
3026 idx = symb_table[2 * elem + 1];
3027 idx += 1 + extra[idx];
3035 if (symb_table[2 * elem] != 0)
3037 /* Compute the index of the byte sequence
3039 idx += 1 + extra[idx];
3040 /* Adjust for the alignment. */
3041 idx = (idx + 3) & ~3;
3043 str[0] = (wchar_t) idx + 4;
3045 else if (symb_table[2 * elem] == 0 && c1 == 1)
3047 /* No valid character. Match it as a
3048 single byte character. */
3049 had_char_class = false;
3051 /* Update the length of characters */
3053 range_start = str[0];
3055 /* Throw away the ] at the end of the
3056 collating symbol. */
3058 /* exit from the switch block. */
3062 FREE_STACK_RETURN (REG_ECOLLATE);
3067 /* Throw away the ] at the end of the equivalence
3068 class (or collating symbol). */
3071 /* Allocate the space for the equivalence class
3072 (or collating symbol) (and '\0' if needed). */
3073 GET_BUFFER_SPACE(datasize);
3074 /* Update the pointer to indicate end of buffer. */
3078 { /* equivalence class */
3079 /* Calculate the offset of char_ranges,
3080 which is next to equivalence_classes. */
3081 offset = laststart[1] + laststart[2]
3084 insert_space(datasize, laststart + offset, b - 1);
3086 /* Write the equivalence_class and \0. */
3087 for (i = 0 ; i < datasize ; i++)
3088 laststart[offset + i] = str[i];
3090 /* Update the length of equivalence_classes. */
3091 laststart[3] += datasize;
3092 had_char_class = true;
3094 else /* delim == '.' */
3095 { /* collating symbol */
3096 /* Calculate the offset of the equivalence_classes,
3097 which is next to collating_symbols. */
3098 offset = laststart[1] + laststart[2] + 6;
3099 /* Insert space and write the collationg_symbol
3101 insert_space(datasize, laststart + offset, b-1);
3102 for (i = 0 ; i < datasize ; i++)
3103 laststart[offset + i] = str[i];
3105 /* In re_match_2_internal if range_start < -1, we
3106 assume -range_start is the offset of the
3107 collating symbol which is specified as
3108 the character of the range start. So we assign
3109 -(laststart[1] + laststart[2] + 6) to
3111 range_start = -(laststart[1] + laststart[2] + 6);
3112 /* Update the length of collating_symbol. */
3113 laststart[2] += datasize;
3114 had_char_class = false;
3124 laststart[5] += 2; /* Update the length of characters */
3125 range_start = delim;
3126 had_char_class = false;
3131 had_char_class = false;
3133 laststart[5]++; /* Update the length of characters */
3139 /* Ensure that we have enough space to push a charset: the
3140 opcode, the length count, and the bitset; 34 bytes in all. */
3141 GET_BUFFER_SPACE (34);
3145 /* We test `*p == '^' twice, instead of using an if
3146 statement, so we only need one BUF_PUSH. */
3147 BUF_PUSH (*p == '^' ? charset_not : charset);
3151 /* Remember the first position in the bracket expression. */
3154 /* Push the number of bytes in the bitmap. */
3155 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
3157 /* Clear the whole map. */
3158 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
3160 /* charset_not matches newline according to a syntax bit. */
3161 if ((re_opcode_t) b[-2] == charset_not
3162 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
3163 SET_LIST_BIT ('\n');
3165 /* Read in characters and ranges, setting map bits. */
3168 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3172 /* \ might escape characters inside [...] and [^...]. */
3173 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
3175 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3183 /* Could be the end of the bracket expression. If it's
3184 not (i.e., when the bracket expression is `[]' so
3185 far), the ']' character bit gets set way below. */
3186 if (c == ']' && p != p1 + 1)
3189 /* Look ahead to see if it's a range when the last thing
3190 was a character class. */
3191 if (had_char_class && c == '-' && *p != ']')
3192 FREE_STACK_RETURN (REG_ERANGE);
3194 /* Look ahead to see if it's a range when the last thing
3195 was a character: if this is a hyphen not at the
3196 beginning or the end of a list, then it's the range
3199 && !(p - 2 >= pattern && p[-2] == '[')
3200 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
3204 = byte_compile_range (range_start, &p, pend, translate,
3206 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3207 range_start = 0xffffffff;
3210 else if (p[0] == '-' && p[1] != ']')
3211 { /* This handles ranges made up of characters only. */
3214 /* Move past the `-'. */
3217 ret = byte_compile_range (c, &p, pend, translate, syntax, b);
3218 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3219 range_start = 0xffffffff;
3222 /* See if we're at the beginning of a possible character
3225 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
3226 { /* Leave room for the null. */
3227 char str[CHAR_CLASS_MAX_LENGTH + 1];
3232 /* If pattern is `[[:'. */
3233 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3238 if ((c == ':' && *p == ']') || p == pend)
3240 if (c1 < CHAR_CLASS_MAX_LENGTH)
3243 /* This is in any case an invalid class name. */
3248 /* If isn't a word bracketed by `[:' and `:]':
3249 undo the ending character, the letters, and leave
3250 the leading `:' and `[' (but set bits for them). */
3251 if (c == ':' && *p == ']')
3253 # if defined _LIBC || WIDE_CHAR_SUPPORT
3254 boolean is_lower = STREQ (str, "lower");
3255 boolean is_upper = STREQ (str, "upper");
3259 wt = IS_CHAR_CLASS (str);
3261 FREE_STACK_RETURN (REG_ECTYPE);
3263 /* Throw away the ] at the end of the character
3267 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3269 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
3272 if (__iswctype (__btowc (ch), wt))
3275 if (iswctype (btowc (ch), wt))
3279 if (translate && (is_upper || is_lower)
3280 && (ISUPPER (ch) || ISLOWER (ch)))
3284 had_char_class = true;
3287 boolean is_alnum = STREQ (str, "alnum");
3288 boolean is_alpha = STREQ (str, "alpha");
3289 boolean is_blank = STREQ (str, "blank");
3290 boolean is_cntrl = STREQ (str, "cntrl");
3291 boolean is_digit = STREQ (str, "digit");
3292 boolean is_graph = STREQ (str, "graph");
3293 boolean is_lower = STREQ (str, "lower");
3294 boolean is_print = STREQ (str, "print");
3295 boolean is_punct = STREQ (str, "punct");
3296 boolean is_space = STREQ (str, "space");
3297 boolean is_upper = STREQ (str, "upper");
3298 boolean is_xdigit = STREQ (str, "xdigit");
3300 if (!IS_CHAR_CLASS (str))
3301 FREE_STACK_RETURN (REG_ECTYPE);
3303 /* Throw away the ] at the end of the character
3307 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3309 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
3311 /* This was split into 3 if's to
3312 avoid an arbitrary limit in some compiler. */
3313 if ( (is_alnum && ISALNUM (ch))
3314 || (is_alpha && ISALPHA (ch))
3315 || (is_blank && ISBLANK (ch))
3316 || (is_cntrl && ISCNTRL (ch)))
3318 if ( (is_digit && ISDIGIT (ch))
3319 || (is_graph && ISGRAPH (ch))
3320 || (is_lower && ISLOWER (ch))
3321 || (is_print && ISPRINT (ch)))
3323 if ( (is_punct && ISPUNCT (ch))
3324 || (is_space && ISSPACE (ch))
3325 || (is_upper && ISUPPER (ch))
3326 || (is_xdigit && ISXDIGIT (ch)))
3328 if ( translate && (is_upper || is_lower)
3329 && (ISUPPER (ch) || ISLOWER (ch)))
3332 had_char_class = true;
3333 # endif /* libc || wctype.h */
3343 had_char_class = false;
3346 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '=')
3348 unsigned char str[MB_LEN_MAX + 1];
3351 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3357 /* If pattern is `[[='. */
3358 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3363 if ((c == '=' && *p == ']') || p == pend)
3365 if (c1 < MB_LEN_MAX)
3368 /* This is in any case an invalid class name. */
3373 if (c == '=' && *p == ']' && str[0] != '\0')
3375 /* If we have no collation data we use the default
3376 collation in which each character is in a class
3377 by itself. It also means that ASCII is the
3378 character set and therefore we cannot have character
3379 with more than one byte in the multibyte
3386 FREE_STACK_RETURN (REG_ECOLLATE);
3388 /* Throw away the ] at the end of the equivalence
3392 /* Set the bit for the character. */
3393 SET_LIST_BIT (str[0]);
3398 /* Try to match the byte sequence in `str' against
3399 those known to the collate implementation.
3400 First find out whether the bytes in `str' are
3401 actually from exactly one character. */
3402 const int32_t *table;
3403 const unsigned char *weights;
3404 const unsigned char *extra;
3405 const int32_t *indirect;
3407 const unsigned char *cp = str;
3410 /* This #include defines a local function! */
3411 # include <locale/weight.h>
3413 table = (const int32_t *)
3414 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
3415 weights = (const unsigned char *)
3416 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
3417 extra = (const unsigned char *)
3418 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
3419 indirect = (const int32_t *)
3420 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
3422 idx = findidx (&cp);
3423 if (idx == 0 || cp < str + c1)
3424 /* This is no valid character. */
3425 FREE_STACK_RETURN (REG_ECOLLATE);
3427 /* Throw away the ] at the end of the equivalence
3431 /* Now we have to go throught the whole table
3432 and find all characters which have the same
3435 XXX Note that this is not entirely correct.
3436 we would have to match multibyte sequences
3437 but this is not possible with the current
3439 for (ch = 1; ch < 256; ++ch)
3440 /* XXX This test would have to be changed if we
3441 would allow matching multibyte sequences. */
3444 int32_t idx2 = table[ch];
3445 size_t len = weights[idx2];
3447 /* Test whether the lenghts match. */
3448 if (weights[idx] == len)
3450 /* They do. New compare the bytes of
3455 && (weights[idx + 1 + cnt]
3456 == weights[idx2 + 1 + cnt]))
3460 /* They match. Mark the character as
3467 had_char_class = true;
3477 had_char_class = false;
3480 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '.')
3482 unsigned char str[128]; /* Should be large enough. */
3485 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3491 /* If pattern is `[[.'. */
3492 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3497 if ((c == '.' && *p == ']') || p == pend)
3499 if (c1 < sizeof (str))
3502 /* This is in any case an invalid class name. */
3507 if (c == '.' && *p == ']' && str[0] != '\0')
3509 /* If we have no collation data we use the default
3510 collation in which each character is the name
3511 for its own class which contains only the one
3512 character. It also means that ASCII is the
3513 character set and therefore we cannot have character
3514 with more than one byte in the multibyte
3521 FREE_STACK_RETURN (REG_ECOLLATE);
3523 /* Throw away the ] at the end of the equivalence
3527 /* Set the bit for the character. */
3528 SET_LIST_BIT (str[0]);
3529 range_start = ((const unsigned char *) str)[0];
3534 /* Try to match the byte sequence in `str' against
3535 those known to the collate implementation.
3536 First find out whether the bytes in `str' are
3537 actually from exactly one character. */
3539 const int32_t *symb_table;
3540 const unsigned char *extra;
3547 _NL_CURRENT_WORD (LC_COLLATE,
3548 _NL_COLLATE_SYMB_HASH_SIZEMB);
3549 symb_table = (const int32_t *)
3550 _NL_CURRENT (LC_COLLATE,
3551 _NL_COLLATE_SYMB_TABLEMB);
3552 extra = (const unsigned char *)
3553 _NL_CURRENT (LC_COLLATE,
3554 _NL_COLLATE_SYMB_EXTRAMB);
3556 /* Locate the character in the hashing table. */
3557 hash = elem_hash (str, c1);
3560 elem = hash % table_size;
3561 second = hash % (table_size - 2);
3562 while (symb_table[2 * elem] != 0)
3564 /* First compare the hashing value. */
3565 if (symb_table[2 * elem] == hash
3566 && c1 == extra[symb_table[2 * elem + 1]]
3568 &extra[symb_table[2 * elem + 1]
3572 /* Yep, this is the entry. */
3573 idx = symb_table[2 * elem + 1];
3574 idx += 1 + extra[idx];
3582 if (symb_table[2 * elem] == 0)
