1 /* Extended regular expression matching and search library, version
2 0.12. (Implements POSIX draft P1003.2/D11.2, except for some of the
3 internationalization features.)
5 Copyright (C) 1993,94,95,96,97,98,99,2000 Free Software Foundation, Inc.
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2, or (at your option)
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
23 - structure the opcode space into opcode+flag.
24 - merge with glibc's regex.[ch].
25 - replace (succeed_n + jump_n + set_number_at) with something that doesn't
26 need to modify the compiled regexp so that re_match can be reentrant.
27 - get rid of on_failure_jump_smart by doing the optimization in re_comp
28 rather than at run-time, so that re_match can be reentrant.
31 /* AIX requires this to be the first thing in the file. */
32 #if defined _AIX && !defined REGEX_MALLOC
43 #if defined STDC_HEADERS && !defined emacs
46 /* We need this for `regex.h', and perhaps for the Emacs include files. */
47 # include <sys/types.h>
50 /* Whether to use ISO C Amendment 1 wide char functions.
51 Those should not be used for Emacs since it uses its own. */
52 #define WIDE_CHAR_SUPPORT \
53 (defined _LIBC || HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC && !emacs)
55 /* For platform which support the ISO C amendement 1 functionality we
56 support user defined character classes. */
58 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
64 /* We have to keep the namespace clean. */
65 # define regfree(preg) __regfree (preg)
66 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
67 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
68 # define regerror(errcode, preg, errbuf, errbuf_size) \
69 __regerror(errcode, preg, errbuf, errbuf_size)
70 # define re_set_registers(bu, re, nu, st, en) \
71 __re_set_registers (bu, re, nu, st, en)
72 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
73 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
74 # define re_match(bufp, string, size, pos, regs) \
75 __re_match (bufp, string, size, pos, regs)
76 # define re_search(bufp, string, size, startpos, range, regs) \
77 __re_search (bufp, string, size, startpos, range, regs)
78 # define re_compile_pattern(pattern, length, bufp) \
79 __re_compile_pattern (pattern, length, bufp)
80 # define re_set_syntax(syntax) __re_set_syntax (syntax)
81 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
82 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
83 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
85 /* Make sure we call libc's function even if the user overrides them. */
86 # define btowc __btowc
87 # define iswctype __iswctype
88 # define wctype __wctype
90 # define WEAK_ALIAS(a,b) weak_alias (a, b)
92 /* We are also using some library internals. */
93 # include <locale/localeinfo.h>
94 # include <locale/elem-hash.h>
95 # include <langinfo.h>
97 # define WEAK_ALIAS(a,b)
100 /* This is for other GNU distributions with internationalized messages. */
101 #if HAVE_LIBINTL_H || defined _LIBC
102 # include <libintl.h>
104 # define gettext(msgid) (msgid)
108 /* This define is so xgettext can find the internationalizable
110 # define gettext_noop(String) String
113 /* The `emacs' switch turns on certain matching commands
114 that make sense only in Emacs. */
120 /* Make syntax table lookup grant data in gl_state. */
121 # define SYNTAX_ENTRY_VIA_PROPERTY
124 # include "charset.h"
125 # include "category.h"
127 # define malloc xmalloc
128 # define realloc xrealloc
131 /* Converts the pointer to the char to BEG-based offset from the start. */
132 # define PTR_TO_OFFSET(d) POS_AS_IN_BUFFER (POINTER_TO_OFFSET (d))
133 # define POS_AS_IN_BUFFER(p) ((p) + (NILP (re_match_object) || BUFFERP (re_match_object)))
135 # define RE_MULTIBYTE_P(bufp) ((bufp)->multibyte)
136 # define RE_STRING_CHAR(p, s) \
137 (multibyte ? (STRING_CHAR (p, s)) : (*(p)))
138 # define RE_STRING_CHAR_AND_LENGTH(p, s, len) \
139 (multibyte ? (STRING_CHAR_AND_LENGTH (p, s, len)) : ((len) = 1, *(p)))
141 /* Set C a (possibly multibyte) character before P. P points into a
142 string which is the virtual concatenation of STR1 (which ends at
143 END1) or STR2 (which ends at END2). */
144 # define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
148 re_char *dtemp = (p) == (str2) ? (end1) : (p); \
149 re_char *dlimit = ((p) > (str2) && (p) <= (end2)) ? (str2) : (str1); \
150 while (dtemp-- > dlimit && !CHAR_HEAD_P (*dtemp)); \
151 c = STRING_CHAR (dtemp, (p) - dtemp); \
154 (c = ((p) == (str2) ? (end1) : (p))[-1]); \
158 #else /* not emacs */
160 /* If we are not linking with Emacs proper,
161 we can't use the relocating allocator
162 even if config.h says that we can. */
165 # if defined STDC_HEADERS || defined _LIBC
172 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
173 If nothing else has been done, use the method below. */
174 # ifdef INHIBIT_STRING_HEADER
175 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
176 # if !defined bzero && !defined bcopy
177 # undef INHIBIT_STRING_HEADER
182 /* This is the normal way of making sure we have memcpy, memcmp and bzero.
