1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993-1999, 2000, 2001 Free Software Foundation, Inc.
6 This file is part of the GNU C Library.
8 The GNU C Library is free software; you can redistribute it and/or
9 modify it under the terms of the GNU Lesser General Public
10 License as published by the Free Software Foundation; either
11 version 2.1 of the License, or (at your option) any later version.
13 The GNU C Library is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 Lesser General Public License for more details.
18 You should have received a copy of the GNU Lesser General Public
19 License along with the GNU C Library; if not, write to the Free
20 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
23 /* 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 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 || 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. */
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));
413 static reg_errcode_t wcs_regex_compile _RE_ARGS ((const char *pattern, size_t size,
415 struct re_pattern_buffer *bufp));
417 static int byte_re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
418 const char *string1, int size1,
419 const char *string2, int size2,
421 struct re_registers *regs,
423 static int wcs_re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
424 const char *cstring1, int csize1,
425 const char *cstring2, int csize2,
427 struct re_registers *regs,
429 wchar_t *string1, int size1,
430 wchar_t *string2, int size2,
431 int *mbs_offset1, int *mbs_offset2));
432 static int byte_re_search_2 PARAMS ((struct re_pattern_buffer *bufp,
433 const char *string1, int size1,
434 const char *string2, int size2,
435 int startpos, int range,
436 struct re_registers *regs, int stop));
437 static int wcs_re_search_2 PARAMS ((struct re_pattern_buffer *bufp,
438 const char *string1, int size1,
439 const char *string2, int size2,
440 int startpos, int range,
441 struct re_registers *regs, int stop));
442 static int byte_re_compile_fastmap PARAMS ((struct re_pattern_buffer *bufp));
443 static int wcs_re_compile_fastmap PARAMS ((struct re_pattern_buffer *bufp));
446 /* These are the command codes that appear in compiled regular
447 expressions. Some opcodes are followed by argument bytes. A
448 command code can specify any interpretation whatsoever for its
449 arguments. Zero bytes may appear in the compiled regular expression. */
455 /* Succeed right away--no more backtracking. */
458 /* Followed by one byte giving n, then by n literal bytes. */
462 /* Same as exactn, but contains binary data. */
466 /* Matches any (more or less) character. */
469 /* Matches any one char belonging to specified set. First
470 following byte is number of bitmap bytes. Then come bytes
471 for a bitmap saying which chars are in. Bits in each byte
472 are ordered low-bit-first. A character is in the set if its
473 bit is 1. A character too large to have a bit in the map is
474 automatically not in the set. */
475 /* ifdef MBS_SUPPORT, following element is length of character
476 classes, length of collating symbols, length of equivalence
477 classes, length of character ranges, and length of characters.
478 Next, character class element, collating symbols elements,
479 equivalence class elements, range elements, and character
481 See regex_compile function. */
484 /* Same parameters as charset, but match any character that is
485 not one of those specified. */
488 /* Start remembering the text that is matched, for storing in a
489 register. Followed by one byte with the register number, in
490 the range 0 to one less than the pattern buffer's re_nsub
491 field. Then followed by one byte with the number of groups
492 inner to this one. (This last has to be part of the
493 start_memory only because we need it in the on_failure_jump
497 /* Stop remembering the text that is matched and store it in a
498 memory register. Followed by one byte with the register
499 number, in the range 0 to one less than `re_nsub' in the
500 pattern buffer, and one byte with the number of inner groups,
501 just like `start_memory'. (We need the number of inner
502 groups here because we don't have any easy way of finding the
503 corresponding start_memory when we're at a stop_memory.) */
506 /* Match a duplicate of something remembered. Followed by one
507 byte containing the register number. */
510 /* Fail unless at beginning of line. */
513 /* Fail unless at end of line. */
516 /* Succeeds if at beginning of buffer (if emacs) or at beginning
517 of string to be matched (if not). */
520 /* Analogously, for end of buffer/string. */
523 /* Followed by two byte relative address to which to jump. */
526 /* Same as jump, but marks the end of an alternative. */
529 /* Followed by two-byte relative address of place to resume at
530 in case of failure. */
531 /* ifdef MBS_SUPPORT, the size of address is 1. */
534 /* Like on_failure_jump, but pushes a placeholder instead of the
535 current string position when executed. */
536 on_failure_keep_string_jump,
538 /* Throw away latest failure point and then jump to following
539 two-byte relative address. */
540 /* ifdef MBS_SUPPORT, the size of address is 1. */
543 /* Change to pop_failure_jump if know won't have to backtrack to
544 match; otherwise change to jump. This is used to jump
545 back to the beginning of a repeat. If what follows this jump
546 clearly won't match what the repeat does, such that we can be
547 sure that there is no use backtracking out of repetitions
548 already matched, then we change it to a pop_failure_jump.
549 Followed by two-byte address. */
550 /* ifdef MBS_SUPPORT, the size of address is 1. */
553 /* Jump to following two-byte address, and push a dummy failure
554 point. This failure point will be thrown away if an attempt
555 is made to use it for a failure. A `+' construct makes this
556 before the first repeat. Also used as an intermediary kind
557 of jump when compiling an alternative. */
558 /* ifdef MBS_SUPPORT, the size of address is 1. */
561 /* Push a dummy failure point and continue. Used at the end of
565 /* Followed by two-byte relative address and two-byte number n.
566 After matching N times, jump to the address upon failure. */
567 /* ifdef MBS_SUPPORT, the size of address is 1. */
570 /* Followed by two-byte relative address, and two-byte number n.
571 Jump to the address N times, then fail. */
572 /* ifdef MBS_SUPPORT, the size of address is 1. */
575 /* Set the following two-byte relative address to the
576 subsequent two-byte number. The address *includes* the two
578 /* ifdef MBS_SUPPORT, the size of address is 1. */
581 wordchar, /* Matches any word-constituent character. */
582 notwordchar, /* Matches any char that is not a word-constituent. */
584 wordbeg, /* Succeeds if at word beginning. */
585 wordend, /* Succeeds if at word end. */
587 wordbound, /* Succeeds if at a word boundary. */
588 notwordbound /* Succeeds if not at a word boundary. */
591 ,before_dot, /* Succeeds if before point. */
592 at_dot, /* Succeeds if at point. */
593 after_dot, /* Succeeds if after point. */
595 /* Matches any character whose syntax is specified. Followed by
596 a byte which contains a syntax code, e.g., Sword. */
599 /* Matches any character whose syntax is not that specified. */
603 #endif /* not INSIDE_RECURSION */
608 # define UCHAR_T unsigned char
609 # define COMPILED_BUFFER_VAR bufp->buffer
610 # define OFFSET_ADDRESS_SIZE 2
611 # define PREFIX(name) byte_##name
612 # define ARG_PREFIX(name) name
613 # define PUT_CHAR(c) putchar (c)
615 # define CHAR_T wchar_t
616 # define UCHAR_T wchar_t
617 # define COMPILED_BUFFER_VAR wc_buffer
618 # define OFFSET_ADDRESS_SIZE 1 /* the size which STORE_NUMBER macro use */
619 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
620 # define PREFIX(name) wcs_##name
621 # define ARG_PREFIX(name) c##name
622 /* Should we use wide stream?? */
623 # define PUT_CHAR(c) printf ("%C", c);
629 # define INSIDE_RECURSION
631 # undef INSIDE_RECURSION
634 # define INSIDE_RECURSION
636 # undef INSIDE_RECURSION
639 #ifdef INSIDE_RECURSION
640 /* Common operations on the compiled pattern. */
642 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
643 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
646 # define STORE_NUMBER(destination, number) \
648 *(destination) = (UCHAR_T)(number); \
651 # define STORE_NUMBER(destination, number) \
653 (destination)[0] = (number) & 0377; \
654 (destination)[1] = (number) >> 8; \
658 /* Same as STORE_NUMBER, except increment DESTINATION to
659 the byte after where the number is stored. Therefore, DESTINATION
660 must be an lvalue. */
661 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
663 # define STORE_NUMBER_AND_INCR(destination, number) \
665 STORE_NUMBER (destination, number); \
666 (destination) += OFFSET_ADDRESS_SIZE; \
669 /* Put into DESTINATION a number stored in two contiguous bytes starting
671 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
674 # define EXTRACT_NUMBER(destination, source) \
676 (destination) = *(source); \
679 # define EXTRACT_NUMBER(destination, source) \
681 (destination) = *(source) & 0377; \
682 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
687 static void PREFIX(extract_number) _RE_ARGS ((int *dest, UCHAR_T *source));
689 PREFIX(extract_number) (dest, source)
696 int temp = SIGN_EXTEND_CHAR (*(source + 1));
697 *dest = *source & 0377;
702 # ifndef EXTRACT_MACROS /* To debug the macros. */
703 # undef EXTRACT_NUMBER
704 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
705 # endif /* not EXTRACT_MACROS */
709 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
710 SOURCE must be an lvalue. */
712 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
714 EXTRACT_NUMBER (destination, source); \
715 (source) += OFFSET_ADDRESS_SIZE; \
719 static void PREFIX(extract_number_and_incr) _RE_ARGS ((int *destination,
722 PREFIX(extract_number_and_incr) (destination, source)
726 PREFIX(extract_number) (destination, *source);
727 *source += OFFSET_ADDRESS_SIZE;
730 # ifndef EXTRACT_MACROS
731 # undef EXTRACT_NUMBER_AND_INCR
732 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
733 PREFIX(extract_number_and_incr) (&dest, &src)
734 # endif /* not EXTRACT_MACROS */
740 /* If DEBUG is defined, Regex prints many voluminous messages about what
741 it is doing (if the variable `debug' is nonzero). If linked with the
742 main program in `iregex.c', you can enter patterns and strings
743 interactively. And if linked with the main program in `main.c' and
744 the other test files, you can run the already-written tests. */
748 # ifndef DEFINED_ONCE
750 /* We use standard I/O for debugging. */
753 /* It is useful to test things that ``must'' be true when debugging. */
758 # define DEBUG_STATEMENT(e) e
759 # define DEBUG_PRINT1(x) if (debug) printf (x)
760 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
761 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
762 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
763 # endif /* not DEFINED_ONCE */
765 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
766 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
767 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
768 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
771 /* Print the fastmap in human-readable form. */
773 # ifndef DEFINED_ONCE
775 print_fastmap (fastmap)
778 unsigned was_a_range = 0;
781 while (i < (1 << BYTEWIDTH))
787 while (i < (1 << BYTEWIDTH) && fastmap[i])
801 # endif /* not DEFINED_ONCE */
804 /* Print a compiled pattern string in human-readable form, starting at
805 the START pointer into it and ending just before the pointer END. */
808 PREFIX(print_partial_compiled_pattern) (start, end)
823 /* Loop over pattern commands. */
827 printf ("%td:\t", p - start);
829 printf ("%ld:\t", (long int) (p - start));
832 switch ((re_opcode_t) *p++)
840 printf ("/exactn/%d", mcnt);
852 printf ("/exactn_bin/%d", mcnt);
855 printf("/%lx", (long int) *p++);
859 # endif /* MBS_SUPPORT */
863 printf ("/start_memory/%d/%ld", mcnt, (long int) *p++);
868 printf ("/stop_memory/%d/%ld", mcnt, (long int) *p++);
872 printf ("/duplicate/%ld", (long int) *p++);
885 printf ("/charset [%s",
886 (re_opcode_t) *(workp - 1) == charset_not ? "^" : "");
888 length = *workp++; /* the length of char_classes */
889 for (i=0 ; i<length ; i++)
890 printf("[:%lx:]", (long int) *p++);
891 length = *workp++; /* the length of collating_symbol */
892 for (i=0 ; i<length ;)
896 PUT_CHAR((i++,*p++));
900 length = *workp++; /* the length of equivalence_class */
901 for (i=0 ; i<length ;)
905 PUT_CHAR((i++,*p++));
909 length = *workp++; /* the length of char_range */
910 for (i=0 ; i<length ; i++)
912 wchar_t range_start = *p++;
913 wchar_t range_end = *p++;
914 printf("%C-%C", range_start, range_end);
916 length = *workp++; /* the length of char */
917 for (i=0 ; i<length ; i++)
921 register int c, last = -100;
922 register int in_range = 0;
924 printf ("/charset [%s",
925 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
927 assert (p + *p < pend);
929 for (c = 0; c < 256; c++)
931 && (p[1 + (c/8)] & (1 << (c % 8))))
933 /* Are we starting a range? */
934 if (last + 1 == c && ! in_range)
939 /* Have we broken a range? */
940 else if (last + 1 != c && in_range)
970 case on_failure_jump:
971 PREFIX(extract_number_and_incr) (&mcnt, &p);
973 printf ("/on_failure_jump to %td", p + mcnt - start);
975 printf ("/on_failure_jump to %ld", (long int) (p + mcnt - start));
979 case on_failure_keep_string_jump:
980 PREFIX(extract_number_and_incr) (&mcnt, &p);
982 printf ("/on_failure_keep_string_jump to %td", p + mcnt - start);
984 printf ("/on_failure_keep_string_jump to %ld",
985 (long int) (p + mcnt - start));
989 case dummy_failure_jump:
990 PREFIX(extract_number_and_incr) (&mcnt, &p);
992 printf ("/dummy_failure_jump to %td", p + mcnt - start);
994 printf ("/dummy_failure_jump to %ld", (long int) (p + mcnt - start));
998 case push_dummy_failure:
999 printf ("/push_dummy_failure");
1002 case maybe_pop_jump:
1003 PREFIX(extract_number_and_incr) (&mcnt, &p);
1005 printf ("/maybe_pop_jump to %td", p + mcnt - start);
1007 printf ("/maybe_pop_jump to %ld", (long int) (p + mcnt - start));
1011 case pop_failure_jump:
1012 PREFIX(extract_number_and_incr) (&mcnt, &p);
1014 printf ("/pop_failure_jump to %td", p + mcnt - start);
1016 printf ("/pop_failure_jump to %ld", (long int) (p + mcnt - start));
1021 PREFIX(extract_number_and_incr) (&mcnt, &p);
1023 printf ("/jump_past_alt to %td", p + mcnt - start);
1025 printf ("/jump_past_alt to %ld", (long int) (p + mcnt - start));
1030 PREFIX(extract_number_and_incr) (&mcnt, &p);
1032 printf ("/jump to %td", p + mcnt - start);
1034 printf ("/jump to %ld", (long int) (p + mcnt - start));
1039 PREFIX(extract_number_and_incr) (&mcnt, &p);
1041 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1043 printf ("/succeed_n to %td, %d times", p1 - start, mcnt2);
1045 printf ("/succeed_n to %ld, %d times",
1046 (long int) (p1 - start), mcnt2);
1051 PREFIX(extract_number_and_incr) (&mcnt, &p);
1053 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1054 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
1058 PREFIX(extract_number_and_incr) (&mcnt, &p);
1060 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1062 printf ("/set_number_at location %td to %d", p1 - start, mcnt2);
1064 printf ("/set_number_at location %ld to %d",
1065 (long int) (p1 - start), mcnt2);
1070 printf ("/wordbound");
1074 printf ("/notwordbound");
1078 printf ("/wordbeg");
1082 printf ("/wordend");
1087 printf ("/before_dot");
1095 printf ("/after_dot");
1099 printf ("/syntaxspec");
1101 printf ("/%d", mcnt);
1105 printf ("/notsyntaxspec");
1107 printf ("/%d", mcnt);
1112 printf ("/wordchar");
1116 printf ("/notwordchar");
1128 printf ("?%ld", (long int) *(p-1));
1135 printf ("%td:\tend of pattern.\n", p - start);
1137 printf ("%ld:\tend of pattern.\n", (long int) (p - start));
1143 PREFIX(print_compiled_pattern) (bufp)
1144 struct re_pattern_buffer *bufp;
1146 UCHAR_T *buffer = (UCHAR_T*) bufp->buffer;
1148 PREFIX(print_partial_compiled_pattern) (buffer, buffer
1149 + bufp->used / sizeof(UCHAR_T));
1150 printf ("%ld bytes used/%ld bytes allocated.\n",
1151 bufp->used, bufp->allocated);
1153 if (bufp->fastmap_accurate && bufp->fastmap)
1155 printf ("fastmap: ");
1156 print_fastmap (bufp->fastmap);
1160 printf ("re_nsub: %Zd\t", bufp->re_nsub);
1162 printf ("re_nsub: %ld\t", (long int) bufp->re_nsub);
1164 printf ("regs_alloc: %d\t", bufp->regs_allocated);
1165 printf ("can_be_null: %d\t", bufp->can_be_null);
1166 printf ("newline_anchor: %d\n", bufp->newline_anchor);
1167 printf ("no_sub: %d\t", bufp->no_sub);
1168 printf ("not_bol: %d\t", bufp->not_bol);
1169 printf ("not_eol: %d\t", bufp->not_eol);
1170 printf ("syntax: %lx\n", bufp->syntax);
1171 /* Perhaps we should print the translate table? */
1176 PREFIX(print_double_string) (where, string1, size1, string2, size2)
1177 const CHAR_T *where;
1178 const CHAR_T *string1;
1179 const CHAR_T *string2;
1191 if (FIRST_STRING_P (where))
1193 for (this_char = where - string1; this_char < size1; this_char++)
1194 PUT_CHAR (string1[this_char]);
1200 for (this_char = where - string2; this_char < size2; this_char++)
1202 PUT_CHAR (string2[this_char]);
1205 fputs ("...", stdout);
1212 # ifndef DEFINED_ONCE
1221 # else /* not DEBUG */
1223 # ifndef DEFINED_ONCE
1227 # define DEBUG_STATEMENT(e)
1228 # define DEBUG_PRINT1(x)
1229 # define DEBUG_PRINT2(x1, x2)
1230 # define DEBUG_PRINT3(x1, x2, x3)
1231 # define DEBUG_PRINT4(x1, x2, x3, x4)
1232 # endif /* not DEFINED_ONCE */
1233 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1234 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1236 # endif /* not DEBUG */
1241 /* This convert a multibyte string to a wide character string.
1242 And write their correspondances to offset_buffer(see below)
1243 and write whether each wchar_t is binary data to is_binary.
1244 This assume invalid multibyte sequences as binary data.
1245 We assume offset_buffer and is_binary is already allocated
1248 static size_t convert_mbs_to_wcs (CHAR_T *dest, const unsigned char* src,
1249 size_t len, int *offset_buffer,
1252 convert_mbs_to_wcs (dest, src, len, offset_buffer, is_binary)
1254 const unsigned char* src;
1255 size_t len; /* the length of multibyte string. */
1257 /* It hold correspondances between src(char string) and
1258 dest(wchar_t string) for optimization.
1260 dest = {'X', 'Y', 'Z'}
1261 (each "xxx", "y" and "zz" represent one multibyte character
1262 corresponding to 'X', 'Y' and 'Z'.)
1263 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1269 wchar_t *pdest = dest;
1270 const unsigned char *psrc = src;
1271 size_t wc_count = 0;
1275 size_t mb_remain = len;
1276 size_t mb_count = 0;
1278 /* Initialize the conversion state. */
1279 memset (&mbs, 0, sizeof (mbstate_t));
1281 offset_buffer[0] = 0;
1282 for( ; mb_remain > 0 ; ++wc_count, ++pdest, mb_remain -= consumed,
1285 consumed = mbrtowc (pdest, psrc, mb_remain, &mbs);
1288 /* failed to convert. maybe src contains binary data.
1289 So we consume 1 byte manualy. */
1293 is_binary[wc_count] = TRUE;
1296 is_binary[wc_count] = FALSE;
1297 /* In sjis encoding, we use yen sign as escape character in
1298 place of reverse solidus. So we convert 0x5c(yen sign in
1299 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1300 solidus in UCS2). */
1301 if (consumed == 1 && (int) *psrc == 0x5c && (int) *pdest == 0xa5)
1302 *pdest = (wchar_t) *psrc;
1304 offset_buffer[wc_count + 1] = mb_count += consumed;
1307 /* Fill remain of the buffer with sentinel. */
1308 for (i = wc_count + 1 ; i <= len ; i++)
1309 offset_buffer[i] = mb_count + 1;
1316 #else /* not INSIDE_RECURSION */
1318 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1319 also be assigned to arbitrarily: each pattern buffer stores its own
1320 syntax, so it can be changed between regex compilations. */
1321 /* This has no initializer because initialized variables in Emacs
1322 become read-only after dumping. */
1323 reg_syntax_t re_syntax_options;
1326 /* Specify the precise syntax of regexps for compilation. This provides
1327 for compatibility for various utilities which historically have
1328 different, incompatible syntaxes.
1330 The argument SYNTAX is a bit mask comprised of the various bits
1331 defined in regex.h. We return the old syntax. */
1334 re_set_syntax (syntax)
1335 reg_syntax_t syntax;
1337 reg_syntax_t ret = re_syntax_options;
1339 re_syntax_options = syntax;
1341 if (syntax & RE_DEBUG)
1343 else if (debug) /* was on but now is not */
1349 weak_alias (__re_set_syntax, re_set_syntax)
1352 /* This table gives an error message for each of the error codes listed
1353 in regex.h. Obviously the order here has to be same as there.
1354 POSIX doesn't require that we do anything for REG_NOERROR,
1355 but why not be nice? */
1357 static const char re_error_msgid[] =
1359 # define REG_NOERROR_IDX 0
1360 gettext_noop ("Success") /* REG_NOERROR */
1362 # define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success")
1363 gettext_noop ("No match") /* REG_NOMATCH */
1365 # define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match")
1366 gettext_noop ("Invalid regular expression") /* REG_BADPAT */
1368 # define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression")
1369 gettext_noop ("Invalid collation character") /* REG_ECOLLATE */
1371 # define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character")
1372 gettext_noop ("Invalid character class name") /* REG_ECTYPE */
1374 # define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name")
1375 gettext_noop ("Trailing backslash") /* REG_EESCAPE */
1377 # define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash")
1378 gettext_noop ("Invalid back reference") /* REG_ESUBREG */
1380 # define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference")
1381 gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */
1383 # define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^")
1384 gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */
1386 # define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(")
1387 gettext_noop ("Unmatched \\{") /* REG_EBRACE */
1389 # define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{")
1390 gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */
1392 # define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}")
1393 gettext_noop ("Invalid range end") /* REG_ERANGE */
1395 # define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end")
1396 gettext_noop ("Memory exhausted") /* REG_ESPACE */
1398 # define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted")
1399 gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */
1401 # define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression")
1402 gettext_noop ("Premature end of regular expression") /* REG_EEND */
1404 # define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression")
1405 gettext_noop ("Regular expression too big") /* REG_ESIZE */
1407 # define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big")
1408 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
1411 static const size_t re_error_msgid_idx[] =
1432 #endif /* INSIDE_RECURSION */
1434 #ifndef DEFINED_ONCE
1435 /* Avoiding alloca during matching, to placate r_alloc. */
1437 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1438 searching and matching functions should not call alloca. On some
1439 systems, alloca is implemented in terms of malloc, and if we're
1440 using the relocating allocator routines, then malloc could cause a
1441 relocation, which might (if the strings being searched are in the
1442 ralloc heap) shift the data out from underneath the regexp
1445 Here's another reason to avoid allocation: Emacs
1446 processes input from X in a signal handler; processing X input may
1447 call malloc; if input arrives while a matching routine is calling
1448 malloc, then we're scrod. But Emacs can't just block input while
1449 calling matching routines; then we don't notice interrupts when
1450 they come in. So, Emacs blocks input around all regexp calls
1451 except the matching calls, which it leaves unprotected, in the
1452 faith that they will not malloc. */
1454 /* Normally, this is fine. */
1455 # define MATCH_MAY_ALLOCATE
1457 /* When using GNU C, we are not REALLY using the C alloca, no matter
1458 what config.h may say. So don't take precautions for it. */
1463 /* The match routines may not allocate if (1) they would do it with malloc
1464 and (2) it's not safe for them to use malloc.