3583 /* This is no valid character. */
3584 FREE_STACK_RETURN (REG_ECOLLATE);
3586 /* Throw away the ] at the end of the equivalence
3590 /* Now add the multibyte character(s) we found
3593 XXX Note that this is not entirely correct.
3594 we would have to match multibyte sequences
3595 but this is not possible with the current
3596 implementation. Also, we have to match
3597 collating symbols, which expand to more than
3598 one file, as a whole and not allow the
3599 individual bytes. */
3602 range_start = extra[idx];
3605 SET_LIST_BIT (extra[idx]);
3610 had_char_class = false;
3620 had_char_class = false;
3625 had_char_class = false;
3631 /* Discard any (non)matching list bytes that are all 0 at the
3632 end of the map. Decrease the map-length byte too. */
3633 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
3642 if (syntax & RE_NO_BK_PARENS)
3649 if (syntax & RE_NO_BK_PARENS)
3656 if (syntax & RE_NEWLINE_ALT)
3663 if (syntax & RE_NO_BK_VBAR)
3670 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
3671 goto handle_interval;
3677 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3679 /* Do not translate the character after the \, so that we can
3680 distinguish, e.g., \B from \b, even if we normally would
3681 translate, e.g., B to b. */
3687 if (syntax & RE_NO_BK_PARENS)
3688 goto normal_backslash;
3694 if (COMPILE_STACK_FULL)
3696 RETALLOC (compile_stack.stack, compile_stack.size << 1,
3697 compile_stack_elt_t);
3698 if (compile_stack.stack == NULL) return REG_ESPACE;
3700 compile_stack.size <<= 1;
3703 /* These are the values to restore when we hit end of this
3704 group. They are all relative offsets, so that if the
3705 whole pattern moves because of realloc, they will still
3707 COMPILE_STACK_TOP.begalt_offset = begalt - COMPILED_BUFFER_VAR;
3708 COMPILE_STACK_TOP.fixup_alt_jump
3709 = fixup_alt_jump ? fixup_alt_jump - COMPILED_BUFFER_VAR + 1 : 0;
3710 COMPILE_STACK_TOP.laststart_offset = b - COMPILED_BUFFER_VAR;
3711 COMPILE_STACK_TOP.regnum = regnum;
3713 /* We will eventually replace the 0 with the number of
3714 groups inner to this one. But do not push a
3715 start_memory for groups beyond the last one we can
3716 represent in the compiled pattern. */
3717 if (regnum <= MAX_REGNUM)
3719 COMPILE_STACK_TOP.inner_group_offset = b
3720 - COMPILED_BUFFER_VAR + 2;
3721 BUF_PUSH_3 (start_memory, regnum, 0);
3724 compile_stack.avail++;
3729 /* If we've reached MAX_REGNUM groups, then this open
3730 won't actually generate any code, so we'll have to
3731 clear pending_exact explicitly. */
3737 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
3739 if (COMPILE_STACK_EMPTY)
3741 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3742 goto normal_backslash;
3744 FREE_STACK_RETURN (REG_ERPAREN);
3749 { /* Push a dummy failure point at the end of the
3750 alternative for a possible future
3751 `pop_failure_jump' to pop. See comments at
3752 `push_dummy_failure' in `re_match_2'. */
3753 BUF_PUSH (push_dummy_failure);
3755 /* We allocated space for this jump when we assigned
3756 to `fixup_alt_jump', in the `handle_alt' case below. */
3757 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
3760 /* See similar code for backslashed left paren above. */
3761 if (COMPILE_STACK_EMPTY)
3763 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3766 FREE_STACK_RETURN (REG_ERPAREN);
3769 /* Since we just checked for an empty stack above, this
3770 ``can't happen''. */
3771 assert (compile_stack.avail != 0);
3773 /* We don't just want to restore into `regnum', because
3774 later groups should continue to be numbered higher,
3775 as in `(ab)c(de)' -- the second group is #2. */
3776 regnum_t this_group_regnum;
3778 compile_stack.avail--;
3779 begalt = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.begalt_offset;
3781 = COMPILE_STACK_TOP.fixup_alt_jump
3782 ? COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.fixup_alt_jump - 1
3784 laststart = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.laststart_offset;
3785 this_group_regnum = COMPILE_STACK_TOP.regnum;
3786 /* If we've reached MAX_REGNUM groups, then this open
3787 won't actually generate any code, so we'll have to
3788 clear pending_exact explicitly. */
3791 /* We're at the end of the group, so now we know how many
3792 groups were inside this one. */
3793 if (this_group_regnum <= MAX_REGNUM)
3795 UCHAR_T *inner_group_loc
3796 = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.inner_group_offset;
3798 *inner_group_loc = regnum - this_group_regnum;
3799 BUF_PUSH_3 (stop_memory, this_group_regnum,
3800 regnum - this_group_regnum);
3806 case '|': /* `\|'. */
3807 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
3808 goto normal_backslash;
3810 if (syntax & RE_LIMITED_OPS)
3813 /* Insert before the previous alternative a jump which
3814 jumps to this alternative if the former fails. */
3815 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3816 INSERT_JUMP (on_failure_jump, begalt,
3817 b + 2 + 2 * OFFSET_ADDRESS_SIZE);
3819 b += 1 + OFFSET_ADDRESS_SIZE;
3821 /* The alternative before this one has a jump after it
3822 which gets executed if it gets matched. Adjust that
3823 jump so it will jump to this alternative's analogous
3824 jump (put in below, which in turn will jump to the next
3825 (if any) alternative's such jump, etc.). The last such
3826 jump jumps to the correct final destination. A picture:
3832 If we are at `b', then fixup_alt_jump right now points to a
3833 three-byte space after `a'. We'll put in the jump, set
3834 fixup_alt_jump to right after `b', and leave behind three
3835 bytes which we'll fill in when we get to after `c'. */
3838 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
3840 /* Mark and leave space for a jump after this alternative,
3841 to be filled in later either by next alternative or
3842 when know we're at the end of a series of alternatives. */
3844 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3845 b += 1 + OFFSET_ADDRESS_SIZE;
3853 /* If \{ is a literal. */
3854 if (!(syntax & RE_INTERVALS)
3855 /* If we're at `\{' and it's not the open-interval
3857 || (syntax & RE_NO_BK_BRACES))
3858 goto normal_backslash;
3862 /* If got here, then the syntax allows intervals. */
3864 /* At least (most) this many matches must be made. */
3865 int lower_bound = -1, upper_bound = -1;
3867 /* Place in the uncompiled pattern (i.e., just after
3868 the '{') to go back to if the interval is invalid. */
3869 const CHAR_T *beg_interval = p;
3872 goto invalid_interval;
3874 GET_UNSIGNED_NUMBER (lower_bound);
3878 GET_UNSIGNED_NUMBER (upper_bound);
3879 if (upper_bound < 0)
3880 upper_bound = RE_DUP_MAX;
3883 /* Interval such as `{1}' => match exactly once. */
3884 upper_bound = lower_bound;
3886 if (! (0 <= lower_bound && lower_bound <= upper_bound))
3887 goto invalid_interval;
3889 if (!(syntax & RE_NO_BK_BRACES))
3891 if (c != '\\' || p == pend)
3892 goto invalid_interval;
3897 goto invalid_interval;
3899 /* If it's invalid to have no preceding re. */
3902 if (syntax & RE_CONTEXT_INVALID_OPS
3903 && !(syntax & RE_INVALID_INTERVAL_ORD))
3904 FREE_STACK_RETURN (REG_BADRPT);
3905 else if (syntax & RE_CONTEXT_INDEP_OPS)
3908 goto unfetch_interval;
3911 /* We just parsed a valid interval. */
3913 if (RE_DUP_MAX < upper_bound)
3914 FREE_STACK_RETURN (REG_BADBR);
3916 /* If the upper bound is zero, don't want to succeed at
3917 all; jump from `laststart' to `b + 3', which will be
3918 the end of the buffer after we insert the jump. */
3919 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3920 instead of 'b + 3'. */
3921 if (upper_bound == 0)
3923 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3924 INSERT_JUMP (jump, laststart, b + 1
3925 + OFFSET_ADDRESS_SIZE);
3926 b += 1 + OFFSET_ADDRESS_SIZE;
3929 /* Otherwise, we have a nontrivial interval. When
3930 we're all done, the pattern will look like:
3931 set_number_at <jump count> <upper bound>
3932 set_number_at <succeed_n count> <lower bound>
3933 succeed_n <after jump addr> <succeed_n count>
3935 jump_n <succeed_n addr> <jump count>
3936 (The upper bound and `jump_n' are omitted if
3937 `upper_bound' is 1, though.) */
3939 { /* If the upper bound is > 1, we need to insert
3940 more at the end of the loop. */
3941 unsigned nbytes = 2 + 4 * OFFSET_ADDRESS_SIZE +
3942 (upper_bound > 1) * (2 + 4 * OFFSET_ADDRESS_SIZE);
3944 GET_BUFFER_SPACE (nbytes);
3946 /* Initialize lower bound of the `succeed_n', even
3947 though it will be set during matching by its
3948 attendant `set_number_at' (inserted next),
3949 because `re_compile_fastmap' needs to know.
3950 Jump to the `jump_n' we might insert below. */
3951 INSERT_JUMP2 (succeed_n, laststart,
3952 b + 1 + 2 * OFFSET_ADDRESS_SIZE
3953 + (upper_bound > 1) * (1 + 2 * OFFSET_ADDRESS_SIZE)
3955 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3957 /* Code to initialize the lower bound. Insert
3958 before the `succeed_n'. The `5' is the last two
3959 bytes of this `set_number_at', plus 3 bytes of
3960 the following `succeed_n'. */
3961 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3962 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3963 of the following `succeed_n'. */
3964 PREFIX(insert_op2) (set_number_at, laststart, 1
3965 + 2 * OFFSET_ADDRESS_SIZE, lower_bound, b);
3966 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3968 if (upper_bound > 1)
3969 { /* More than one repetition is allowed, so
3970 append a backward jump to the `succeed_n'
3971 that starts this interval.
3973 When we've reached this during matching,
3974 we'll have matched the interval once, so
3975 jump back only `upper_bound - 1' times. */
3976 STORE_JUMP2 (jump_n, b, laststart
3977 + 2 * OFFSET_ADDRESS_SIZE + 1,
3979 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3981 /* The location we want to set is the second
3982 parameter of the `jump_n'; that is `b-2' as
3983 an absolute address. `laststart' will be
3984 the `set_number_at' we're about to insert;
3985 `laststart+3' the number to set, the source
3986 for the relative address. But we are
3987 inserting into the middle of the pattern --
3988 so everything is getting moved up by 5.
3989 Conclusion: (b - 2) - (laststart + 3) + 5,
3990 i.e., b - laststart.
3992 We insert this at the beginning of the loop
3993 so that if we fail during matching, we'll
3994 reinitialize the bounds. */
3995 PREFIX(insert_op2) (set_number_at, laststart,
3997 upper_bound - 1, b);
3998 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4005 if (!(syntax & RE_INVALID_INTERVAL_ORD))
4006 FREE_STACK_RETURN (p == pend ? REG_EBRACE : REG_BADBR);
4008 /* Match the characters as literals. */
4011 if (syntax & RE_NO_BK_BRACES)
4014 goto normal_backslash;
4018 /* There is no way to specify the before_dot and after_dot
4019 operators. rms says this is ok. --karl */
4027 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
4033 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
4039 if (syntax & RE_NO_GNU_OPS)
4042 BUF_PUSH (wordchar);
4047 if (syntax & RE_NO_GNU_OPS)
4050 BUF_PUSH (notwordchar);
4055 if (syntax & RE_NO_GNU_OPS)
4061 if (syntax & RE_NO_GNU_OPS)
4067 if (syntax & RE_NO_GNU_OPS)
4069 BUF_PUSH (wordbound);
4073 if (syntax & RE_NO_GNU_OPS)
4075 BUF_PUSH (notwordbound);
4079 if (syntax & RE_NO_GNU_OPS)
4085 if (syntax & RE_NO_GNU_OPS)
4090 case '1': case '2': case '3': case '4': case '5':
4091 case '6': case '7': case '8': case '9':
4092 if (syntax & RE_NO_BK_REFS)
4098 FREE_STACK_RETURN (REG_ESUBREG);
4100 /* Can't back reference to a subexpression if inside of it. */
4101 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
4105 BUF_PUSH_2 (duplicate, c1);
4111 if (syntax & RE_BK_PLUS_QM)
4114 goto normal_backslash;
4118 /* You might think it would be useful for \ to mean
4119 not to translate; but if we don't translate it
4120 it will never match anything. */
4128 /* Expects the character in `c'. */
4130 /* If no exactn currently being built. */
4133 /* If last exactn handle binary(or character) and
4134 new exactn handle character(or binary). */
4135 || is_exactn_bin != is_binary[p - 1 - pattern]
4138 /* If last exactn not at current position. */
4139 || pending_exact + *pending_exact + 1 != b
4141 /* We have only one byte following the exactn for the count. */
4142 || *pending_exact == (1 << BYTEWIDTH) - 1
4144 /* If followed by a repetition operator. */
4145 || *p == '*' || *p == '^'
4146 || ((syntax & RE_BK_PLUS_QM)
4147 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
4148 : (*p == '+' || *p == '?'))
4149 || ((syntax & RE_INTERVALS)
4150 && ((syntax & RE_NO_BK_BRACES)
4152 : (p[0] == '\\' && p[1] == '{'))))
4154 /* Start building a new exactn. */
4159 /* Is this exactn binary data or character? */
4160 is_exactn_bin = is_binary[p - 1 - pattern];
4162 BUF_PUSH_2 (exactn_bin, 0);
4164 BUF_PUSH_2 (exactn, 0);
4166 BUF_PUSH_2 (exactn, 0);
4168 pending_exact = b - 1;
4175 } /* while p != pend */
4178 /* Through the pattern now. */
4181 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
4183 if (!COMPILE_STACK_EMPTY)
4184 FREE_STACK_RETURN (REG_EPAREN);
4186 /* If we don't want backtracking, force success
4187 the first time we reach the end of the compiled pattern. */
4188 if (syntax & RE_NO_POSIX_BACKTRACKING)
4196 free (compile_stack.stack);
4198 /* We have succeeded; set the length of the buffer. */
4200 bufp->used = (uintptr_t) b - (uintptr_t) COMPILED_BUFFER_VAR;
4202 bufp->used = b - bufp->buffer;
4208 DEBUG_PRINT1 ("\nCompiled pattern: \n");
4209 PREFIX(print_compiled_pattern) (bufp);
4213 #ifndef MATCH_MAY_ALLOCATE
4214 /* Initialize the failure stack to the largest possible stack. This
4215 isn't necessary unless we're trying to avoid calling alloca in
4216 the search and match routines. */
4218 int num_regs = bufp->re_nsub + 1;
4220 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4221 is strictly greater than re_max_failures, the largest possible stack
4222 is 2 * re_max_failures failure points. */
4223 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
4225 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
4228 if (! fail_stack.stack)
4230 = (PREFIX(fail_stack_elt_t) *) xmalloc (fail_stack.size
4231 * sizeof (PREFIX(fail_stack_elt_t)));
4234 = (PREFIX(fail_stack_elt_t) *) xrealloc (fail_stack.stack,
4236 * sizeof (PREFIX(fail_stack_elt_t))));
4237 # else /* not emacs */
4238 if (! fail_stack.stack)
4240 = (PREFIX(fail_stack_elt_t) *) malloc (fail_stack.size
4241 * sizeof (PREFIX(fail_stack_elt_t)));
4244 = (PREFIX(fail_stack_elt_t) *) realloc (fail_stack.stack,
4246 * sizeof (PREFIX(fail_stack_elt_t))));
4247 # endif /* not emacs */
4250 PREFIX(regex_grow_registers) (num_regs);
4252 #endif /* not MATCH_MAY_ALLOCATE */
4255 } /* regex_compile */
4257 /* Subroutines for `regex_compile'. */
4259 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4260 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4263 PREFIX(store_op1) (op, loc, arg)
4268 *loc = (UCHAR_T) op;
4269 STORE_NUMBER (loc + 1, arg);
4273 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4274 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4277 PREFIX(store_op2) (op, loc, arg1, arg2)
4282 *loc = (UCHAR_T) op;
4283 STORE_NUMBER (loc + 1, arg1);
4284 STORE_NUMBER (loc + 1 + OFFSET_ADDRESS_SIZE, arg2);
4288 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4289 for OP followed by two-byte integer parameter ARG. */
4290 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4293 PREFIX(insert_op1) (op, loc, arg, end)
4299 register UCHAR_T *pfrom = end;
4300 register UCHAR_T *pto = end + 1 + OFFSET_ADDRESS_SIZE;
4302 while (pfrom != loc)
4305 PREFIX(store_op1) (op, loc, arg);
4309 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4310 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4313 PREFIX(insert_op2) (op, loc, arg1, arg2, end)
4319 register UCHAR_T *pfrom = end;
4320 register UCHAR_T *pto = end + 1 + 2 * OFFSET_ADDRESS_SIZE;
4322 while (pfrom != loc)
4325 PREFIX(store_op2) (op, loc, arg1, arg2);
4329 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4330 after an alternative or a begin-subexpression. We assume there is at
4331 least one character before the ^. */
4334 PREFIX(at_begline_loc_p) (pattern, p, syntax)
4335 const CHAR_T *pattern, *p;
4336 reg_syntax_t syntax;
4338 const CHAR_T *prev = p - 2;
4339 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
4342 /* After a subexpression? */
4343 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
4344 /* After an alternative? */
4345 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
4349 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4350 at least one character after the $, i.e., `P < PEND'. */
4353 PREFIX(at_endline_loc_p) (p, pend, syntax)
4354 const CHAR_T *p, *pend;
4355 reg_syntax_t syntax;
4357 const CHAR_T *next = p;
4358 boolean next_backslash = *next == '\\';
4359 const CHAR_T *next_next = p + 1 < pend ? p + 1 : 0;
4362 /* Before a subexpression? */
4363 (syntax & RE_NO_BK_PARENS ? *next == ')'
4364 : next_backslash && next_next && *next_next == ')')
4365 /* Before an alternative? */
4366 || (syntax & RE_NO_BK_VBAR ? *next == '|'
4367 : next_backslash && next_next && *next_next == '|');
4370 #else /* not INSIDE_RECURSION */
4372 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4373 false if it's not. */
4376 group_in_compile_stack (compile_stack, regnum)
4377 compile_stack_type compile_stack;
4382 for (this_element = compile_stack.avail - 1;
4385 if (compile_stack.stack[this_element].regnum == regnum)
4390 #endif /* not INSIDE_RECURSION */
4392 #ifdef INSIDE_RECURSION
4395 /* This insert space, which size is "num", into the pattern at "loc".