183 This is used in most programs--a few other programs avoid this
184 by defining INHIBIT_STRING_HEADER. */
185 # ifndef INHIBIT_STRING_HEADER
186 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
190 # define bzero(s, n) (memset (s, '\0', n), (s))
192 # define bzero(s, n) __bzero (s, n)
196 # include <strings.h>
198 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
201 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
206 /* Define the syntax stuff for \<, \>, etc. */
208 /* Sword must be nonzero for the wordchar pattern commands in re_match_2. */
209 enum syntaxcode { Swhitespace = 0, Sword = 1 };
211 # ifdef SWITCH_ENUM_BUG
212 # define SWITCH_ENUM_CAST(x) ((int)(x))
214 # define SWITCH_ENUM_CAST(x) (x)
217 /* Dummy macros for non-Emacs environments. */
218 # define BASE_LEADING_CODE_P(c) (0)
219 # define CHAR_CHARSET(c) 0
220 # define CHARSET_LEADING_CODE_BASE(c) 0
221 # define MAX_MULTIBYTE_LENGTH 1
222 # define RE_MULTIBYTE_P(x) 0
223 # define WORD_BOUNDARY_P(c1, c2) (0)
224 # define CHAR_HEAD_P(p) (1)
225 # define SINGLE_BYTE_CHAR_P(c) (1)
226 # define SAME_CHARSET_P(c1, c2) (1)
227 # define MULTIBYTE_FORM_LENGTH(p, s) (1)
228 # define STRING_CHAR(p, s) (*(p))
229 # define RE_STRING_CHAR STRING_CHAR
230 # define CHAR_STRING(c, s) (*(s) = (c), 1)
231 # define STRING_CHAR_AND_LENGTH(p, s, actual_len) ((actual_len) = 1, *(p))
232 # define RE_STRING_CHAR_AND_LENGTH STRING_CHAR_AND_LENGTH
233 # define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
234 (c = ((p) == (str2) ? *((end1) - 1) : *((p) - 1)))
235 # define MAKE_CHAR(charset, c1, c2) (c1)
236 #endif /* not emacs */
239 # define RE_TRANSLATE(TBL, C) ((unsigned char)(TBL)[C])
240 # define RE_TRANSLATE_P(TBL) (TBL)
243 /* Get the interface, including the syntax bits. */
246 /* isalpha etc. are used for the character classes. */
251 /* 1 if C is an ASCII character. */
252 # define IS_REAL_ASCII(c) ((c) < 0200)
254 /* 1 if C is a unibyte character. */
255 # define ISUNIBYTE(c) (SINGLE_BYTE_CHAR_P ((c)))
257 /* The Emacs definitions should not be directly affected by locales. */
259 /* In Emacs, these are only used for single-byte characters. */
260 # define ISDIGIT(c) ((c) >= '0' && (c) <= '9')
261 # define ISCNTRL(c) ((c) < ' ')
262 # define ISXDIGIT(c) (((c) >= '0' && (c) <= '9') \
263 || ((c) >= 'a' && (c) <= 'f') \
264 || ((c) >= 'A' && (c) <= 'F'))
266 /* This is only used for single-byte characters. */
267 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
269 /* The rest must handle multibyte characters. */
271 # define ISGRAPH(c) (SINGLE_BYTE_CHAR_P (c) \
272 ? (c) > ' ' && !((c) >= 0177 && (c) <= 0237) \
275 # define ISPRINT(c) (SINGLE_BYTE_CHAR_P (c) \
276 ? (c) >= ' ' && !((c) >= 0177 && (c) <= 0237) \
279 # define ISALNUM(c) (IS_REAL_ASCII (c) \
280 ? (((c) >= 'a' && (c) <= 'z') \
281 || ((c) >= 'A' && (c) <= 'Z') \
282 || ((c) >= '0' && (c) <= '9')) \
283 : SYNTAX (c) == Sword)
285 # define ISALPHA(c) (IS_REAL_ASCII (c) \
286 ? (((c) >= 'a' && (c) <= 'z') \
287 || ((c) >= 'A' && (c) <= 'Z')) \
288 : SYNTAX (c) == Sword)
290 # define ISLOWER(c) (LOWERCASEP (c))
292 # define ISPUNCT(c) (IS_REAL_ASCII (c) \
293 ? ((c) > ' ' && (c) < 0177 \
294 && !(((c) >= 'a' && (c) <= 'z') \
295 || ((c) >= 'A' && (c) <= 'Z') \
296 || ((c) >= '0' && (c) <= '9'))) \
297 : SYNTAX (c) != Sword)
299 # define ISSPACE(c) (SYNTAX (c) == Swhitespace)
301 # define ISUPPER(c) (UPPERCASEP (c))
303 # define ISWORD(c) (SYNTAX (c) == Sword)
305 #else /* not emacs */
307 /* Jim Meyering writes:
309 "... Some ctype macros are valid only for character codes that
310 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
311 using /bin/cc or gcc but without giving an ansi option). So, all
312 ctype uses should be through macros like ISPRINT... If
313 STDC_HEADERS is defined, then autoconf has verified that the ctype
314 macros don't need to be guarded with references to isascii. ...
315 Defining isascii to 1 should let any compiler worth its salt
316 eliminate the && through constant folding."
317 Solaris defines some of these symbols so we must undefine them first. */
320 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
321 # define ISASCII(c) 1
323 # define ISASCII(c) isascii(c)
326 /* 1 if C is an ASCII character. */
327 # define IS_REAL_ASCII(c) ((c) < 0200)
329 /* This distinction is not meaningful, except in Emacs. */
330 # define ISUNIBYTE(c) 1
333 # define ISBLANK(c) (ISASCII (c) && isblank (c))
335 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
338 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
340 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
344 # define ISPRINT(c) (ISASCII (c) && isprint (c))
345 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
346 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
347 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
348 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
349 # define ISLOWER(c) (ISASCII (c) && islower (c))
350 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
351 # define ISSPACE(c) (ISASCII (c) && isspace (c))
352 # define ISUPPER(c) (ISASCII (c) && isupper (c))
353 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
355 # define ISWORD(c) ISALPHA(c)
358 # define TOLOWER(c) _tolower(c)
360 # define TOLOWER(c) tolower(c)
363 /* How many characters in the character set. */
364 # define CHAR_SET_SIZE 256
368 extern char *re_syntax_table;
370 # else /* not SYNTAX_TABLE */
372 static char re_syntax_table[CHAR_SET_SIZE];
383 bzero (re_syntax_table, sizeof re_syntax_table);
385 for (c = 0; c < CHAR_SET_SIZE; ++c)
387 re_syntax_table[c] = Sword;
389 re_syntax_table['_'] = Sword;
394 # endif /* not SYNTAX_TABLE */
396 # define SYNTAX(c) re_syntax_table[(c)]
398 #endif /* not emacs */
401 # define NULL (void *)0
404 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
405 since ours (we hope) works properly with all combinations of
406 machines, compilers, `char' and `unsigned char' argument types.