1465 Note that if REL_ALLOC is defined, matching would not use malloc for the
1466 failure stack, but we would still use it for the register vectors;
1467 so REL_ALLOC should not affect this. */
1468 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1469 # undef MATCH_MAY_ALLOCATE
1471 #endif /* not DEFINED_ONCE */
1473 #ifdef INSIDE_RECURSION
1474 /* Failure stack declarations and macros; both re_compile_fastmap and
1475 re_match_2 use a failure stack. These have to be macros because of
1476 REGEX_ALLOCATE_STACK. */
1479 /* Number of failure points for which to initially allocate space
1480 when matching. If this number is exceeded, we allocate more
1481 space, so it is not a hard limit. */
1482 # ifndef INIT_FAILURE_ALLOC
1483 # define INIT_FAILURE_ALLOC 5
1486 /* Roughly the maximum number of failure points on the stack. Would be
1487 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1488 This is a variable only so users of regex can assign to it; we never
1489 change it ourselves. */
1491 # ifdef INT_IS_16BIT
1493 # ifndef DEFINED_ONCE
1494 # if defined MATCH_MAY_ALLOCATE
1495 /* 4400 was enough to cause a crash on Alpha OSF/1,
1496 whose default stack limit is 2mb. */
1497 long int re_max_failures = 4000;
1499 long int re_max_failures = 2000;
1503 union PREFIX(fail_stack_elt)
1509 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1513 PREFIX(fail_stack_elt_t) *stack;
1514 unsigned long int size;
1515 unsigned long int avail; /* Offset of next open position. */
1516 } PREFIX(fail_stack_type);
1518 # else /* not INT_IS_16BIT */
1520 # ifndef DEFINED_ONCE
1521 # if defined MATCH_MAY_ALLOCATE
1522 /* 4400 was enough to cause a crash on Alpha OSF/1,
1523 whose default stack limit is 2mb. */
1524 int re_max_failures = 4000;
1526 int re_max_failures = 2000;
1530 union PREFIX(fail_stack_elt)
1536 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1540 PREFIX(fail_stack_elt_t) *stack;
1542 unsigned avail; /* Offset of next open position. */
1543 } PREFIX(fail_stack_type);
1545 # endif /* INT_IS_16BIT */
1547 # ifndef DEFINED_ONCE
1548 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1549 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1550 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1554 /* Define macros to initialize and free the failure stack.
1555 Do `return -2' if the alloc fails. */
1557 # ifdef MATCH_MAY_ALLOCATE
1558 # define INIT_FAIL_STACK() \
1560 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1561 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1563 if (fail_stack.stack == NULL) \
1566 fail_stack.size = INIT_FAILURE_ALLOC; \
1567 fail_stack.avail = 0; \
1570 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1572 # define INIT_FAIL_STACK() \
1574 fail_stack.avail = 0; \
1577 # define RESET_FAIL_STACK()
1581 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1583 Return 1 if succeeds, and 0 if either ran out of memory
1584 allocating space for it or it was already too large.
1586 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1588 # define DOUBLE_FAIL_STACK(fail_stack) \
1589 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1591 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1592 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1593 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1594 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1596 (fail_stack).stack == NULL \
1598 : ((fail_stack).size <<= 1, \
1602 /* Push pointer POINTER on FAIL_STACK.
1603 Return 1 if was able to do so and 0 if ran out of memory allocating
1605 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1606 ((FAIL_STACK_FULL () \
1607 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1609 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1612 /* Push a pointer value onto the failure stack.
1613 Assumes the variable `fail_stack'. Probably should only
1614 be called from within `PUSH_FAILURE_POINT'. */
1615 # define PUSH_FAILURE_POINTER(item) \
1616 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1618 /* This pushes an integer-valued item onto the failure stack.
1619 Assumes the variable `fail_stack'. Probably should only
1620 be called from within `PUSH_FAILURE_POINT'. */
1621 # define PUSH_FAILURE_INT(item) \
1622 fail_stack.stack[fail_stack.avail++].integer = (item)
1624 /* Push a fail_stack_elt_t value onto the failure stack.
1625 Assumes the variable `fail_stack'. Probably should only
1626 be called from within `PUSH_FAILURE_POINT'. */
1627 # define PUSH_FAILURE_ELT(item) \
1628 fail_stack.stack[fail_stack.avail++] = (item)
1630 /* These three POP... operations complement the three PUSH... operations.
1631 All assume that `fail_stack' is nonempty. */
1632 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1633 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1634 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1636 /* Used to omit pushing failure point id's when we're not debugging. */
1638 # define DEBUG_PUSH PUSH_FAILURE_INT
1639 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1641 # define DEBUG_PUSH(item)
1642 # define DEBUG_POP(item_addr)
1646 /* Push the information about the state we will need
1647 if we ever fail back to it.
1649 Requires variables fail_stack, regstart, regend, reg_info, and
1650 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1653 Does `return FAILURE_CODE' if runs out of memory. */
1655 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1657 char *destination; \
1658 /* Must be int, so when we don't save any registers, the arithmetic \
1659 of 0 + -1 isn't done as unsigned. */ \
1660 /* Can't be int, since there is not a shred of a guarantee that int \
1661 is wide enough to hold a value of something to which pointer can \
1663 active_reg_t this_reg; \
1665 DEBUG_STATEMENT (failure_id++); \
1666 DEBUG_STATEMENT (nfailure_points_pushed++); \
1667 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1668 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1669 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1671 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1672 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1674 /* Ensure we have enough space allocated for what we will push. */ \
1675 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1677 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1678 return failure_code; \
1680 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1681 (fail_stack).size); \
1682 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1685 /* Push the info, starting with the registers. */ \
1686 DEBUG_PRINT1 ("\n"); \
1689 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1692 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1693 DEBUG_STATEMENT (num_regs_pushed++); \
1695 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1696 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1698 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1699 PUSH_FAILURE_POINTER (regend[this_reg]); \
1701 DEBUG_PRINT2 (" info: %p\n ", \
1702 reg_info[this_reg].word.pointer); \
1703 DEBUG_PRINT2 (" match_null=%d", \
1704 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1705 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1706 DEBUG_PRINT2 (" matched_something=%d", \
1707 MATCHED_SOMETHING (reg_info[this_reg])); \
1708 DEBUG_PRINT2 (" ever_matched=%d", \
1709 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1710 DEBUG_PRINT1 ("\n"); \
1711 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1714 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1715 PUSH_FAILURE_INT (lowest_active_reg); \
1717 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1718 PUSH_FAILURE_INT (highest_active_reg); \
1720 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1721 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1722 PUSH_FAILURE_POINTER (pattern_place); \
1724 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1725 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1727 DEBUG_PRINT1 ("'\n"); \
1728 PUSH_FAILURE_POINTER (string_place); \
1730 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1731 DEBUG_PUSH (failure_id); \
1734 # ifndef DEFINED_ONCE
1735 /* This is the number of items that are pushed and popped on the stack
1736 for each register. */
1737 # define NUM_REG_ITEMS 3
1739 /* Individual items aside from the registers. */
1741 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1743 # define NUM_NONREG_ITEMS 4
1746 /* We push at most this many items on the stack. */
1747 /* We used to use (num_regs - 1), which is the number of registers
1748 this regexp will save; but that was changed to 5
1749 to avoid stack overflow for a regexp with lots of parens. */
1750 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1752 /* We actually push this many items. */
1753 # define NUM_FAILURE_ITEMS \
1755 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1759 /* How many items can still be added to the stack without overflowing it. */
1760 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1761 # endif /* not DEFINED_ONCE */
1764 /* Pops what PUSH_FAIL_STACK pushes.
1766 We restore into the parameters, all of which should be lvalues:
1767 STR -- the saved data position.
1768 PAT -- the saved pattern position.
1769 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1770 REGSTART, REGEND -- arrays of string positions.
1771 REG_INFO -- array of information about each subexpression.
1773 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1774 `pend', `string1', `size1', `string2', and `size2'. */
1775 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1777 DEBUG_STATEMENT (unsigned failure_id;) \
1778 active_reg_t this_reg; \
1779 const UCHAR_T *string_temp; \
1781 assert (!FAIL_STACK_EMPTY ()); \
1783 /* Remove failure points and point to how many regs pushed. */ \
1784 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1785 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1786 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1788 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1790 DEBUG_POP (&failure_id); \
1791 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1793 /* If the saved string location is NULL, it came from an \
1794 on_failure_keep_string_jump opcode, and we want to throw away the \
1795 saved NULL, thus retaining our current position in the string. */ \
1796 string_temp = POP_FAILURE_POINTER (); \
1797 if (string_temp != NULL) \
1798 str = (const CHAR_T *) string_temp; \
1800 DEBUG_PRINT2 (" Popping string %p: `", str); \
1801 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1802 DEBUG_PRINT1 ("'\n"); \
1804 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1805 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1806 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1808 /* Restore register info. */ \
1809 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1810 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1812 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1813 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1816 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1818 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1820 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1821 DEBUG_PRINT2 (" info: %p\n", \
1822 reg_info[this_reg].word.pointer); \
1824 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1825 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1827 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1828 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1832 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1834 reg_info[this_reg].word.integer = 0; \
1835 regend[this_reg] = 0; \
1836 regstart[this_reg] = 0; \
1838 highest_active_reg = high_reg; \
1841 set_regs_matched_done = 0; \
1842 DEBUG_STATEMENT (nfailure_points_popped++); \
1843 } /* POP_FAILURE_POINT */
1845 /* Structure for per-register (a.k.a. per-group) information.
1846 Other register information, such as the
1847 starting and ending positions (which are addresses), and the list of
1848 inner groups (which is a bits list) are maintained in separate
1851 We are making a (strictly speaking) nonportable assumption here: that
1852 the compiler will pack our bit fields into something that fits into
1853 the type of `word', i.e., is something that fits into one item on the
1857 /* Declarations and macros for re_match_2. */
1861 PREFIX(fail_stack_elt_t) word;
1864 /* This field is one if this group can match the empty string,
1865 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1866 # define MATCH_NULL_UNSET_VALUE 3
1867 unsigned match_null_string_p : 2;
1868 unsigned is_active : 1;
1869 unsigned matched_something : 1;
1870 unsigned ever_matched_something : 1;
1872 } PREFIX(register_info_type);
1874 # ifndef DEFINED_ONCE
1875 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1876 # define IS_ACTIVE(R) ((R).bits.is_active)
1877 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1878 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1881 /* Call this when have matched a real character; it sets `matched' flags
1882 for the subexpressions which we are currently inside. Also records
1883 that those subexprs have matched. */
1884 # define SET_REGS_MATCHED() \
1887 if (!set_regs_matched_done) \
1890 set_regs_matched_done = 1; \
1891 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1893 MATCHED_SOMETHING (reg_info[r]) \
1894 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1900 # endif /* not DEFINED_ONCE */
1902 /* Registers are set to a sentinel when they haven't yet matched. */
1903 static CHAR_T PREFIX(reg_unset_dummy);
1904 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1905 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1907 /* Subroutine declarations and macros for regex_compile. */
1908 static void PREFIX(store_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc, int arg));
1909 static void PREFIX(store_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1910 int arg1, int arg2));
1911 static void PREFIX(insert_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1912 int arg, UCHAR_T *end));
1913 static void PREFIX(insert_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1914 int arg1, int arg2, UCHAR_T *end));
1915 static boolean PREFIX(at_begline_loc_p) _RE_ARGS ((const CHAR_T *pattern,
1917 reg_syntax_t syntax));
1918 static boolean PREFIX(at_endline_loc_p) _RE_ARGS ((const CHAR_T *p,
1920 reg_syntax_t syntax));
1922 static reg_errcode_t wcs_compile_range _RE_ARGS ((CHAR_T range_start,
1923 const CHAR_T **p_ptr,
1926 reg_syntax_t syntax,
1929 static void insert_space _RE_ARGS ((int num, CHAR_T *loc, CHAR_T *end));
1931 static reg_errcode_t byte_compile_range _RE_ARGS ((unsigned int range_start,
1935 reg_syntax_t syntax,
1939 /* Fetch the next character in the uncompiled pattern---translating it
1940 if necessary. Also cast from a signed character in the constant
1941 string passed to us by the user to an unsigned char that we can use
1942 as an array index (in, e.g., `translate'). */
1943 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1944 because it is impossible to allocate 4GB array for some encodings
1945 which have 4 byte character_set like UCS4. */
1948 # define PATFETCH(c) \
1949 do {if (p == pend) return REG_EEND; \
1950 c = (UCHAR_T) *p++; \
1951 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1954 # define PATFETCH(c) \
1955 do {if (p == pend) return REG_EEND; \
1956 c = (unsigned char) *p++; \
1957 if (translate) c = (unsigned char) translate[c]; \
1962 /* Fetch the next character in the uncompiled pattern, with no
1964 # define PATFETCH_RAW(c) \
1965 do {if (p == pend) return REG_EEND; \
1966 c = (UCHAR_T) *p++; \
1969 /* Go backwards one character in the pattern. */
1970 # define PATUNFETCH p--
1973 /* If `translate' is non-null, return translate[D], else just D. We
1974 cast the subscript to translate because some data is declared as
1975 `char *', to avoid warnings when a string constant is passed. But
1976 when we use a character as a subscript we must make it unsigned. */
1977 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1978 because it is impossible to allocate 4GB array for some encodings
1979 which have 4 byte character_set like UCS4. */
1983 # define TRANSLATE(d) \
1984 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1985 ? (char) translate[(unsigned char) (d)] : (d))
1987 # define TRANSLATE(d) \
1988 (translate ? (char) translate[(unsigned char) (d)] : (d))
1993 /* Macros for outputting the compiled pattern into `buffer'. */
1995 /* If the buffer isn't allocated when it comes in, use this. */
1996 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1998 /* Make sure we have at least N more bytes of space in buffer. */
2000 # define GET_BUFFER_SPACE(n) \
2001 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
2002 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
2005 # define GET_BUFFER_SPACE(n) \
2006 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
2010 /* Make sure we have one more byte of buffer space and then add C to it. */
2011 # define BUF_PUSH(c) \
2013 GET_BUFFER_SPACE (1); \
2014 *b++ = (UCHAR_T) (c); \
2018 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
2019 # define BUF_PUSH_2(c1, c2) \
2021 GET_BUFFER_SPACE (2); \
2022 *b++ = (UCHAR_T) (c1); \
2023 *b++ = (UCHAR_T) (c2); \
2027 /* As with BUF_PUSH_2, except for three bytes. */
2028 # define BUF_PUSH_3(c1, c2, c3) \
2030 GET_BUFFER_SPACE (3); \
2031 *b++ = (UCHAR_T) (c1); \
2032 *b++ = (UCHAR_T) (c2); \
2033 *b++ = (UCHAR_T) (c3); \
2036 /* Store a jump with opcode OP at LOC to location TO. We store a
2037 relative address offset by the three bytes the jump itself occupies. */
2038 # define STORE_JUMP(op, loc, to) \
2039 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
2041 /* Likewise, for a two-argument jump. */
2042 # define STORE_JUMP2(op, loc, to, arg) \
2043 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
2045 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
2046 # define INSERT_JUMP(op, loc, to) \
2047 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
2049 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
2050 # define INSERT_JUMP2(op, loc, to, arg) \
2051 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
2054 /* This is not an arbitrary limit: the arguments which represent offsets
2055 into the pattern are two bytes long. So if 2^16 bytes turns out to
2056 be too small, many things would have to change. */
2057 /* Any other compiler which, like MSC, has allocation limit below 2^16
2058 bytes will have to use approach similar to what was done below for
2059 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2060 reallocating to 0 bytes. Such thing is not going to work too well.
2061 You have been warned!! */
2062 # ifndef DEFINED_ONCE
2063 # if defined _MSC_VER && !defined WIN32
2064 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2065 The REALLOC define eliminates a flurry of conversion warnings,
2066 but is not required. */
2067 # define MAX_BUF_SIZE 65500L
2068 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2070 # define MAX_BUF_SIZE (1L << 16)
2071 # define REALLOC(p,s) realloc ((p), (s))
2074 /* Extend the buffer by twice its current size via realloc and
2075 reset the pointers that pointed into the old block to point to the
2076 correct places in the new one. If extending the buffer results in it
2077 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2078 # if __BOUNDED_POINTERS__
2079 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2080 # define MOVE_BUFFER_POINTER(P) \
2081 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2082 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2085 SET_HIGH_BOUND (b); \
2086 SET_HIGH_BOUND (begalt); \
2087 if (fixup_alt_jump) \
2088 SET_HIGH_BOUND (fixup_alt_jump); \
2090 SET_HIGH_BOUND (laststart); \
2091 if (pending_exact) \
2092 SET_HIGH_BOUND (pending_exact); \
2095 # define MOVE_BUFFER_POINTER(P) (P) += incr
2096 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2098 # endif /* not DEFINED_ONCE */
2101 # define EXTEND_BUFFER() \
2103 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2105 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2107 bufp->allocated <<= 1; \
2108 if (bufp->allocated > MAX_BUF_SIZE) \
2109 bufp->allocated = MAX_BUF_SIZE; \
2110 /* How many characters the new buffer can have? */ \
2111 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2112 if (wchar_count == 0) wchar_count = 1; \
2113 /* Truncate the buffer to CHAR_T align. */ \
2114 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2115 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2116 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2117 if (COMPILED_BUFFER_VAR == NULL) \
2118 return REG_ESPACE; \
2119 /* If the buffer moved, move all the pointers into it. */ \
2120 if (old_buffer != COMPILED_BUFFER_VAR) \
2122 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2123 MOVE_BUFFER_POINTER (b); \
2124 MOVE_BUFFER_POINTER (begalt); \
2125 if (fixup_alt_jump) \
2126 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2128 MOVE_BUFFER_POINTER (laststart); \
2129 if (pending_exact) \
2130 MOVE_BUFFER_POINTER (pending_exact); \
2132 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2135 # define EXTEND_BUFFER() \
2137 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2138 if (bufp->allocated == MAX_BUF_SIZE) \
2140 bufp->allocated <<= 1; \
2141 if (bufp->allocated > MAX_BUF_SIZE) \
2142 bufp->allocated = MAX_BUF_SIZE; \
2143 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2145 if (COMPILED_BUFFER_VAR == NULL) \
2146 return REG_ESPACE; \
2147 /* If the buffer moved, move all the pointers into it. */ \
2148 if (old_buffer != COMPILED_BUFFER_VAR) \
2150 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2151 MOVE_BUFFER_POINTER (b); \
2152 MOVE_BUFFER_POINTER (begalt); \
2153 if (fixup_alt_jump) \
2154 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2156 MOVE_BUFFER_POINTER (laststart); \
2157 if (pending_exact) \
2158 MOVE_BUFFER_POINTER (pending_exact); \
2160 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2164 # ifndef DEFINED_ONCE
2165 /* Since we have one byte reserved for the register number argument to
2166 {start,stop}_memory, the maximum number of groups we can report
2167 things about is what fits in that byte. */
2168 # define MAX_REGNUM 255
2170 /* But patterns can have more than `MAX_REGNUM' registers. We just
2171 ignore the excess. */
2172 typedef unsigned regnum_t;
2175 /* Macros for the compile stack. */
2177 /* Since offsets can go either forwards or backwards, this type needs to
2178 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2179 /* int may be not enough when sizeof(int) == 2. */
2180 typedef long pattern_offset_t;
2184 pattern_offset_t begalt_offset;
2185 pattern_offset_t fixup_alt_jump;
2186 pattern_offset_t inner_group_offset;
2187 pattern_offset_t laststart_offset;
2189 } compile_stack_elt_t;
2194 compile_stack_elt_t *stack;
2196 unsigned avail; /* Offset of next open position. */
2197 } compile_stack_type;
2200 # define INIT_COMPILE_STACK_SIZE 32
2202 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2203 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2205 /* The next available element. */
2206 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2208 # endif /* not DEFINED_ONCE */
2210 /* Set the bit for character C in a list. */
2211 # ifndef DEFINED_ONCE
2212 # define SET_LIST_BIT(c) \
2213 (b[((unsigned char) (c)) / BYTEWIDTH] \
2214 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2215 # endif /* DEFINED_ONCE */
2217 /* Get the next unsigned number in the uncompiled pattern. */
2218 # define GET_UNSIGNED_NUMBER(num) \
2223 if (c < '0' || c > '9') \
2225 if (num <= RE_DUP_MAX) \
2229 num = num * 10 + c - '0'; \
2234 # ifndef DEFINED_ONCE
2235 # if defined _LIBC || WIDE_CHAR_SUPPORT
2236 /* The GNU C library provides support for user-defined character classes
2237 and the functions from ISO C amendement 1. */
2238 # ifdef CHARCLASS_NAME_MAX
2239 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2241 /* This shouldn't happen but some implementation might still have this
2242 problem. Use a reasonable default value. */
2243 # define CHAR_CLASS_MAX_LENGTH 256
2247 # define IS_CHAR_CLASS(string) __wctype (string)
2249 # define IS_CHAR_CLASS(string) wctype (string)
2252 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
2254 # define IS_CHAR_CLASS(string) \
2255 (STREQ (string, "alpha") || STREQ (string, "upper") \
2256 || STREQ (string, "lower") || STREQ (string, "digit") \
2257 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
2258 || STREQ (string, "space") || STREQ (string, "print") \
2259 || STREQ (string, "punct") || STREQ (string, "graph") \
2260 || STREQ (string, "cntrl") || STREQ (string, "blank"))
2262 # endif /* DEFINED_ONCE */
2264 # ifndef MATCH_MAY_ALLOCATE
2266 /* If we cannot allocate large objects within re_match_2_internal,
2267 we make the fail stack and register vectors global.
2268 The fail stack, we grow to the maximum size when a regexp
2270 The register vectors, we adjust in size each time we
2271 compile a regexp, according to the number of registers it needs. */
2273 static PREFIX(fail_stack_type) fail_stack;
2275 /* Size with which the following vectors are currently allocated.