4396 "end" must point the end of the allocated buffer. */
4398 insert_space (num, loc, end)
4403 register CHAR_T *pto = end;
4404 register CHAR_T *pfrom = end - num;
4406 while (pfrom >= loc)
4412 static reg_errcode_t
4413 wcs_compile_range (range_start_char, p_ptr, pend, translate, syntax, b,
4415 CHAR_T range_start_char;
4416 const CHAR_T **p_ptr, *pend;
4417 CHAR_T *char_set, *b;
4418 RE_TRANSLATE_TYPE translate;
4419 reg_syntax_t syntax;
4421 const CHAR_T *p = *p_ptr;
4422 CHAR_T range_start, range_end;
4426 uint32_t start_val, end_val;
4432 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
4435 const char *collseq = (const char *) _NL_CURRENT(LC_COLLATE,
4436 _NL_COLLATE_COLLSEQWC);
4437 const unsigned char *extra = (const unsigned char *)
4438 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
4440 if (range_start_char < -1)
4442 /* range_start is a collating symbol. */
4444 /* Retreive the index and get collation sequence value. */
4445 wextra = (int32_t*)(extra + char_set[-range_start_char]);
4446 start_val = wextra[1 + *wextra];
4449 start_val = collseq_table_lookup(collseq, TRANSLATE(range_start_char));
4451 end_val = collseq_table_lookup (collseq, TRANSLATE (p[0]));
4453 /* Report an error if the range is empty and the syntax prohibits
4455 ret = ((syntax & RE_NO_EMPTY_RANGES)
4456 && (start_val > end_val))? REG_ERANGE : REG_NOERROR;
4458 /* Insert space to the end of the char_ranges. */
4459 insert_space(2, b - char_set[5] - 2, b - 1);
4460 *(b - char_set[5] - 2) = (wchar_t)start_val;
4461 *(b - char_set[5] - 1) = (wchar_t)end_val;
4462 char_set[4]++; /* ranges_index */
4467 range_start = (range_start_char >= 0)? TRANSLATE (range_start_char):
4469 range_end = TRANSLATE (p[0]);
4470 /* Report an error if the range is empty and the syntax prohibits
4472 ret = ((syntax & RE_NO_EMPTY_RANGES)
4473 && (range_start > range_end))? REG_ERANGE : REG_NOERROR;
4475 /* Insert space to the end of the char_ranges. */
4476 insert_space(2, b - char_set[5] - 2, b - 1);
4477 *(b - char_set[5] - 2) = range_start;
4478 *(b - char_set[5] - 1) = range_end;
4479 char_set[4]++; /* ranges_index */
4481 /* Have to increment the pointer into the pattern string, so the
4482 caller isn't still at the ending character. */
4488 /* Read the ending character of a range (in a bracket expression) from the
4489 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4490 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4491 Then we set the translation of all bits between the starting and
4492 ending characters (inclusive) in the compiled pattern B.
4494 Return an error code.
4496 We use these short variable names so we can use the same macros as
4497 `regex_compile' itself. */
4499 static reg_errcode_t
4500 byte_compile_range (range_start_char, p_ptr, pend, translate, syntax, b)
4501 unsigned int range_start_char;
4502 const char **p_ptr, *pend;
4503 RE_TRANSLATE_TYPE translate;
4504 reg_syntax_t syntax;
4508 const char *p = *p_ptr;
4511 const unsigned char *collseq;
4512 unsigned int start_colseq;
4513 unsigned int end_colseq;
4521 /* Have to increment the pointer into the pattern string, so the
4522 caller isn't still at the ending character. */
4525 /* Report an error if the range is empty and the syntax prohibits this. */
4526 ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
4529 collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
4530 _NL_COLLATE_COLLSEQMB);
4532 start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
4533 end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
4534 for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
4536 unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
4538 if (start_colseq <= this_colseq && this_colseq <= end_colseq)
4540 SET_LIST_BIT (TRANSLATE (this_char));
4545 /* Here we see why `this_char' has to be larger than an `unsigned
4546 char' -- we would otherwise go into an infinite loop, since all
4547 characters <= 0xff. */
4548 range_start_char = TRANSLATE (range_start_char);
4549 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4550 and some compilers cast it to int implicitly, so following for_loop
4551 may fall to (almost) infinite loop.
4552 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4553 To avoid this, we cast p[0] to unsigned int and truncate it. */
4554 end_char = ((unsigned)TRANSLATE(p[0]) & ((1 << BYTEWIDTH) - 1));
4556 for (this_char = range_start_char; this_char <= end_char; ++this_char)
4558 SET_LIST_BIT (TRANSLATE (this_char));
4567 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4568 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4569 characters can start a string that matches the pattern. This fastmap
4570 is used by re_search to skip quickly over impossible starting points.
4572 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4573 area as BUFP->fastmap.
4575 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4578 Returns 0 if we succeed, -2 if an internal error. */
4581 /* local function for re_compile_fastmap.
4582 truncate wchar_t character to char. */
4583 static unsigned char truncate_wchar (CHAR_T c);
4585 static unsigned char
4589 unsigned char buf[MB_CUR_MAX];
4592 memset (&state, '\0', sizeof (state));
4594 retval = __wcrtomb (buf, c, &state);
4596 retval = wcrtomb (buf, c, &state);
4598 return retval > 0 ? buf[0] : (unsigned char) c;
4603 PREFIX(re_compile_fastmap) (bufp)
4604 struct re_pattern_buffer *bufp;
4607 #ifdef MATCH_MAY_ALLOCATE
4608 PREFIX(fail_stack_type) fail_stack;
4610 #ifndef REGEX_MALLOC
4614 register char *fastmap = bufp->fastmap;
4617 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4618 pattern to (char*) in regex_compile. */
4619 UCHAR_T *pattern = (UCHAR_T*)bufp->buffer;
4620 register UCHAR_T *pend = (UCHAR_T*) (bufp->buffer + bufp->used);
4622 UCHAR_T *pattern = bufp->buffer;
4623 register UCHAR_T *pend = pattern + bufp->used;
4625 UCHAR_T *p = pattern;
4628 /* This holds the pointer to the failure stack, when
4629 it is allocated relocatably. */
4630 fail_stack_elt_t *failure_stack_ptr;
4633 /* Assume that each path through the pattern can be null until
4634 proven otherwise. We set this false at the bottom of switch
4635 statement, to which we get only if a particular path doesn't
4636 match the empty string. */
4637 boolean path_can_be_null = true;
4639 /* We aren't doing a `succeed_n' to begin with. */
4640 boolean succeed_n_p = false;
4642 assert (fastmap != NULL && p != NULL);
4645 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
4646 bufp->fastmap_accurate = 1; /* It will be when we're done. */
4647 bufp->can_be_null = 0;
4651 if (p == pend || *p == (UCHAR_T) succeed)
4653 /* We have reached the (effective) end of pattern. */
4654 if (!FAIL_STACK_EMPTY ())
4656 bufp->can_be_null |= path_can_be_null;
4658 /* Reset for next path. */
4659 path_can_be_null = true;
4661 p = fail_stack.stack[--fail_stack.avail].pointer;
4669 /* We should never be about to go beyond the end of the pattern. */
4672 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4675 /* I guess the idea here is to simply not bother with a fastmap
4676 if a backreference is used, since it's too hard to figure out
4677 the fastmap for the corresponding group. Setting
4678 `can_be_null' stops `re_search_2' from using the fastmap, so
4679 that is all we do. */
4681 bufp->can_be_null = 1;
4685 /* Following are the cases which match a character. These end
4690 fastmap[truncate_wchar(p[1])] = 1;
4704 /* It is hard to distinguish fastmap from (multi byte) characters
4705 which depends on current locale. */
4710 bufp->can_be_null = 1;
4714 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4715 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
4721 /* Chars beyond end of map must be allowed. */
4722 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
4725 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4726 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
4732 for (j = 0; j < (1 << BYTEWIDTH); j++)
4733 if (SYNTAX (j) == Sword)
4739 for (j = 0; j < (1 << BYTEWIDTH); j++)
4740 if (SYNTAX (j) != Sword)
4747 int fastmap_newline = fastmap['\n'];
4749 /* `.' matches anything ... */
4750 for (j = 0; j < (1 << BYTEWIDTH); j++)
4753 /* ... except perhaps newline. */
4754 if (!(bufp->syntax & RE_DOT_NEWLINE))
4755 fastmap['\n'] = fastmap_newline;
4757 /* Return if we have already set `can_be_null'; if we have,
4758 then the fastmap is irrelevant. Something's wrong here. */
4759 else if (bufp->can_be_null)
4762 /* Otherwise, have to check alternative paths. */
4769 for (j = 0; j < (1 << BYTEWIDTH); j++)
4770 if (SYNTAX (j) == (enum syntaxcode) k)
4777 for (j = 0; j < (1 << BYTEWIDTH); j++)
4778 if (SYNTAX (j) != (enum syntaxcode) k)
4783 /* All cases after this match the empty string. These end with
4803 case push_dummy_failure:
4808 case pop_failure_jump:
4809 case maybe_pop_jump:
4812 case dummy_failure_jump:
4813 EXTRACT_NUMBER_AND_INCR (j, p);
4818 /* Jump backward implies we just went through the body of a
4819 loop and matched nothing. Opcode jumped to should be
4820 `on_failure_jump' or `succeed_n'. Just treat it like an
4821 ordinary jump. For a * loop, it has pushed its failure
4822 point already; if so, discard that as redundant. */
4823 if ((re_opcode_t) *p != on_failure_jump
4824 && (re_opcode_t) *p != succeed_n)
4828 EXTRACT_NUMBER_AND_INCR (j, p);
4831 /* If what's on the stack is where we are now, pop it. */
4832 if (!FAIL_STACK_EMPTY ()
4833 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
4839 case on_failure_jump:
4840 case on_failure_keep_string_jump:
4841 handle_on_failure_jump:
4842 EXTRACT_NUMBER_AND_INCR (j, p);
4844 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4845 end of the pattern. We don't want to push such a point,
4846 since when we restore it above, entering the switch will
4847 increment `p' past the end of the pattern. We don't need
4848 to push such a point since we obviously won't find any more
4849 fastmap entries beyond `pend'. Such a pattern can match
4850 the null string, though. */
4853 if (!PUSH_PATTERN_OP (p + j, fail_stack))
4855 RESET_FAIL_STACK ();
4860 bufp->can_be_null = 1;
4864 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
4865 succeed_n_p = false;
4872 /* Get to the number of times to succeed. */
4873 p += OFFSET_ADDRESS_SIZE;
4875 /* Increment p past the n for when k != 0. */
4876 EXTRACT_NUMBER_AND_INCR (k, p);
4879 p -= 2 * OFFSET_ADDRESS_SIZE;
4880 succeed_n_p = true; /* Spaghetti code alert. */
4881 goto handle_on_failure_jump;
4887 p += 2 * OFFSET_ADDRESS_SIZE;
4898 abort (); /* We have listed all the cases. */
4901 /* Getting here means we have found the possible starting
4902 characters for one path of the pattern -- and that the empty
4903 string does not match. We need not follow this path further.
4904 Instead, look at the next alternative (remembered on the
4905 stack), or quit if no more. The test at the top of the loop
4906 does these things. */
4907 path_can_be_null = false;
4911 /* Set `can_be_null' for the last path (also the first path, if the
4912 pattern is empty). */
4913 bufp->can_be_null |= path_can_be_null;
4916 RESET_FAIL_STACK ();
4920 #else /* not INSIDE_RECURSION */
4923 re_compile_fastmap (bufp)
4924 struct re_pattern_buffer *bufp;
4927 if (MB_CUR_MAX != 1)
4928 return wcs_re_compile_fastmap(bufp);
4931 return byte_re_compile_fastmap(bufp);
4932 } /* re_compile_fastmap */
4934 weak_alias (__re_compile_fastmap, re_compile_fastmap)
4938 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4939 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4940 this memory for recording register information. STARTS and ENDS
4941 must be allocated using the malloc library routine, and must each
4942 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4944 If NUM_REGS == 0, then subsequent matches should allocate their own
4947 Unless this function is called, the first search or match using
4948 PATTERN_BUFFER will allocate its own register data, without
4949 freeing the old data. */
4952 re_set_registers (bufp, regs, num_regs, starts, ends)
4953 struct re_pattern_buffer *bufp;
4954 struct re_registers *regs;
4956 regoff_t *starts, *ends;
4960 bufp->regs_allocated = REGS_REALLOCATE;
4961 regs->num_regs = num_regs;
4962 regs->start = starts;
4967 bufp->regs_allocated = REGS_UNALLOCATED;
4969 regs->start = regs->end = (regoff_t *) 0;
4973 weak_alias (__re_set_registers, re_set_registers)
4976 /* Searching routines. */
4978 /* Like re_search_2, below, but only one string is specified, and
4979 doesn't let you say where to stop matching. */
4982 re_search (bufp, string, size, startpos, range, regs)
4983 struct re_pattern_buffer *bufp;
4985 int size, startpos, range;
4986 struct re_registers *regs;
4988 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
4992 weak_alias (__re_search, re_search)
4996 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4997 virtual concatenation of STRING1 and STRING2, starting first at index
4998 STARTPOS, then at STARTPOS + 1, and so on.
5000 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
5002 RANGE is how far to scan while trying to match. RANGE = 0 means try
5003 only at STARTPOS; in general, the last start tried is STARTPOS +
5006 In REGS, return the indices of the virtual concatenation of STRING1
5007 and STRING2 that matched the entire BUFP->buffer and its contained
5010 Do not consider matching one past the index STOP in the virtual
5011 concatenation of STRING1 and STRING2.