407 (Per Bothner suggested the basic approach.) */
408 #undef SIGN_EXTEND_CHAR
410 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
411 #else /* not __STDC__ */
412 /* As in Harbison and Steele. */
413 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
416 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
417 use `alloca' instead of `malloc'. This is because using malloc in
418 re_search* or re_match* could cause memory leaks when C-g is used in
419 Emacs; also, malloc is slower and causes storage fragmentation. On
420 the other hand, malloc is more portable, and easier to debug.
422 Because we sometimes use alloca, some routines have to be macros,
423 not functions -- `alloca'-allocated space disappears at the end of the
424 function it is called in. */
428 # define REGEX_ALLOCATE malloc
429 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
430 # define REGEX_FREE free
432 #else /* not REGEX_MALLOC */
434 /* Emacs already defines alloca, sometimes. */
437 /* Make alloca work the best possible way. */
439 # define alloca __builtin_alloca
440 # else /* not __GNUC__ */
443 # endif /* HAVE_ALLOCA_H */
444 # endif /* not __GNUC__ */
446 # endif /* not alloca */
448 # define REGEX_ALLOCATE alloca
450 /* Assumes a `char *destination' variable. */
451 # define REGEX_REALLOCATE(source, osize, nsize) \
452 (destination = (char *) alloca (nsize), \
453 memcpy (destination, source, osize))
455 /* No need to do anything to free, after alloca. */
456 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
458 #endif /* not REGEX_MALLOC */
460 /* Define how to allocate the failure stack. */
462 #if defined REL_ALLOC && defined REGEX_MALLOC
464 # define REGEX_ALLOCATE_STACK(size) \
465 r_alloc (&failure_stack_ptr, (size))
466 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
467 r_re_alloc (&failure_stack_ptr, (nsize))
468 # define REGEX_FREE_STACK(ptr) \
469 r_alloc_free (&failure_stack_ptr)
471 #else /* not using relocating allocator */
475 # define REGEX_ALLOCATE_STACK malloc
476 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
477 # define REGEX_FREE_STACK free
479 # else /* not REGEX_MALLOC */
481 # define REGEX_ALLOCATE_STACK alloca
483 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
484 REGEX_REALLOCATE (source, osize, nsize)
485 /* No need to explicitly free anything. */
486 # define REGEX_FREE_STACK(arg) ((void)0)
488 # endif /* not REGEX_MALLOC */
489 #endif /* not using relocating allocator */
492 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
493 `string1' or just past its end. This works if PTR is NULL, which is
495 #define FIRST_STRING_P(ptr) \
496 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
498 /* (Re)Allocate N items of type T using malloc, or fail. */
499 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
500 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
501 #define RETALLOC_IF(addr, n, t) \
502 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
503 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
505 #define BYTEWIDTH 8 /* In bits. */
507 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
511 #define MAX(a, b) ((a) > (b) ? (a) : (b))
512 #define MIN(a, b) ((a) < (b) ? (a) : (b))
514 /* Type of source-pattern and string chars. */
515 typedef const unsigned char re_char;
517 typedef char boolean;
521 static int re_match_2_internal _RE_ARGS ((struct re_pattern_buffer *bufp,
522 re_char *string1, int size1,
523 re_char *string2, int size2,
525 struct re_registers *regs,
528 /* These are the command codes that appear in compiled regular
529 expressions. Some opcodes are followed by argument bytes. A
530 command code can specify any interpretation whatsoever for its
531 arguments. Zero bytes may appear in the compiled regular expression. */
537 /* Succeed right away--no more backtracking. */
540 /* Followed by one byte giving n, then by n literal bytes. */
543 /* Matches any (more or less) character. */
546 /* Matches any one char belonging to specified set. First
547 following byte is number of bitmap bytes. Then come bytes
548 for a bitmap saying which chars are in. Bits in each byte
549 are ordered low-bit-first. A character is in the set if its
550 bit is 1. A character too large to have a bit in the map is
551 automatically not in the set.
553 If the length byte has the 0x80 bit set, then that stuff
554 is followed by a range table:
555 2 bytes of flags for character sets (low 8 bits, high 8 bits)
556 See RANGE_TABLE_WORK_BITS below.
557 2 bytes, the number of pairs that follow (upto 32767)
558 pairs, each 2 multibyte characters,
559 each multibyte character represented as 3 bytes. */
562 /* Same parameters as charset, but match any character that is
563 not one of those specified. */
566 /* Start remembering the text that is matched, for storing in a
567 register. Followed by one byte with the register number, in
568 the range 0 to one less than the pattern buffer's re_nsub
572 /* Stop remembering the text that is matched and store it in a
573 memory register. Followed by one byte with the register
574 number, in the range 0 to one less than `re_nsub' in the
578 /* Match a duplicate of something remembered. Followed by one
579 byte containing the register number. */
582 /* Fail unless at beginning of line. */
585 /* Fail unless at end of line. */
588 /* Succeeds if at beginning of buffer (if emacs) or at beginning
589 of string to be matched (if not). */
592 /* Analogously, for end of buffer/string. */
595 /* Followed by two byte relative address to which to jump. */
598 /* Followed by two-byte relative address of place to resume at
599 in case of failure. */
602 /* Like on_failure_jump, but pushes a placeholder instead of the
603 current string position when executed. */
604 on_failure_keep_string_jump,
606 /* Just like `on_failure_jump', except that it checks that we
607 don't get stuck in an infinite loop (matching an empty string
609 on_failure_jump_loop,
611 /* Just like `on_failure_jump_loop', except that it checks for
612 a different kind of loop (the kind that shows up with non-greedy
613 operators). This operation has to be immediately preceded
615 on_failure_jump_nastyloop,
617 /* A smart `on_failure_jump' used for greedy * and + operators.
618 It analyses the loop before which it is put and if the
619 loop does not require backtracking, it changes itself to
620 `on_failure_keep_string_jump' and short-circuits the loop,
621 else it just defaults to changing itself into `on_failure_jump'.
622 It assumes that it is pointing to just past a `jump'. */
623 on_failure_jump_smart,
625 /* Followed by two-byte relative address and two-byte number n.