2276 That is so we can make them bigger as needed,
2277 but never make them smaller. */
2278 # ifdef DEFINED_ONCE
2279 static int regs_allocated_size;
2281 static const char ** regstart, ** regend;
2282 static const char ** old_regstart, ** old_regend;
2283 static const char **best_regstart, **best_regend;
2284 static const char **reg_dummy;
2285 # endif /* DEFINED_ONCE */
2287 static PREFIX(register_info_type) *PREFIX(reg_info);
2288 static PREFIX(register_info_type) *PREFIX(reg_info_dummy);
2290 /* Make the register vectors big enough for NUM_REGS registers,
2291 but don't make them smaller. */
2294 PREFIX(regex_grow_registers) (num_regs)
2297 if (num_regs > regs_allocated_size)
2299 RETALLOC_IF (regstart, num_regs, const char *);
2300 RETALLOC_IF (regend, num_regs, const char *);
2301 RETALLOC_IF (old_regstart, num_regs, const char *);
2302 RETALLOC_IF (old_regend, num_regs, const char *);
2303 RETALLOC_IF (best_regstart, num_regs, const char *);
2304 RETALLOC_IF (best_regend, num_regs, const char *);
2305 RETALLOC_IF (PREFIX(reg_info), num_regs, PREFIX(register_info_type));
2306 RETALLOC_IF (reg_dummy, num_regs, const char *);
2307 RETALLOC_IF (PREFIX(reg_info_dummy), num_regs, PREFIX(register_info_type));
2309 regs_allocated_size = num_regs;
2313 # endif /* not MATCH_MAY_ALLOCATE */
2315 # ifndef DEFINED_ONCE
2316 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
2319 # endif /* not DEFINED_ONCE */
2321 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2322 Returns one of error codes defined in `regex.h', or zero for success.
2324 Assumes the `allocated' (and perhaps `buffer') and `translate'
2325 fields are set in BUFP on entry.
2327 If it succeeds, results are put in BUFP (if it returns an error, the
2328 contents of BUFP are undefined):
2329 `buffer' is the compiled pattern;
2330 `syntax' is set to SYNTAX;
2331 `used' is set to the length of the compiled pattern;
2332 `fastmap_accurate' is zero;
2333 `re_nsub' is the number of subexpressions in PATTERN;
2334 `not_bol' and `not_eol' are zero;
2336 The `fastmap' and `newline_anchor' fields are neither
2337 examined nor set. */
2339 /* Return, freeing storage we allocated. */
2341 # define FREE_STACK_RETURN(value) \
2342 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2344 # define FREE_STACK_RETURN(value) \
2345 return (free (compile_stack.stack), value)
2348 static reg_errcode_t
2349 PREFIX(regex_compile) (ARG_PREFIX(pattern), ARG_PREFIX(size), syntax, bufp)
2350 const char *ARG_PREFIX(pattern);
2351 size_t ARG_PREFIX(size);
2352 reg_syntax_t syntax;
2353 struct re_pattern_buffer *bufp;
2355 /* We fetch characters from PATTERN here. Even though PATTERN is
2356 `char *' (i.e., signed), we declare these variables as unsigned, so
2357 they can be reliably used as array indices. */
2358 register UCHAR_T c, c1;
2361 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2362 CHAR_T *pattern, *COMPILED_BUFFER_VAR;
2364 /* offset buffer for optimization. See convert_mbs_to_wc. */
2365 int *mbs_offset = NULL;
2366 /* It hold whether each wchar_t is binary data or not. */
2367 char *is_binary = NULL;
2368 /* A flag whether exactn is handling binary data or not. */
2369 char is_exactn_bin = FALSE;
2372 /* A random temporary spot in PATTERN. */
2375 /* Points to the end of the buffer, where we should append. */
2376 register UCHAR_T *b;
2378 /* Keeps track of unclosed groups. */
2379 compile_stack_type compile_stack;
2381 /* Points to the current (ending) position in the pattern. */
2386 const CHAR_T *p = pattern;
2387 const CHAR_T *pend = pattern + size;
2390 /* How to translate the characters in the pattern. */
2391 RE_TRANSLATE_TYPE translate = bufp->translate;
2393 /* Address of the count-byte of the most recently inserted `exactn'
2394 command. This makes it possible to tell if a new exact-match
2395 character can be added to that command or if the character requires
2396 a new `exactn' command. */
2397 UCHAR_T *pending_exact = 0;
2399 /* Address of start of the most recently finished expression.
2400 This tells, e.g., postfix * where to find the start of its
2401 operand. Reset at the beginning of groups and alternatives. */
2402 UCHAR_T *laststart = 0;
2404 /* Address of beginning of regexp, or inside of last group. */
2407 /* Address of the place where a forward jump should go to the end of
2408 the containing expression. Each alternative of an `or' -- except the
2409 last -- ends with a forward jump of this sort. */
2410 UCHAR_T *fixup_alt_jump = 0;
2412 /* Counts open-groups as they are encountered. Remembered for the
2413 matching close-group on the compile stack, so the same register
2414 number is put in the stop_memory as the start_memory. */
2415 regnum_t regnum = 0;
2418 /* Initialize the wchar_t PATTERN and offset_buffer. */
2419 p = pend = pattern = TALLOC(csize + 1, CHAR_T);
2420 mbs_offset = TALLOC(csize + 1, int);
2421 is_binary = TALLOC(csize + 1, char);
2422 if (pattern == NULL || mbs_offset == NULL || is_binary == NULL)
2429 pattern[csize] = L'\0'; /* sentinel */
2430 size = convert_mbs_to_wcs(pattern, cpattern, csize, mbs_offset, is_binary);
2442 DEBUG_PRINT1 ("\nCompiling pattern: ");
2445 unsigned debug_count;
2447 for (debug_count = 0; debug_count < size; debug_count++)
2448 PUT_CHAR (pattern[debug_count]);
2453 /* Initialize the compile stack. */
2454 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2455 if (compile_stack.stack == NULL)
2465 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2466 compile_stack.avail = 0;
2468 /* Initialize the pattern buffer. */
2469 bufp->syntax = syntax;
2470 bufp->fastmap_accurate = 0;
2471 bufp->not_bol = bufp->not_eol = 0;
2473 /* Set `used' to zero, so that if we return an error, the pattern
2474 printer (for debugging) will think there's no pattern. We reset it
2478 /* Always count groups, whether or not bufp->no_sub is set. */
2481 #if !defined emacs && !defined SYNTAX_TABLE
2482 /* Initialize the syntax table. */
2483 init_syntax_once ();
2486 if (bufp->allocated == 0)
2489 { /* If zero allocated, but buffer is non-null, try to realloc
2490 enough space. This loses if buffer's address is bogus, but
2491 that is the user's responsibility. */
2493 /* Free bufp->buffer and allocate an array for wchar_t pattern
2496 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE/sizeof(UCHAR_T),
2499 RETALLOC (COMPILED_BUFFER_VAR, INIT_BUF_SIZE, UCHAR_T);
2503 { /* Caller did not allocate a buffer. Do it for them. */
2504 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE / sizeof(UCHAR_T),
2508 if (!COMPILED_BUFFER_VAR) FREE_STACK_RETURN (REG_ESPACE);
2510 bufp->buffer = (char*)COMPILED_BUFFER_VAR;
2512 bufp->allocated = INIT_BUF_SIZE;
2516 COMPILED_BUFFER_VAR = (UCHAR_T*) bufp->buffer;
2519 begalt = b = COMPILED_BUFFER_VAR;
2521 /* Loop through the uncompiled pattern until we're at the end. */
2530 if ( /* If at start of pattern, it's an operator. */
2532 /* If context independent, it's an operator. */
2533 || syntax & RE_CONTEXT_INDEP_ANCHORS
2534 /* Otherwise, depends on what's come before. */
2535 || PREFIX(at_begline_loc_p) (pattern, p, syntax))
2545 if ( /* If at end of pattern, it's an operator. */
2547 /* If context independent, it's an operator. */
2548 || syntax & RE_CONTEXT_INDEP_ANCHORS
2549 /* Otherwise, depends on what's next. */
2550 || PREFIX(at_endline_loc_p) (p, pend, syntax))
2560 if ((syntax & RE_BK_PLUS_QM)
2561 || (syntax & RE_LIMITED_OPS))
2565 /* If there is no previous pattern... */
2568 if (syntax & RE_CONTEXT_INVALID_OPS)
2569 FREE_STACK_RETURN (REG_BADRPT);
2570 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2575 /* Are we optimizing this jump? */
2576 boolean keep_string_p = false;
2578 /* 1 means zero (many) matches is allowed. */
2579 char zero_times_ok = 0, many_times_ok = 0;
2581 /* If there is a sequence of repetition chars, collapse it
2582 down to just one (the right one). We can't combine
2583 interval operators with these because of, e.g., `a{2}*',
2584 which should only match an even number of `a's. */
2588 zero_times_ok |= c != '+';
2589 many_times_ok |= c != '?';
2597 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2600 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2602 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2605 if (!(c1 == '+' || c1 == '?'))
2620 /* If we get here, we found another repeat character. */
2623 /* Star, etc. applied to an empty pattern is equivalent
2624 to an empty pattern. */
2628 /* Now we know whether or not zero matches is allowed
2629 and also whether or not two or more matches is allowed. */
2631 { /* More than one repetition is allowed, so put in at the
2632 end a backward relative jump from `b' to before the next
2633 jump we're going to put in below (which jumps from
2634 laststart to after this jump).
2636 But if we are at the `*' in the exact sequence `.*\n',
2637 insert an unconditional jump backwards to the .,
2638 instead of the beginning of the loop. This way we only
2639 push a failure point once, instead of every time
2640 through the loop. */
2641 assert (p - 1 > pattern);
2643 /* Allocate the space for the jump. */
2644 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2646 /* We know we are not at the first character of the pattern,
2647 because laststart was nonzero. And we've already
2648 incremented `p', by the way, to be the character after
2649 the `*'. Do we have to do something analogous here
2650 for null bytes, because of RE_DOT_NOT_NULL? */
2651 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2653 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2654 && !(syntax & RE_DOT_NEWLINE))
2655 { /* We have .*\n. */
2656 STORE_JUMP (jump, b, laststart);
2657 keep_string_p = true;
2660 /* Anything else. */
2661 STORE_JUMP (maybe_pop_jump, b, laststart -
2662 (1 + OFFSET_ADDRESS_SIZE));
2664 /* We've added more stuff to the buffer. */
2665 b += 1 + OFFSET_ADDRESS_SIZE;
2668 /* On failure, jump from laststart to b + 3, which will be the
2669 end of the buffer after this jump is inserted. */
2670 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2672 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2673 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2675 laststart, b + 1 + OFFSET_ADDRESS_SIZE);
2677 b += 1 + OFFSET_ADDRESS_SIZE;
2681 /* At least one repetition is required, so insert a
2682 `dummy_failure_jump' before the initial
2683 `on_failure_jump' instruction of the loop. This
2684 effects a skip over that instruction the first time
2685 we hit that loop. */
2686 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2687 INSERT_JUMP (dummy_failure_jump, laststart, laststart +
2688 2 + 2 * OFFSET_ADDRESS_SIZE);
2689 b += 1 + OFFSET_ADDRESS_SIZE;
2703 boolean had_char_class = false;
2705 CHAR_T range_start = 0xffffffff;
2707 unsigned int range_start = 0xffffffff;
2709 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2712 /* We assume a charset(_not) structure as a wchar_t array.
2713 charset[0] = (re_opcode_t) charset(_not)
2714 charset[1] = l (= length of char_classes)
2715 charset[2] = m (= length of collating_symbols)
2716 charset[3] = n (= length of equivalence_classes)
2717 charset[4] = o (= length of char_ranges)
2718 charset[5] = p (= length of chars)
2720 charset[6] = char_class (wctype_t)
2721 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2723 charset[l+5] = char_class (wctype_t)
2725 charset[l+6] = collating_symbol (wchar_t)
2727 charset[l+m+5] = collating_symbol (wchar_t)
2728 ifdef _LIBC we use the index if
2729 _NL_COLLATE_SYMB_EXTRAMB instead of
2732 charset[l+m+6] = equivalence_classes (wchar_t)
2734 charset[l+m+n+5] = equivalence_classes (wchar_t)
2735 ifdef _LIBC we use the index in
2736 _NL_COLLATE_WEIGHT instead of
2739 charset[l+m+n+6] = range_start
2740 charset[l+m+n+7] = range_end
2742 charset[l+m+n+2o+4] = range_start
2743 charset[l+m+n+2o+5] = range_end
2744 ifdef _LIBC we use the value looked up
2745 in _NL_COLLATE_COLLSEQ instead of
2748 charset[l+m+n+2o+6] = char
2750 charset[l+m+n+2o+p+5] = char
2754 /* We need at least 6 spaces: the opcode, the length of
2755 char_classes, the length of collating_symbols, the length of
2756 equivalence_classes, the length of char_ranges, the length of
2758 GET_BUFFER_SPACE (6);
2760 /* Save b as laststart. And We use laststart as the pointer
2761 to the first element of the charset here.
2762 In other words, laststart[i] indicates charset[i]. */
2765 /* We test `*p == '^' twice, instead of using an if
2766 statement, so we only need one BUF_PUSH. */
2767 BUF_PUSH (*p == '^' ? charset_not : charset);
2771 /* Push the length of char_classes, the length of
2772 collating_symbols, the length of equivalence_classes, the
2773 length of char_ranges and the length of chars. */
2774 BUF_PUSH_3 (0, 0, 0);
2777 /* Remember the first position in the bracket expression. */
2780 /* charset_not matches newline according to a syntax bit. */
2781 if ((re_opcode_t) b[-6] == charset_not
2782 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2785 laststart[5]++; /* Update the length of characters */
2788 /* Read in characters and ranges, setting map bits. */
2791 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2795 /* \ might escape characters inside [...] and [^...]. */
2796 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2798 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2802 laststart[5]++; /* Update the length of chars */
2807 /* Could be the end of the bracket expression. If it's
2808 not (i.e., when the bracket expression is `[]' so
2809 far), the ']' character bit gets set way below. */
2810 if (c == ']' && p != p1 + 1)
2813 /* Look ahead to see if it's a range when the last thing
2814 was a character class. */
2815 if (had_char_class && c == '-' && *p != ']')
2816 FREE_STACK_RETURN (REG_ERANGE);
2818 /* Look ahead to see if it's a range when the last thing
2819 was a character: if this is a hyphen not at the
2820 beginning or the end of a list, then it's the range
2823 && !(p - 2 >= pattern && p[-2] == '[')
2824 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2828 /* Allocate the space for range_start and range_end. */
2829 GET_BUFFER_SPACE (2);
2830 /* Update the pointer to indicate end of buffer. */
2832 ret = wcs_compile_range (range_start, &p, pend, translate,
2833 syntax, b, laststart);
2834 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2835 range_start = 0xffffffff;
2837 else if (p[0] == '-' && p[1] != ']')
2838 { /* This handles ranges made up of characters only. */
2841 /* Move past the `-'. */
2843 /* Allocate the space for range_start and range_end. */
2844 GET_BUFFER_SPACE (2);
2845 /* Update the pointer to indicate end of buffer. */
2847 ret = wcs_compile_range (c, &p, pend, translate, syntax, b,
2849 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2850 range_start = 0xffffffff;
2853 /* See if we're at the beginning of a possible character
2855 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2856 { /* Leave room for the null. */
2857 char str[CHAR_CLASS_MAX_LENGTH + 1];
2862 /* If pattern is `[[:'. */
2863 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2868 if ((c == ':' && *p == ']') || p == pend)
2870 if (c1 < CHAR_CLASS_MAX_LENGTH)
2873 /* This is in any case an invalid class name. */
2878 /* If isn't a word bracketed by `[:' and `:]':
2879 undo the ending character, the letters, and leave
2880 the leading `:' and `[' (but store them as character). */
2881 if (c == ':' && *p == ']')
2886 /* Query the character class as wctype_t. */
2887 wt = IS_CHAR_CLASS (str);
2889 FREE_STACK_RETURN (REG_ECTYPE);
2891 /* Throw away the ] at the end of the character
2895 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2897 /* Allocate the space for character class. */
2898 GET_BUFFER_SPACE(CHAR_CLASS_SIZE);
2899 /* Update the pointer to indicate end of buffer. */
2900 b += CHAR_CLASS_SIZE;
2901 /* Move data which follow character classes
2902 not to violate the data. */
2903 insert_space(CHAR_CLASS_SIZE,
2904 laststart + 6 + laststart[1],
2906 alignedp = ((uintptr_t)(laststart + 6 + laststart[1])
2907 + __alignof__(wctype_t) - 1)
2908 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
2909 /* Store the character class. */
2910 *((wctype_t*)alignedp) = wt;
2911 /* Update length of char_classes */
2912 laststart[1] += CHAR_CLASS_SIZE;
2914 had_char_class = true;
2923 laststart[5] += 2; /* Update the length of characters */
2925 had_char_class = false;
2928 else if (syntax & RE_CHAR_CLASSES && c == '[' && (*p == '='
2931 CHAR_T str[128]; /* Should be large enough. */
2932 CHAR_T delim = *p; /* '=' or '.' */
2935 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
2940 /* If pattern is `[[=' or '[[.'. */
2941 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2946 if ((c == delim && *p == ']') || p == pend)
2948 if (c1 < sizeof (str) - 1)
2951 /* This is in any case an invalid class name. */
2956 if (c == delim && *p == ']' && str[0] != '\0')
2958 unsigned int i, offset;
2959 /* If we have no collation data we use the default
2960 collation in which each character is in a class
2961 by itself. It also means that ASCII is the
2962 character set and therefore we cannot have character
2963 with more than one byte in the multibyte
2966 /* If not defined _LIBC, we push the name and
2967 `\0' for the sake of matching performance. */
2968 int datasize = c1 + 1;
2976 FREE_STACK_RETURN (REG_ECOLLATE);
2981 const int32_t *table;
2982 const int32_t *weights;
2983 const int32_t *extra;
2984 const int32_t *indirect;
2987 /* This #include defines a local function! */
2988 # include <locale/weightwc.h>
2992 /* We push the index for equivalence class. */
2995 table = (const int32_t *)
2996 _NL_CURRENT (LC_COLLATE,
2997 _NL_COLLATE_TABLEWC);
2998 weights = (const int32_t *)
2999 _NL_CURRENT (LC_COLLATE,
3000 _NL_COLLATE_WEIGHTWC);
3001 extra = (const int32_t *)
3002 _NL_CURRENT (LC_COLLATE,
3003 _NL_COLLATE_EXTRAWC);
3004 indirect = (const int32_t *)
3005 _NL_CURRENT (LC_COLLATE,
3006 _NL_COLLATE_INDIRECTWC);
3008 idx = findidx ((const wint_t**)&cp);
3009 if (idx == 0 || cp < (wint_t*) str + c1)
3010 /* This is no valid character. */
3011 FREE_STACK_RETURN (REG_ECOLLATE);
3013 str[0] = (wchar_t)idx;
3015 else /* delim == '.' */
3017 /* We push collation sequence value
3018 for collating symbol. */
3020 const int32_t *symb_table;
3021 const unsigned char *extra;
3028 /* We have to convert the name to a single-byte
3029 string. This is possible since the names
3030 consist of ASCII characters and the internal
3031 representation is UCS4. */
3032 for (i = 0; i < c1; ++i)
3033 char_str[i] = str[i];
3036 _NL_CURRENT_WORD (LC_COLLATE,
3037 _NL_COLLATE_SYMB_HASH_SIZEMB);
3038 symb_table = (const int32_t *)
3039 _NL_CURRENT (LC_COLLATE,
3040 _NL_COLLATE_SYMB_TABLEMB);
3041 extra = (const unsigned char *)
3042 _NL_CURRENT (LC_COLLATE,
3043 _NL_COLLATE_SYMB_EXTRAMB);
3045 /* Locate the character in the hashing table. */
3046 hash = elem_hash (char_str, c1);
3049 elem = hash % table_size;
3050 second = hash % (table_size - 2);
3051 while (symb_table[2 * elem] != 0)
3053 /* First compare the hashing value. */
3054 if (symb_table[2 * elem] == hash
3055 && c1 == extra[symb_table[2 * elem + 1]]
3057 &extra[symb_table[2 * elem + 1]
3060 /* Yep, this is the entry. */
3061 idx = symb_table[2 * elem + 1];
3062 idx += 1 + extra[idx];
3070 if (symb_table[2 * elem] != 0)
3072 /* Compute the index of the byte sequence
3074 idx += 1 + extra[idx];
3075 /* Adjust for the alignment. */
3076 idx = (idx + 3) & ~4;
3078 str[0] = (wchar_t) idx + 4;
3080 else if (symb_table[2 * elem] == 0 && c1 == 1)
3082 /* No valid character. Match it as a
3083 single byte character. */
3084 had_char_class = false;
3086 /* Update the length of characters */
3088 range_start = str[0];
3090 /* Throw away the ] at the end of the
3091 collating symbol. */
3093 /* exit from the switch block. */
3097 FREE_STACK_RETURN (REG_ECOLLATE);
3102 /* Throw away the ] at the end of the equivalence
3103 class (or collating symbol). */
3106 /* Allocate the space for the equivalence class
3107 (or collating symbol) (and '\0' if needed). */
3108 GET_BUFFER_SPACE(datasize);
3109 /* Update the pointer to indicate end of buffer. */
3113 { /* equivalence class */
3114 /* Calculate the offset of char_ranges,
3115 which is next to equivalence_classes. */
3116 offset = laststart[1] + laststart[2]
3119 insert_space(datasize, laststart + offset, b - 1);
3121 /* Write the equivalence_class and \0. */
3122 for (i = 0 ; i < datasize ; i++)
3123 laststart[offset + i] = str[i];
3125 /* Update the length of equivalence_classes. */
3126 laststart[3] += datasize;
3127 had_char_class = true;
3129 else /* delim == '.' */
3130 { /* collating symbol */
3131 /* Calculate the offset of the equivalence_classes,
3132 which is next to collating_symbols. */
3133 offset = laststart[1] + laststart[2] + 6;
3134 /* Insert space and write the collationg_symbol
3136 insert_space(datasize, laststart + offset, b-1);
3137 for (i = 0 ; i < datasize ; i++)
3138 laststart[offset + i] = str[i];
3140 /* In re_match_2_internal if range_start < -1, we
3141 assume -range_start is the offset of the
3142 collating symbol which is specified as
3143 the character of the range start. So we assign
3144 -(laststart[1] + laststart[2] + 6) to
3146 range_start = -(laststart[1] + laststart[2] + 6);
3147 /* Update the length of collating_symbol. */
3148 laststart[2] += datasize;
3149 had_char_class = false;
3159 laststart[5] += 2; /* Update the length of characters */
3160 range_start = delim;
3161 had_char_class = false;
3166 had_char_class = false;
3168 laststart[5]++; /* Update the length of characters */
3174 /* Ensure that we have enough space to push a charset: the
3175 opcode, the length count, and the bitset; 34 bytes in all. */
3176 GET_BUFFER_SPACE (34);
3180 /* We test `*p == '^' twice, instead of using an if
3181 statement, so we only need one BUF_PUSH. */
3182 BUF_PUSH (*p == '^' ? charset_not : charset);
3186 /* Remember the first position in the bracket expression. */
3189 /* Push the number of bytes in the bitmap. */
3190 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
3192 /* Clear the whole map. */
3193 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
3195 /* charset_not matches newline according to a syntax bit. */
3196 if ((re_opcode_t) b[-2] == charset_not
3197 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
3198 SET_LIST_BIT ('\n');
3200 /* Read in characters and ranges, setting map bits. */
3203 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3207 /* \ might escape characters inside [...] and [^...]. */
3208 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
3210 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3218 /* Could be the end of the bracket expression. If it's
3219 not (i.e., when the bracket expression is `[]' so
3220 far), the ']' character bit gets set way below. */
3221 if (c == ']' && p != p1 + 1)
3224 /* Look ahead to see if it's a range when the last thing
3225 was a character class. */
3226 if (had_char_class && c == '-' && *p != ']')
3227 FREE_STACK_RETURN (REG_ERANGE);
3229 /* Look ahead to see if it's a range when the last thing
3230 was a character: if this is a hyphen not at the
3231 beginning or the end of a list, then it's the range
3234 && !(p - 2 >= pattern && p[-2] == '[')
3235 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
3239 = byte_compile_range (range_start, &p, pend, translate,
3241 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3242 range_start = 0xffffffff;
3245 else if (p[0] == '-' && p[1] != ']')
3246 { /* This handles ranges made up of characters only. */
3249 /* Move past the `-'. */
3252 ret = byte_compile_range (c, &p, pend, translate, syntax, b);
3253 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3254 range_start = 0xffffffff;
3257 /* See if we're at the beginning of a possible character
3260 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
3261 { /* Leave room for the null. */
3262 char str[CHAR_CLASS_MAX_LENGTH + 1];
3267 /* If pattern is `[[:'. */
3268 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3273 if ((c == ':' && *p == ']') || p == pend)
3275 if (c1 < CHAR_CLASS_MAX_LENGTH)
3278 /* This is in any case an invalid class name. */
3283 /* If isn't a word bracketed by `[:' and `:]':
3284 undo the ending character, the letters, and leave
3285 the leading `:' and `[' (but set bits for them). */
3286 if (c == ':' && *p == ']')
3288 # if defined _LIBC || WIDE_CHAR_SUPPORT
3289 boolean is_lower = STREQ (str, "lower");
3290 boolean is_upper = STREQ (str, "upper");
3294 wt = IS_CHAR_CLASS (str);
3296 FREE_STACK_RETURN (REG_ECTYPE);
3298 /* Throw away the ] at the end of the character
3302 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3304 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
3307 if (__iswctype (__btowc (ch), wt))
3310 if (iswctype (btowc (ch), wt))
3314 if (translate && (is_upper || is_lower)
3315 && (ISUPPER (ch) || ISLOWER (ch)))
3319 had_char_class = true;
3322 boolean is_alnum = STREQ (str, "alnum");
3323 boolean is_alpha = STREQ (str, "alpha");
3324 boolean is_blank = STREQ (str, "blank");
3325 boolean is_cntrl = STREQ (str, "cntrl");
3326 boolean is_digit = STREQ (str, "digit");
3327 boolean is_graph = STREQ (str, "graph");
3328 boolean is_lower = STREQ (str, "lower");
3329 boolean is_print = STREQ (str, "print");
3330 boolean is_punct = STREQ (str, "punct");
3331 boolean is_space = STREQ (str, "space");
3332 boolean is_upper = STREQ (str, "upper");
3333 boolean is_xdigit = STREQ (str, "xdigit");
3335 if (!IS_CHAR_CLASS (str))
3336 FREE_STACK_RETURN (REG_ECTYPE);
3338 /* Throw away the ] at the end of the character
3342 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3344 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
3346 /* This was split into 3 if's to
3347 avoid an arbitrary limit in some compiler. */
3348 if ( (is_alnum && ISALNUM (ch))
3349 || (is_alpha && ISALPHA (ch))
3350 || (is_blank && ISBLANK (ch))
3351 || (is_cntrl && ISCNTRL (ch)))
3353 if ( (is_digit && ISDIGIT (ch))
3354 || (is_graph && ISGRAPH (ch))
3355 || (is_lower && ISLOWER (ch))
3356 || (is_print && ISPRINT (ch)))
3358 if ( (is_punct && ISPUNCT (ch))
3359 || (is_space && ISSPACE (ch))
3360 || (is_upper && ISUPPER (ch))
3361 || (is_xdigit && ISXDIGIT (ch)))
3363 if ( translate && (is_upper || is_lower)
3364 && (ISUPPER (ch) || ISLOWER (ch)))
3367 had_char_class = true;
3368 # endif /* libc || wctype.h */
3378 had_char_class = false;
3381 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '=')
3383 unsigned char str[MB_LEN_MAX + 1];
3386 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3392 /* If pattern is `[[='. */
3393 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3398 if ((c == '=' && *p == ']') || p == pend)
3400 if (c1 < MB_LEN_MAX)
3403 /* This is in any case an invalid class name. */
3408 if (c == '=' && *p == ']' && str[0] != '\0')
3410 /* If we have no collation data we use the default
3411 collation in which each character is in a class
3412 by itself. It also means that ASCII is the
3413 character set and therefore we cannot have character
3414 with more than one byte in the multibyte
3421 FREE_STACK_RETURN (REG_ECOLLATE);
3423 /* Throw away the ] at the end of the equivalence
3427 /* Set the bit for the character. */
3428 SET_LIST_BIT (str[0]);
3433 /* Try to match the byte sequence in `str' against
3434 those known to the collate implementation.