5013 We return either the position in the strings at which the match was
5014 found, -1 if no match, or -2 if error (such as failure
5018 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
5019 struct re_pattern_buffer *bufp;
5020 const char *string1, *string2;
5024 struct re_registers *regs;
5028 if (MB_CUR_MAX != 1)
5029 return wcs_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5033 return byte_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5037 weak_alias (__re_search_2, re_search_2)
5040 #endif /* not INSIDE_RECURSION */
5042 #ifdef INSIDE_RECURSION
5044 #ifdef MATCH_MAY_ALLOCATE
5045 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
5047 # define FREE_VAR(var) if (var) free (var); var = NULL
5051 # define MAX_ALLOCA_SIZE 2000
5053 # define FREE_WCS_BUFFERS() \
5055 if (size1 > MAX_ALLOCA_SIZE) \
5057 free (wcs_string1); \
5058 free (mbs_offset1); \
5062 FREE_VAR (wcs_string1); \
5063 FREE_VAR (mbs_offset1); \
5065 if (size2 > MAX_ALLOCA_SIZE) \
5067 free (wcs_string2); \
5068 free (mbs_offset2); \
5072 FREE_VAR (wcs_string2); \
5073 FREE_VAR (mbs_offset2); \
5081 PREFIX(re_search_2) (bufp, string1, size1, string2, size2, startpos, range,
5083 struct re_pattern_buffer *bufp;
5084 const char *string1, *string2;
5088 struct re_registers *regs;
5092 register char *fastmap = bufp->fastmap;
5093 register RE_TRANSLATE_TYPE translate = bufp->translate;
5094 int total_size = size1 + size2;
5095 int endpos = startpos + range;
5097 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5098 wchar_t *wcs_string1 = NULL, *wcs_string2 = NULL;
5099 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5100 int wcs_size1 = 0, wcs_size2 = 0;
5101 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5102 int *mbs_offset1 = NULL, *mbs_offset2 = NULL;
5103 /* They hold whether each wchar_t is binary data or not. */
5104 char *is_binary = NULL;
5107 /* Check for out-of-range STARTPOS. */
5108 if (startpos < 0 || startpos > total_size)
5111 /* Fix up RANGE if it might eventually take us outside
5112 the virtual concatenation of STRING1 and STRING2.
5113 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5115 range = 0 - startpos;
5116 else if (endpos > total_size)
5117 range = total_size - startpos;
5119 /* If the search isn't to be a backwards one, don't waste time in a
5120 search for a pattern that must be anchored. */
5121 if (bufp->used > 0 && range > 0
5122 && ((re_opcode_t) bufp->buffer[0] == begbuf
5123 /* `begline' is like `begbuf' if it cannot match at newlines. */
5124 || ((re_opcode_t) bufp->buffer[0] == begline
5125 && !bufp->newline_anchor)))
5134 /* In a forward search for something that starts with \=.
5135 don't keep searching past point. */
5136 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
5138 range = PT - startpos;
5144 /* Update the fastmap now if not correct already. */
5145 if (fastmap && !bufp->fastmap_accurate)
5146 if (re_compile_fastmap (bufp) == -2)
5150 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5151 fill them with converted string. */
5154 if (size1 > MAX_ALLOCA_SIZE)
5156 wcs_string1 = TALLOC (size1 + 1, CHAR_T);
5157 mbs_offset1 = TALLOC (size1 + 1, int);
5158 is_binary = TALLOC (size1 + 1, char);
5162 wcs_string1 = REGEX_TALLOC (size1 + 1, CHAR_T);
5163 mbs_offset1 = REGEX_TALLOC (size1 + 1, int);
5164 is_binary = REGEX_TALLOC (size1 + 1, char);
5166 if (!wcs_string1 || !mbs_offset1 || !is_binary)
5168 if (size1 > MAX_ALLOCA_SIZE)
5176 FREE_VAR (wcs_string1);
5177 FREE_VAR (mbs_offset1);
5178 FREE_VAR (is_binary);
5182 wcs_size1 = convert_mbs_to_wcs(wcs_string1, string1, size1,
5183 mbs_offset1, is_binary);
5184 wcs_string1[wcs_size1] = L'\0'; /* for a sentinel */
5185 if (size1 > MAX_ALLOCA_SIZE)
5188 FREE_VAR (is_binary);
5192 if (size2 > MAX_ALLOCA_SIZE)
5194 wcs_string2 = TALLOC (size2 + 1, CHAR_T);
5195 mbs_offset2 = TALLOC (size2 + 1, int);
5196 is_binary = TALLOC (size2 + 1, char);
5200 wcs_string2 = REGEX_TALLOC (size2 + 1, CHAR_T);
5201 mbs_offset2 = REGEX_TALLOC (size2 + 1, int);
5202 is_binary = REGEX_TALLOC (size2 + 1, char);
5204 if (!wcs_string2 || !mbs_offset2 || !is_binary)
5206 FREE_WCS_BUFFERS ();
5207 if (size2 > MAX_ALLOCA_SIZE)
5210 FREE_VAR (is_binary);
5213 wcs_size2 = convert_mbs_to_wcs(wcs_string2, string2, size2,
5214 mbs_offset2, is_binary);
5215 wcs_string2[wcs_size2] = L'\0'; /* for a sentinel */
5216 if (size2 > MAX_ALLOCA_SIZE)
5219 FREE_VAR (is_binary);
5224 /* Loop through the string, looking for a place to start matching. */
5227 /* If a fastmap is supplied, skip quickly over characters that
5228 cannot be the start of a match. If the pattern can match the
5229 null string, however, we don't need to skip characters; we want
5230 the first null string. */
5231 if (fastmap && startpos < total_size && !bufp->can_be_null)
5233 if (range > 0) /* Searching forwards. */
5235 register const char *d;
5236 register int lim = 0;
5239 if (startpos < size1 && startpos + range >= size1)
5240 lim = range - (size1 - startpos);
5242 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
5244 /* Written out as an if-else to avoid testing `translate'
5248 && !fastmap[(unsigned char)
5249 translate[(unsigned char) *d++]])
5252 while (range > lim && !fastmap[(unsigned char) *d++])
5255 startpos += irange - range;
5257 else /* Searching backwards. */
5259 register CHAR_T c = (size1 == 0 || startpos >= size1
5260 ? string2[startpos - size1]
5261 : string1[startpos]);
5263 if (!fastmap[(unsigned char) TRANSLATE (c)])
5268 /* If can't match the null string, and that's all we have left, fail. */
5269 if (range >= 0 && startpos == total_size && fastmap
5270 && !bufp->can_be_null)
5273 FREE_WCS_BUFFERS ();
5279 val = wcs_re_match_2_internal (bufp, string1, size1, string2,
5280 size2, startpos, regs, stop,
5281 wcs_string1, wcs_size1,
5282 wcs_string2, wcs_size2,
5283 mbs_offset1, mbs_offset2);
5285 val = byte_re_match_2_internal (bufp, string1, size1, string2,
5286 size2, startpos, regs, stop);
5289 #ifndef REGEX_MALLOC
5298 FREE_WCS_BUFFERS ();
5306 FREE_WCS_BUFFERS ();
5326 FREE_WCS_BUFFERS ();
5332 /* This converts PTR, a pointer into one of the search wchar_t strings
5333 `string1' and `string2' into an multibyte string offset from the
5334 beginning of that string. We use mbs_offset to optimize.
5335 See convert_mbs_to_wcs. */
5336 # define POINTER_TO_OFFSET(ptr) \
5337 (FIRST_STRING_P (ptr) \
5338 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5339 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5342 /* This converts PTR, a pointer into one of the search strings `string1'
5343 and `string2' into an offset from the beginning of that string. */
5344 # define POINTER_TO_OFFSET(ptr) \
5345 (FIRST_STRING_P (ptr) \
5346 ? ((regoff_t) ((ptr) - string1)) \
5347 : ((regoff_t) ((ptr) - string2 + size1)))
5350 /* Macros for dealing with the split strings in re_match_2. */
5352 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5354 /* Call before fetching a character with *d. This switches over to
5355 string2 if necessary. */
5356 #define PREFETCH() \
5359 /* End of string2 => fail. */ \
5360 if (dend == end_match_2) \
5362 /* End of string1 => advance to string2. */ \
5364 dend = end_match_2; \
5367 /* Test if at very beginning or at very end of the virtual concatenation
5368 of `string1' and `string2'. If only one string, it's `string2'. */
5369 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5370 #define AT_STRINGS_END(d) ((d) == end2)
5373 /* Test if D points to a character which is word-constituent. We have
5374 two special cases to check for: if past the end of string1, look at
5375 the first character in string2; and if before the beginning of
5376 string2, look at the last character in string1. */
5378 /* Use internationalized API instead of SYNTAX. */
5379 # define WORDCHAR_P(d) \
5380 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5381 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5382 || ((d) == end1 ? *string2 \
5383 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) == L'_')
5385 # define WORDCHAR_P(d) \
5386 (SYNTAX ((d) == end1 ? *string2 \
5387 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5391 /* Disabled due to a compiler bug -- see comment at case wordbound */
5393 /* Test if the character before D and the one at D differ with respect
5394 to being word-constituent. */
5395 #define AT_WORD_BOUNDARY(d) \
5396 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5397 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5400 /* Free everything we malloc. */
5401 #ifdef MATCH_MAY_ALLOCATE
5403 # define FREE_VARIABLES() \
5405 REGEX_FREE_STACK (fail_stack.stack); \
5406 FREE_VAR (regstart); \
5407 FREE_VAR (regend); \
5408 FREE_VAR (old_regstart); \
5409 FREE_VAR (old_regend); \
5410 FREE_VAR (best_regstart); \
5411 FREE_VAR (best_regend); \
5412 FREE_VAR (reg_info); \
5413 FREE_VAR (reg_dummy); \
5414 FREE_VAR (reg_info_dummy); \
5415 if (!cant_free_wcs_buf) \
5417 FREE_VAR (string1); \
5418 FREE_VAR (string2); \
5419 FREE_VAR (mbs_offset1); \
5420 FREE_VAR (mbs_offset2); \
5424 # define FREE_VARIABLES() \
5426 REGEX_FREE_STACK (fail_stack.stack); \
5427 FREE_VAR (regstart); \
5428 FREE_VAR (regend); \
5429 FREE_VAR (old_regstart); \
5430 FREE_VAR (old_regend); \
5431 FREE_VAR (best_regstart); \
5432 FREE_VAR (best_regend); \
5433 FREE_VAR (reg_info); \
5434 FREE_VAR (reg_dummy); \
5435 FREE_VAR (reg_info_dummy); \
5440 # define FREE_VARIABLES() \
5442 if (!cant_free_wcs_buf) \
5444 FREE_VAR (string1); \
5445 FREE_VAR (string2); \
5446 FREE_VAR (mbs_offset1); \
5447 FREE_VAR (mbs_offset2); \
5451 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
5453 #endif /* not MATCH_MAY_ALLOCATE */
5455 /* These values must meet several constraints. They must not be valid
5456 register values; since we have a limit of 255 registers (because
5457 we use only one byte in the pattern for the register number), we can
5458 use numbers larger than 255. They must differ by 1, because of
5459 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5460 be larger than the value for the highest register, so we do not try
5461 to actually save any registers when none are active. */
5462 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5463 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5465 #else /* not INSIDE_RECURSION */
5466 /* Matching routines. */
5468 #ifndef emacs /* Emacs never uses this. */
5469 /* re_match is like re_match_2 except it takes only a single string. */
5472 re_match (bufp, string, size, pos, regs)
5473 struct re_pattern_buffer *bufp;
5476 struct re_registers *regs;
5480 if (MB_CUR_MAX != 1)
5481 result = wcs_re_match_2_internal (bufp, NULL, 0, string, size,
5483 NULL, 0, NULL, 0, NULL, NULL);
5486 result = byte_re_match_2_internal (bufp, NULL, 0, string, size,
5488 # ifndef REGEX_MALLOC
5496 weak_alias (__re_match, re_match)
5498 #endif /* not emacs */
5500 #endif /* not INSIDE_RECURSION */
5502 #ifdef INSIDE_RECURSION
5503 static boolean PREFIX(group_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5505 PREFIX(register_info_type) *reg_info));
5506 static boolean PREFIX(alt_match_null_string_p) _RE_ARGS ((UCHAR_T *p,
5508 PREFIX(register_info_type) *reg_info));
5509 static boolean PREFIX(common_op_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5511 PREFIX(register_info_type) *reg_info));
5512 static int PREFIX(bcmp_translate) _RE_ARGS ((const CHAR_T *s1, const CHAR_T *s2,
5513 int len, char *translate));
5514 #else /* not INSIDE_RECURSION */
5516 /* re_match_2 matches the compiled pattern in BUFP against the
5517 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5518 and SIZE2, respectively). We start matching at POS, and stop
5521 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5522 store offsets for the substring each group matched in REGS. See the
5523 documentation for exactly how many groups we fill.
5525 We return -1 if no match, -2 if an internal error (such as the
5526 failure stack overflowing). Otherwise, we return the length of the
5527 matched substring. */
5530 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
5531 struct re_pattern_buffer *bufp;
5532 const char *string1, *string2;
5535 struct re_registers *regs;
5540 if (MB_CUR_MAX != 1)
5541 result = wcs_re_match_2_internal (bufp, string1, size1, string2, size2,
5543 NULL, 0, NULL, 0, NULL, NULL);
5546 result = byte_re_match_2_internal (bufp, string1, size1, string2, size2,
5549 #ifndef REGEX_MALLOC
5557 weak_alias (__re_match_2, re_match_2)
5560 #endif /* not INSIDE_RECURSION */
5562 #ifdef INSIDE_RECURSION
5565 static int count_mbs_length PARAMS ((int *, int));
5567 /* This check the substring (from 0, to length) of the multibyte string,
5568 to which offset_buffer correspond. And count how many wchar_t_characters
5569 the substring occupy. We use offset_buffer to optimization.
5570 See convert_mbs_to_wcs. */
5573 count_mbs_length(offset_buffer, length)
5579 /* Check whether the size is valid. */
5583 if (offset_buffer == NULL)
5586 /* If there are no multibyte character, offset_buffer[i] == i.