626 After matching N times, jump to the address upon failure.
627 Does not work if N starts at 0: use on_failure_jump_loop
631 /* Followed by two-byte relative address, and two-byte number n.
632 Jump to the address N times, then fail. */
635 /* Set the following two-byte relative address to the
636 subsequent two-byte number. The address *includes* the two
640 wordbeg, /* Succeeds if at word beginning. */
641 wordend, /* Succeeds if at word end. */
643 wordbound, /* Succeeds if at a word boundary. */
644 notwordbound, /* Succeeds if not at a word boundary. */
646 /* Matches any character whose syntax is specified. Followed by
647 a byte which contains a syntax code, e.g., Sword. */
650 /* Matches any character whose syntax is not that specified. */
654 ,before_dot, /* Succeeds if before point. */
655 at_dot, /* Succeeds if at point. */
656 after_dot, /* Succeeds if after point. */
658 /* Matches any character whose category-set contains the specified
659 category. The operator is followed by a byte which contains a
660 category code (mnemonic ASCII character). */
663 /* Matches any character whose category-set does not contain the
664 specified category. The operator is followed by a byte which
665 contains the category code (mnemonic ASCII character). */
670 /* Common operations on the compiled pattern. */
672 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
674 #define STORE_NUMBER(destination, number) \
676 (destination)[0] = (number) & 0377; \
677 (destination)[1] = (number) >> 8; \
680 /* Same as STORE_NUMBER, except increment DESTINATION to
681 the byte after where the number is stored. Therefore, DESTINATION
682 must be an lvalue. */
684 #define STORE_NUMBER_AND_INCR(destination, number) \
686 STORE_NUMBER (destination, number); \
687 (destination) += 2; \
690 /* Put into DESTINATION a number stored in two contiguous bytes starting
693 #define EXTRACT_NUMBER(destination, source) \
695 (destination) = *(source) & 0377; \
696 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
700 static void extract_number _RE_ARGS ((int *dest, re_char *source));
702 extract_number (dest, source)
706 int temp = SIGN_EXTEND_CHAR (*(source + 1));
707 *dest = *source & 0377;
711 # ifndef EXTRACT_MACROS /* To debug the macros. */
712 # undef EXTRACT_NUMBER
713 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
714 # endif /* not EXTRACT_MACROS */
718 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
719 SOURCE must be an lvalue. */
721 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
723 EXTRACT_NUMBER (destination, source); \
728 static void extract_number_and_incr _RE_ARGS ((int *destination,
731 extract_number_and_incr (destination, source)
735 extract_number (destination, *source);
739 # ifndef EXTRACT_MACROS
740 # undef EXTRACT_NUMBER_AND_INCR
741 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
742 extract_number_and_incr (&dest, &src)
743 # endif /* not EXTRACT_MACROS */
747 /* Store a multibyte character in three contiguous bytes starting
748 DESTINATION, and increment DESTINATION to the byte after where the
749 character is stored. Therefore, DESTINATION must be an lvalue. */
751 #define STORE_CHARACTER_AND_INCR(destination, character) \
753 (destination)[0] = (character) & 0377; \
754 (destination)[1] = ((character) >> 8) & 0377; \
755 (destination)[2] = (character) >> 16; \
756 (destination) += 3; \
759 /* Put into DESTINATION a character stored in three contiguous bytes
760 starting at SOURCE. */
762 #define EXTRACT_CHARACTER(destination, source) \
764 (destination) = ((source)[0] \
765 | ((source)[1] << 8) \
766 | ((source)[2] << 16)); \
770 /* Macros for charset. */
772 /* Size of bitmap of charset P in bytes. P is a start of charset,
773 i.e. *P is (re_opcode_t) charset or (re_opcode_t) charset_not. */
774 #define CHARSET_BITMAP_SIZE(p) ((p)[1] & 0x7F)
776 /* Nonzero if charset P has range table. */
777 #define CHARSET_RANGE_TABLE_EXISTS_P(p) ((p)[1] & 0x80)
779 /* Return the address of range table of charset P. But not the start
780 of table itself, but the before where the number of ranges is
781 stored. `2 +' means to skip re_opcode_t and size of bitmap,
782 and the 2 bytes of flags at the start of the range table. */
783 #define CHARSET_RANGE_TABLE(p) (&(p)[4 + CHARSET_BITMAP_SIZE (p)])
785 /* Extract the bit flags that start a range table. */
786 #define CHARSET_RANGE_TABLE_BITS(p) \
787 ((p)[2 + CHARSET_BITMAP_SIZE (p)] \
788 + (p)[3 + CHARSET_BITMAP_SIZE (p)] * 0x100)
790 /* Test if C is listed in the bitmap of charset P. */
791 #define CHARSET_LOOKUP_BITMAP(p, c) \
792 ((c) < CHARSET_BITMAP_SIZE (p) * BYTEWIDTH \
793 && (p)[2 + (c) / BYTEWIDTH] & (1 << ((c) % BYTEWIDTH)))
795 /* Return the address of end of RANGE_TABLE. COUNT is number of
796 ranges (which is a pair of (start, end)) in the RANGE_TABLE. `* 2'
797 is start of range and end of range. `* 3' is size of each start
799 #define CHARSET_RANGE_TABLE_END(range_table, count) \
800 ((range_table) + (count) * 2 * 3)
802 /* Test if C is in RANGE_TABLE. A flag NOT is negated if C is in.