3435 First find out whether the bytes in `str' are
3436 actually from exactly one character. */
3437 const int32_t *table;
3438 const unsigned char *weights;
3439 const unsigned char *extra;
3440 const int32_t *indirect;
3442 const unsigned char *cp = str;
3445 /* This #include defines a local function! */
3446 # include <locale/weight.h>
3448 table = (const int32_t *)
3449 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
3450 weights = (const unsigned char *)
3451 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
3452 extra = (const unsigned char *)
3453 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
3454 indirect = (const int32_t *)
3455 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
3457 idx = findidx (&cp);
3458 if (idx == 0 || cp < str + c1)
3459 /* This is no valid character. */
3460 FREE_STACK_RETURN (REG_ECOLLATE);
3462 /* Throw away the ] at the end of the equivalence
3466 /* Now we have to go throught the whole table
3467 and find all characters which have the same
3470 XXX Note that this is not entirely correct.
3471 we would have to match multibyte sequences
3472 but this is not possible with the current
3474 for (ch = 1; ch < 256; ++ch)
3475 /* XXX This test would have to be changed if we
3476 would allow matching multibyte sequences. */
3479 int32_t idx2 = table[ch];
3480 size_t len = weights[idx2];
3482 /* Test whether the lenghts match. */
3483 if (weights[idx] == len)
3485 /* They do. New compare the bytes of
3490 && (weights[idx + 1 + cnt]
3491 == weights[idx2 + 1 + cnt]))
3495 /* They match. Mark the character as
3502 had_char_class = true;
3512 had_char_class = false;
3515 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '.')
3517 unsigned char str[128]; /* Should be large enough. */
3520 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3526 /* If pattern is `[[.'. */
3527 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3532 if ((c == '.' && *p == ']') || p == pend)
3534 if (c1 < sizeof (str))
3537 /* This is in any case an invalid class name. */
3542 if (c == '.' && *p == ']' && str[0] != '\0')
3544 /* If we have no collation data we use the default
3545 collation in which each character is the name
3546 for its own class which contains only the one
3547 character. It also means that ASCII is the
3548 character set and therefore we cannot have character
3549 with more than one byte in the multibyte
3556 FREE_STACK_RETURN (REG_ECOLLATE);
3558 /* Throw away the ] at the end of the equivalence
3562 /* Set the bit for the character. */
3563 SET_LIST_BIT (str[0]);
3564 range_start = ((const unsigned char *) str)[0];
3569 /* Try to match the byte sequence in `str' against
3570 those known to the collate implementation.
3571 First find out whether the bytes in `str' are
3572 actually from exactly one character. */
3574 const int32_t *symb_table;
3575 const unsigned char *extra;
3582 _NL_CURRENT_WORD (LC_COLLATE,
3583 _NL_COLLATE_SYMB_HASH_SIZEMB);
3584 symb_table = (const int32_t *)
3585 _NL_CURRENT (LC_COLLATE,
3586 _NL_COLLATE_SYMB_TABLEMB);
3587 extra = (const unsigned char *)
3588 _NL_CURRENT (LC_COLLATE,
3589 _NL_COLLATE_SYMB_EXTRAMB);
3591 /* Locate the character in the hashing table. */
3592 hash = elem_hash (str, c1);
3595 elem = hash % table_size;
3596 second = hash % (table_size - 2);
3597 while (symb_table[2 * elem] != 0)
3599 /* First compare the hashing value. */
3600 if (symb_table[2 * elem] == hash
3601 && c1 == extra[symb_table[2 * elem + 1]]
3603 &extra[symb_table[2 * elem + 1]
3607 /* Yep, this is the entry. */
3608 idx = symb_table[2 * elem + 1];
3609 idx += 1 + extra[idx];
3617 if (symb_table[2 * elem] == 0)
3618 /* This is no valid character. */
3619 FREE_STACK_RETURN (REG_ECOLLATE);
3621 /* Throw away the ] at the end of the equivalence
3625 /* Now add the multibyte character(s) we found
3628 XXX Note that this is not entirely correct.
3629 we would have to match multibyte sequences
3630 but this is not possible with the current
3631 implementation. Also, we have to match
3632 collating symbols, which expand to more than
3633 one file, as a whole and not allow the
3634 individual bytes. */
3637 range_start = extra[idx];
3640 SET_LIST_BIT (extra[idx]);
3645 had_char_class = false;
3655 had_char_class = false;
3660 had_char_class = false;
3666 /* Discard any (non)matching list bytes that are all 0 at the
3667 end of the map. Decrease the map-length byte too. */
3668 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
3677 if (syntax & RE_NO_BK_PARENS)
3684 if (syntax & RE_NO_BK_PARENS)
3691 if (syntax & RE_NEWLINE_ALT)
3698 if (syntax & RE_NO_BK_VBAR)
3705 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
3706 goto handle_interval;
3712 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3714 /* Do not translate the character after the \, so that we can
3715 distinguish, e.g., \B from \b, even if we normally would
3716 translate, e.g., B to b. */
3722 if (syntax & RE_NO_BK_PARENS)
3723 goto normal_backslash;
3729 if (COMPILE_STACK_FULL)
3731 RETALLOC (compile_stack.stack, compile_stack.size << 1,
3732 compile_stack_elt_t);
3733 if (compile_stack.stack == NULL) return REG_ESPACE;
3735 compile_stack.size <<= 1;
3738 /* These are the values to restore when we hit end of this
3739 group. They are all relative offsets, so that if the
3740 whole pattern moves because of realloc, they will still
3742 COMPILE_STACK_TOP.begalt_offset = begalt - COMPILED_BUFFER_VAR;
3743 COMPILE_STACK_TOP.fixup_alt_jump
3744 = fixup_alt_jump ? fixup_alt_jump - COMPILED_BUFFER_VAR + 1 : 0;
3745 COMPILE_STACK_TOP.laststart_offset = b - COMPILED_BUFFER_VAR;
3746 COMPILE_STACK_TOP.regnum = regnum;
3748 /* We will eventually replace the 0 with the number of
3749 groups inner to this one. But do not push a
3750 start_memory for groups beyond the last one we can
3751 represent in the compiled pattern. */
3752 if (regnum <= MAX_REGNUM)
3754 COMPILE_STACK_TOP.inner_group_offset = b
3755 - COMPILED_BUFFER_VAR + 2;
3756 BUF_PUSH_3 (start_memory, regnum, 0);
3759 compile_stack.avail++;
3764 /* If we've reached MAX_REGNUM groups, then this open
3765 won't actually generate any code, so we'll have to
3766 clear pending_exact explicitly. */
3772 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
3774 if (COMPILE_STACK_EMPTY)
3776 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3777 goto normal_backslash;
3779 FREE_STACK_RETURN (REG_ERPAREN);
3784 { /* Push a dummy failure point at the end of the
3785 alternative for a possible future
3786 `pop_failure_jump' to pop. See comments at
3787 `push_dummy_failure' in `re_match_2'. */
3788 BUF_PUSH (push_dummy_failure);
3790 /* We allocated space for this jump when we assigned
3791 to `fixup_alt_jump', in the `handle_alt' case below. */
3792 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
3795 /* See similar code for backslashed left paren above. */
3796 if (COMPILE_STACK_EMPTY)
3798 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3801 FREE_STACK_RETURN (REG_ERPAREN);
3804 /* Since we just checked for an empty stack above, this
3805 ``can't happen''. */
3806 assert (compile_stack.avail != 0);
3808 /* We don't just want to restore into `regnum', because
3809 later groups should continue to be numbered higher,
3810 as in `(ab)c(de)' -- the second group is #2. */
3811 regnum_t this_group_regnum;
3813 compile_stack.avail--;
3814 begalt = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.begalt_offset;
3816 = COMPILE_STACK_TOP.fixup_alt_jump
3817 ? COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.fixup_alt_jump - 1
3819 laststart = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.laststart_offset;
3820 this_group_regnum = COMPILE_STACK_TOP.regnum;
3821 /* If we've reached MAX_REGNUM groups, then this open
3822 won't actually generate any code, so we'll have to
3823 clear pending_exact explicitly. */
3826 /* We're at the end of the group, so now we know how many
3827 groups were inside this one. */
3828 if (this_group_regnum <= MAX_REGNUM)
3830 UCHAR_T *inner_group_loc
3831 = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.inner_group_offset;
3833 *inner_group_loc = regnum - this_group_regnum;
3834 BUF_PUSH_3 (stop_memory, this_group_regnum,
3835 regnum - this_group_regnum);
3841 case '|': /* `\|'. */
3842 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
3843 goto normal_backslash;
3845 if (syntax & RE_LIMITED_OPS)
3848 /* Insert before the previous alternative a jump which
3849 jumps to this alternative if the former fails. */
3850 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3851 INSERT_JUMP (on_failure_jump, begalt,
3852 b + 2 + 2 * OFFSET_ADDRESS_SIZE);
3854 b += 1 + OFFSET_ADDRESS_SIZE;
3856 /* The alternative before this one has a jump after it
3857 which gets executed if it gets matched. Adjust that
3858 jump so it will jump to this alternative's analogous
3859 jump (put in below, which in turn will jump to the next
3860 (if any) alternative's such jump, etc.). The last such
3861 jump jumps to the correct final destination. A picture:
3867 If we are at `b', then fixup_alt_jump right now points to a
3868 three-byte space after `a'. We'll put in the jump, set
3869 fixup_alt_jump to right after `b', and leave behind three
3870 bytes which we'll fill in when we get to after `c'. */
3873 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
3875 /* Mark and leave space for a jump after this alternative,
3876 to be filled in later either by next alternative or
3877 when know we're at the end of a series of alternatives. */
3879 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3880 b += 1 + OFFSET_ADDRESS_SIZE;
3888 /* If \{ is a literal. */
3889 if (!(syntax & RE_INTERVALS)
3890 /* If we're at `\{' and it's not the open-interval
3892 || (syntax & RE_NO_BK_BRACES))
3893 goto normal_backslash;
3897 /* If got here, then the syntax allows intervals. */
3899 /* At least (most) this many matches must be made. */
3900 int lower_bound = -1, upper_bound = -1;
3902 /* Place in the uncompiled pattern (i.e., just after
3903 the '{') to go back to if the interval is invalid. */
3904 const CHAR_T *beg_interval = p;
3907 goto invalid_interval;
3909 GET_UNSIGNED_NUMBER (lower_bound);
3913 GET_UNSIGNED_NUMBER (upper_bound);
3914 if (upper_bound < 0)
3915 upper_bound = RE_DUP_MAX;
3918 /* Interval such as `{1}' => match exactly once. */
3919 upper_bound = lower_bound;
3921 if (! (0 <= lower_bound && lower_bound <= upper_bound))
3922 goto invalid_interval;
3924 if (!(syntax & RE_NO_BK_BRACES))
3926 if (c != '\\' || p == pend)
3927 goto invalid_interval;
3932 goto invalid_interval;
3934 /* If it's invalid to have no preceding re. */
3937 if (syntax & RE_CONTEXT_INVALID_OPS
3938 && !(syntax & RE_INVALID_INTERVAL_ORD))
3939 FREE_STACK_RETURN (REG_BADRPT);
3940 else if (syntax & RE_CONTEXT_INDEP_OPS)
3943 goto unfetch_interval;
3946 /* We just parsed a valid interval. */
3948 if (RE_DUP_MAX < upper_bound)
3949 FREE_STACK_RETURN (REG_BADBR);
3951 /* If the upper bound is zero, don't want to succeed at
3952 all; jump from `laststart' to `b + 3', which will be
3953 the end of the buffer after we insert the jump. */
3954 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3955 instead of 'b + 3'. */
3956 if (upper_bound == 0)
3958 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3959 INSERT_JUMP (jump, laststart, b + 1
3960 + OFFSET_ADDRESS_SIZE);
3961 b += 1 + OFFSET_ADDRESS_SIZE;
3964 /* Otherwise, we have a nontrivial interval. When
3965 we're all done, the pattern will look like:
3966 set_number_at <jump count> <upper bound>
3967 set_number_at <succeed_n count> <lower bound>
3968 succeed_n <after jump addr> <succeed_n count>
3970 jump_n <succeed_n addr> <jump count>
3971 (The upper bound and `jump_n' are omitted if
3972 `upper_bound' is 1, though.) */
3974 { /* If the upper bound is > 1, we need to insert
3975 more at the end of the loop. */
3976 unsigned nbytes = 2 + 4 * OFFSET_ADDRESS_SIZE +
3977 (upper_bound > 1) * (2 + 4 * OFFSET_ADDRESS_SIZE);
3979 GET_BUFFER_SPACE (nbytes);
3981 /* Initialize lower bound of the `succeed_n', even
3982 though it will be set during matching by its
3983 attendant `set_number_at' (inserted next),
3984 because `re_compile_fastmap' needs to know.
3985 Jump to the `jump_n' we might insert below. */
3986 INSERT_JUMP2 (succeed_n, laststart,
3987 b + 1 + 2 * OFFSET_ADDRESS_SIZE
3988 + (upper_bound > 1) * (1 + 2 * OFFSET_ADDRESS_SIZE)
3990 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3992 /* Code to initialize the lower bound. Insert
3993 before the `succeed_n'. The `5' is the last two
3994 bytes of this `set_number_at', plus 3 bytes of
3995 the following `succeed_n'. */
3996 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3997 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3998 of the following `succeed_n'. */
3999 PREFIX(insert_op2) (set_number_at, laststart, 1
4000 + 2 * OFFSET_ADDRESS_SIZE, lower_bound, b);
4001 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4003 if (upper_bound > 1)
4004 { /* More than one repetition is allowed, so
4005 append a backward jump to the `succeed_n'
4006 that starts this interval.
4008 When we've reached this during matching,
4009 we'll have matched the interval once, so
4010 jump back only `upper_bound - 1' times. */
4011 STORE_JUMP2 (jump_n, b, laststart
4012 + 2 * OFFSET_ADDRESS_SIZE + 1,
4014 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4016 /* The location we want to set is the second
4017 parameter of the `jump_n'; that is `b-2' as
4018 an absolute address. `laststart' will be
4019 the `set_number_at' we're about to insert;
4020 `laststart+3' the number to set, the source
4021 for the relative address. But we are
4022 inserting into the middle of the pattern --
4023 so everything is getting moved up by 5.
4024 Conclusion: (b - 2) - (laststart + 3) + 5,
4025 i.e., b - laststart.
4027 We insert this at the beginning of the loop
4028 so that if we fail during matching, we'll
4029 reinitialize the bounds. */
4030 PREFIX(insert_op2) (set_number_at, laststart,
4032 upper_bound - 1, b);
4033 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4040 if (!(syntax & RE_INVALID_INTERVAL_ORD))
4041 FREE_STACK_RETURN (p == pend ? REG_EBRACE : REG_BADBR);
4043 /* Match the characters as literals. */
4046 if (syntax & RE_NO_BK_BRACES)
4049 goto normal_backslash;
4053 /* There is no way to specify the before_dot and after_dot
4054 operators. rms says this is ok. --karl */
4062 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
4068 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
4074 if (syntax & RE_NO_GNU_OPS)
4077 BUF_PUSH (wordchar);
4082 if (syntax & RE_NO_GNU_OPS)
4085 BUF_PUSH (notwordchar);
4090 if (syntax & RE_NO_GNU_OPS)
4096 if (syntax & RE_NO_GNU_OPS)
4102 if (syntax & RE_NO_GNU_OPS)
4104 BUF_PUSH (wordbound);
4108 if (syntax & RE_NO_GNU_OPS)
4110 BUF_PUSH (notwordbound);
4114 if (syntax & RE_NO_GNU_OPS)
4120 if (syntax & RE_NO_GNU_OPS)
4125 case '1': case '2': case '3': case '4': case '5':
4126 case '6': case '7': case '8': case '9':
4127 if (syntax & RE_NO_BK_REFS)
4133 FREE_STACK_RETURN (REG_ESUBREG);
4135 /* Can't back reference to a subexpression if inside of it. */
4136 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
4140 BUF_PUSH_2 (duplicate, c1);
4146 if (syntax & RE_BK_PLUS_QM)
4149 goto normal_backslash;
4153 /* You might think it would be useful for \ to mean
4154 not to translate; but if we don't translate it
4155 it will never match anything. */
4163 /* Expects the character in `c'. */
4165 /* If no exactn currently being built. */
4168 /* If last exactn handle binary(or character) and
4169 new exactn handle character(or binary). */
4170 || is_exactn_bin != is_binary[p - 1 - pattern]
4173 /* If last exactn not at current position. */
4174 || pending_exact + *pending_exact + 1 != b
4176 /* We have only one byte following the exactn for the count. */
4177 || *pending_exact == (1 << BYTEWIDTH) - 1
4179 /* If followed by a repetition operator. */
4180 || *p == '*' || *p == '^'
4181 || ((syntax & RE_BK_PLUS_QM)
4182 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
4183 : (*p == '+' || *p == '?'))