5587 Optmize for this case. */
5588 if (offset_buffer[length] == length)
5591 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5597 int middle = (lower + upper) / 2;
5598 if (middle == lower || middle == upper)
5600 if (offset_buffer[middle] > length)
5602 else if (offset_buffer[middle] < length)
5612 /* This is a separate function so that we can force an alloca cleanup
5616 wcs_re_match_2_internal (bufp, cstring1, csize1, cstring2, csize2, pos,
5617 regs, stop, string1, size1, string2, size2,
5618 mbs_offset1, mbs_offset2)
5619 struct re_pattern_buffer *bufp;
5620 const char *cstring1, *cstring2;
5623 struct re_registers *regs;
5625 /* string1 == string2 == NULL means string1/2, size1/2 and
5626 mbs_offset1/2 need seting up in this function. */
5627 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5628 wchar_t *string1, *string2;
5629 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5631 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5632 int *mbs_offset1, *mbs_offset2;
5635 byte_re_match_2_internal (bufp, string1, size1,string2, size2, pos,
5637 struct re_pattern_buffer *bufp;
5638 const char *string1, *string2;
5641 struct re_registers *regs;
5645 /* General temporaries. */
5649 /* They hold whether each wchar_t is binary data or not. */
5650 char *is_binary = NULL;
5651 /* If true, we can't free string1/2, mbs_offset1/2. */
5652 int cant_free_wcs_buf = 1;
5655 /* Just past the end of the corresponding string. */
5656 const CHAR_T *end1, *end2;
5658 /* Pointers into string1 and string2, just past the last characters in
5659 each to consider matching. */
5660 const CHAR_T *end_match_1, *end_match_2;
5662 /* Where we are in the data, and the end of the current string. */
5663 const CHAR_T *d, *dend;
5665 /* Where we are in the pattern, and the end of the pattern. */
5667 UCHAR_T *pattern, *p;
5668 register UCHAR_T *pend;
5670 UCHAR_T *p = bufp->buffer;
5671 register UCHAR_T *pend = p + bufp->used;
5674 /* Mark the opcode just after a start_memory, so we can test for an
5675 empty subpattern when we get to the stop_memory. */
5676 UCHAR_T *just_past_start_mem = 0;
5678 /* We use this to map every character in the string. */
5679 RE_TRANSLATE_TYPE translate = bufp->translate;
5681 /* Failure point stack. Each place that can handle a failure further
5682 down the line pushes a failure point on this stack. It consists of
5683 restart, regend, and reg_info for all registers corresponding to
5684 the subexpressions we're currently inside, plus the number of such
5685 registers, and, finally, two char *'s. The first char * is where
5686 to resume scanning the pattern; the second one is where to resume
5687 scanning the strings. If the latter is zero, the failure point is
5688 a ``dummy''; if a failure happens and the failure point is a dummy,
5689 it gets discarded and the next next one is tried. */
5690 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5691 PREFIX(fail_stack_type) fail_stack;
5694 static unsigned failure_id;
5695 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
5699 /* This holds the pointer to the failure stack, when
5700 it is allocated relocatably. */
5701 fail_stack_elt_t *failure_stack_ptr;
5704 /* We fill all the registers internally, independent of what we
5705 return, for use in backreferences. The number here includes
5706 an element for register zero. */
5707 size_t num_regs = bufp->re_nsub + 1;
5709 /* The currently active registers. */
5710 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
5711 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
5713 /* Information on the contents of registers. These are pointers into
5714 the input strings; they record just what was matched (on this
5715 attempt) by a subexpression part of the pattern, that is, the
5716 regnum-th regstart pointer points to where in the pattern we began
5717 matching and the regnum-th regend points to right after where we
5718 stopped matching the regnum-th subexpression. (The zeroth register
5719 keeps track of what the whole pattern matches.) */
5720 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5721 const CHAR_T **regstart, **regend;
5724 /* If a group that's operated upon by a repetition operator fails to
5725 match anything, then the register for its start will need to be
5726 restored because it will have been set to wherever in the string we
5727 are when we last see its open-group operator. Similarly for a
5729 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5730 const CHAR_T **old_regstart, **old_regend;
5733 /* The is_active field of reg_info helps us keep track of which (possibly
5734 nested) subexpressions we are currently in. The matched_something
5735 field of reg_info[reg_num] helps us tell whether or not we have
5736 matched any of the pattern so far this time through the reg_num-th
5737 subexpression. These two fields get reset each time through any
5738 loop their register is in. */
5739 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5740 PREFIX(register_info_type) *reg_info;
5743 /* The following record the register info as found in the above
5744 variables when we find a match better than any we've seen before.
5745 This happens as we backtrack through the failure points, which in
5746 turn happens only if we have not yet matched the entire string. */
5747 unsigned best_regs_set = false;
5748 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5749 const CHAR_T **best_regstart, **best_regend;
5752 /* Logically, this is `best_regend[0]'. But we don't want to have to
5753 allocate space for that if we're not allocating space for anything
5754 else (see below). Also, we never need info about register 0 for
5755 any of the other register vectors, and it seems rather a kludge to
5756 treat `best_regend' differently than the rest. So we keep track of
5757 the end of the best match so far in a separate variable. We
5758 initialize this to NULL so that when we backtrack the first time
5759 and need to test it, it's not garbage. */
5760 const CHAR_T *match_end = NULL;
5762 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5763 int set_regs_matched_done = 0;
5765 /* Used when we pop values we don't care about. */
5766 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5767 const CHAR_T **reg_dummy;
5768 PREFIX(register_info_type) *reg_info_dummy;
5772 /* Counts the total number of registers pushed. */
5773 unsigned num_regs_pushed = 0;
5776 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
5780 #ifdef MATCH_MAY_ALLOCATE
5781 /* Do not bother to initialize all the register variables if there are
5782 no groups in the pattern, as it takes a fair amount of time. If
5783 there are groups, we include space for register 0 (the whole
5784 pattern), even though we never use it, since it simplifies the
5785 array indexing. We should fix this. */
5788 regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5789 regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5790 old_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5791 old_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5792 best_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5793 best_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5794 reg_info = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5795 reg_dummy = REGEX_TALLOC (num_regs, const CHAR_T *);
5796 reg_info_dummy = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5798 if (!(regstart && regend && old_regstart && old_regend && reg_info
5799 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
5807 /* We must initialize all our variables to NULL, so that
5808 `FREE_VARIABLES' doesn't try to free them. */
5809 regstart = regend = old_regstart = old_regend = best_regstart
5810 = best_regend = reg_dummy = NULL;
5811 reg_info = reg_info_dummy = (PREFIX(register_info_type) *) NULL;
5813 #endif /* MATCH_MAY_ALLOCATE */
5815 /* The starting position is bogus. */
5817 if (pos < 0 || pos > csize1 + csize2)
5819 if (pos < 0 || pos > size1 + size2)
5827 /* Allocate wchar_t array for string1 and string2 and
5828 fill them with converted string. */
5829 if (string1 == NULL && string2 == NULL)
5831 /* We need seting up buffers here. */
5833 /* We must free wcs buffers in this function. */
5834 cant_free_wcs_buf = 0;
5838 string1 = REGEX_TALLOC (csize1 + 1, CHAR_T);
5839 mbs_offset1 = REGEX_TALLOC (csize1 + 1, int);
5840 is_binary = REGEX_TALLOC (csize1 + 1, char);
5841 if (!string1 || !mbs_offset1 || !is_binary)
5844 FREE_VAR (mbs_offset1);
5845 FREE_VAR (is_binary);
5851 string2 = REGEX_TALLOC (csize2 + 1, CHAR_T);
5852 mbs_offset2 = REGEX_TALLOC (csize2 + 1, int);
5853 is_binary = REGEX_TALLOC (csize2 + 1, char);
5854 if (!string2 || !mbs_offset2 || !is_binary)
5857 FREE_VAR (mbs_offset1);
5859 FREE_VAR (mbs_offset2);
5860 FREE_VAR (is_binary);
5863 size2 = convert_mbs_to_wcs(string2, cstring2, csize2,
5864 mbs_offset2, is_binary);
5865 string2[size2] = L'\0'; /* for a sentinel */
5866 FREE_VAR (is_binary);
5870 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5871 pattern to (char*) in regex_compile. */
5872 p = pattern = (CHAR_T*)bufp->buffer;
5873 pend = (CHAR_T*)(bufp->buffer + bufp->used);
5877 /* Initialize subexpression text positions to -1 to mark ones that no
5878 start_memory/stop_memory has been seen for. Also initialize the
5879 register information struct. */
5880 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
5882 regstart[mcnt] = regend[mcnt]
5883 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
5885 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
5886 IS_ACTIVE (reg_info[mcnt]) = 0;
5887 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5888 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5891 /* We move `string1' into `string2' if the latter's empty -- but not if
5892 `string1' is null. */
5893 if (size2 == 0 && string1 != NULL)
5900 mbs_offset2 = mbs_offset1;
5906 end1 = string1 + size1;
5907 end2 = string2 + size2;
5909 /* Compute where to stop matching, within the two strings. */
5913 mcnt = count_mbs_length(mbs_offset1, stop);
5914 end_match_1 = string1 + mcnt;
5915 end_match_2 = string2;
5919 if (stop > csize1 + csize2)
5920 stop = csize1 + csize2;
5922 mcnt = count_mbs_length(mbs_offset2, stop-csize1);
5923 end_match_2 = string2 + mcnt;
5926 { /* count_mbs_length return error. */
5933 end_match_1 = string1 + stop;
5934 end_match_2 = string2;
5939 end_match_2 = string2 + stop - size1;
5943 /* `p' scans through the pattern as `d' scans through the data.
5944 `dend' is the end of the input string that `d' points within. `d'
5945 is advanced into the following input string whenever necessary, but
5946 this happens before fetching; therefore, at the beginning of the
5947 loop, `d' can be pointing at the end of a string, but it cannot
5950 if (size1 > 0 && pos <= csize1)
5952 mcnt = count_mbs_length(mbs_offset1, pos);
5958 mcnt = count_mbs_length(mbs_offset2, pos-csize1);
5964 { /* count_mbs_length return error. */
5969 if (size1 > 0 && pos <= size1)
5976 d = string2 + pos - size1;
5981 DEBUG_PRINT1 ("The compiled pattern is:\n");
5982 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
5983 DEBUG_PRINT1 ("The string to match is: `");
5984 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
5985 DEBUG_PRINT1 ("'\n");
5987 /* This loops over pattern commands. It exits by returning from the
5988 function if the match is complete, or it drops through if the match
5989 fails at this starting point in the input data. */
5993 DEBUG_PRINT2 ("\n%p: ", p);
5995 DEBUG_PRINT2 ("\n0x%x: ", p);
5999 { /* End of pattern means we might have succeeded. */
6000 DEBUG_PRINT1 ("end of pattern ... ");
6002 /* If we haven't matched the entire string, and we want the
6003 longest match, try backtracking. */
6004 if (d != end_match_2)
6006 /* 1 if this match ends in the same string (string1 or string2)
6007 as the best previous match. */
6008 boolean same_str_p = (FIRST_STRING_P (match_end)
6009 == MATCHING_IN_FIRST_STRING);
6010 /* 1 if this match is the best seen so far. */
6011 boolean best_match_p;
6013 /* AIX compiler got confused when this was combined
6014 with the previous declaration. */
6016 best_match_p = d > match_end;
6018 best_match_p = !MATCHING_IN_FIRST_STRING;
6020 DEBUG_PRINT1 ("backtracking.\n");
6022 if (!FAIL_STACK_EMPTY ())
6023 { /* More failure points to try. */
6025 /* If exceeds best match so far, save it. */
6026 if (!best_regs_set || best_match_p)
6028 best_regs_set = true;
6031 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
6033 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6035 best_regstart[mcnt] = regstart[mcnt];
6036 best_regend[mcnt] = regend[mcnt];
6042 /* If no failure points, don't restore garbage. And if
6043 last match is real best match, don't restore second
6045 else if (best_regs_set && !best_match_p)
6048 /* Restore best match. It may happen that `dend ==
6049 end_match_1' while the restored d is in string2.
6050 For example, the pattern `x.*y.*z' against the
6051 strings `x-' and `y-z-', if the two strings are
6052 not consecutive in memory. */
6053 DEBUG_PRINT1 ("Restoring best registers.\n");
6056 dend = ((d >= string1 && d <= end1)
6057 ? end_match_1 : end_match_2);
6059 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6061 regstart[mcnt] = best_regstart[mcnt];
6062 regend[mcnt] = best_regend[mcnt];
6065 } /* d != end_match_2 */
6068 DEBUG_PRINT1 ("Accepting match.\n");
6069 /* If caller wants register contents data back, do it. */
6070 if (regs && !bufp->no_sub)
6072 /* Have the register data arrays been allocated? */
6073 if (bufp->regs_allocated == REGS_UNALLOCATED)
6074 { /* No. So allocate them with malloc. We need one
6075 extra element beyond `num_regs' for the `-1' marker
6077 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
6078 regs->start = TALLOC (regs->num_regs, regoff_t);
6079 regs->end = TALLOC (regs->num_regs, regoff_t);
6080 if (regs->start == NULL || regs->end == NULL)
6085 bufp->regs_allocated = REGS_REALLOCATE;
6087 else if (bufp->regs_allocated == REGS_REALLOCATE)
6088 { /* Yes. If we need more elements than were already
6089 allocated, reallocate them. If we need fewer, just
6091 if (regs->num_regs < num_regs + 1)
6093 regs->num_regs = num_regs + 1;
6094 RETALLOC (regs->start, regs->num_regs, regoff_t);
6095 RETALLOC (regs->end, regs->num_regs, regoff_t);
6096 if (regs->start == NULL || regs->end == NULL)
6105 /* These braces fend off a "empty body in an else-statement"
6106 warning under GCC when assert expands to nothing. */
6107 assert (bufp->regs_allocated == REGS_FIXED);
6110 /* Convert the pointer data in `regstart' and `regend' to
6111 indices. Register zero has to be set differently,
6112 since we haven't kept track of any info for it. */
6113 if (regs->num_regs > 0)
6115 regs->start[0] = pos;
6117 if (MATCHING_IN_FIRST_STRING)
6118 regs->end[0] = mbs_offset1 != NULL ?
6119 mbs_offset1[d-string1] : 0;
6121 regs->end[0] = csize1 + (mbs_offset2 != NULL ?
6122 mbs_offset2[d-string2] : 0);
6124 regs->end[0] = (MATCHING_IN_FIRST_STRING
6125 ? ((regoff_t) (d - string1))
6126 : ((regoff_t) (d - string2 + size1)));
6130 /* Go through the first `min (num_regs, regs->num_regs)'
6131 registers, since that is all we initialized. */
6132 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
6135 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
6136 regs->start[mcnt] = regs->end[mcnt] = -1;
6140 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
6142 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
6146 /* If the regs structure we return has more elements than
6147 were in the pattern, set the extra elements to -1. If
6148 we (re)allocated the registers, this is the case,
6149 because we always allocate enough to have at least one
6151 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
6152 regs->start[mcnt] = regs->end[mcnt] = -1;
6153 } /* regs && !bufp->no_sub */
6155 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
6156 nfailure_points_pushed, nfailure_points_popped,
6157 nfailure_points_pushed - nfailure_points_popped);
6158 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
6161 if (MATCHING_IN_FIRST_STRING)
6162 mcnt = mbs_offset1 != NULL ? mbs_offset1[d-string1] : 0;
6164 mcnt = (mbs_offset2 != NULL ? mbs_offset2[d-string2] : 0) +
6168 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
6173 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
6179 /* Otherwise match next pattern command. */
6180 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
6182 /* Ignore these. Used to ignore the n of succeed_n's which
6183 currently have n == 0. */
6185 DEBUG_PRINT1 ("EXECUTING no_op.\n");
6189 DEBUG_PRINT1 ("EXECUTING succeed.\n");
6192 /* Match the next n pattern characters exactly. The following
6193 byte in the pattern defines n, and the n bytes after that
6194 are the characters to match. */
6200 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
6202 /* This is written out as an if-else so we don't waste time
6203 testing `translate' inside the loop. */
6212 if ((UCHAR_T) translate[(unsigned char) *d++]
6218 if (*d++ != (CHAR_T) *p++)
6222 if ((UCHAR_T) translate[(unsigned char) *d++]
6234 if (*d++ != (CHAR_T) *p++) goto fail;
6238 SET_REGS_MATCHED ();
6242 /* Match any character except possibly a newline or a null. */
6244 DEBUG_PRINT1 ("EXECUTING anychar.\n");
6248 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
6249 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
6252 SET_REGS_MATCHED ();
6253 DEBUG_PRINT2 (" Matched `%ld'.\n", (long int) *d);
6263 unsigned int i, char_class_length, coll_symbol_length,
6264 equiv_class_length, ranges_length, chars_length, length;
6265 CHAR_T *workp, *workp2, *charset_top;
6266 #define WORK_BUFFER_SIZE 128
6267 CHAR_T str_buf[WORK_BUFFER_SIZE];
6272 boolean not = (re_opcode_t) *(p - 1) == charset_not;
6274 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
6276 c = TRANSLATE (*d); /* The character to match. */
6279 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
6281 charset_top = p - 1;
6282 char_class_length = *p++;
6283 coll_symbol_length = *p++;
6284 equiv_class_length = *p++;
6285 ranges_length = *p++;
6286 chars_length = *p++;
6287 /* p points charset[6], so the address of the next instruction
6288 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6289 where l=length of char_classes, m=length of collating_symbol,
6290 n=equivalence_class, o=length of char_range,
6291 p'=length of character. */
6293 /* Update p to indicate the next instruction. */
6294 p += char_class_length + coll_symbol_length+ equiv_class_length +
6295 2*ranges_length + chars_length;
6297 /* match with char_class? */
6298 for (i = 0; i < char_class_length ; i += CHAR_CLASS_SIZE)
6301 uintptr_t alignedp = ((uintptr_t)workp
6302 + __alignof__(wctype_t) - 1)
6303 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
6304 wctype = *((wctype_t*)alignedp);
6305 workp += CHAR_CLASS_SIZE;
6307 if (__iswctype((wint_t)c, wctype))
6308 goto char_set_matched;
6310 if (iswctype((wint_t)c, wctype))
6311 goto char_set_matched;
6315 /* match with collating_symbol? */
6319 const unsigned char *extra = (const unsigned char *)
6320 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
6322 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;
6326 wextra = (int32_t*)(extra + *workp++);
6327 for (i = 0; i < *wextra; ++i)
6328 if (TRANSLATE(d[i]) != wextra[1 + i])
6333 /* Update d, however d will be incremented at
6334 char_set_matched:, we decrement d here. */
6336 goto char_set_matched;
6340 else /* (nrules == 0) */
6342 /* If we can't look up collation data, we use wcscoll
6345 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;)
6347 const CHAR_T *backup_d = d, *backup_dend = dend;
6349 length = __wcslen (workp);
6351 length = wcslen (workp);
6354 /* If wcscoll(the collating symbol, whole string) > 0,
6355 any substring of the string never match with the
6356 collating symbol. */
6358 if (__wcscoll (workp, d) > 0)
6360 if (wcscoll (workp, d) > 0)
6363 workp += length + 1;
6367 /* First, we compare the collating symbol with
6368 the first character of the string.