803 COUNT is number of ranges in RANGE_TABLE. */
804 #define CHARSET_LOOKUP_RANGE_TABLE_RAW(not, c, range_table, count) \
807 re_wchar_t range_start, range_end; \
809 re_char *range_table_end \
810 = CHARSET_RANGE_TABLE_END ((range_table), (count)); \
812 for (p = (range_table); p < range_table_end; p += 2 * 3) \
814 EXTRACT_CHARACTER (range_start, p); \
815 EXTRACT_CHARACTER (range_end, p + 3); \
817 if (range_start <= (c) && (c) <= range_end) \
826 /* Test if C is in range table of CHARSET. The flag NOT is negated if
827 C is listed in it. */
828 #define CHARSET_LOOKUP_RANGE_TABLE(not, c, charset) \
831 /* Number of ranges in range table. */ \
833 re_char *range_table = CHARSET_RANGE_TABLE (charset); \
835 EXTRACT_NUMBER_AND_INCR (count, range_table); \
836 CHARSET_LOOKUP_RANGE_TABLE_RAW ((not), (c), range_table, count); \
840 /* If DEBUG is defined, Regex prints many voluminous messages about what
841 it is doing (if the variable `debug' is nonzero). If linked with the
842 main program in `iregex.c', you can enter patterns and strings
843 interactively. And if linked with the main program in `main.c' and
844 the other test files, you can run the already-written tests. */
848 /* We use standard I/O for debugging. */
851 /* It is useful to test things that ``must'' be true when debugging. */
854 static int debug = -100000;
856 # define DEBUG_STATEMENT(e) e
857 # define DEBUG_PRINT1(x) if (debug > 0) printf (x)
858 # define DEBUG_PRINT2(x1, x2) if (debug > 0) printf (x1, x2)
859 # define DEBUG_PRINT3(x1, x2, x3) if (debug > 0) printf (x1, x2, x3)
860 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug > 0) printf (x1, x2, x3, x4)
861 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
862 if (debug > 0) print_partial_compiled_pattern (s, e)
863 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
864 if (debug > 0) print_double_string (w, s1, sz1, s2, sz2)
867 /* Print the fastmap in human-readable form. */
870 print_fastmap (fastmap)
873 unsigned was_a_range = 0;
876 while (i < (1 << BYTEWIDTH))
882 while (i < (1 << BYTEWIDTH) && fastmap[i])
898 /* Print a compiled pattern string in human-readable form, starting at
899 the START pointer into it and ending just before the pointer END. */
902 print_partial_compiled_pattern (start, end)
916 /* Loop over pattern commands. */
919 printf ("%d:\t", p - start);
921 switch ((re_opcode_t) *p++)
933 printf ("/exactn/%d", mcnt);
943 printf ("/start_memory/%d", *p++);
947 printf ("/stop_memory/%d", *p++);
951 printf ("/duplicate/%d", *p++);
961 register int c, last = -100;
962 register int in_range = 0;
963 int length = CHARSET_BITMAP_SIZE (p - 1);
964 int has_range_table = CHARSET_RANGE_TABLE_EXISTS_P (p - 1);
966 printf ("/charset [%s",
967 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
969 assert (p + *p < pend);
971 for (c = 0; c < 256; c++)
973 && (p[1 + (c/8)] & (1 << (c % 8))))
975 /* Are we starting a range? */
976 if (last + 1 == c && ! in_range)
981 /* Have we broken a range? */
982 else if (last + 1 != c && in_range)
1001 if (has_range_table)
1004 printf ("has-range-table");
1006 /* ??? Should print the range table; for now, just skip it. */
1007 p += 2; /* skip range table bits */
1008 EXTRACT_NUMBER_AND_INCR (count, p);
1009 p = CHARSET_RANGE_TABLE_END (p, count);
1015 printf ("/begline");
1019 printf ("/endline");
1022 case on_failure_jump:
1023 extract_number_and_incr (&mcnt, &p);
1024 printf ("/on_failure_jump to %d", p + mcnt - start);
1027 case on_failure_keep_string_jump:
1028 extract_number_and_incr (&mcnt, &p);
1029 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
1032 case on_failure_jump_nastyloop:
1033 extract_number_and_incr (&mcnt, &p);
1034 printf ("/on_failure_jump_nastyloop to %d", p + mcnt - start);
1037 case on_failure_jump_loop:
1038 extract_number_and_incr (&mcnt, &p);
1039 printf ("/on_failure_jump_loop to %d", p + mcnt - start);
1042 case on_failure_jump_smart:
1043 extract_number_and_incr (&mcnt, &p);
1044 printf ("/on_failure_jump_smart to %d", p + mcnt - start);
1048 extract_number_and_incr (&mcnt, &p);
1049 printf ("/jump to %d", p + mcnt - start);
1053 extract_number_and_incr (&mcnt, &p);
1054 extract_number_and_incr (&mcnt2, &p);
1055 printf ("/succeed_n to %d, %d times", p - 2 + mcnt - start, mcnt2);
1059 extract_number_and_incr (&mcnt, &p);
1060 extract_number_and_incr (&mcnt2, &p);
1061 printf ("/jump_n to %d, %d times", p - 2 + mcnt - start, mcnt2);
1065 extract_number_and_incr (&mcnt, &p);
1066 extract_number_and_incr (&mcnt2, &p);
1067 printf ("/set_number_at location %d to %d", p - 2 + mcnt - start, mcnt2);
1071 printf ("/wordbound");
1075 printf ("/notwordbound");
1079 printf ("/wordbeg");
1083 printf ("/wordend");
1086 printf ("/syntaxspec");
1088 printf ("/%d", mcnt);
1092 printf ("/notsyntaxspec");
1094 printf ("/%d", mcnt);
1099 printf ("/before_dot");
1107 printf ("/after_dot");
1111 printf ("/categoryspec");
1113 printf ("/%d", mcnt);
1116 case notcategoryspec:
1117 printf ("/notcategoryspec");
1119 printf ("/%d", mcnt);
1132 printf ("?%d", *(p-1));
1138 printf ("%d:\tend of pattern.\n", p - start);
1143 print_compiled_pattern (bufp)
1144 struct re_pattern_buffer *bufp;
1146 re_char *buffer = bufp->buffer;
1148 print_partial_compiled_pattern (buffer, buffer + bufp->used);
1149 printf ("%ld bytes used/%ld bytes allocated.\n",
1150 bufp->used, bufp->allocated);
1152 if (bufp->fastmap_accurate && bufp->fastmap)
1154 printf ("fastmap: ");
1155 print_fastmap (bufp->fastmap);
1158 printf ("re_nsub: %d\t", bufp->re_nsub);
1159 printf ("regs_alloc: %d\t", bufp->regs_allocated);
1160 printf ("can_be_null: %d\t", bufp->can_be_null);
1161 printf ("no_sub: %d\t", bufp->no_sub);
1162 printf ("not_bol: %d\t", bufp->not_bol);
1163 printf ("not_eol: %d\t", bufp->not_eol);
1164 printf ("syntax: %lx\n", bufp->syntax);
1166 /* Perhaps we should print the translate table? */
1171 print_double_string (where, string1, size1, string2, size2)
1184 if (FIRST_STRING_P (where))
1186 for (this_char = where - string1; this_char < size1; this_char++)
1187 putchar (string1[this_char]);
1192 for (this_char = where - string2; this_char < size2; this_char++)
1193 putchar (string2[this_char]);
1197 #else /* not DEBUG */
1202 # define DEBUG_STATEMENT(e)
1203 # define DEBUG_PRINT1(x)
1204 # define DEBUG_PRINT2(x1, x2)
1205 # define DEBUG_PRINT3(x1, x2, x3)
1206 # define DEBUG_PRINT4(x1, x2, x3, x4)
1207 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1208 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1210 #endif /* not DEBUG */
1212 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1213 also be assigned to arbitrarily: each pattern buffer stores its own
1214 syntax, so it can be changed between regex compilations. */
1215 /* This has no initializer because initialized variables in Emacs
1216 become read-only after dumping. */
1217 reg_syntax_t re_syntax_options;
1220 /* Specify the precise syntax of regexps for compilation. This provides
1221 for compatibility for various utilities which historically have
1222 different, incompatible syntaxes.