4184 || ((syntax & RE_INTERVALS)
4185 && ((syntax & RE_NO_BK_BRACES)
4187 : (p[0] == '\\' && p[1] == '{'))))
4189 /* Start building a new exactn. */
4194 /* Is this exactn binary data or character? */
4195 is_exactn_bin = is_binary[p - 1 - pattern];
4197 BUF_PUSH_2 (exactn_bin, 0);
4199 BUF_PUSH_2 (exactn, 0);
4201 BUF_PUSH_2 (exactn, 0);
4203 pending_exact = b - 1;
4210 } /* while p != pend */
4213 /* Through the pattern now. */
4216 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
4218 if (!COMPILE_STACK_EMPTY)
4219 FREE_STACK_RETURN (REG_EPAREN);
4221 /* If we don't want backtracking, force success
4222 the first time we reach the end of the compiled pattern. */
4223 if (syntax & RE_NO_POSIX_BACKTRACKING)
4231 free (compile_stack.stack);
4233 /* We have succeeded; set the length of the buffer. */
4235 bufp->used = (uintptr_t) b - (uintptr_t) COMPILED_BUFFER_VAR;
4237 bufp->used = b - bufp->buffer;
4243 DEBUG_PRINT1 ("\nCompiled pattern: \n");
4244 PREFIX(print_compiled_pattern) (bufp);
4248 #ifndef MATCH_MAY_ALLOCATE
4249 /* Initialize the failure stack to the largest possible stack. This
4250 isn't necessary unless we're trying to avoid calling alloca in
4251 the search and match routines. */
4253 int num_regs = bufp->re_nsub + 1;
4255 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4256 is strictly greater than re_max_failures, the largest possible stack
4257 is 2 * re_max_failures failure points. */
4258 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
4260 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
4263 if (! fail_stack.stack)
4265 = (PREFIX(fail_stack_elt_t) *) xmalloc (fail_stack.size
4266 * sizeof (PREFIX(fail_stack_elt_t)));
4269 = (PREFIX(fail_stack_elt_t) *) xrealloc (fail_stack.stack,
4271 * sizeof (PREFIX(fail_stack_elt_t))));
4272 # else /* not emacs */
4273 if (! fail_stack.stack)
4275 = (PREFIX(fail_stack_elt_t) *) malloc (fail_stack.size
4276 * sizeof (PREFIX(fail_stack_elt_t)));
4279 = (PREFIX(fail_stack_elt_t) *) realloc (fail_stack.stack,
4281 * sizeof (PREFIX(fail_stack_elt_t))));
4282 # endif /* not emacs */
4285 PREFIX(regex_grow_registers) (num_regs);
4287 #endif /* not MATCH_MAY_ALLOCATE */
4290 } /* regex_compile */
4292 /* Subroutines for `regex_compile'. */
4294 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4295 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4298 PREFIX(store_op1) (op, loc, arg)
4303 *loc = (UCHAR_T) op;
4304 STORE_NUMBER (loc + 1, arg);
4308 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4309 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4312 PREFIX(store_op2) (op, loc, arg1, arg2)
4317 *loc = (UCHAR_T) op;
4318 STORE_NUMBER (loc + 1, arg1);
4319 STORE_NUMBER (loc + 1 + OFFSET_ADDRESS_SIZE, arg2);
4323 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4324 for OP followed by two-byte integer parameter ARG. */
4325 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4328 PREFIX(insert_op1) (op, loc, arg, end)
4334 register UCHAR_T *pfrom = end;
4335 register UCHAR_T *pto = end + 1 + OFFSET_ADDRESS_SIZE;
4337 while (pfrom != loc)
4340 PREFIX(store_op1) (op, loc, arg);
4344 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4345 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4348 PREFIX(insert_op2) (op, loc, arg1, arg2, end)
4354 register UCHAR_T *pfrom = end;
4355 register UCHAR_T *pto = end + 1 + 2 * OFFSET_ADDRESS_SIZE;
4357 while (pfrom != loc)
4360 PREFIX(store_op2) (op, loc, arg1, arg2);
4364 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4365 after an alternative or a begin-subexpression. We assume there is at
4366 least one character before the ^. */
4369 PREFIX(at_begline_loc_p) (pattern, p, syntax)
4370 const CHAR_T *pattern, *p;
4371 reg_syntax_t syntax;
4373 const CHAR_T *prev = p - 2;
4374 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
4377 /* After a subexpression? */
4378 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
4379 /* After an alternative? */
4380 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
4384 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4385 at least one character after the $, i.e., `P < PEND'. */
4388 PREFIX(at_endline_loc_p) (p, pend, syntax)
4389 const CHAR_T *p, *pend;
4390 reg_syntax_t syntax;
4392 const CHAR_T *next = p;
4393 boolean next_backslash = *next == '\\';
4394 const CHAR_T *next_next = p + 1 < pend ? p + 1 : 0;
4397 /* Before a subexpression? */
4398 (syntax & RE_NO_BK_PARENS ? *next == ')'
4399 : next_backslash && next_next && *next_next == ')')
4400 /* Before an alternative? */
4401 || (syntax & RE_NO_BK_VBAR ? *next == '|'
4402 : next_backslash && next_next && *next_next == '|');
4405 #else /* not INSIDE_RECURSION */
4407 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4408 false if it's not. */
4411 group_in_compile_stack (compile_stack, regnum)
4412 compile_stack_type compile_stack;
4417 for (this_element = compile_stack.avail - 1;
4420 if (compile_stack.stack[this_element].regnum == regnum)
4425 #endif /* not INSIDE_RECURSION */
4427 #ifdef INSIDE_RECURSION
4430 /* This insert space, which size is "num", into the pattern at "loc".
4431 "end" must point the end of the allocated buffer. */
4433 insert_space (num, loc, end)
4438 register CHAR_T *pto = end;
4439 register CHAR_T *pfrom = end - num;
4441 while (pfrom >= loc)
4447 static reg_errcode_t
4448 wcs_compile_range (range_start_char, p_ptr, pend, translate, syntax, b,
4450 CHAR_T range_start_char;
4451 const CHAR_T **p_ptr, *pend;
4452 CHAR_T *char_set, *b;
4453 RE_TRANSLATE_TYPE translate;
4454 reg_syntax_t syntax;
4456 const CHAR_T *p = *p_ptr;
4457 CHAR_T range_start, range_end;
4461 uint32_t start_val, end_val;
4467 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
4470 const char *collseq = (const char *) _NL_CURRENT(LC_COLLATE,
4471 _NL_COLLATE_COLLSEQWC);
4472 const unsigned char *extra = (const unsigned char *)
4473 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
4475 if (range_start_char < -1)
4477 /* range_start is a collating symbol. */
4479 /* Retreive the index and get collation sequence value. */
4480 wextra = (int32_t*)(extra + char_set[-range_start_char]);
4481 start_val = wextra[1 + *wextra];
4484 start_val = collseq_table_lookup(collseq, TRANSLATE(range_start_char));
4486 end_val = collseq_table_lookup (collseq, TRANSLATE (p[0]));
4488 /* Report an error if the range is empty and the syntax prohibits
4490 ret = ((syntax & RE_NO_EMPTY_RANGES)
4491 && (start_val > end_val))? REG_ERANGE : REG_NOERROR;
4493 /* Insert space to the end of the char_ranges. */
4494 insert_space(2, b - char_set[5] - 2, b - 1);
4495 *(b - char_set[5] - 2) = (wchar_t)start_val;
4496 *(b - char_set[5] - 1) = (wchar_t)end_val;
4497 char_set[4]++; /* ranges_index */
4502 range_start = (range_start_char >= 0)? TRANSLATE (range_start_char):
4504 range_end = TRANSLATE (p[0]);
4505 /* Report an error if the range is empty and the syntax prohibits
4507 ret = ((syntax & RE_NO_EMPTY_RANGES)
4508 && (range_start > range_end))? REG_ERANGE : REG_NOERROR;
4510 /* Insert space to the end of the char_ranges. */
4511 insert_space(2, b - char_set[5] - 2, b - 1);
4512 *(b - char_set[5] - 2) = range_start;
4513 *(b - char_set[5] - 1) = range_end;
4514 char_set[4]++; /* ranges_index */
4516 /* Have to increment the pointer into the pattern string, so the
4517 caller isn't still at the ending character. */
4523 /* Read the ending character of a range (in a bracket expression) from the
4524 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4525 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4526 Then we set the translation of all bits between the starting and
4527 ending characters (inclusive) in the compiled pattern B.
4529 Return an error code.
4531 We use these short variable names so we can use the same macros as
4532 `regex_compile' itself. */
4534 static reg_errcode_t
4535 byte_compile_range (range_start_char, p_ptr, pend, translate, syntax, b)
4536 unsigned int range_start_char;
4537 const char **p_ptr, *pend;
4538 RE_TRANSLATE_TYPE translate;
4539 reg_syntax_t syntax;
4543 const char *p = *p_ptr;
4546 const unsigned char *collseq;
4547 unsigned int start_colseq;
4548 unsigned int end_colseq;
4556 /* Have to increment the pointer into the pattern string, so the
4557 caller isn't still at the ending character. */
4560 /* Report an error if the range is empty and the syntax prohibits this. */
4561 ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
4564 collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
4565 _NL_COLLATE_COLLSEQMB);
4567 start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
4568 end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
4569 for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
4571 unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
4573 if (start_colseq <= this_colseq && this_colseq <= end_colseq)
4575 SET_LIST_BIT (TRANSLATE (this_char));
4580 /* Here we see why `this_char' has to be larger than an `unsigned
4581 char' -- we would otherwise go into an infinite loop, since all
4582 characters <= 0xff. */
4583 range_start_char = TRANSLATE (range_start_char);
4584 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4585 and some compilers cast it to int implicitly, so following for_loop
4586 may fall to (almost) infinite loop.
4587 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4588 To avoid this, we cast p[0] to unsigned int and truncate it. */
4589 end_char = ((unsigned)TRANSLATE(p[0]) & ((1 << BYTEWIDTH) - 1));
4591 for (this_char = range_start_char; this_char <= end_char; ++this_char)
4593 SET_LIST_BIT (TRANSLATE (this_char));
4602 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4603 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4604 characters can start a string that matches the pattern. This fastmap
4605 is used by re_search to skip quickly over impossible starting points.
4607 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4608 area as BUFP->fastmap.
4610 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4613 Returns 0 if we succeed, -2 if an internal error. */
4616 /* local function for re_compile_fastmap.
4617 truncate wchar_t character to char. */
4618 static unsigned char truncate_wchar (CHAR_T c);
4620 static unsigned char
4624 unsigned char buf[MB_LEN_MAX];
4625 int retval = wctomb(buf, c);
4626 return retval > 0 ? buf[0] : (unsigned char)c;
4631 PREFIX(re_compile_fastmap) (bufp)
4632 struct re_pattern_buffer *bufp;
4635 #ifdef MATCH_MAY_ALLOCATE
4636 PREFIX(fail_stack_type) fail_stack;
4638 #ifndef REGEX_MALLOC
4642 register char *fastmap = bufp->fastmap;
4645 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4646 pattern to (char*) in regex_compile. */
4647 UCHAR_T *pattern = (UCHAR_T*)bufp->buffer;
4648 register UCHAR_T *pend = (UCHAR_T*) (bufp->buffer + bufp->used);
4650 UCHAR_T *pattern = bufp->buffer;
4651 register UCHAR_T *pend = pattern + bufp->used;
4653 UCHAR_T *p = pattern;
4656 /* This holds the pointer to the failure stack, when
4657 it is allocated relocatably. */
4658 fail_stack_elt_t *failure_stack_ptr;
4661 /* Assume that each path through the pattern can be null until
4662 proven otherwise. We set this false at the bottom of switch
4663 statement, to which we get only if a particular path doesn't
4664 match the empty string. */
4665 boolean path_can_be_null = true;
4667 /* We aren't doing a `succeed_n' to begin with. */
4668 boolean succeed_n_p = false;
4670 assert (fastmap != NULL && p != NULL);
4673 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
4674 bufp->fastmap_accurate = 1; /* It will be when we're done. */
4675 bufp->can_be_null = 0;
4679 if (p == pend || *p == succeed)
4681 /* We have reached the (effective) end of pattern. */
4682 if (!FAIL_STACK_EMPTY ())
4684 bufp->can_be_null |= path_can_be_null;
4686 /* Reset for next path. */
4687 path_can_be_null = true;
4689 p = fail_stack.stack[--fail_stack.avail].pointer;
4697 /* We should never be about to go beyond the end of the pattern. */
4700 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4703 /* I guess the idea here is to simply not bother with a fastmap
4704 if a backreference is used, since it's too hard to figure out
4705 the fastmap for the corresponding group. Setting
4706 `can_be_null' stops `re_search_2' from using the fastmap, so
4707 that is all we do. */
4709 bufp->can_be_null = 1;
4713 /* Following are the cases which match a character. These end
4718 fastmap[truncate_wchar(p[1])] = 1;
4732 /* It is hard to distinguish fastmap from (multi byte) characters
4733 which depends on current locale. */
4738 bufp->can_be_null = 1;
4742 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4743 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
4749 /* Chars beyond end of map must be allowed. */
4750 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
4753 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4754 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
4760 for (j = 0; j < (1 << BYTEWIDTH); j++)
4761 if (SYNTAX (j) == Sword)
4767 for (j = 0; j < (1 << BYTEWIDTH); j++)
4768 if (SYNTAX (j) != Sword)
4775 int fastmap_newline = fastmap['\n'];
4777 /* `.' matches anything ... */
4778 for (j = 0; j < (1 << BYTEWIDTH); j++)
4781 /* ... except perhaps newline. */
4782 if (!(bufp->syntax & RE_DOT_NEWLINE))
4783 fastmap['\n'] = fastmap_newline;
4785 /* Return if we have already set `can_be_null'; if we have,
4786 then the fastmap is irrelevant. Something's wrong here. */
4787 else if (bufp->can_be_null)
4790 /* Otherwise, have to check alternative paths. */
4797 for (j = 0; j < (1 << BYTEWIDTH); j++)
4798 if (SYNTAX (j) == (enum syntaxcode) k)
4805 for (j = 0; j < (1 << BYTEWIDTH); j++)
4806 if (SYNTAX (j) != (enum syntaxcode) k)
4811 /* All cases after this match the empty string. These end with
4831 case push_dummy_failure:
4836 case pop_failure_jump:
4837 case maybe_pop_jump:
4840 case dummy_failure_jump:
4841 EXTRACT_NUMBER_AND_INCR (j, p);
4846 /* Jump backward implies we just went through the body of a
4847 loop and matched nothing. Opcode jumped to should be
4848 `on_failure_jump' or `succeed_n'. Just treat it like an
4849 ordinary jump. For a * loop, it has pushed its failure
4850 point already; if so, discard that as redundant. */
4851 if ((re_opcode_t) *p != on_failure_jump
4852 && (re_opcode_t) *p != succeed_n)
4856 EXTRACT_NUMBER_AND_INCR (j, p);
4859 /* If what's on the stack is where we are now, pop it. */
4860 if (!FAIL_STACK_EMPTY ()
4861 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
4867 case on_failure_jump:
4868 case on_failure_keep_string_jump:
4869 handle_on_failure_jump:
4870 EXTRACT_NUMBER_AND_INCR (j, p);
4872 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4873 end of the pattern. We don't want to push such a point,
4874 since when we restore it above, entering the switch will
4875 increment `p' past the end of the pattern. We don't need
4876 to push such a point since we obviously won't find any more
4877 fastmap entries beyond `pend'. Such a pattern can match
4878 the null string, though. */
4881 if (!PUSH_PATTERN_OP (p + j, fail_stack))
4883 RESET_FAIL_STACK ();
4888 bufp->can_be_null = 1;
4892 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
4893 succeed_n_p = false;
4900 /* Get to the number of times to succeed. */
4901 p += OFFSET_ADDRESS_SIZE;
4903 /* Increment p past the n for when k != 0. */
4904 EXTRACT_NUMBER_AND_INCR (k, p);
4907 p -= 2 * OFFSET_ADDRESS_SIZE;
4908 succeed_n_p = true; /* Spaghetti code alert. */
4909 goto handle_on_failure_jump;
4915 p += 2 * OFFSET_ADDRESS_SIZE;
4926 abort (); /* We have listed all the cases. */
4929 /* Getting here means we have found the possible starting
4930 characters for one path of the pattern -- and that the empty
4931 string does not match. We need not follow this path further.
4932 Instead, look at the next alternative (remembered on the
4933 stack), or quit if no more. The test at the top of the loop
4934 does these things. */
4935 path_can_be_null = false;
4939 /* Set `can_be_null' for the last path (also the first path, if the
4940 pattern is empty). */
4941 bufp->can_be_null |= path_can_be_null;
4944 RESET_FAIL_STACK ();
4948 #else /* not INSIDE_RECURSION */
4951 re_compile_fastmap (bufp)
4952 struct re_pattern_buffer *bufp;
4955 if (MB_CUR_MAX != 1)
4956 return wcs_re_compile_fastmap(bufp);
4959 return byte_re_compile_fastmap(bufp);
4960 } /* re_compile_fastmap */
4962 weak_alias (__re_compile_fastmap, re_compile_fastmap)
4966 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4967 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4968 this memory for recording register information. STARTS and ENDS
4969 must be allocated using the malloc library routine, and must each
4970 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4972 If NUM_REGS == 0, then subsequent matches should allocate their own
4975 Unless this function is called, the first search or match using
4976 PATTERN_BUFFER will allocate its own register data, without
4977 freeing the old data. */
4980 re_set_registers (bufp, regs, num_regs, starts, ends)
4981 struct re_pattern_buffer *bufp;
4982 struct re_registers *regs;
4984 regoff_t *starts, *ends;
4988 bufp->regs_allocated = REGS_REALLOCATE;
4989 regs->num_regs = num_regs;
4990 regs->start = starts;
4995 bufp->regs_allocated = REGS_UNALLOCATED;
4997 regs->start = regs->end = (regoff_t *) 0;
5001 weak_alias (__re_set_registers, re_set_registers)
5004 /* Searching routines. */
5006 /* Like re_search_2, below, but only one string is specified, and
5007 doesn't let you say where to stop matching. */
5010 re_search (bufp, string, size, startpos, range, regs)
5011 struct re_pattern_buffer *bufp;
5013 int size, startpos, range;
5014 struct re_registers *regs;
5016 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
5020 weak_alias (__re_search, re_search)
5024 /* Using the compiled pattern in BUFP->buffer, first tries to match the
5025 virtual concatenation of STRING1 and STRING2, starting first at index
5026 STARTPOS, then at STARTPOS + 1, and so on.
5028 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
5030 RANGE is how far to scan while trying to match. RANGE = 0 means try
5031 only at STARTPOS; in general, the last start tried is STARTPOS +
5034 In REGS, return the indices of the virtual concatenation of STRING1
5035 and STRING2 that matched the entire BUFP->buffer and its contained
5038 Do not consider matching one past the index STOP in the virtual
5039 concatenation of STRING1 and STRING2.
5041 We return either the position in the strings at which the match was
5042 found, -1 if no match, or -2 if error (such as failure
5046 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
5047 struct re_pattern_buffer *bufp;
5048 const char *string1, *string2;
5052 struct re_registers *regs;
5056 if (MB_CUR_MAX != 1)
5057 return wcs_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5061 return byte_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5065 weak_alias (__re_search_2, re_search_2)
5068 #endif /* not INSIDE_RECURSION */
5070 #ifdef INSIDE_RECURSION
5072 #ifdef MATCH_MAY_ALLOCATE
5073 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
5075 # define FREE_VAR(var) if (var) free (var); var = NULL
5079 # define FREE_WCS_BUFFERS() \
5081 FREE_VAR (string1); \
5082 FREE_VAR (string2); \
5083 FREE_VAR (mbs_offset1); \
5084 FREE_VAR (mbs_offset2); \
5090 PREFIX(re_search_2) (bufp, string1, size1, string2, size2, startpos, range,
5092 struct re_pattern_buffer *bufp;
5093 const char *string1, *string2;
5097 struct re_registers *regs;
5101 register char *fastmap = bufp->fastmap;
5102 register RE_TRANSLATE_TYPE translate = bufp->translate;
5103 int total_size = size1 + size2;
5104 int endpos = startpos + range;
5106 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5107 wchar_t *wcs_string1 = NULL, *wcs_string2 = NULL;
5108 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5109 int wcs_size1 = 0, wcs_size2 = 0;
5110 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5111 int *mbs_offset1 = NULL, *mbs_offset2 = NULL;
5112 /* They hold whether each wchar_t is binary data or not. */
5113 char *is_binary = NULL;
5116 /* Check for out-of-range STARTPOS. */
5117 if (startpos < 0 || startpos > total_size)
5120 /* Fix up RANGE if it might eventually take us outside
5121 the virtual concatenation of STRING1 and STRING2.
5122 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5124 range = 0 - startpos;
5125 else if (endpos > total_size)
5126 range = total_size - startpos;
5128 /* If the search isn't to be a backwards one, don't waste time in a
5129 search for a pattern that must be anchored. */
5130 if (bufp->used > 0 && range > 0
5131 && ((re_opcode_t) bufp->buffer[0] == begbuf
5132 /* `begline' is like `begbuf' if it cannot match at newlines. */
5133 || ((re_opcode_t) bufp->buffer[0] == begline
5134 && !bufp->newline_anchor)))
5143 /* In a forward search for something that starts with \=.
5144 don't keep searching past point. */
5145 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
5147 range = PT - startpos;
5153 /* Update the fastmap now if not correct already. */
5154 if (fastmap && !bufp->fastmap_accurate)
5155 if (re_compile_fastmap (bufp) == -2)
5159 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5160 fill them with converted string. */
5163 wcs_string1 = REGEX_TALLOC (size1 + 1, CHAR_T);
5164 mbs_offset1 = REGEX_TALLOC (size1 + 1, int);
5165 is_binary = REGEX_TALLOC (size1 + 1, char);
5166 if (!wcs_string1 || !mbs_offset1 || !is_binary)
5168 FREE_VAR (wcs_string1);
5169 FREE_VAR (mbs_offset1);
5170 FREE_VAR (is_binary);
5173 wcs_size1 = convert_mbs_to_wcs(wcs_string1, string1, size1,
5174 mbs_offset1, is_binary);
5175 wcs_string1[wcs_size1] = L'\0'; /* for a sentinel */
5176 FREE_VAR (is_binary);
5180 wcs_string2 = REGEX_TALLOC (size2 + 1, CHAR_T);
5181 mbs_offset2 = REGEX_TALLOC (size2 + 1, int);
5182 is_binary = REGEX_TALLOC (size2 + 1, char);
5183 if (!wcs_string2 || !mbs_offset2 || !is_binary)
5185 FREE_WCS_BUFFERS ();
5186 FREE_VAR (is_binary);
5189 wcs_size2 = convert_mbs_to_wcs(wcs_string2, string2, size2,
5190 mbs_offset2, is_binary);
5191 wcs_string2[wcs_size2] = L'\0'; /* for a sentinel */
5192 FREE_VAR (is_binary);
5197 /* Loop through the string, looking for a place to start matching. */
5200 /* If a fastmap is supplied, skip quickly over characters that
5201 cannot be the start of a match. If the pattern can match the
5202 null string, however, we don't need to skip characters; we want
5203 the first null string. */
5204 if (fastmap && startpos < total_size && !bufp->can_be_null)
5206 if (range > 0) /* Searching forwards. */
5208 register const char *d;
5209 register int lim = 0;
5212 if (startpos < size1 && startpos + range >= size1)
5213 lim = range - (size1 - startpos);
5215 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
5217 /* Written out as an if-else to avoid testing `translate'
5221 && !fastmap[(unsigned char)
5222 translate[(unsigned char) *d++]])
5225 while (range > lim && !fastmap[(unsigned char) *d++])
5228 startpos += irange - range;
5230 else /* Searching backwards. */
5232 register CHAR_T c = (size1 == 0 || startpos >= size1
5233 ? string2[startpos - size1]
5234 : string1[startpos]);
5236 if (!fastmap[(unsigned char) TRANSLATE (c)])
5241 /* If can't match the null string, and that's all we have left, fail. */
5242 if (range >= 0 && startpos == total_size && fastmap
5243 && !bufp->can_be_null)
5246 FREE_WCS_BUFFERS ();
5252 val = wcs_re_match_2_internal (bufp, string1, size1, string2,
5253 size2, startpos, regs, stop,
5254 wcs_string1, wcs_size1,
5255 wcs_string2, wcs_size2,
5256 mbs_offset1, mbs_offset2);
5258 val = byte_re_match_2_internal (bufp, string1, size1, string2,
5259 size2, startpos, regs, stop);
5262 #ifndef REGEX_MALLOC
5271 FREE_WCS_BUFFERS ();
5279 FREE_WCS_BUFFERS ();
5299 FREE_WCS_BUFFERS ();
5305 /* This converts PTR, a pointer into one of the search wchar_t strings
5306 `string1' and `string2' into an multibyte string offset from the
5307 beginning of that string. We use mbs_offset to optimize.