6369 If it don't match, we add the next character to
6370 the compare buffer in turn. */
6371 for (i = 0 ; i < WORK_BUFFER_SIZE-1 ; i++, d++)
6376 if (dend == end_match_2)
6382 /* add next character to the compare buffer. */
6383 str_buf[i] = TRANSLATE(*d);
6384 str_buf[i+1] = '\0';
6387 match = __wcscoll (workp, str_buf);
6389 match = wcscoll (workp, str_buf);
6392 goto char_set_matched;
6395 /* (str_buf > workp) indicate (str_buf + X > workp),
6396 because for all X (str_buf + X > str_buf).
6397 So we don't need continue this loop. */
6400 /* Otherwise(str_buf < workp),
6401 (str_buf+next_character) may equals (workp).
6402 So we continue this loop. */
6407 workp += length + 1;
6410 /* match with equivalence_class? */
6414 const CHAR_T *backup_d = d, *backup_dend = dend;
6415 /* Try to match the equivalence class against
6416 those known to the collate implementation. */
6417 const int32_t *table;
6418 const int32_t *weights;
6419 const int32_t *extra;
6420 const int32_t *indirect;
6425 /* This #include defines a local function! */
6426 # include <locale/weightwc.h>
6428 table = (const int32_t *)
6429 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEWC);
6430 weights = (const wint_t *)
6431 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTWC);
6432 extra = (const wint_t *)
6433 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAWC);
6434 indirect = (const int32_t *)
6435 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTWC);
6437 /* Write 1 collating element to str_buf, and
6441 for (i = 0 ; idx2 == 0 && i < WORK_BUFFER_SIZE - 1; i++)
6443 cp = (wint_t*)str_buf;
6446 if (dend == end_match_2)
6451 str_buf[i] = TRANSLATE(*(d+i));
6452 str_buf[i+1] = '\0'; /* sentinel */
6453 idx2 = findidx ((const wint_t**)&cp);
6456 /* Update d, however d will be incremented at
6457 char_set_matched:, we decrement d here. */
6458 d = backup_d + ((wchar_t*)cp - (wchar_t*)str_buf - 1);
6461 if (dend == end_match_2)
6470 len = weights[idx2];
6472 for (workp2 = workp + equiv_class_length ; workp < workp2 ;
6475 idx = (int32_t)*workp;
6476 /* We already checked idx != 0 in regex_compile. */
6478 if (idx2 != 0 && len == weights[idx])
6481 while (cnt < len && (weights[idx + 1 + cnt]
6482 == weights[idx2 + 1 + cnt]))
6486 goto char_set_matched;
6493 else /* (nrules == 0) */
6495 /* If we can't look up collation data, we use wcscoll
6498 for (workp2 = workp + equiv_class_length ; workp < workp2 ;)
6500 const CHAR_T *backup_d = d, *backup_dend = dend;
6502 length = __wcslen (workp);
6504 length = wcslen (workp);
6507 /* If wcscoll(the collating symbol, whole string) > 0,
6508 any substring of the string never match with the
6509 collating symbol. */
6511 if (__wcscoll (workp, d) > 0)
6513 if (wcscoll (workp, d) > 0)
6516 workp += length + 1;
6520 /* First, we compare the equivalence class with
6521 the first character of the string.
6522 If it don't match, we add the next character to
6523 the compare buffer in turn. */
6524 for (i = 0 ; i < WORK_BUFFER_SIZE - 1 ; i++, d++)
6529 if (dend == end_match_2)
6535 /* add next character to the compare buffer. */
6536 str_buf[i] = TRANSLATE(*d);
6537 str_buf[i+1] = '\0';
6540 match = __wcscoll (workp, str_buf);
6542 match = wcscoll (workp, str_buf);
6546 goto char_set_matched;
6549 /* (str_buf > workp) indicate (str_buf + X > workp),
6550 because for all X (str_buf + X > str_buf).
6551 So we don't need continue this loop. */
6554 /* Otherwise(str_buf < workp),
6555 (str_buf+next_character) may equals (workp).
6556 So we continue this loop. */
6561 workp += length + 1;
6565 /* match with char_range? */
6569 uint32_t collseqval;
6570 const char *collseq = (const char *)
6571 _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQWC);
6573 collseqval = collseq_table_lookup (collseq, c);
6575 for (; workp < p - chars_length ;)
6577 uint32_t start_val, end_val;
6579 /* We already compute the collation sequence value
6580 of the characters (or collating symbols). */
6581 start_val = (uint32_t) *workp++; /* range_start */
6582 end_val = (uint32_t) *workp++; /* range_end */
6584 if (start_val <= collseqval && collseqval <= end_val)
6585 goto char_set_matched;
6591 /* We set range_start_char at str_buf[0], range_end_char
6592 at str_buf[4], and compared char at str_buf[2]. */
6597 for (; workp < p - chars_length ;)
6599 wchar_t *range_start_char, *range_end_char;
6601 /* match if (range_start_char <= c <= range_end_char). */
6603 /* If range_start(or end) < 0, we assume -range_start(end)
6604 is the offset of the collating symbol which is specified
6605 as the character of the range start(end). */
6609 range_start_char = charset_top - (*workp++);
6612 str_buf[0] = *workp++;
6613 range_start_char = str_buf;
6618 range_end_char = charset_top - (*workp++);
6621 str_buf[4] = *workp++;
6622 range_end_char = str_buf + 4;
6626 if (__wcscoll (range_start_char, str_buf+2) <= 0
6627 && __wcscoll (str_buf+2, range_end_char) <= 0)
6629 if (wcscoll (range_start_char, str_buf+2) <= 0
6630 && wcscoll (str_buf+2, range_end_char) <= 0)
6632 goto char_set_matched;
6636 /* match with char? */
6637 for (; workp < p ; workp++)
6639 goto char_set_matched;
6646 /* Cast to `unsigned' instead of `unsigned char' in case the
6647 bit list is a full 32 bytes long. */
6648 if (c < (unsigned) (*p * BYTEWIDTH)
6649 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
6654 if (!not) goto fail;
6655 #undef WORK_BUFFER_SIZE
6657 SET_REGS_MATCHED ();
6663 /* The beginning of a group is represented by start_memory.
6664 The arguments are the register number in the next byte, and the
6665 number of groups inner to this one in the next. The text
6666 matched within the group is recorded (in the internal
6667 registers data structure) under the register number. */
6669 DEBUG_PRINT3 ("EXECUTING start_memory %ld (%ld):\n",
6670 (long int) *p, (long int) p[1]);
6672 /* Find out if this group can match the empty string. */
6673 p1 = p; /* To send to group_match_null_string_p. */
6675 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
6676 REG_MATCH_NULL_STRING_P (reg_info[*p])
6677 = PREFIX(group_match_null_string_p) (&p1, pend, reg_info);
6679 /* Save the position in the string where we were the last time
6680 we were at this open-group operator in case the group is
6681 operated upon by a repetition operator, e.g., with `(a*)*b'
6682 against `ab'; then we want to ignore where we are now in
6683 the string in case this attempt to match fails. */
6684 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6685 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
6687 DEBUG_PRINT2 (" old_regstart: %d\n",
6688 POINTER_TO_OFFSET (old_regstart[*p]));
6691 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
6693 IS_ACTIVE (reg_info[*p]) = 1;
6694 MATCHED_SOMETHING (reg_info[*p]) = 0;
6696 /* Clear this whenever we change the register activity status. */
6697 set_regs_matched_done = 0;
6699 /* This is the new highest active register. */
6700 highest_active_reg = *p;
6702 /* If nothing was active before, this is the new lowest active
6704 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
6705 lowest_active_reg = *p;
6707 /* Move past the register number and inner group count. */
6709 just_past_start_mem = p;
6714 /* The stop_memory opcode represents the end of a group. Its
6715 arguments are the same as start_memory's: the register
6716 number, and the number of inner groups. */
6718 DEBUG_PRINT3 ("EXECUTING stop_memory %ld (%ld):\n",
6719 (long int) *p, (long int) p[1]);
6721 /* We need to save the string position the last time we were at
6722 this close-group operator in case the group is operated
6723 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6724 against `aba'; then we want to ignore where we are now in
6725 the string in case this attempt to match fails. */
6726 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6727 ? REG_UNSET (regend[*p]) ? d : regend[*p]
6729 DEBUG_PRINT2 (" old_regend: %d\n",
6730 POINTER_TO_OFFSET (old_regend[*p]));
6733 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
6735 /* This register isn't active anymore. */
6736 IS_ACTIVE (reg_info[*p]) = 0;
6738 /* Clear this whenever we change the register activity status. */
6739 set_regs_matched_done = 0;
6741 /* If this was the only register active, nothing is active
6743 if (lowest_active_reg == highest_active_reg)
6745 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6746 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6749 { /* We must scan for the new highest active register, since
6750 it isn't necessarily one less than now: consider
6751 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6752 new highest active register is 1. */
6754 while (r > 0 && !IS_ACTIVE (reg_info[r]))
6757 /* If we end up at register zero, that means that we saved
6758 the registers as the result of an `on_failure_jump', not
6759 a `start_memory', and we jumped to past the innermost
6760 `stop_memory'. For example, in ((.)*) we save
6761 registers 1 and 2 as a result of the *, but when we pop
6762 back to the second ), we are at the stop_memory 1.
6763 Thus, nothing is active. */
6766 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6767 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6770 highest_active_reg = r;
6773 /* If just failed to match something this time around with a
6774 group that's operated on by a repetition operator, try to
6775 force exit from the ``loop'', and restore the register
6776 information for this group that we had before trying this
6778 if ((!MATCHED_SOMETHING (reg_info[*p])
6779 || just_past_start_mem == p - 1)
6782 boolean is_a_jump_n = false;
6786 switch ((re_opcode_t) *p1++)
6790 case pop_failure_jump:
6791 case maybe_pop_jump:
6793 case dummy_failure_jump:
6794 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6796 p1 += OFFSET_ADDRESS_SIZE;
6804 /* If the next operation is a jump backwards in the pattern
6805 to an on_failure_jump right before the start_memory
6806 corresponding to this stop_memory, exit from the loop
6807 by forcing a failure after pushing on the stack the
6808 on_failure_jump's jump in the pattern, and d. */
6809 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
6810 && (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == start_memory
6811 && p1[2+OFFSET_ADDRESS_SIZE] == *p)
6813 /* If this group ever matched anything, then restore
6814 what its registers were before trying this last
6815 failed match, e.g., with `(a*)*b' against `ab' for
6816 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6817 against `aba' for regend[3].
6819 Also restore the registers for inner groups for,
6820 e.g., `((a*)(b*))*' against `aba' (register 3 would
6821 otherwise get trashed). */
6823 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
6827 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
6829 /* Restore this and inner groups' (if any) registers. */
6830 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
6833 regstart[r] = old_regstart[r];
6835 /* xx why this test? */
6836 if (old_regend[r] >= regstart[r])
6837 regend[r] = old_regend[r];
6841 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6842 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
6848 /* Move past the register number and the inner group count. */
6853 /* \<digit> has been turned into a `duplicate' command which is
6854 followed by the numeric value of <digit> as the register number. */
6857 register const CHAR_T *d2, *dend2;
6858 int regno = *p++; /* Get which register to match against. */
6859 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
6861 /* Can't back reference a group which we've never matched. */
6862 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
6865 /* Where in input to try to start matching. */
6866 d2 = regstart[regno];
6868 /* Where to stop matching; if both the place to start and
6869 the place to stop matching are in the same string, then
6870 set to the place to stop, otherwise, for now have to use
6871 the end of the first string. */
6873 dend2 = ((FIRST_STRING_P (regstart[regno])
6874 == FIRST_STRING_P (regend[regno]))
6875 ? regend[regno] : end_match_1);
6878 /* If necessary, advance to next segment in register
6882 if (dend2 == end_match_2) break;
6883 if (dend2 == regend[regno]) break;
6885 /* End of string1 => advance to string2. */
6887 dend2 = regend[regno];
6889 /* At end of register contents => success */
6890 if (d2 == dend2) break;
6892 /* If necessary, advance to next segment in data. */
6895 /* How many characters left in this segment to match. */
6898 /* Want how many consecutive characters we can match in
6899 one shot, so, if necessary, adjust the count. */
6900 if (mcnt > dend2 - d2)
6903 /* Compare that many; failure if mismatch, else move
6906 ? PREFIX(bcmp_translate) (d, d2, mcnt, translate)
6907 : memcmp (d, d2, mcnt*sizeof(UCHAR_T)))
6909 d += mcnt, d2 += mcnt;
6911 /* Do this because we've match some characters. */
6912 SET_REGS_MATCHED ();
6918 /* begline matches the empty string at the beginning of the string
6919 (unless `not_bol' is set in `bufp'), and, if
6920 `newline_anchor' is set, after newlines. */
6922 DEBUG_PRINT1 ("EXECUTING begline.\n");
6924 if (AT_STRINGS_BEG (d))
6926 if (!bufp->not_bol) break;
6928 else if (d[-1] == '\n' && bufp->newline_anchor)
6932 /* In all other cases, we fail. */
6936 /* endline is the dual of begline. */
6938 DEBUG_PRINT1 ("EXECUTING endline.\n");
6940 if (AT_STRINGS_END (d))
6942 if (!bufp->not_eol) break;
6945 /* We have to ``prefetch'' the next character. */
6946 else if ((d == end1 ? *string2 : *d) == '\n'
6947 && bufp->newline_anchor)
6954 /* Match at the very beginning of the data. */
6956 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
6957 if (AT_STRINGS_BEG (d))
6962 /* Match at the very end of the data. */
6964 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
6965 if (AT_STRINGS_END (d))
6970 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6971 pushes NULL as the value for the string on the stack. Then
6972 `pop_failure_point' will keep the current value for the
6973 string, instead of restoring it. To see why, consider
6974 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6975 then the . fails against the \n. But the next thing we want
6976 to do is match the \n against the \n; if we restored the
6977 string value, we would be back at the foo.