1224 The argument SYNTAX is a bit mask comprised of the various bits
1225 defined in regex.h. We return the old syntax. */
1228 re_set_syntax (syntax)
1229 reg_syntax_t syntax;
1231 reg_syntax_t ret = re_syntax_options;
1233 re_syntax_options = syntax;
1236 WEAK_ALIAS (__re_set_syntax, re_set_syntax)
1238 /* This table gives an error message for each of the error codes listed
1239 in regex.h. Obviously the order here has to be same as there.
1240 POSIX doesn't require that we do anything for REG_NOERROR,
1241 but why not be nice? */
1243 static const char *re_error_msgid[] =
1245 gettext_noop ("Success"), /* REG_NOERROR */
1246 gettext_noop ("No match"), /* REG_NOMATCH */
1247 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
1248 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
1249 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
1250 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
1251 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1252 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1253 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1254 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1255 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1256 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1257 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1258 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1259 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1260 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1261 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1264 /* Avoiding alloca during matching, to placate r_alloc. */
1266 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1267 searching and matching functions should not call alloca. On some
1268 systems, alloca is implemented in terms of malloc, and if we're
1269 using the relocating allocator routines, then malloc could cause a
1270 relocation, which might (if the strings being searched are in the
1271 ralloc heap) shift the data out from underneath the regexp
1274 Here's another reason to avoid allocation: Emacs
1275 processes input from X in a signal handler; processing X input may
1276 call malloc; if input arrives while a matching routine is calling
1277 malloc, then we're scrod. But Emacs can't just block input while
1278 calling matching routines; then we don't notice interrupts when
1279 they come in. So, Emacs blocks input around all regexp calls
1280 except the matching calls, which it leaves unprotected, in the
1281 faith that they will not malloc. */
1283 /* Normally, this is fine. */
1284 #define MATCH_MAY_ALLOCATE
1286 /* When using GNU C, we are not REALLY using the C alloca, no matter
1287 what config.h may say. So don't take precautions for it. */
1292 /* The match routines may not allocate if (1) they would do it with malloc
1293 and (2) it's not safe for them to use malloc.
1294 Note that if REL_ALLOC is defined, matching would not use malloc for the
1295 failure stack, but we would still use it for the register vectors;
1296 so REL_ALLOC should not affect this. */
1297 #if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1298 # undef MATCH_MAY_ALLOCATE
1302 /* Failure stack declarations and macros; both re_compile_fastmap and
1303 re_match_2 use a failure stack. These have to be macros because of
1304 REGEX_ALLOCATE_STACK. */
1307 /* Approximate number of failure points for which to initially allocate space
1308 when matching. If this number is exceeded, we allocate more
1309 space, so it is not a hard limit. */
1310 #ifndef INIT_FAILURE_ALLOC
1311 # define INIT_FAILURE_ALLOC 20
1314 /* Roughly the maximum number of failure points on the stack. Would be
1315 exactly that if always used TYPICAL_FAILURE_SIZE items each time we failed.
1316 This is a variable only so users of regex can assign to it; we never
1317 change it ourselves. */
1318 # if defined MATCH_MAY_ALLOCATE
1319 /* Note that 4400 was enough to cause a crash on Alpha OSF/1,
1320 whose default stack limit is 2mb. In order for a larger
1321 value to work reliably, you have to try to make it accord
1322 with the process stack limit. */
1323 size_t re_max_failures = 40000;
1325 size_t re_max_failures = 4000;
1328 union fail_stack_elt
1331 /* This should be the biggest `int' that's no bigger than a pointer. */
1335 typedef union fail_stack_elt fail_stack_elt_t;
1339 fail_stack_elt_t *stack;
1341 size_t avail; /* Offset of next open position. */
1342 size_t frame; /* Offset of the cur constructed frame. */
1345 #define FAIL_STACK_EMPTY() (fail_stack.frame == 0)
1346 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1349 /* Define macros to initialize and free the failure stack.
1350 Do `return -2' if the alloc fails. */
1352 #ifdef MATCH_MAY_ALLOCATE
1353 # define INIT_FAIL_STACK() \
1355 fail_stack.stack = (fail_stack_elt_t *) \
1356 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * TYPICAL_FAILURE_SIZE \
1357 * sizeof (fail_stack_elt_t)); \
1359 if (fail_stack.stack == NULL) \
1362 fail_stack.size = INIT_FAILURE_ALLOC; \
1363 fail_stack.avail = 0; \
1364 fail_stack.frame = 0; \
1367 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1369 # define INIT_FAIL_STACK() \
1371 fail_stack.avail = 0; \
1372 fail_stack.frame = 0; \
1375 # define RESET_FAIL_STACK() ((void)0)
1379 /* Double the size of FAIL_STACK, up to a limit
1380 which allows approximately `re_max_failures' items.