5308 See convert_mbs_to_wcs. */
5309 # define POINTER_TO_OFFSET(ptr) \
5310 (FIRST_STRING_P (ptr) \
5311 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5312 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5315 /* This converts PTR, a pointer into one of the search strings `string1'
5316 and `string2' into an offset from the beginning of that string. */
5317 # define POINTER_TO_OFFSET(ptr) \
5318 (FIRST_STRING_P (ptr) \
5319 ? ((regoff_t) ((ptr) - string1)) \
5320 : ((regoff_t) ((ptr) - string2 + size1)))
5323 /* Macros for dealing with the split strings in re_match_2. */
5325 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5327 /* Call before fetching a character with *d. This switches over to
5328 string2 if necessary. */
5329 #define PREFETCH() \
5332 /* End of string2 => fail. */ \
5333 if (dend == end_match_2) \
5335 /* End of string1 => advance to string2. */ \
5337 dend = end_match_2; \
5340 /* Test if at very beginning or at very end of the virtual concatenation
5341 of `string1' and `string2'. If only one string, it's `string2'. */
5342 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5343 #define AT_STRINGS_END(d) ((d) == end2)
5346 /* Test if D points to a character which is word-constituent. We have
5347 two special cases to check for: if past the end of string1, look at
5348 the first character in string2; and if before the beginning of
5349 string2, look at the last character in string1. */
5351 /* Use internationalized API instead of SYNTAX. */
5352 # define WORDCHAR_P(d) \
5353 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5354 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5355 || ((d) == end1 ? *string2 \
5356 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) == L'_')
5358 # define WORDCHAR_P(d) \
5359 (SYNTAX ((d) == end1 ? *string2 \
5360 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5364 /* Disabled due to a compiler bug -- see comment at case wordbound */
5366 /* Test if the character before D and the one at D differ with respect
5367 to being word-constituent. */
5368 #define AT_WORD_BOUNDARY(d) \
5369 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5370 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5373 /* Free everything we malloc. */
5374 #ifdef MATCH_MAY_ALLOCATE
5376 # define FREE_VARIABLES() \
5378 REGEX_FREE_STACK (fail_stack.stack); \
5379 FREE_VAR (regstart); \
5380 FREE_VAR (regend); \
5381 FREE_VAR (old_regstart); \
5382 FREE_VAR (old_regend); \
5383 FREE_VAR (best_regstart); \
5384 FREE_VAR (best_regend); \
5385 FREE_VAR (reg_info); \
5386 FREE_VAR (reg_dummy); \
5387 FREE_VAR (reg_info_dummy); \
5388 if (!cant_free_wcs_buf) \
5390 FREE_VAR (string1); \
5391 FREE_VAR (string2); \
5392 FREE_VAR (mbs_offset1); \
5393 FREE_VAR (mbs_offset2); \
5397 # define FREE_VARIABLES() \
5399 REGEX_FREE_STACK (fail_stack.stack); \
5400 FREE_VAR (regstart); \
5401 FREE_VAR (regend); \
5402 FREE_VAR (old_regstart); \
5403 FREE_VAR (old_regend); \
5404 FREE_VAR (best_regstart); \
5405 FREE_VAR (best_regend); \
5406 FREE_VAR (reg_info); \
5407 FREE_VAR (reg_dummy); \
5408 FREE_VAR (reg_info_dummy); \
5413 # define FREE_VARIABLES() \
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() ((void)0) /* Do nothing! But inhibit gcc warning. */
5426 #endif /* not MATCH_MAY_ALLOCATE */
5428 /* These values must meet several constraints. They must not be valid
5429 register values; since we have a limit of 255 registers (because
5430 we use only one byte in the pattern for the register number), we can
5431 use numbers larger than 255. They must differ by 1, because of
5432 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5433 be larger than the value for the highest register, so we do not try
5434 to actually save any registers when none are active. */
5435 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5436 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5438 #else /* not INSIDE_RECURSION */
5439 /* Matching routines. */
5441 #ifndef emacs /* Emacs never uses this. */
5442 /* re_match is like re_match_2 except it takes only a single string. */
5445 re_match (bufp, string, size, pos, regs)
5446 struct re_pattern_buffer *bufp;
5449 struct re_registers *regs;
5453 if (MB_CUR_MAX != 1)
5454 result = wcs_re_match_2_internal (bufp, NULL, 0, string, size,
5456 NULL, 0, NULL, 0, NULL, NULL);
5459 result = byte_re_match_2_internal (bufp, NULL, 0, string, size,
5461 # ifndef REGEX_MALLOC
5469 weak_alias (__re_match, re_match)
5471 #endif /* not emacs */
5473 #endif /* not INSIDE_RECURSION */
5475 #ifdef INSIDE_RECURSION
5476 static boolean PREFIX(group_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5478 PREFIX(register_info_type) *reg_info));
5479 static boolean PREFIX(alt_match_null_string_p) _RE_ARGS ((UCHAR_T *p,
5481 PREFIX(register_info_type) *reg_info));
5482 static boolean PREFIX(common_op_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5484 PREFIX(register_info_type) *reg_info));
5485 static int PREFIX(bcmp_translate) _RE_ARGS ((const CHAR_T *s1, const CHAR_T *s2,
5486 int len, char *translate));
5487 #else /* not INSIDE_RECURSION */
5489 /* re_match_2 matches the compiled pattern in BUFP against the
5490 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5491 and SIZE2, respectively). We start matching at POS, and stop
5494 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5495 store offsets for the substring each group matched in REGS. See the
5496 documentation for exactly how many groups we fill.
5498 We return -1 if no match, -2 if an internal error (such as the
5499 failure stack overflowing). Otherwise, we return the length of the
5500 matched substring. */
5503 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
5504 struct re_pattern_buffer *bufp;
5505 const char *string1, *string2;
5508 struct re_registers *regs;
5513 if (MB_CUR_MAX != 1)
5514 result = wcs_re_match_2_internal (bufp, string1, size1, string2, size2,
5516 NULL, 0, NULL, 0, NULL, NULL);
5519 result = byte_re_match_2_internal (bufp, string1, size1, string2, size2,
5522 #ifndef REGEX_MALLOC
5530 weak_alias (__re_match_2, re_match_2)
5533 #endif /* not INSIDE_RECURSION */
5535 #ifdef INSIDE_RECURSION
5538 static int count_mbs_length PARAMS ((int *, int));
5540 /* This check the substring (from 0, to length) of the multibyte string,
5541 to which offset_buffer correspond. And count how many wchar_t_characters
5542 the substring occupy. We use offset_buffer to optimization.
5543 See convert_mbs_to_wcs. */
5546 count_mbs_length(offset_buffer, length)
5552 /* Check whether the size is valid. */
5556 if (offset_buffer == NULL)
5559 /* If there are no multibyte character, offset_buffer[i] == i.
5560 Optmize for this case. */
5561 if (offset_buffer[length] == length)
5564 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5570 int middle = (lower + upper) / 2;
5571 if (middle == lower || middle == upper)
5573 if (offset_buffer[middle] > length)
5575 else if (offset_buffer[middle] < length)
5585 /* This is a separate function so that we can force an alloca cleanup
5589 wcs_re_match_2_internal (bufp, cstring1, csize1, cstring2, csize2, pos,
5590 regs, stop, string1, size1, string2, size2,
5591 mbs_offset1, mbs_offset2)
5592 struct re_pattern_buffer *bufp;
5593 const char *cstring1, *cstring2;
5596 struct re_registers *regs;
5598 /* string1 == string2 == NULL means string1/2, size1/2 and
5599 mbs_offset1/2 need seting up in this function. */
5600 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5601 wchar_t *string1, *string2;
5602 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5604 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5605 int *mbs_offset1, *mbs_offset2;
5608 byte_re_match_2_internal (bufp, string1, size1,string2, size2, pos,
5610 struct re_pattern_buffer *bufp;
5611 const char *string1, *string2;
5614 struct re_registers *regs;
5618 /* General temporaries. */
5622 /* They hold whether each wchar_t is binary data or not. */
5623 char *is_binary = NULL;
5624 /* If true, we can't free string1/2, mbs_offset1/2. */
5625 int cant_free_wcs_buf = 1;
5628 /* Just past the end of the corresponding string. */
5629 const CHAR_T *end1, *end2;
5631 /* Pointers into string1 and string2, just past the last characters in
5632 each to consider matching. */
5633 const CHAR_T *end_match_1, *end_match_2;
5635 /* Where we are in the data, and the end of the current string. */
5636 const CHAR_T *d, *dend;
5638 /* Where we are in the pattern, and the end of the pattern. */
5640 UCHAR_T *pattern, *p;
5641 register UCHAR_T *pend;
5643 UCHAR_T *p = bufp->buffer;
5644 register UCHAR_T *pend = p + bufp->used;
5647 /* Mark the opcode just after a start_memory, so we can test for an
5648 empty subpattern when we get to the stop_memory. */
5649 UCHAR_T *just_past_start_mem = 0;
5651 /* We use this to map every character in the string. */
5652 RE_TRANSLATE_TYPE translate = bufp->translate;
5654 /* Failure point stack. Each place that can handle a failure further
5655 down the line pushes a failure point on this stack. It consists of
5656 restart, regend, and reg_info for all registers corresponding to
5657 the subexpressions we're currently inside, plus the number of such
5658 registers, and, finally, two char *'s. The first char * is where
5659 to resume scanning the pattern; the second one is where to resume
5660 scanning the strings. If the latter is zero, the failure point is
5661 a ``dummy''; if a failure happens and the failure point is a dummy,
5662 it gets discarded and the next next one is tried. */
5663 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5664 PREFIX(fail_stack_type) fail_stack;
5667 static unsigned failure_id;
5668 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
5672 /* This holds the pointer to the failure stack, when
5673 it is allocated relocatably. */
5674 fail_stack_elt_t *failure_stack_ptr;
5677 /* We fill all the registers internally, independent of what we
5678 return, for use in backreferences. The number here includes
5679 an element for register zero. */
5680 size_t num_regs = bufp->re_nsub + 1;
5682 /* The currently active registers. */
5683 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
5684 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
5686 /* Information on the contents of registers. These are pointers into
5687 the input strings; they record just what was matched (on this
5688 attempt) by a subexpression part of the pattern, that is, the
5689 regnum-th regstart pointer points to where in the pattern we began
5690 matching and the regnum-th regend points to right after where we
5691 stopped matching the regnum-th subexpression. (The zeroth register
5692 keeps track of what the whole pattern matches.) */
5693 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5694 const CHAR_T **regstart, **regend;
5697 /* If a group that's operated upon by a repetition operator fails to
5698 match anything, then the register for its start will need to be
5699 restored because it will have been set to wherever in the string we
5700 are when we last see its open-group operator. Similarly for a
5702 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5703 const CHAR_T **old_regstart, **old_regend;
5706 /* The is_active field of reg_info helps us keep track of which (possibly
5707 nested) subexpressions we are currently in. The matched_something
5708 field of reg_info[reg_num] helps us tell whether or not we have
5709 matched any of the pattern so far this time through the reg_num-th
5710 subexpression. These two fields get reset each time through any
5711 loop their register is in. */
5712 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5713 PREFIX(register_info_type) *reg_info;
5716 /* The following record the register info as found in the above
5717 variables when we find a match better than any we've seen before.
5718 This happens as we backtrack through the failure points, which in
5719 turn happens only if we have not yet matched the entire string. */
5720 unsigned best_regs_set = false;
5721 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5722 const CHAR_T **best_regstart, **best_regend;
5725 /* Logically, this is `best_regend[0]'. But we don't want to have to
5726 allocate space for that if we're not allocating space for anything
5727 else (see below). Also, we never need info about register 0 for
5728 any of the other register vectors, and it seems rather a kludge to
5729 treat `best_regend' differently than the rest. So we keep track of
5730 the end of the best match so far in a separate variable. We
5731 initialize this to NULL so that when we backtrack the first time
5732 and need to test it, it's not garbage. */
5733 const CHAR_T *match_end = NULL;
5735 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5736 int set_regs_matched_done = 0;
5738 /* Used when we pop values we don't care about. */
5739 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5740 const CHAR_T **reg_dummy;
5741 PREFIX(register_info_type) *reg_info_dummy;
5745 /* Counts the total number of registers pushed. */
5746 unsigned num_regs_pushed = 0;
5749 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
5753 #ifdef MATCH_MAY_ALLOCATE
5754 /* Do not bother to initialize all the register variables if there are
5755 no groups in the pattern, as it takes a fair amount of time. If
5756 there are groups, we include space for register 0 (the whole
5757 pattern), even though we never use it, since it simplifies the
5758 array indexing. We should fix this. */
5761 regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5762 regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5763 old_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5764 old_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5765 best_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5766 best_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5767 reg_info = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5768 reg_dummy = REGEX_TALLOC (num_regs, const CHAR_T *);
5769 reg_info_dummy = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5771 if (!(regstart && regend && old_regstart && old_regend && reg_info
5772 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
5780 /* We must initialize all our variables to NULL, so that
5781 `FREE_VARIABLES' doesn't try to free them. */
5782 regstart = regend = old_regstart = old_regend = best_regstart
5783 = best_regend = reg_dummy = NULL;
5784 reg_info = reg_info_dummy = (PREFIX(register_info_type) *) NULL;
5786 #endif /* MATCH_MAY_ALLOCATE */
5788 /* The starting position is bogus. */
5790 if (pos < 0 || pos > csize1 + csize2)
5792 if (pos < 0 || pos > size1 + size2)
5800 /* Allocate wchar_t array for string1 and string2 and
5801 fill them with converted string. */
5802 if (string1 == NULL && string2 == NULL)
5804 /* We need seting up buffers here. */
5806 /* We must free wcs buffers in this function. */
5807 cant_free_wcs_buf = 0;
5811 string1 = REGEX_TALLOC (csize1 + 1, CHAR_T);
5812 mbs_offset1 = REGEX_TALLOC (csize1 + 1, int);
5813 is_binary = REGEX_TALLOC (csize1 + 1, char);
5814 if (!string1 || !mbs_offset1 || !is_binary)
5817 FREE_VAR (mbs_offset1);
5818 FREE_VAR (is_binary);
5824 string2 = REGEX_TALLOC (csize2 + 1, CHAR_T);
5825 mbs_offset2 = REGEX_TALLOC (csize2 + 1, int);
5826 is_binary = REGEX_TALLOC (csize2 + 1, char);
5827 if (!string2 || !mbs_offset2 || !is_binary)
5830 FREE_VAR (mbs_offset1);
5832 FREE_VAR (mbs_offset2);
5833 FREE_VAR (is_binary);
5836 size2 = convert_mbs_to_wcs(string2, cstring2, csize2,
5837 mbs_offset2, is_binary);
5838 string2[size2] = L'\0'; /* for a sentinel */
5839 FREE_VAR (is_binary);
5843 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5844 pattern to (char*) in regex_compile. */
5845 p = pattern = (CHAR_T*)bufp->buffer;
5846 pend = (CHAR_T*)(bufp->buffer + bufp->used);
5850 /* Initialize subexpression text positions to -1 to mark ones that no
5851 start_memory/stop_memory has been seen for. Also initialize the
5852 register information struct. */
5853 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
5855 regstart[mcnt] = regend[mcnt]
5856 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
5858 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
5859 IS_ACTIVE (reg_info[mcnt]) = 0;
5860 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5861 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5864 /* We move `string1' into `string2' if the latter's empty -- but not if
5865 `string1' is null. */
5866 if (size2 == 0 && string1 != NULL)
5873 mbs_offset2 = mbs_offset1;
5879 end1 = string1 + size1;
5880 end2 = string2 + size2;
5882 /* Compute where to stop matching, within the two strings. */
5886 mcnt = count_mbs_length(mbs_offset1, stop);
5887 end_match_1 = string1 + mcnt;
5888 end_match_2 = string2;
5892 if (stop > csize1 + csize2)
5893 stop = csize1 + csize2;
5895 mcnt = count_mbs_length(mbs_offset2, stop-csize1);
5896 end_match_2 = string2 + mcnt;
5899 { /* count_mbs_length return error. */
5906 end_match_1 = string1 + stop;
5907 end_match_2 = string2;
5912 end_match_2 = string2 + stop - size1;
5916 /* `p' scans through the pattern as `d' scans through the data.
5917 `dend' is the end of the input string that `d' points within. `d'
5918 is advanced into the following input string whenever necessary, but
5919 this happens before fetching; therefore, at the beginning of the
5920 loop, `d' can be pointing at the end of a string, but it cannot
5923 if (size1 > 0 && pos <= csize1)
5925 mcnt = count_mbs_length(mbs_offset1, pos);
5931 mcnt = count_mbs_length(mbs_offset2, pos-csize1);
5937 { /* count_mbs_length return error. */
5942 if (size1 > 0 && pos <= size1)
5949 d = string2 + pos - size1;
5954 DEBUG_PRINT1 ("The compiled pattern is:\n");
5955 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
5956 DEBUG_PRINT1 ("The string to match is: `");
5957 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
5958 DEBUG_PRINT1 ("'\n");
5960 /* This loops over pattern commands. It exits by returning from the
5961 function if the match is complete, or it drops through if the match
5962 fails at this starting point in the input data. */
5966 DEBUG_PRINT2 ("\n%p: ", p);
5968 DEBUG_PRINT2 ("\n0x%x: ", p);
5972 { /* End of pattern means we might have succeeded. */
5973 DEBUG_PRINT1 ("end of pattern ... ");
5975 /* If we haven't matched the entire string, and we want the
5976 longest match, try backtracking. */
5977 if (d != end_match_2)
5979 /* 1 if this match ends in the same string (string1 or string2)
5980 as the best previous match. */
5981 boolean same_str_p = (FIRST_STRING_P (match_end)
5982 == MATCHING_IN_FIRST_STRING);
5983 /* 1 if this match is the best seen so far. */
5984 boolean best_match_p;
5986 /* AIX compiler got confused when this was combined
5987 with the previous declaration. */
5989 best_match_p = d > match_end;
5991 best_match_p = !MATCHING_IN_FIRST_STRING;
5993 DEBUG_PRINT1 ("backtracking.\n");
5995 if (!FAIL_STACK_EMPTY ())
5996 { /* More failure points to try. */
5998 /* If exceeds best match so far, save it. */
5999 if (!best_regs_set || best_match_p)
6001 best_regs_set = true;
6004 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
6006 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6008 best_regstart[mcnt] = regstart[mcnt];
6009 best_regend[mcnt] = regend[mcnt];
6015 /* If no failure points, don't restore garbage. And if
6016 last match is real best match, don't restore second
6018 else if (best_regs_set && !best_match_p)
6021 /* Restore best match. It may happen that `dend ==
6022 end_match_1' while the restored d is in string2.
6023 For example, the pattern `x.*y.*z' against the
6024 strings `x-' and `y-z-', if the two strings are
6025 not consecutive in memory. */
6026 DEBUG_PRINT1 ("Restoring best registers.\n");
6029 dend = ((d >= string1 && d <= end1)
6030 ? end_match_1 : end_match_2);
6032 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6034 regstart[mcnt] = best_regstart[mcnt];
6035 regend[mcnt] = best_regend[mcnt];
6038 } /* d != end_match_2 */
6041 DEBUG_PRINT1 ("Accepting match.\n");
6042 /* If caller wants register contents data back, do it. */
6043 if (regs && !bufp->no_sub)
6045 /* Have the register data arrays been allocated? */
6046 if (bufp->regs_allocated == REGS_UNALLOCATED)
6047 { /* No. So allocate them with malloc. We need one
6048 extra element beyond `num_regs' for the `-1' marker
6050 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
6051 regs->start = TALLOC (regs->num_regs, regoff_t);
6052 regs->end = TALLOC (regs->num_regs, regoff_t);
6053 if (regs->start == NULL || regs->end == NULL)
6058 bufp->regs_allocated = REGS_REALLOCATE;
6060 else if (bufp->regs_allocated == REGS_REALLOCATE)
6061 { /* Yes. If we need more elements than were already
6062 allocated, reallocate them. If we need fewer, just
6064 if (regs->num_regs < num_regs + 1)
6066 regs->num_regs = num_regs + 1;
6067 RETALLOC (regs->start, regs->num_regs, regoff_t);
6068 RETALLOC (regs->end, regs->num_regs, regoff_t);
6069 if (regs->start == NULL || regs->end == NULL)
6078 /* These braces fend off a "empty body in an else-statement"
6079 warning under GCC when assert expands to nothing. */
6080 assert (bufp->regs_allocated == REGS_FIXED);
6083 /* Convert the pointer data in `regstart' and `regend' to
6084 indices. Register zero has to be set differently,
6085 since we haven't kept track of any info for it. */
6086 if (regs->num_regs > 0)
6088 regs->start[0] = pos;
6090 if (MATCHING_IN_FIRST_STRING)
6091 regs->end[0] = mbs_offset1 != NULL ?
6092 mbs_offset1[d-string1] : 0;
6094 regs->end[0] = csize1 + (mbs_offset2 != NULL ?