6979 Because this is used only in specific cases, we don't need to
6980 check all the things that `on_failure_jump' does, to make
6981 sure the right things get saved on the stack. Hence we don't
6982 share its code. The only reason to push anything on the
6983 stack at all is that otherwise we would have to change
6984 `anychar's code to do something besides goto fail in this
6985 case; that seems worse than this. */
6986 case on_failure_keep_string_jump:
6987 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
6989 EXTRACT_NUMBER_AND_INCR (mcnt, p);
6991 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
6993 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
6996 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
7000 /* Uses of on_failure_jump:
7002 Each alternative starts with an on_failure_jump that points
7003 to the beginning of the next alternative. Each alternative
7004 except the last ends with a jump that in effect jumps past
7005 the rest of the alternatives. (They really jump to the
7006 ending jump of the following alternative, because tensioning
7007 these jumps is a hassle.)
7009 Repeats start with an on_failure_jump that points past both
7010 the repetition text and either the following jump or
7011 pop_failure_jump back to this on_failure_jump. */
7012 case on_failure_jump:
7014 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
7016 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7018 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
7020 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
7023 /* If this on_failure_jump comes right before a group (i.e.,
7024 the original * applied to a group), save the information
7025 for that group and all inner ones, so that if we fail back
7026 to this point, the group's information will be correct.
7027 For example, in \(a*\)*\1, we need the preceding group,
7028 and in \(zz\(a*\)b*\)\2, we need the inner group. */
7030 /* We can't use `p' to check ahead because we push
7031 a failure point to `p + mcnt' after we do this. */
7034 /* We need to skip no_op's before we look for the
7035 start_memory in case this on_failure_jump is happening as
7036 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
7038 while (p1 < pend && (re_opcode_t) *p1 == no_op)
7041 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
7043 /* We have a new highest active register now. This will
7044 get reset at the start_memory we are about to get to,
7045 but we will have saved all the registers relevant to
7046 this repetition op, as described above. */
7047 highest_active_reg = *(p1 + 1) + *(p1 + 2);
7048 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
7049 lowest_active_reg = *(p1 + 1);
7052 DEBUG_PRINT1 (":\n");
7053 PUSH_FAILURE_POINT (p + mcnt, d, -2);
7057 /* A smart repeat ends with `maybe_pop_jump'.
7058 We change it to either `pop_failure_jump' or `jump'. */
7059 case maybe_pop_jump:
7060 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7061 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
7063 register UCHAR_T *p2 = p;
7065 /* Compare the beginning of the repeat with what in the
7066 pattern follows its end. If we can establish that there
7067 is nothing that they would both match, i.e., that we
7068 would have to backtrack because of (as in, e.g., `a*a')
7069 then we can change to pop_failure_jump, because we'll
7070 never have to backtrack.
7072 This is not true in the case of alternatives: in
7073 `(a|ab)*' we do need to backtrack to the `ab' alternative
7074 (e.g., if the string was `ab'). But instead of trying to
7075 detect that here, the alternative has put on a dummy
7076 failure point which is what we will end up popping. */
7078 /* Skip over open/close-group commands.
7079 If what follows this loop is a ...+ construct,
7080 look at what begins its body, since we will have to
7081 match at least one of that. */
7085 && ((re_opcode_t) *p2 == stop_memory
7086 || (re_opcode_t) *p2 == start_memory))
7088 else if (p2 + 2 + 2 * OFFSET_ADDRESS_SIZE < pend
7089 && (re_opcode_t) *p2 == dummy_failure_jump)
7090 p2 += 2 + 2 * OFFSET_ADDRESS_SIZE;
7096 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7097 to the `maybe_finalize_jump' of this case. Examine what
7100 /* If we're at the end of the pattern, we can change. */
7103 /* Consider what happens when matching ":\(.*\)"
7104 against ":/". I don't really understand this code
7106 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7109 (" End of pattern: change to `pop_failure_jump'.\n");
7112 else if ((re_opcode_t) *p2 == exactn
7114 || (re_opcode_t) *p2 == exactn_bin
7116 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
7119 = *p2 == (UCHAR_T) endline ? '\n' : p2[2];
7121 if (((re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn
7123 || (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn_bin
7125 ) && p1[3+OFFSET_ADDRESS_SIZE] != c)
7127 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7130 DEBUG_PRINT3 (" %C != %C => pop_failure_jump.\n",
7132 (wint_t) p1[3+OFFSET_ADDRESS_SIZE]);
7134 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
7136 (char) p1[3+OFFSET_ADDRESS_SIZE]);
7141 else if ((re_opcode_t) p1[3] == charset
7142 || (re_opcode_t) p1[3] == charset_not)
7144 int not = (re_opcode_t) p1[3] == charset_not;
7146 if (c < (unsigned) (p1[4] * BYTEWIDTH)
7147 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
7150 /* `not' is equal to 1 if c would match, which means
7151 that we can't change to pop_failure_jump. */
7154 p[-3] = (unsigned char) pop_failure_jump;
7155 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7158 #endif /* not WCHAR */
7161 else if ((re_opcode_t) *p2 == charset)
7163 /* We win if the first character of the loop is not part
7165 if ((re_opcode_t) p1[3] == exactn
7166 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
7167 && (p2[2 + p1[5] / BYTEWIDTH]
7168 & (1 << (p1[5] % BYTEWIDTH)))))
7170 p[-3] = (unsigned char) pop_failure_jump;
7171 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7174 else if ((re_opcode_t) p1[3] == charset_not)
7177 /* We win if the charset_not inside the loop
7178 lists every character listed in the charset after. */
7179 for (idx = 0; idx < (int) p2[1]; idx++)
7180 if (! (p2[2 + idx] == 0
7181 || (idx < (int) p1[4]
7182 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
7187 p[-3] = (unsigned char) pop_failure_jump;
7188 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7191 else if ((re_opcode_t) p1[3] == charset)
7194 /* We win if the charset inside the loop
7195 has no overlap with the one after the loop. */
7197 idx < (int) p2[1] && idx < (int) p1[4];
7199 if ((p2[2 + idx] & p1[5 + idx]) != 0)
7202 if (idx == p2[1] || idx == p1[4])
7204 p[-3] = (unsigned char) pop_failure_jump;
7205 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7209 #endif /* not WCHAR */
7211 p -= OFFSET_ADDRESS_SIZE; /* Point at relative address again. */
7212 if ((re_opcode_t) p[-1] != pop_failure_jump)
7214 p[-1] = (UCHAR_T) jump;
7215 DEBUG_PRINT1 (" Match => jump.\n");
7216 goto unconditional_jump;
7218 /* Note fall through. */
7221 /* The end of a simple repeat has a pop_failure_jump back to
7222 its matching on_failure_jump, where the latter will push a
7223 failure point. The pop_failure_jump takes off failure
7224 points put on by this pop_failure_jump's matching
7225 on_failure_jump; we got through the pattern to here from the
7226 matching on_failure_jump, so didn't fail. */
7227 case pop_failure_jump:
7229 /* We need to pass separate storage for the lowest and
7230 highest registers, even though we don't care about the
7231 actual values. Otherwise, we will restore only one
7232 register from the stack, since lowest will == highest in
7233 `pop_failure_point'. */
7234 active_reg_t dummy_low_reg, dummy_high_reg;
7235 UCHAR_T *pdummy = NULL;
7236 const CHAR_T *sdummy = NULL;
7238 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
7239 POP_FAILURE_POINT (sdummy, pdummy,
7240 dummy_low_reg, dummy_high_reg,
7241 reg_dummy, reg_dummy, reg_info_dummy);
7243 /* Note fall through. */
7247 DEBUG_PRINT2 ("\n%p: ", p);
7249 DEBUG_PRINT2 ("\n0x%x: ", p);
7251 /* Note fall through. */
7253 /* Unconditionally jump (without popping any failure points). */
7255 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
7256 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
7257 p += mcnt; /* Do the jump. */
7259 DEBUG_PRINT2 ("(to %p).\n", p);
7261 DEBUG_PRINT2 ("(to 0x%x).\n", p);
7266 /* We need this opcode so we can detect where alternatives end
7267 in `group_match_null_string_p' et al. */
7269 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
7270 goto unconditional_jump;
7273 /* Normally, the on_failure_jump pushes a failure point, which
7274 then gets popped at pop_failure_jump. We will end up at
7275 pop_failure_jump, also, and with a pattern of, say, `a+', we
7276 are skipping over the on_failure_jump, so we have to push
7277 something meaningless for pop_failure_jump to pop. */
7278 case dummy_failure_jump:
7279 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
7280 /* It doesn't matter what we push for the string here. What
7281 the code at `fail' tests is the value for the pattern. */
7282 PUSH_FAILURE_POINT (NULL, NULL, -2);
7283 goto unconditional_jump;
7286 /* At the end of an alternative, we need to push a dummy failure
7287 point in case we are followed by a `pop_failure_jump', because
7288 we don't want the failure point for the alternative to be
7289 popped. For example, matching `(a|ab)*' against `aab'
7290 requires that we match the `ab' alternative. */
7291 case push_dummy_failure:
7292 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
7293 /* See comments just above at `dummy_failure_jump' about the
7295 PUSH_FAILURE_POINT (NULL, NULL, -2);
7298 /* Have to succeed matching what follows at least n times.
7299 After that, handle like `on_failure_jump'. */
7301 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7302 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
7305 /* Originally, this is how many times we HAVE to succeed. */
7309 p += OFFSET_ADDRESS_SIZE;
7310 STORE_NUMBER_AND_INCR (p, mcnt);
7312 DEBUG_PRINT3 (" Setting %p to %d.\n", p - OFFSET_ADDRESS_SIZE
7315 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - OFFSET_ADDRESS_SIZE
7322 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n",
7323 p + OFFSET_ADDRESS_SIZE);
7325 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n",
7326 p + OFFSET_ADDRESS_SIZE);
7330 p[1] = (UCHAR_T) no_op;
7332 p[2] = (UCHAR_T) no_op;
7333 p[3] = (UCHAR_T) no_op;
7340 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7341 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
7343 /* Originally, this is how many times we CAN jump. */
7347 STORE_NUMBER (p + OFFSET_ADDRESS_SIZE, mcnt);
7350 DEBUG_PRINT3 (" Setting %p to %d.\n", p + OFFSET_ADDRESS_SIZE,
7353 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + OFFSET_ADDRESS_SIZE,
7356 goto unconditional_jump;
7358 /* If don't have to jump any more, skip over the rest of command. */
7360 p += 2 * OFFSET_ADDRESS_SIZE;
7365 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
7367 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7369 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7371 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
7373 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
7375 STORE_NUMBER (p1, mcnt);
7380 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7381 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7382 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7383 macro and introducing temporary variables works around the bug. */
7386 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7387 if (AT_WORD_BOUNDARY (d))
7392 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7393 if (AT_WORD_BOUNDARY (d))
7399 boolean prevchar, thischar;
7401 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7402 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7405 prevchar = WORDCHAR_P (d - 1);
7406 thischar = WORDCHAR_P (d);
7407 if (prevchar != thischar)
7414 boolean prevchar, thischar;
7416 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7417 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7420 prevchar = WORDCHAR_P (d - 1);
7421 thischar = WORDCHAR_P (d);
7422 if (prevchar != thischar)
7429 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
7430 if (!AT_STRINGS_END (d) && WORDCHAR_P (d)
7431 && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
7436 DEBUG_PRINT1 ("EXECUTING wordend.\n");
7437 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
7438 && (AT_STRINGS_END (d) || !WORDCHAR_P (d)))
7444 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
7445 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
7450 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
7451 if (PTR_CHAR_POS ((unsigned char *) d) != point)
7456 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
7457 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
7462 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
7467 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
7471 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7473 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
7475 SET_REGS_MATCHED ();
7479 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
7481 goto matchnotsyntax;
7484 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
7488 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7490 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
7492 SET_REGS_MATCHED ();
7495 #else /* not emacs */
7497 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
7499 if (!WORDCHAR_P (d))
7501 SET_REGS_MATCHED ();
7506 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
7510 SET_REGS_MATCHED ();
7513 #endif /* not emacs */
7518 continue; /* Successfully executed one pattern command; keep going. */
7521 /* We goto here if a matching operation fails. */
7523 if (!FAIL_STACK_EMPTY ())
7524 { /* A restart point is known. Restore to that state. */
7525 DEBUG_PRINT1 ("\nFAIL:\n");
7526 POP_FAILURE_POINT (d, p,
7527 lowest_active_reg, highest_active_reg,
7528 regstart, regend, reg_info);
7530 /* If this failure point is a dummy, try the next one. */
7534 /* If we failed to the end of the pattern, don't examine *p. */
7538 boolean is_a_jump_n = false;
7540 /* If failed to a backwards jump that's part of a repetition
7541 loop, need to pop this failure point and use the next one. */
7542 switch ((re_opcode_t) *p)
7546 case maybe_pop_jump:
7547 case pop_failure_jump:
7550 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7553 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
7555 && (re_opcode_t) *p1 == on_failure_jump))
7563 if (d >= string1 && d <= end1)
7567 break; /* Matching at this starting point really fails. */
7571 goto restore_best_regs;
7575 return -1; /* Failure to match. */
7578 /* Subroutine definitions for re_match_2. */
7581 /* We are passed P pointing to a register number after a start_memory.
7583 Return true if the pattern up to the corresponding stop_memory can
7584 match the empty string, and false otherwise.
7586 If we find the matching stop_memory, sets P to point to one past its number.
7587 Otherwise, sets P to an undefined byte less than or equal to END.
7589 We don't handle duplicates properly (yet). */
7592 PREFIX(group_match_null_string_p) (p, end, reg_info)
7594 PREFIX(register_info_type) *reg_info;
7597 /* Point to after the args to the start_memory. */
7598 UCHAR_T *p1 = *p + 2;
7602 /* Skip over opcodes that can match nothing, and return true or
7603 false, as appropriate, when we get to one that can't, or to the
7604 matching stop_memory. */
7606 switch ((re_opcode_t) *p1)
7608 /* Could be either a loop or a series of alternatives. */
7609 case on_failure_jump:
7611 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7613 /* If the next operation is not a jump backwards in the
7618 /* Go through the on_failure_jumps of the alternatives,
7619 seeing if any of the alternatives cannot match nothing.