1382 Return 1 if succeeds, and 0 if either ran out of memory
1383 allocating space for it or it was already too large.
1385 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1387 /* Factor to increase the failure stack size by
1388 when we increase it.
1389 This used to be 2, but 2 was too wasteful
1390 because the old discarded stacks added up to as much space
1391 were as ultimate, maximum-size stack. */
1392 #define FAIL_STACK_GROWTH_FACTOR 4
1394 #define GROW_FAIL_STACK(fail_stack) \
1395 (((fail_stack).size * sizeof (fail_stack_elt_t) \
1396 >= re_max_failures * TYPICAL_FAILURE_SIZE) \
1398 : ((fail_stack).stack \
1399 = (fail_stack_elt_t *) \
1400 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1401 (fail_stack).size * sizeof (fail_stack_elt_t), \
1402 MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
1403 ((fail_stack).size * sizeof (fail_stack_elt_t) \
1404 * FAIL_STACK_GROWTH_FACTOR))), \
1406 (fail_stack).stack == NULL \
1408 : ((fail_stack).size \
1409 = (MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
1410 ((fail_stack).size * sizeof (fail_stack_elt_t) \
1411 * FAIL_STACK_GROWTH_FACTOR)) \
1412 / sizeof (fail_stack_elt_t)), \
1416 /* Push a pointer value onto the failure stack.
1417 Assumes the variable `fail_stack'. Probably should only
1418 be called from within `PUSH_FAILURE_POINT'. */
1419 #define PUSH_FAILURE_POINTER(item) \
1420 fail_stack.stack[fail_stack.avail++].pointer = (item)
1422 /* This pushes an integer-valued item onto the failure stack.
1423 Assumes the variable `fail_stack'. Probably should only
1424 be called from within `PUSH_FAILURE_POINT'. */
1425 #define PUSH_FAILURE_INT(item) \
1426 fail_stack.stack[fail_stack.avail++].integer = (item)
1428 /* Push a fail_stack_elt_t value onto the failure stack.
1429 Assumes the variable `fail_stack'. Probably should only
1430 be called from within `PUSH_FAILURE_POINT'. */
1431 #define PUSH_FAILURE_ELT(item) \
1432 fail_stack.stack[fail_stack.avail++] = (item)
1434 /* These three POP... operations complement the three PUSH... operations.
1435 All assume that `fail_stack' is nonempty. */
1436 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1437 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1438 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1440 /* Individual items aside from the registers. */
1441 #define NUM_NONREG_ITEMS 3
1443 /* Used to examine the stack (to detect infinite loops). */
1444 #define FAILURE_PAT(h) fail_stack.stack[(h) - 1].pointer
1445 #define FAILURE_STR(h) (fail_stack.stack[(h) - 2].pointer)
1446 #define NEXT_FAILURE_HANDLE(h) fail_stack.stack[(h) - 3].integer
1447 #define TOP_FAILURE_HANDLE() fail_stack.frame
1450 #define ENSURE_FAIL_STACK(space) \
1451 while (REMAINING_AVAIL_SLOTS <= space) { \
1452 if (!GROW_FAIL_STACK (fail_stack)) \
1454 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", (fail_stack).size);\
1455 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1458 /* Push register NUM onto the stack. */
1459 #define PUSH_FAILURE_REG(num) \
1461 char *destination; \
1462 ENSURE_FAIL_STACK(3); \
1463 DEBUG_PRINT4 (" Push reg %d (spanning %p -> %p)\n", \
1464 num, regstart[num], regend[num]); \
1465 PUSH_FAILURE_POINTER (regstart[num]); \
1466 PUSH_FAILURE_POINTER (regend[num]); \
1467 PUSH_FAILURE_INT (num); \
1470 /* Change the counter's value to VAL, but make sure that it will
1471 be reset when backtracking. */
1472 #define PUSH_NUMBER(ptr,val) \
1474 char *destination; \
1476 ENSURE_FAIL_STACK(3); \
1477 EXTRACT_NUMBER (c, ptr); \
1478 DEBUG_PRINT4 (" Push number %p = %d -> %d\n", ptr, c, val); \
1479 PUSH_FAILURE_INT (c); \
1480 PUSH_FAILURE_POINTER (ptr); \
1481 PUSH_FAILURE_INT (-1); \
1482 STORE_NUMBER (ptr, val); \
1485 /* Pop a saved register off the stack. */
1486 #define POP_FAILURE_REG_OR_COUNT() \
1488 int reg = POP_FAILURE_INT (); \
1491 /* It's a counter. */ \
1492 /* Here, we discard `const', making re_match non-reentrant. */ \
1493 unsigned char *ptr = (unsigned char*) POP_FAILURE_POINTER (); \
1494 reg = POP_FAILURE_INT (); \
1495 STORE_NUMBER (ptr, reg); \
1496 DEBUG_PRINT3 (" Pop counter %p = %d\n", ptr, reg); \
1500 regend[reg] = POP_FAILURE_POINTER (); \
1501 regstart[reg] = POP_FAILURE_POINTER (); \
1502 DEBUG_PRINT4 (" Pop reg %d (spanning %p -> %p)\n", \
1503 reg, regstart[reg], regend[reg]); \
1507 /* Check that we are not stuck in an infinite loop. */
1508 #define CHECK_INFINITE_LOOP(pat_cur, string_place) \
1510 int failure = TOP_FAILURE_HANDLE(); \
1511 /* Check for infinite matching loops */ \
1512 while (failure > 0 && \
1513 (FAILURE_STR (failure) == string_place \
1514 || FAILURE_STR (failure) == NULL)) \
1516 assert (FAILURE_PAT (failure) >= bufp->buffer \
1517 && FAILURE_PAT (failure) <= bufp->buffer + bufp->used); \
1518 if (FAILURE_PAT (failure) == pat_cur) \
1520 DEBUG_PRINT2 (" Other pattern: %p\n", FAILURE_PAT (failure)); \
1521 failure = NEXT_FAILURE_HANDLE(failure); \
1523 DEBUG_PRINT2 (" Other string: %p\n", FAILURE_STR (failure)); \
1526 /* Push the information about the state we will need
1527 if we ever fail back to it.