6095 mbs_offset2[d-string2] : 0);
6097 regs->end[0] = (MATCHING_IN_FIRST_STRING
6098 ? ((regoff_t) (d - string1))
6099 : ((regoff_t) (d - string2 + size1)));
6103 /* Go through the first `min (num_regs, regs->num_regs)'
6104 registers, since that is all we initialized. */
6105 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
6108 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
6109 regs->start[mcnt] = regs->end[mcnt] = -1;
6113 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
6115 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
6119 /* If the regs structure we return has more elements than
6120 were in the pattern, set the extra elements to -1. If
6121 we (re)allocated the registers, this is the case,
6122 because we always allocate enough to have at least one
6124 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
6125 regs->start[mcnt] = regs->end[mcnt] = -1;
6126 } /* regs && !bufp->no_sub */
6128 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
6129 nfailure_points_pushed, nfailure_points_popped,
6130 nfailure_points_pushed - nfailure_points_popped);
6131 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
6134 if (MATCHING_IN_FIRST_STRING)
6135 mcnt = mbs_offset1 != NULL ? mbs_offset1[d-string1] : 0;
6137 mcnt = (mbs_offset2 != NULL ? mbs_offset2[d-string2] : 0) +
6141 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
6146 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
6152 /* Otherwise match next pattern command. */
6153 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
6155 /* Ignore these. Used to ignore the n of succeed_n's which
6156 currently have n == 0. */
6158 DEBUG_PRINT1 ("EXECUTING no_op.\n");
6162 DEBUG_PRINT1 ("EXECUTING succeed.\n");
6165 /* Match the next n pattern characters exactly. The following
6166 byte in the pattern defines n, and the n bytes after that
6167 are the characters to match. */
6173 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
6175 /* This is written out as an if-else so we don't waste time
6176 testing `translate' inside the loop. */
6185 if ((UCHAR_T) translate[(unsigned char) *d++]
6191 if (*d++ != (CHAR_T) *p++)
6195 if ((UCHAR_T) translate[(unsigned char) *d++]
6207 if (*d++ != (CHAR_T) *p++) goto fail;
6211 SET_REGS_MATCHED ();
6215 /* Match any character except possibly a newline or a null. */
6217 DEBUG_PRINT1 ("EXECUTING anychar.\n");
6221 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
6222 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
6225 SET_REGS_MATCHED ();
6226 DEBUG_PRINT2 (" Matched `%ld'.\n", (long int) *d);
6236 unsigned int i, char_class_length, coll_symbol_length,
6237 equiv_class_length, ranges_length, chars_length, length;
6238 CHAR_T *workp, *workp2, *charset_top;
6239 #define WORK_BUFFER_SIZE 128
6240 CHAR_T str_buf[WORK_BUFFER_SIZE];
6245 boolean not = (re_opcode_t) *(p - 1) == charset_not;
6247 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
6249 c = TRANSLATE (*d); /* The character to match. */
6252 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
6254 charset_top = p - 1;
6255 char_class_length = *p++;
6256 coll_symbol_length = *p++;
6257 equiv_class_length = *p++;
6258 ranges_length = *p++;
6259 chars_length = *p++;
6260 /* p points charset[6], so the address of the next instruction
6261 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6262 where l=length of char_classes, m=length of collating_symbol,
6263 n=equivalence_class, o=length of char_range,
6264 p'=length of character. */
6266 /* Update p to indicate the next instruction. */
6267 p += char_class_length + coll_symbol_length+ equiv_class_length +
6268 2*ranges_length + chars_length;
6270 /* match with char_class? */
6271 for (i = 0; i < char_class_length ; i += CHAR_CLASS_SIZE)
6274 uintptr_t alignedp = ((uintptr_t)workp
6275 + __alignof__(wctype_t) - 1)
6276 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
6277 wctype = *((wctype_t*)alignedp);
6278 workp += CHAR_CLASS_SIZE;
6279 if (iswctype((wint_t)c, wctype))
6280 goto char_set_matched;
6283 /* match with collating_symbol? */
6287 const unsigned char *extra = (const unsigned char *)
6288 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
6290 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;
6294 wextra = (int32_t*)(extra + *workp++);
6295 for (i = 0; i < *wextra; ++i)
6296 if (TRANSLATE(d[i]) != wextra[1 + i])
6301 /* Update d, however d will be incremented at
6302 char_set_matched:, we decrement d here. */
6304 goto char_set_matched;
6308 else /* (nrules == 0) */
6310 /* If we can't look up collation data, we use wcscoll
6313 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;)
6315 const CHAR_T *backup_d = d, *backup_dend = dend;
6316 length = wcslen(workp);
6318 /* If wcscoll(the collating symbol, whole string) > 0,
6319 any substring of the string never match with the
6320 collating symbol. */
6321 if (wcscoll(workp, d) > 0)
6323 workp += length + 1;
6327 /* First, we compare the collating symbol with
6328 the first character of the string.
6329 If it don't match, we add the next character to
6330 the compare buffer in turn. */
6331 for (i = 0 ; i < WORK_BUFFER_SIZE-1 ; i++, d++)
6336 if (dend == end_match_2)
6342 /* add next character to the compare buffer. */
6343 str_buf[i] = TRANSLATE(*d);
6344 str_buf[i+1] = '\0';
6346 match = wcscoll(workp, str_buf);
6348 goto char_set_matched;
6351 /* (str_buf > workp) indicate (str_buf + X > workp),
6352 because for all X (str_buf + X > str_buf).
6353 So we don't need continue this loop. */
6356 /* Otherwise(str_buf < workp),
6357 (str_buf+next_character) may equals (workp).
6358 So we continue this loop. */
6363 workp += length + 1;
6366 /* match with equivalence_class? */
6370 const CHAR_T *backup_d = d, *backup_dend = dend;
6371 /* Try to match the equivalence class against
6372 those known to the collate implementation. */
6373 const int32_t *table;
6374 const int32_t *weights;
6375 const int32_t *extra;
6376 const int32_t *indirect;
6381 /* This #include defines a local function! */
6382 # include <locale/weightwc.h>
6384 table = (const int32_t *)
6385 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEWC);
6386 weights = (const wint_t *)
6387 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTWC);
6388 extra = (const wint_t *)
6389 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAWC);
6390 indirect = (const int32_t *)
6391 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTWC);
6393 /* Write 1 collating element to str_buf, and
6397 for (i = 0 ; idx2 == 0 && i < WORK_BUFFER_SIZE - 1; i++)
6399 cp = (wint_t*)str_buf;
6402 if (dend == end_match_2)
6407 str_buf[i] = TRANSLATE(*(d+i));
6408 str_buf[i+1] = '\0'; /* sentinel */
6409 idx2 = findidx ((const wint_t**)&cp);
6412 /* Update d, however d will be incremented at
6413 char_set_matched:, we decrement d here. */
6414 d = backup_d + ((wchar_t*)cp - (wchar_t*)str_buf - 1);
6417 if (dend == end_match_2)
6426 len = weights[idx2];
6428 for (workp2 = workp + equiv_class_length ; workp < workp2 ;
6431 idx = (int32_t)*workp;
6432 /* We already checked idx != 0 in regex_compile. */
6434 if (idx2 != 0 && len == weights[idx])
6437 while (cnt < len && (weights[idx + 1 + cnt]
6438 == weights[idx2 + 1 + cnt]))
6442 goto char_set_matched;
6449 else /* (nrules == 0) */
6451 /* If we can't look up collation data, we use wcscoll
6454 for (workp2 = workp + equiv_class_length ; workp < workp2 ;)
6456 const CHAR_T *backup_d = d, *backup_dend = dend;
6457 length = wcslen(workp);
6459 /* If wcscoll(the collating symbol, whole string) > 0,
6460 any substring of the string never match with the
6461 collating symbol. */
6462 if (wcscoll(workp, d) > 0)
6464 workp += length + 1;
6468 /* First, we compare the equivalence class with
6469 the first character of the string.
6470 If it don't match, we add the next character to
6471 the compare buffer in turn. */
6472 for (i = 0 ; i < WORK_BUFFER_SIZE - 1 ; i++, d++)
6477 if (dend == end_match_2)
6483 /* add next character to the compare buffer. */
6484 str_buf[i] = TRANSLATE(*d);
6485 str_buf[i+1] = '\0';
6487 match = wcscoll(workp, str_buf);
6490 goto char_set_matched;
6493 /* (str_buf > workp) indicate (str_buf + X > workp),
6494 because for all X (str_buf + X > str_buf).
6495 So we don't need continue this loop. */
6498 /* Otherwise(str_buf < workp),
6499 (str_buf+next_character) may equals (workp).
6500 So we continue this loop. */
6505 workp += length + 1;
6509 /* match with char_range? */
6513 uint32_t collseqval;
6514 const char *collseq = (const char *)
6515 _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQWC);
6517 collseqval = collseq_table_lookup (collseq, c);
6519 for (; workp < p - chars_length ;)
6521 uint32_t start_val, end_val;
6523 /* We already compute the collation sequence value
6524 of the characters (or collating symbols). */
6525 start_val = (uint32_t) *workp++; /* range_start */
6526 end_val = (uint32_t) *workp++; /* range_end */
6528 if (start_val <= collseqval && collseqval <= end_val)
6529 goto char_set_matched;
6535 /* We set range_start_char at str_buf[0], range_end_char
6536 at str_buf[4], and compared char at str_buf[2]. */
6541 for (; workp < p - chars_length ;)
6543 wchar_t *range_start_char, *range_end_char;
6545 /* match if (range_start_char <= c <= range_end_char). */
6547 /* If range_start(or end) < 0, we assume -range_start(end)
6548 is the offset of the collating symbol which is specified
6549 as the character of the range start(end). */
6553 range_start_char = charset_top - (*workp++);
6556 str_buf[0] = *workp++;
6557 range_start_char = str_buf;
6562 range_end_char = charset_top - (*workp++);
6565 str_buf[4] = *workp++;
6566 range_end_char = str_buf + 4;
6569 if (wcscoll(range_start_char, str_buf+2) <= 0 &&
6570 wcscoll(str_buf+2, range_end_char) <= 0)
6572 goto char_set_matched;
6576 /* match with char? */
6577 for (; workp < p ; workp++)
6579 goto char_set_matched;
6586 /* Cast to `unsigned' instead of `unsigned char' in case the
6587 bit list is a full 32 bytes long. */
6588 if (c < (unsigned) (*p * BYTEWIDTH)
6589 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
6594 if (!not) goto fail;
6595 #undef WORK_BUFFER_SIZE
6597 SET_REGS_MATCHED ();
6603 /* The beginning of a group is represented by start_memory.
6604 The arguments are the register number in the next byte, and the
6605 number of groups inner to this one in the next. The text
6606 matched within the group is recorded (in the internal
6607 registers data structure) under the register number. */
6609 DEBUG_PRINT3 ("EXECUTING start_memory %ld (%ld):\n",
6610 (long int) *p, (long int) p[1]);
6612 /* Find out if this group can match the empty string. */
6613 p1 = p; /* To send to group_match_null_string_p. */
6615 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
6616 REG_MATCH_NULL_STRING_P (reg_info[*p])
6617 = PREFIX(group_match_null_string_p) (&p1, pend, reg_info);
6619 /* Save the position in the string where we were the last time
6620 we were at this open-group operator in case the group is
6621 operated upon by a repetition operator, e.g., with `(a*)*b'
6622 against `ab'; then we want to ignore where we are now in
6623 the string in case this attempt to match fails. */
6624 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6625 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
6627 DEBUG_PRINT2 (" old_regstart: %d\n",
6628 POINTER_TO_OFFSET (old_regstart[*p]));
6631 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
6633 IS_ACTIVE (reg_info[*p]) = 1;
6634 MATCHED_SOMETHING (reg_info[*p]) = 0;
6636 /* Clear this whenever we change the register activity status. */
6637 set_regs_matched_done = 0;
6639 /* This is the new highest active register. */
6640 highest_active_reg = *p;
6642 /* If nothing was active before, this is the new lowest active
6644 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
6645 lowest_active_reg = *p;
6647 /* Move past the register number and inner group count. */
6649 just_past_start_mem = p;
6654 /* The stop_memory opcode represents the end of a group. Its
6655 arguments are the same as start_memory's: the register
6656 number, and the number of inner groups. */
6658 DEBUG_PRINT3 ("EXECUTING stop_memory %ld (%ld):\n",
6659 (long int) *p, (long int) p[1]);
6661 /* We need to save the string position the last time we were at
6662 this close-group operator in case the group is operated
6663 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6664 against `aba'; then we want to ignore where we are now in
6665 the string in case this attempt to match fails. */
6666 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6667 ? REG_UNSET (regend[*p]) ? d : regend[*p]
6669 DEBUG_PRINT2 (" old_regend: %d\n",
6670 POINTER_TO_OFFSET (old_regend[*p]));
6673 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
6675 /* This register isn't active anymore. */
6676 IS_ACTIVE (reg_info[*p]) = 0;
6678 /* Clear this whenever we change the register activity status. */
6679 set_regs_matched_done = 0;
6681 /* If this was the only register active, nothing is active
6683 if (lowest_active_reg == highest_active_reg)
6685 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6686 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6689 { /* We must scan for the new highest active register, since
6690 it isn't necessarily one less than now: consider
6691 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6692 new highest active register is 1. */
6694 while (r > 0 && !IS_ACTIVE (reg_info[r]))
6697 /* If we end up at register zero, that means that we saved
6698 the registers as the result of an `on_failure_jump', not
6699 a `start_memory', and we jumped to past the innermost
6700 `stop_memory'. For example, in ((.)*) we save
6701 registers 1 and 2 as a result of the *, but when we pop
6702 back to the second ), we are at the stop_memory 1.
6703 Thus, nothing is active. */
6706 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6707 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6710 highest_active_reg = r;
6713 /* If just failed to match something this time around with a
6714 group that's operated on by a repetition operator, try to
6715 force exit from the ``loop'', and restore the register
6716 information for this group that we had before trying this
6718 if ((!MATCHED_SOMETHING (reg_info[*p])
6719 || just_past_start_mem == p - 1)
6722 boolean is_a_jump_n = false;
6726 switch ((re_opcode_t) *p1++)
6730 case pop_failure_jump:
6731 case maybe_pop_jump:
6733 case dummy_failure_jump:
6734 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6736 p1 += OFFSET_ADDRESS_SIZE;
6744 /* If the next operation is a jump backwards in the pattern
6745 to an on_failure_jump right before the start_memory
6746 corresponding to this stop_memory, exit from the loop
6747 by forcing a failure after pushing on the stack the
6748 on_failure_jump's jump in the pattern, and d. */
6749 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
6750 && (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == start_memory
6751 && p1[2+OFFSET_ADDRESS_SIZE] == *p)
6753 /* If this group ever matched anything, then restore
6754 what its registers were before trying this last
6755 failed match, e.g., with `(a*)*b' against `ab' for
6756 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6757 against `aba' for regend[3].
6759 Also restore the registers for inner groups for,
6760 e.g., `((a*)(b*))*' against `aba' (register 3 would
6761 otherwise get trashed). */
6763 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
6767 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
6769 /* Restore this and inner groups' (if any) registers. */
6770 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
6773 regstart[r] = old_regstart[r];
6775 /* xx why this test? */
6776 if (old_regend[r] >= regstart[r])
6777 regend[r] = old_regend[r];
6781 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6782 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
6788 /* Move past the register number and the inner group count. */
6793 /* \<digit> has been turned into a `duplicate' command which is
6794 followed by the numeric value of <digit> as the register number. */
6797 register const CHAR_T *d2, *dend2;
6798 int regno = *p++; /* Get which register to match against. */
6799 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
6801 /* Can't back reference a group which we've never matched. */
6802 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
6805 /* Where in input to try to start matching. */
6806 d2 = regstart[regno];
6808 /* Where to stop matching; if both the place to start and
6809 the place to stop matching are in the same string, then
6810 set to the place to stop, otherwise, for now have to use
6811 the end of the first string. */
6813 dend2 = ((FIRST_STRING_P (regstart[regno])
6814 == FIRST_STRING_P (regend[regno]))
6815 ? regend[regno] : end_match_1);
6818 /* If necessary, advance to next segment in register
6822 if (dend2 == end_match_2) break;
6823 if (dend2 == regend[regno]) break;
6825 /* End of string1 => advance to string2. */
6827 dend2 = regend[regno];
6829 /* At end of register contents => success */
6830 if (d2 == dend2) break;
6832 /* If necessary, advance to next segment in data. */
6835 /* How many characters left in this segment to match. */
6838 /* Want how many consecutive characters we can match in
6839 one shot, so, if necessary, adjust the count. */
6840 if (mcnt > dend2 - d2)
6843 /* Compare that many; failure if mismatch, else move
6846 ? PREFIX(bcmp_translate) (d, d2, mcnt, translate)
6847 : memcmp (d, d2, mcnt*sizeof(UCHAR_T)))
6849 d += mcnt, d2 += mcnt;
6851 /* Do this because we've match some characters. */
6852 SET_REGS_MATCHED ();
6858 /* begline matches the empty string at the beginning of the string
6859 (unless `not_bol' is set in `bufp'), and, if
6860 `newline_anchor' is set, after newlines. */
6862 DEBUG_PRINT1 ("EXECUTING begline.\n");
6864 if (AT_STRINGS_BEG (d))
6866 if (!bufp->not_bol) break;
6868 else if (d[-1] == '\n' && bufp->newline_anchor)
6872 /* In all other cases, we fail. */
6876 /* endline is the dual of begline. */
6878 DEBUG_PRINT1 ("EXECUTING endline.\n");
6880 if (AT_STRINGS_END (d))
6882 if (!bufp->not_eol) break;
6885 /* We have to ``prefetch'' the next character. */
6886 else if ((d == end1 ? *string2 : *d) == '\n'
6887 && bufp->newline_anchor)
6894 /* Match at the very beginning of the data. */
6896 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
6897 if (AT_STRINGS_BEG (d))
6902 /* Match at the very end of the data. */
6904 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
6905 if (AT_STRINGS_END (d))
6910 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6911 pushes NULL as the value for the string on the stack. Then
6912 `pop_failure_point' will keep the current value for the
6913 string, instead of restoring it. To see why, consider
6914 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6915 then the . fails against the \n. But the next thing we want
6916 to do is match the \n against the \n; if we restored the
6917 string value, we would be back at the foo.
6919 Because this is used only in specific cases, we don't need to
6920 check all the things that `on_failure_jump' does, to make
6921 sure the right things get saved on the stack. Hence we don't
6922 share its code. The only reason to push anything on the
6923 stack at all is that otherwise we would have to change
6924 `anychar's code to do something besides goto fail in this
6925 case; that seems worse than this. */
6926 case on_failure_keep_string_jump:
6927 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
6929 EXTRACT_NUMBER_AND_INCR (mcnt, p);
6931 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
6933 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
6936 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
6940 /* Uses of on_failure_jump:
6942 Each alternative starts with an on_failure_jump that points
6943 to the beginning of the next alternative. Each alternative
6944 except the last ends with a jump that in effect jumps past
6945 the rest of the alternatives. (They really jump to the
6946 ending jump of the following alternative, because tensioning
6947 these jumps is a hassle.)
6949 Repeats start with an on_failure_jump that points past both
6950 the repetition text and either the following jump or
6951 pop_failure_jump back to this on_failure_jump. */
6952 case on_failure_jump:
6954 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
6956 EXTRACT_NUMBER_AND_INCR (mcnt, p);
6958 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
6960 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
6963 /* If this on_failure_jump comes right before a group (i.e.,
6964 the original * applied to a group), save the information
6965 for that group and all inner ones, so that if we fail back
6966 to this point, the group's information will be correct.
6967 For example, in \(a*\)*\1, we need the preceding group,
6968 and in \(zz\(a*\)b*\)\2, we need the inner group. */
6970 /* We can't use `p' to check ahead because we push
6971 a failure point to `p + mcnt' after we do this. */
6974 /* We need to skip no_op's before we look for the
6975 start_memory in case this on_failure_jump is happening as
6976 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
6978 while (p1 < pend && (re_opcode_t) *p1 == no_op)
6981 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
6983 /* We have a new highest active register now. This will
6984 get reset at the start_memory we are about to get to,
6985 but we will have saved all the registers relevant to
6986 this repetition op, as described above. */
6987 highest_active_reg = *(p1 + 1) + *(p1 + 2);
6988 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
6989 lowest_active_reg = *(p1 + 1);
6992 DEBUG_PRINT1 (":\n");
6993 PUSH_FAILURE_POINT (p + mcnt, d, -2);
6997 /* A smart repeat ends with `maybe_pop_jump'.
6998 We change it to either `pop_failure_jump' or `jump'. */
6999 case maybe_pop_jump:
7000 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7001 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
7003 register UCHAR_T *p2 = p;
7005 /* Compare the beginning of the repeat with what in the
7006 pattern follows its end. If we can establish that there
7007 is nothing that they would both match, i.e., that we
7008 would have to backtrack because of (as in, e.g., `a*a')
7009 then we can change to pop_failure_jump, because we'll
7010 never have to backtrack.
7012 This is not true in the case of alternatives: in
7013 `(a|ab)*' we do need to backtrack to the `ab' alternative
7014 (e.g., if the string was `ab'). But instead of trying to
7015 detect that here, the alternative has put on a dummy
7016 failure point which is what we will end up popping. */
7018 /* Skip over open/close-group commands.