7620 The last alternative starts with only a jump,
7621 whereas the rest start with on_failure_jump and end
7622 with a jump, e.g., here is the pattern for `a|b|c':
7624 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7625 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7628 So, we have to first go through the first (n-1)
7629 alternatives and then deal with the last one separately. */
7632 /* Deal with the first (n-1) alternatives, which start
7633 with an on_failure_jump (see above) that jumps to right
7634 past a jump_past_alt. */
7636 while ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] ==
7639 /* `mcnt' holds how many bytes long the alternative
7640 is, including the ending `jump_past_alt' and
7643 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt -
7644 (1 + OFFSET_ADDRESS_SIZE),
7648 /* Move to right after this alternative, including the
7652 /* Break if it's the beginning of an n-th alternative
7653 that doesn't begin with an on_failure_jump. */
7654 if ((re_opcode_t) *p1 != on_failure_jump)
7657 /* Still have to check that it's not an n-th
7658 alternative that starts with an on_failure_jump. */
7660 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7661 if ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] !=
7664 /* Get to the beginning of the n-th alternative. */
7665 p1 -= 1 + OFFSET_ADDRESS_SIZE;
7670 /* Deal with the last alternative: go back and get number
7671 of the `jump_past_alt' just before it. `mcnt' contains
7672 the length of the alternative. */
7673 EXTRACT_NUMBER (mcnt, p1 - OFFSET_ADDRESS_SIZE);
7675 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt, reg_info))
7678 p1 += mcnt; /* Get past the n-th alternative. */
7684 assert (p1[1] == **p);
7690 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7693 } /* while p1 < end */
7696 } /* group_match_null_string_p */
7699 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7700 It expects P to be the first byte of a single alternative and END one
7701 byte past the last. The alternative can contain groups. */
7704 PREFIX(alt_match_null_string_p) (p, end, reg_info)
7706 PREFIX(register_info_type) *reg_info;
7713 /* Skip over opcodes that can match nothing, and break when we get
7714 to one that can't. */
7716 switch ((re_opcode_t) *p1)
7719 case on_failure_jump:
7721 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7726 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7729 } /* while p1 < end */
7732 } /* alt_match_null_string_p */
7735 /* Deals with the ops common to group_match_null_string_p and
7736 alt_match_null_string_p.
7738 Sets P to one after the op and its arguments, if any. */
7741 PREFIX(common_op_match_null_string_p) (p, end, reg_info)
7743 PREFIX(register_info_type) *reg_info;
7750 switch ((re_opcode_t) *p1++)
7770 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
7771 ret = PREFIX(group_match_null_string_p) (&p1, end, reg_info);
7773 /* Have to set this here in case we're checking a group which
7774 contains a group and a back reference to it. */
7776 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
7777 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
7783 /* If this is an optimized succeed_n for zero times, make the jump. */
7785 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7793 /* Get to the number of times to succeed. */
7794 p1 += OFFSET_ADDRESS_SIZE;
7795 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7799 p1 -= 2 * OFFSET_ADDRESS_SIZE;
7800 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7808 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
7813 p1 += 2 * OFFSET_ADDRESS_SIZE;
7816 /* All other opcodes mean we cannot match the empty string. */
7822 } /* common_op_match_null_string_p */
7825 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7826 bytes; nonzero otherwise. */
7829 PREFIX(bcmp_translate) (s1, s2, len, translate)
7830 const CHAR_T *s1, *s2;
7832 RE_TRANSLATE_TYPE translate;
7834 register const UCHAR_T *p1 = (const UCHAR_T *) s1;
7835 register const UCHAR_T *p2 = (const UCHAR_T *) s2;
7839 if (((*p1<=0xff)?translate[*p1++]:*p1++)
7840 != ((*p2<=0xff)?translate[*p2++]:*p2++))
7843 if (translate[*p1++] != translate[*p2++]) return 1;
7851 #else /* not INSIDE_RECURSION */
7853 /* Entry points for GNU code. */
7855 /* re_compile_pattern is the GNU regular expression compiler: it
7856 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7857 Returns 0 if the pattern was valid, otherwise an error string.
7859 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7860 are set in BUFP on entry.
7862 We call regex_compile to do the actual compilation. */
7865 re_compile_pattern (pattern, length, bufp)
7866 const char *pattern;
7868 struct re_pattern_buffer *bufp;
7872 /* GNU code is written to assume at least RE_NREGS registers will be set
7873 (and at least one extra will be -1). */
7874 bufp->regs_allocated = REGS_UNALLOCATED;
7876 /* And GNU code determines whether or not to get register information
7877 by passing null for the REGS argument to re_match, etc., not by
7881 /* Match anchors at newline. */
7882 bufp->newline_anchor = 1;
7885 if (MB_CUR_MAX != 1)
7886 ret = wcs_regex_compile (pattern, length, re_syntax_options, bufp);
7889 ret = byte_regex_compile (pattern, length, re_syntax_options, bufp);
7893 return gettext (re_error_msgid[(int) ret]);
7896 weak_alias (__re_compile_pattern, re_compile_pattern)
7899 /* Entry points compatible with 4.2 BSD regex library. We don't define
7900 them unless specifically requested. */
7902 #if defined _REGEX_RE_COMP || defined _LIBC
7904 /* BSD has one and only one pattern buffer. */
7905 static struct re_pattern_buffer re_comp_buf;
7909 /* Make these definitions weak in libc, so POSIX programs can redefine
7910 these names if they don't use our functions, and still use
7911 regcomp/regexec below without link errors. */
7921 if (!re_comp_buf.buffer)
7922 return gettext ("No previous regular expression");
7926 if (!re_comp_buf.buffer)
7928 re_comp_buf.buffer = (unsigned char *) malloc (200);
7929 if (re_comp_buf.buffer == NULL)
7930 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
7931 re_comp_buf.allocated = 200;
7933 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
7934 if (re_comp_buf.fastmap == NULL)
7935 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
7938 /* Since `re_exec' always passes NULL for the `regs' argument, we
7939 don't need to initialize the pattern buffer fields which affect it. */
7941 /* Match anchors at newlines. */
7942 re_comp_buf.newline_anchor = 1;
7945 if (MB_CUR_MAX != 1)
7946 ret = wcs_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7949 ret = byte_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7954 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7955 return (char *) gettext (re_error_msgid[(int) ret]);
7966 const int len = strlen (s);
7968 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
7971 #endif /* _REGEX_RE_COMP */
7973 /* POSIX.2 functions. Don't define these for Emacs. */
7977 /* regcomp takes a regular expression as a string and compiles it.
7979 PREG is a regex_t *. We do not expect any fields to be initialized,
7980 since POSIX says we shouldn't. Thus, we set
7982 `buffer' to the compiled pattern;
7983 `used' to the length of the compiled pattern;
7984 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7985 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7986 RE_SYNTAX_POSIX_BASIC;
7987 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7988 `fastmap' to an allocated space for the fastmap;
7989 `fastmap_accurate' to zero;
7990 `re_nsub' to the number of subexpressions in PATTERN.
7992 PATTERN is the address of the pattern string.
7994 CFLAGS is a series of bits which affect compilation.
7996 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7997 use POSIX basic syntax.
7999 If REG_NEWLINE is set, then . and [^...] don't match newline.
8000 Also, regexec will try a match beginning after every newline.
8002 If REG_ICASE is set, then we considers upper- and lowercase
8003 versions of letters to be equivalent when matching.
8005 If REG_NOSUB is set, then when PREG is passed to regexec, that
8006 routine will report only success or failure, and nothing about the
8009 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
8010 the return codes and their meanings.) */
8013 regcomp (preg, pattern, cflags)
8015 const char *pattern;
8020 = (cflags & REG_EXTENDED) ?
8021 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
8023 /* regex_compile will allocate the space for the compiled pattern. */
8025 preg->allocated = 0;
8028 /* Try to allocate space for the fastmap. */
8029 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
8031 if (cflags & REG_ICASE)
8036 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
8037 * sizeof (*(RE_TRANSLATE_TYPE)0));
8038 if (preg->translate == NULL)
8039 return (int) REG_ESPACE;
8041 /* Map uppercase characters to corresponding lowercase ones. */
8042 for (i = 0; i < CHAR_SET_SIZE; i++)
8043 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
8046 preg->translate = NULL;
8048 /* If REG_NEWLINE is set, newlines are treated differently. */
8049 if (cflags & REG_NEWLINE)
8050 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
8051 syntax &= ~RE_DOT_NEWLINE;
8052 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
8053 /* It also changes the matching behavior. */
8054 preg->newline_anchor = 1;
8057 preg->newline_anchor = 0;
8059 preg->no_sub = !!(cflags & REG_NOSUB);
8061 /* POSIX says a null character in the pattern terminates it, so we
8062 can use strlen here in compiling the pattern. */
8064 if (MB_CUR_MAX != 1)
8065 ret = wcs_regex_compile (pattern, strlen (pattern), syntax, preg);
8068 ret = byte_regex_compile (pattern, strlen (pattern), syntax, preg);
8070 /* POSIX doesn't distinguish between an unmatched open-group and an
8071 unmatched close-group: both are REG_EPAREN. */
8072 if (ret == REG_ERPAREN) ret = REG_EPAREN;
8074 if (ret == REG_NOERROR && preg->fastmap)
8076 /* Compute the fastmap now, since regexec cannot modify the pattern
8078 if (re_compile_fastmap (preg) == -2)
8080 /* Some error occurred while computing the fastmap, just forget
8082 free (preg->fastmap);
8083 preg->fastmap = NULL;
8090 weak_alias (__regcomp, regcomp)
8094 /* regexec searches for a given pattern, specified by PREG, in the
8097 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8098 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8099 least NMATCH elements, and we set them to the offsets of the
8100 corresponding matched substrings.
8102 EFLAGS specifies `execution flags' which affect matching: if
8103 REG_NOTBOL is set, then ^ does not match at the beginning of the
8104 string; if REG_NOTEOL is set, then $ does not match at the end.
8106 We return 0 if we find a match and REG_NOMATCH if not. */
8109 regexec (preg, string, nmatch, pmatch, eflags)
8110 const regex_t *preg;
8113 regmatch_t pmatch[];
8117 struct re_registers regs;
8118 regex_t private_preg;
8119 int len = strlen (string);
8120 boolean want_reg_info = !preg->no_sub && nmatch > 0;
8122 private_preg = *preg;
8124 private_preg.not_bol = !!(eflags & REG_NOTBOL);
8125 private_preg.not_eol = !!(eflags & REG_NOTEOL);
8127 /* The user has told us exactly how many registers to return
8128 information about, via `nmatch'. We have to pass that on to the
8129 matching routines. */
8130 private_preg.regs_allocated = REGS_FIXED;
8134 regs.num_regs = nmatch;
8135 regs.start = TALLOC (nmatch * 2, regoff_t);
8136 if (regs.start == NULL)
8137 return (int) REG_NOMATCH;
8138 regs.end = regs.start + nmatch;
8141 /* Perform the searching operation. */
8142 ret = re_search (&private_preg, string, len,
8143 /* start: */ 0, /* range: */ len,
8144 want_reg_info ? ®s : (struct re_registers *) 0);
8146 /* Copy the register information to the POSIX structure. */
8153 for (r = 0; r < nmatch; r++)
8155 pmatch[r].rm_so = regs.start[r];
8156 pmatch[r].rm_eo = regs.end[r];
8160 /* If we needed the temporary register info, free the space now. */
8164 /* We want zero return to mean success, unlike `re_search'. */
8165 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
8168 weak_alias (__regexec, regexec)
8172 /* Returns a message corresponding to an error code, ERRCODE, returned
8173 from either regcomp or regexec. We don't use PREG here. */
8176 regerror (errcode, preg, errbuf, errbuf_size)
8178 const regex_t *preg;
8186 || errcode >= (int) (sizeof (re_error_msgid)
8187 / sizeof (re_error_msgid[0])))
8188 /* Only error codes returned by the rest of the code should be passed
8189 to this routine. If we are given anything else, or if other regex
8190 code generates an invalid error code, then the program has a bug.
8191 Dump core so we can fix it. */
8194 msg = gettext (re_error_msgid[errcode]);
8196 msg_size = strlen (msg) + 1; /* Includes the null. */
8198 if (errbuf_size != 0)
8200 if (msg_size > errbuf_size)
8202 #if defined HAVE_MEMPCPY || defined _LIBC
8203 *((char *) mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
8205 memcpy (errbuf, msg, errbuf_size - 1);
8206 errbuf[errbuf_size - 1] = 0;
8210 memcpy (errbuf, msg, msg_size);
8216 weak_alias (__regerror, regerror)
8220 /* Free dynamically allocated space used by PREG. */
8226 if (preg->buffer != NULL)
8227 free (preg->buffer);
8228 preg->buffer = NULL;
8230 preg->allocated = 0;
8233 if (preg->fastmap != NULL)
8234 free (preg->fastmap);
8235 preg->fastmap = NULL;
8236 preg->fastmap_accurate = 0;
8238 if (preg->translate != NULL)
8239 free (preg->translate);
8240 preg->translate = NULL;
8243 weak_alias (__regfree, regfree)
8246 #endif /* not emacs */
8248 #endif /* not INSIDE_RECURSION */
8252 #undef STORE_NUMBER_AND_INCR
8253 #undef EXTRACT_NUMBER
8254 #undef EXTRACT_NUMBER_AND_INCR
8256 #undef DEBUG_PRINT_COMPILED_PATTERN
8257 #undef DEBUG_PRINT_DOUBLE_STRING
8259 #undef INIT_FAIL_STACK
8260 #undef RESET_FAIL_STACK
8261 #undef DOUBLE_FAIL_STACK
8262 #undef PUSH_PATTERN_OP
8263 #undef PUSH_FAILURE_POINTER
8264 #undef PUSH_FAILURE_INT
8265 #undef PUSH_FAILURE_ELT
8266 #undef POP_FAILURE_POINTER
8267 #undef POP_FAILURE_INT
8268 #undef POP_FAILURE_ELT
8271 #undef PUSH_FAILURE_POINT
8272 #undef POP_FAILURE_POINT
8274 #undef REG_UNSET_VALUE
8282 #undef INIT_BUF_SIZE
8283 #undef GET_BUFFER_SPACE
8291 #undef EXTEND_BUFFER
8292 #undef GET_UNSIGNED_NUMBER
8293 #undef FREE_STACK_RETURN
8295 # undef POINTER_TO_OFFSET
8296 # undef MATCHING_IN_FRST_STRING
8298 # undef AT_STRINGS_BEG
8299 # undef AT_STRINGS_END
8302 # undef FREE_VARIABLES
8303 # undef NO_HIGHEST_ACTIVE_REG
8304 # undef NO_LOWEST_ACTIVE_REG
8308 # undef COMPILED_BUFFER_VAR
8309 # undef OFFSET_ADDRESS_SIZE
8310 # undef CHAR_CLASS_SIZE
8317 # define DEFINED_ONCE
OSDN Git Service