7019 If what follows this loop is a ...+ construct,
7020 look at what begins its body, since we will have to
7021 match at least one of that. */
7025 && ((re_opcode_t) *p2 == stop_memory
7026 || (re_opcode_t) *p2 == start_memory))
7028 else if (p2 + 2 + 2 * OFFSET_ADDRESS_SIZE < pend
7029 && (re_opcode_t) *p2 == dummy_failure_jump)
7030 p2 += 2 + 2 * OFFSET_ADDRESS_SIZE;
7036 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7037 to the `maybe_finalize_jump' of this case. Examine what
7040 /* If we're at the end of the pattern, we can change. */
7043 /* Consider what happens when matching ":\(.*\)"
7044 against ":/". I don't really understand this code
7046 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7049 (" End of pattern: change to `pop_failure_jump'.\n");
7052 else if ((re_opcode_t) *p2 == exactn
7054 || (re_opcode_t) *p2 == exactn_bin
7056 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
7059 = *p2 == (UCHAR_T) endline ? '\n' : p2[2];
7061 if (((re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn
7063 || (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn_bin
7065 ) && p1[3+OFFSET_ADDRESS_SIZE] != c)
7067 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7070 DEBUG_PRINT3 (" %C != %C => pop_failure_jump.\n",
7072 (wint_t) p1[3+OFFSET_ADDRESS_SIZE]);
7074 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
7076 (char) p1[3+OFFSET_ADDRESS_SIZE]);
7081 else if ((re_opcode_t) p1[3] == charset
7082 || (re_opcode_t) p1[3] == charset_not)
7084 int not = (re_opcode_t) p1[3] == charset_not;
7086 if (c < (unsigned) (p1[4] * BYTEWIDTH)
7087 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
7090 /* `not' is equal to 1 if c would match, which means
7091 that we can't change to pop_failure_jump. */
7094 p[-3] = (unsigned char) pop_failure_jump;
7095 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7098 #endif /* not WCHAR */
7101 else if ((re_opcode_t) *p2 == charset)
7103 /* We win if the first character of the loop is not part
7105 if ((re_opcode_t) p1[3] == exactn
7106 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
7107 && (p2[2 + p1[5] / BYTEWIDTH]
7108 & (1 << (p1[5] % BYTEWIDTH)))))
7110 p[-3] = (unsigned char) pop_failure_jump;
7111 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7114 else if ((re_opcode_t) p1[3] == charset_not)
7117 /* We win if the charset_not inside the loop
7118 lists every character listed in the charset after. */
7119 for (idx = 0; idx < (int) p2[1]; idx++)
7120 if (! (p2[2 + idx] == 0
7121 || (idx < (int) p1[4]
7122 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
7127 p[-3] = (unsigned char) pop_failure_jump;
7128 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7131 else if ((re_opcode_t) p1[3] == charset)
7134 /* We win if the charset inside the loop
7135 has no overlap with the one after the loop. */
7137 idx < (int) p2[1] && idx < (int) p1[4];
7139 if ((p2[2 + idx] & p1[5 + idx]) != 0)
7142 if (idx == p2[1] || idx == p1[4])
7144 p[-3] = (unsigned char) pop_failure_jump;
7145 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7149 #endif /* not WCHAR */
7151 p -= OFFSET_ADDRESS_SIZE; /* Point at relative address again. */
7152 if ((re_opcode_t) p[-1] != pop_failure_jump)
7154 p[-1] = (UCHAR_T) jump;
7155 DEBUG_PRINT1 (" Match => jump.\n");
7156 goto unconditional_jump;
7158 /* Note fall through. */
7161 /* The end of a simple repeat has a pop_failure_jump back to
7162 its matching on_failure_jump, where the latter will push a
7163 failure point. The pop_failure_jump takes off failure
7164 points put on by this pop_failure_jump's matching
7165 on_failure_jump; we got through the pattern to here from the
7166 matching on_failure_jump, so didn't fail. */
7167 case pop_failure_jump:
7169 /* We need to pass separate storage for the lowest and
7170 highest registers, even though we don't care about the
7171 actual values. Otherwise, we will restore only one
7172 register from the stack, since lowest will == highest in
7173 `pop_failure_point'. */
7174 active_reg_t dummy_low_reg, dummy_high_reg;
7175 UCHAR_T *pdummy = NULL;
7176 const CHAR_T *sdummy = NULL;
7178 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
7179 POP_FAILURE_POINT (sdummy, pdummy,
7180 dummy_low_reg, dummy_high_reg,
7181 reg_dummy, reg_dummy, reg_info_dummy);
7183 /* Note fall through. */
7187 DEBUG_PRINT2 ("\n%p: ", p);
7189 DEBUG_PRINT2 ("\n0x%x: ", p);
7191 /* Note fall through. */
7193 /* Unconditionally jump (without popping any failure points). */
7195 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
7196 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
7197 p += mcnt; /* Do the jump. */
7199 DEBUG_PRINT2 ("(to %p).\n", p);
7201 DEBUG_PRINT2 ("(to 0x%x).\n", p);
7206 /* We need this opcode so we can detect where alternatives end
7207 in `group_match_null_string_p' et al. */
7209 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
7210 goto unconditional_jump;
7213 /* Normally, the on_failure_jump pushes a failure point, which
7214 then gets popped at pop_failure_jump. We will end up at
7215 pop_failure_jump, also, and with a pattern of, say, `a+', we
7216 are skipping over the on_failure_jump, so we have to push
7217 something meaningless for pop_failure_jump to pop. */
7218 case dummy_failure_jump:
7219 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
7220 /* It doesn't matter what we push for the string here. What
7221 the code at `fail' tests is the value for the pattern. */
7222 PUSH_FAILURE_POINT (NULL, NULL, -2);
7223 goto unconditional_jump;
7226 /* At the end of an alternative, we need to push a dummy failure
7227 point in case we are followed by a `pop_failure_jump', because
7228 we don't want the failure point for the alternative to be
7229 popped. For example, matching `(a|ab)*' against `aab'
7230 requires that we match the `ab' alternative. */
7231 case push_dummy_failure:
7232 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
7233 /* See comments just above at `dummy_failure_jump' about the
7235 PUSH_FAILURE_POINT (NULL, NULL, -2);
7238 /* Have to succeed matching what follows at least n times.
7239 After that, handle like `on_failure_jump'. */
7241 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7242 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
7245 /* Originally, this is how many times we HAVE to succeed. */
7249 p += OFFSET_ADDRESS_SIZE;
7250 STORE_NUMBER_AND_INCR (p, mcnt);
7252 DEBUG_PRINT3 (" Setting %p to %d.\n", p - OFFSET_ADDRESS_SIZE
7255 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - OFFSET_ADDRESS_SIZE
7262 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n",
7263 p + OFFSET_ADDRESS_SIZE);
7265 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n",
7266 p + OFFSET_ADDRESS_SIZE);
7270 p[1] = (UCHAR_T) no_op;
7272 p[2] = (UCHAR_T) no_op;
7273 p[3] = (UCHAR_T) no_op;
7280 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7281 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
7283 /* Originally, this is how many times we CAN jump. */
7287 STORE_NUMBER (p + OFFSET_ADDRESS_SIZE, mcnt);
7290 DEBUG_PRINT3 (" Setting %p to %d.\n", p + OFFSET_ADDRESS_SIZE,
7293 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + OFFSET_ADDRESS_SIZE,
7296 goto unconditional_jump;
7298 /* If don't have to jump any more, skip over the rest of command. */
7300 p += 2 * OFFSET_ADDRESS_SIZE;
7305 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
7307 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7309 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7311 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
7313 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
7315 STORE_NUMBER (p1, mcnt);
7320 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7321 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7322 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7323 macro and introducing temporary variables works around the bug. */
7326 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7327 if (AT_WORD_BOUNDARY (d))
7332 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7333 if (AT_WORD_BOUNDARY (d))
7339 boolean prevchar, thischar;
7341 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7342 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7345 prevchar = WORDCHAR_P (d - 1);
7346 thischar = WORDCHAR_P (d);
7347 if (prevchar != thischar)
7354 boolean prevchar, thischar;
7356 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7357 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7360 prevchar = WORDCHAR_P (d - 1);
7361 thischar = WORDCHAR_P (d);
7362 if (prevchar != thischar)
7369 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
7370 if (!AT_STRINGS_END (d) && WORDCHAR_P (d)
7371 && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
7376 DEBUG_PRINT1 ("EXECUTING wordend.\n");
7377 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
7378 && (AT_STRINGS_END (d) || !WORDCHAR_P (d)))
7384 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
7385 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
7390 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
7391 if (PTR_CHAR_POS ((unsigned char *) d) != point)
7396 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
7397 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
7402 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
7407 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
7411 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7413 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
7415 SET_REGS_MATCHED ();
7419 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
7421 goto matchnotsyntax;
7424 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
7428 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7430 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
7432 SET_REGS_MATCHED ();
7435 #else /* not emacs */
7437 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
7439 if (!WORDCHAR_P (d))
7441 SET_REGS_MATCHED ();
7446 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
7450 SET_REGS_MATCHED ();
7453 #endif /* not emacs */
7458 continue; /* Successfully executed one pattern command; keep going. */
7461 /* We goto here if a matching operation fails. */
7463 if (!FAIL_STACK_EMPTY ())
7464 { /* A restart point is known. Restore to that state. */
7465 DEBUG_PRINT1 ("\nFAIL:\n");
7466 POP_FAILURE_POINT (d, p,
7467 lowest_active_reg, highest_active_reg,
7468 regstart, regend, reg_info);
7470 /* If this failure point is a dummy, try the next one. */
7474 /* If we failed to the end of the pattern, don't examine *p. */
7478 boolean is_a_jump_n = false;
7480 /* If failed to a backwards jump that's part of a repetition
7481 loop, need to pop this failure point and use the next one. */
7482 switch ((re_opcode_t) *p)
7486 case maybe_pop_jump:
7487 case pop_failure_jump:
7490 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7493 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
7495 && (re_opcode_t) *p1 == on_failure_jump))
7503 if (d >= string1 && d <= end1)
7507 break; /* Matching at this starting point really fails. */
7511 goto restore_best_regs;
7515 return -1; /* Failure to match. */
7518 /* Subroutine definitions for re_match_2. */
7521 /* We are passed P pointing to a register number after a start_memory.
7523 Return true if the pattern up to the corresponding stop_memory can
7524 match the empty string, and false otherwise.
7526 If we find the matching stop_memory, sets P to point to one past its number.
7527 Otherwise, sets P to an undefined byte less than or equal to END.
7529 We don't handle duplicates properly (yet). */
7532 PREFIX(group_match_null_string_p) (p, end, reg_info)
7534 PREFIX(register_info_type) *reg_info;
7537 /* Point to after the args to the start_memory. */
7538 UCHAR_T *p1 = *p + 2;
7542 /* Skip over opcodes that can match nothing, and return true or
7543 false, as appropriate, when we get to one that can't, or to the
7544 matching stop_memory. */
7546 switch ((re_opcode_t) *p1)
7548 /* Could be either a loop or a series of alternatives. */
7549 case on_failure_jump:
7551 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7553 /* If the next operation is not a jump backwards in the
7558 /* Go through the on_failure_jumps of the alternatives,
7559 seeing if any of the alternatives cannot match nothing.
7560 The last alternative starts with only a jump,
7561 whereas the rest start with on_failure_jump and end
7562 with a jump, e.g., here is the pattern for `a|b|c':
7564 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7565 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7568 So, we have to first go through the first (n-1)
7569 alternatives and then deal with the last one separately. */
7572 /* Deal with the first (n-1) alternatives, which start
7573 with an on_failure_jump (see above) that jumps to right
7574 past a jump_past_alt. */
7576 while ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] ==
7579 /* `mcnt' holds how many bytes long the alternative
7580 is, including the ending `jump_past_alt' and
7583 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt -
7584 (1 + OFFSET_ADDRESS_SIZE),
7588 /* Move to right after this alternative, including the
7592 /* Break if it's the beginning of an n-th alternative
7593 that doesn't begin with an on_failure_jump. */
7594 if ((re_opcode_t) *p1 != on_failure_jump)
7597 /* Still have to check that it's not an n-th
7598 alternative that starts with an on_failure_jump. */
7600 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7601 if ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] !=
7604 /* Get to the beginning of the n-th alternative. */
7605 p1 -= 1 + OFFSET_ADDRESS_SIZE;
7610 /* Deal with the last alternative: go back and get number
7611 of the `jump_past_alt' just before it. `mcnt' contains
7612 the length of the alternative. */
7613 EXTRACT_NUMBER (mcnt, p1 - OFFSET_ADDRESS_SIZE);
7615 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt, reg_info))
7618 p1 += mcnt; /* Get past the n-th alternative. */
7624 assert (p1[1] == **p);
7630 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7633 } /* while p1 < end */
7636 } /* group_match_null_string_p */
7639 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7640 It expects P to be the first byte of a single alternative and END one
7641 byte past the last. The alternative can contain groups. */
7644 PREFIX(alt_match_null_string_p) (p, end, reg_info)
7646 PREFIX(register_info_type) *reg_info;
7653 /* Skip over opcodes that can match nothing, and break when we get
7654 to one that can't. */
7656 switch ((re_opcode_t) *p1)
7659 case on_failure_jump:
7661 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7666 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7669 } /* while p1 < end */
7672 } /* alt_match_null_string_p */
7675 /* Deals with the ops common to group_match_null_string_p and
7676 alt_match_null_string_p.
7678 Sets P to one after the op and its arguments, if any. */
7681 PREFIX(common_op_match_null_string_p) (p, end, reg_info)
7683 PREFIX(register_info_type) *reg_info;
7690 switch ((re_opcode_t) *p1++)
7710 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
7711 ret = PREFIX(group_match_null_string_p) (&p1, end, reg_info);
7713 /* Have to set this here in case we're checking a group which
7714 contains a group and a back reference to it. */
7716 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
7717 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
7723 /* If this is an optimized succeed_n for zero times, make the jump. */
7725 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7733 /* Get to the number of times to succeed. */
7734 p1 += OFFSET_ADDRESS_SIZE;
7735 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7739 p1 -= 2 * OFFSET_ADDRESS_SIZE;
7740 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7748 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
7753 p1 += 2 * OFFSET_ADDRESS_SIZE;
7756 /* All other opcodes mean we cannot match the empty string. */
7762 } /* common_op_match_null_string_p */
7765 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7766 bytes; nonzero otherwise. */
7769 PREFIX(bcmp_translate) (s1, s2, len, translate)
7770 const CHAR_T *s1, *s2;
7772 RE_TRANSLATE_TYPE translate;
7774 register const UCHAR_T *p1 = (const UCHAR_T *) s1;
7775 register const UCHAR_T *p2 = (const UCHAR_T *) s2;
7779 if (((*p1<=0xff)?translate[*p1++]:*p1++)
7780 != ((*p2<=0xff)?translate[*p2++]:*p2++))
7783 if (translate[*p1++] != translate[*p2++]) return 1;
7791 #else /* not INSIDE_RECURSION */
7793 /* Entry points for GNU code. */
7795 /* re_compile_pattern is the GNU regular expression compiler: it
7796 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7797 Returns 0 if the pattern was valid, otherwise an error string.
7799 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7800 are set in BUFP on entry.
7802 We call regex_compile to do the actual compilation. */
7805 re_compile_pattern (pattern, length, bufp)
7806 const char *pattern;
7808 struct re_pattern_buffer *bufp;
7812 /* GNU code is written to assume at least RE_NREGS registers will be set
7813 (and at least one extra will be -1). */
7814 bufp->regs_allocated = REGS_UNALLOCATED;
7816 /* And GNU code determines whether or not to get register information
7817 by passing null for the REGS argument to re_match, etc., not by
7821 /* Match anchors at newline. */
7822 bufp->newline_anchor = 1;
7825 if (MB_CUR_MAX != 1)
7826 ret = wcs_regex_compile (pattern, length, re_syntax_options, bufp);
7829 ret = byte_regex_compile (pattern, length, re_syntax_options, bufp);
7833 return gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
7836 weak_alias (__re_compile_pattern, re_compile_pattern)
7839 /* Entry points compatible with 4.2 BSD regex library. We don't define
7840 them unless specifically requested. */
7842 #if defined _REGEX_RE_COMP || defined _LIBC
7844 /* BSD has one and only one pattern buffer. */
7845 static struct re_pattern_buffer re_comp_buf;
7849 /* Make these definitions weak in libc, so POSIX programs can redefine
7850 these names if they don't use our functions, and still use
7851 regcomp/regexec below without link errors. */
7861 if (!re_comp_buf.buffer)
7862 return gettext ("No previous regular expression");
7866 if (!re_comp_buf.buffer)
7868 re_comp_buf.buffer = (unsigned char *) malloc (200);
7869 if (re_comp_buf.buffer == NULL)
7870 return (char *) gettext (re_error_msgid
7871 + re_error_msgid_idx[(int) REG_ESPACE]);
7872 re_comp_buf.allocated = 200;
7874 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
7875 if (re_comp_buf.fastmap == NULL)
7876 return (char *) gettext (re_error_msgid
7877 + re_error_msgid_idx[(int) REG_ESPACE]);
7880 /* Since `re_exec' always passes NULL for the `regs' argument, we
7881 don't need to initialize the pattern buffer fields which affect it. */
7883 /* Match anchors at newlines. */
7884 re_comp_buf.newline_anchor = 1;
7887 if (MB_CUR_MAX != 1)
7888 ret = wcs_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7891 ret = byte_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7896 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7897 return (char *) gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
7908 const int len = strlen (s);
7910 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
7913 #endif /* _REGEX_RE_COMP */
7915 /* POSIX.2 functions. Don't define these for Emacs. */
7919 /* regcomp takes a regular expression as a string and compiles it.
7921 PREG is a regex_t *. We do not expect any fields to be initialized,
7922 since POSIX says we shouldn't. Thus, we set
7924 `buffer' to the compiled pattern;
7925 `used' to the length of the compiled pattern;
7926 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7927 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7928 RE_SYNTAX_POSIX_BASIC;
7929 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7930 `fastmap' to an allocated space for the fastmap;
7931 `fastmap_accurate' to zero;
7932 `re_nsub' to the number of subexpressions in PATTERN.
7934 PATTERN is the address of the pattern string.
7936 CFLAGS is a series of bits which affect compilation.
7938 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7939 use POSIX basic syntax.
7941 If REG_NEWLINE is set, then . and [^...] don't match newline.
7942 Also, regexec will try a match beginning after every newline.
7944 If REG_ICASE is set, then we considers upper- and lowercase
7945 versions of letters to be equivalent when matching.
7947 If REG_NOSUB is set, then when PREG is passed to regexec, that
7948 routine will report only success or failure, and nothing about the
7951 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
7952 the return codes and their meanings.) */
7955 regcomp (preg, pattern, cflags)
7957 const char *pattern;
7962 = (cflags & REG_EXTENDED) ?
7963 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
7965 /* regex_compile will allocate the space for the compiled pattern. */
7967 preg->allocated = 0;
7970 /* Try to allocate space for the fastmap. */
7971 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
7973 if (cflags & REG_ICASE)
7978 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
7979 * sizeof (*(RE_TRANSLATE_TYPE)0));
7980 if (preg->translate == NULL)
7981 return (int) REG_ESPACE;
7983 /* Map uppercase characters to corresponding lowercase ones. */
7984 for (i = 0; i < CHAR_SET_SIZE; i++)
7985 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
7988 preg->translate = NULL;
7990 /* If REG_NEWLINE is set, newlines are treated differently. */
7991 if (cflags & REG_NEWLINE)
7992 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
7993 syntax &= ~RE_DOT_NEWLINE;
7994 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
7995 /* It also changes the matching behavior. */
7996 preg->newline_anchor = 1;
7999 preg->newline_anchor = 0;
8001 preg->no_sub = !!(cflags & REG_NOSUB);
8003 /* POSIX says a null character in the pattern terminates it, so we
8004 can use strlen here in compiling the pattern. */
8006 if (MB_CUR_MAX != 1)
8007 ret = wcs_regex_compile (pattern, strlen (pattern), syntax, preg);
8010 ret = byte_regex_compile (pattern, strlen (pattern), syntax, preg);
8012 /* POSIX doesn't distinguish between an unmatched open-group and an
8013 unmatched close-group: both are REG_EPAREN. */
8014 if (ret == REG_ERPAREN) ret = REG_EPAREN;
8016 if (ret == REG_NOERROR && preg->fastmap)
8018 /* Compute the fastmap now, since regexec cannot modify the pattern
8020 if (re_compile_fastmap (preg) == -2)
8022 /* Some error occurred while computing the fastmap, just forget
8024 free (preg->fastmap);
8025 preg->fastmap = NULL;
8032 weak_alias (__regcomp, regcomp)
8036 /* regexec searches for a given pattern, specified by PREG, in the
8039 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8040 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8041 least NMATCH elements, and we set them to the offsets of the
8042 corresponding matched substrings.
8044 EFLAGS specifies `execution flags' which affect matching: if
8045 REG_NOTBOL is set, then ^ does not match at the beginning of the
8046 string; if REG_NOTEOL is set, then $ does not match at the end.
8048 We return 0 if we find a match and REG_NOMATCH if not. */
8051 regexec (preg, string, nmatch, pmatch, eflags)
8052 const regex_t *preg;
8055 regmatch_t pmatch[];
8059 struct re_registers regs;
8060 regex_t private_preg;
8061 int len = strlen (string);
8062 boolean want_reg_info = !preg->no_sub && nmatch > 0;
8064 private_preg = *preg;
8066 private_preg.not_bol = !!(eflags & REG_NOTBOL);
8067 private_preg.not_eol = !!(eflags & REG_NOTEOL);
8069 /* The user has told us exactly how many registers to return
8070 information about, via `nmatch'. We have to pass that on to the
8071 matching routines. */
8072 private_preg.regs_allocated = REGS_FIXED;
8076 regs.num_regs = nmatch;
8077 regs.start = TALLOC (nmatch * 2, regoff_t);
8078 if (regs.start == NULL)
8079 return (int) REG_NOMATCH;
8080 regs.end = regs.start + nmatch;
8083 /* Perform the searching operation. */
8084 ret = re_search (&private_preg, string, len,
8085 /* start: */ 0, /* range: */ len,
8086 want_reg_info ? ®s : (struct re_registers *) 0);
8088 /* Copy the register information to the POSIX structure. */
8095 for (r = 0; r < nmatch; r++)
8097 pmatch[r].rm_so = regs.start[r];
8098 pmatch[r].rm_eo = regs.end[r];
8102 /* If we needed the temporary register info, free the space now. */
8106 /* We want zero return to mean success, unlike `re_search'. */
8107 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
8110 weak_alias (__regexec, regexec)
8114 /* Returns a message corresponding to an error code, ERRCODE, returned
8115 from either regcomp or regexec. We don't use PREG here. */
8118 regerror (errcode, preg, errbuf, errbuf_size)
8120 const regex_t *preg;
8128 || errcode >= (int) (sizeof (re_error_msgid_idx)
8129 / sizeof (re_error_msgid_idx[0])))
8130 /* Only error codes returned by the rest of the code should be passed
8131 to this routine. If we are given anything else, or if other regex
8132 code generates an invalid error code, then the program has a bug.
8133 Dump core so we can fix it. */
8136 msg = gettext (re_error_msgid + re_error_msgid_idx[errcode]);
8138 msg_size = strlen (msg) + 1; /* Includes the null. */
8140 if (errbuf_size != 0)
8142 if (msg_size > errbuf_size)
8144 #if defined HAVE_MEMPCPY || defined _LIBC
8145 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
8147 memcpy (errbuf, msg, errbuf_size - 1);
8148 errbuf[errbuf_size - 1] = 0;
8152 memcpy (errbuf, msg, msg_size);
8158 weak_alias (__regerror, regerror)
8162 /* Free dynamically allocated space used by PREG. */
8168 if (preg->buffer != NULL)
8169 free (preg->buffer);
8170 preg->buffer = NULL;
8172 preg->allocated = 0;
8175 if (preg->fastmap != NULL)
8176 free (preg->fastmap);
8177 preg->fastmap = NULL;
8178 preg->fastmap_accurate = 0;
8180 if (preg->translate != NULL)
8181 free (preg->translate);
8182 preg->translate = NULL;
8185 weak_alias (__regfree, regfree)
8188 #endif /* not emacs */
8190 #endif /* not INSIDE_RECURSION */
8194 #undef STORE_NUMBER_AND_INCR
8195 #undef EXTRACT_NUMBER
8196 #undef EXTRACT_NUMBER_AND_INCR
8198 #undef DEBUG_PRINT_COMPILED_PATTERN
8199 #undef DEBUG_PRINT_DOUBLE_STRING
8201 #undef INIT_FAIL_STACK
8202 #undef RESET_FAIL_STACK
8203 #undef DOUBLE_FAIL_STACK
8204 #undef PUSH_PATTERN_OP
8205 #undef PUSH_FAILURE_POINTER
8206 #undef PUSH_FAILURE_INT
8207 #undef PUSH_FAILURE_ELT
8208 #undef POP_FAILURE_POINTER
8209 #undef POP_FAILURE_INT
8210 #undef POP_FAILURE_ELT
8213 #undef PUSH_FAILURE_POINT
8214 #undef POP_FAILURE_POINT
8216 #undef REG_UNSET_VALUE
8224 #undef INIT_BUF_SIZE
8225 #undef GET_BUFFER_SPACE
8233 #undef EXTEND_BUFFER
8234 #undef GET_UNSIGNED_NUMBER
8235 #undef FREE_STACK_RETURN
8237 # undef POINTER_TO_OFFSET
8238 # undef MATCHING_IN_FRST_STRING
8240 # undef AT_STRINGS_BEG
8241 # undef AT_STRINGS_END
8244 # undef FREE_VARIABLES
8245 # undef NO_HIGHEST_ACTIVE_REG
8246 # undef NO_LOWEST_ACTIVE_REG
8250 # undef COMPILED_BUFFER_VAR
8251 # undef OFFSET_ADDRESS_SIZE
8252 # undef CHAR_CLASS_SIZE
8259 # define DEFINED_ONCE
OSDN Git Service