1529 Requires variables fail_stack, regstart, regend and
1530 num_regs be declared. GROW_FAIL_STACK requires `destination' be
1533 Does `return FAILURE_CODE' if runs out of memory. */
1535 #define PUSH_FAILURE_POINT(pattern, string_place) \
1537 char *destination; \
1538 /* Must be int, so when we don't save any registers, the arithmetic \
1539 of 0 + -1 isn't done as unsigned. */ \
1541 DEBUG_STATEMENT (nfailure_points_pushed++); \
1542 DEBUG_PRINT1 ("\nPUSH_FAILURE_POINT:\n"); \
1543 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail); \
1544 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1546 ENSURE_FAIL_STACK (NUM_NONREG_ITEMS); \
1548 DEBUG_PRINT1 ("\n"); \
1550 DEBUG_PRINT2 (" Push frame index: %d\n", fail_stack.frame); \
1551 PUSH_FAILURE_INT (fail_stack.frame); \
1553 DEBUG_PRINT2 (" Push string %p: `", string_place); \
1554 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, size2);\
1555 DEBUG_PRINT1 ("'\n"); \
1556 PUSH_FAILURE_POINTER (string_place); \
1558 DEBUG_PRINT2 (" Push pattern %p: ", pattern); \
1559 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern, pend); \
1560 PUSH_FAILURE_POINTER (pattern); \
1562 /* Close the frame by moving the frame pointer past it. */ \
1563 fail_stack.frame = fail_stack.avail; \
1566 /* Estimate the size of data pushed by a typical failure stack entry.
1567 An estimate is all we need, because all we use this for
1568 is to choose a limit for how big to make the failure stack. */
1570 #define TYPICAL_FAILURE_SIZE 20
1572 /* How many items can still be added to the stack without overflowing it. */
1573 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1576 /* Pops what PUSH_FAIL_STACK pushes.
1578 We restore into the parameters, all of which should be lvalues:
1579 STR -- the saved data position.
1580 PAT -- the saved pattern position.
1581 REGSTART, REGEND -- arrays of string positions.
1583 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1584 `pend', `string1', `size1', `string2', and `size2'. */
1586 #define POP_FAILURE_POINT(str, pat) \
1588 assert (!FAIL_STACK_EMPTY ()); \
1590 /* Remove failure points and point to how many regs pushed. */ \
1591 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1592 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1593 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1595 /* Pop the saved registers. */ \
1596 while (fail_stack.frame < fail_stack.avail) \
1597 POP_FAILURE_REG_OR_COUNT (); \
1599 pat = POP_FAILURE_POINTER (); \
1600 DEBUG_PRINT2 (" Popping pattern %p: ", pat); \
1601 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1603 /* If the saved string location is NULL, it came from an \
1604 on_failure_keep_string_jump opcode, and we want to throw away the \
1605 saved NULL, thus retaining our current position in the string. */ \
1606 str = POP_FAILURE_POINTER (); \
1607 DEBUG_PRINT2 (" Popping string %p: `", str); \
1608 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1609 DEBUG_PRINT1 ("'\n"); \
1611 fail_stack.frame = POP_FAILURE_INT (); \
1612 DEBUG_PRINT2 (" Popping frame index: %d\n", fail_stack.frame); \
1614 assert (fail_stack.avail >= 0); \
1615 assert (fail_stack.frame <= fail_stack.avail); \
1617 DEBUG_STATEMENT (nfailure_points_popped++); \
1618 } while (0) /* POP_FAILURE_POINT */
1622 /* Registers are set to a sentinel when they haven't yet matched. */
1623 #define REG_UNSET(e) ((e) == NULL)
1625 /* Subroutine declarations and macros for regex_compile. */
1627 static reg_errcode_t regex_compile _RE_ARGS ((re_char *pattern, size_t size,
1628 reg_syntax_t syntax,
1629 struct re_pattern_buffer *bufp));
1630 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1631 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1632 int arg1, int arg2));
1633 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1634 int arg, unsigned char *end));
1635 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1636 int arg1, int arg2, unsigned char *end));
1637 static boolean at_begline_loc_p _RE_ARGS ((re_char *pattern,
1639 reg_syntax_t syntax));
1640 static boolean at_endline_loc_p _RE_ARGS ((re_char *p,
1642 reg_syntax_t syntax));
1643 static re_char *skip_one_char _RE_ARGS ((re_char *p));
1644 static int analyse_first _RE_ARGS ((re_char *p, re_char *pend,
1645 char *fastmap, const int multibyte));
1647 /* Fetch the next character in the uncompiled pattern---translating it
1649 #define PATFETCH(c) \
1652 c = TRANSLATE (c); \
1655 /* Fetch the next character in the uncompiled pattern, with no
1657 #define PATFETCH_RAW(c) \
1660 if (p == pend) return REG_EEND; \
1661 c = RE_STRING_CHAR_AND_LENGTH (p, pend - p, len); \
1666 /* If `translate' is non-null, return translate[D], else just D. We
1667 cast the subscript to translate because some data is declared as
1668 `char *', to avoid warnings when a string constant is passed. But
1669 when we use a character as a subscript we must make it unsigned. */
1671 # define TRANSLATE(d) \
1672 (RE_TRANSLATE_P (translate) ? RE_TRANSLATE (translate, (d)) : (d))
1676 /* Macros for outputting the compiled pattern into `buffer'. */
1678 /* If the buffer isn't allocated when it comes in, use this. */
1679 #define INIT_BUF_SIZE 32
1681 /* Make sure we have at least N more bytes of space in buffer. */
1682 #define GET_BUFFER_SPACE(n) \
1683 while ((size_t) (b - bufp->buffer + (n)) > bufp->allocated) \
1686 /* Make sure we have one more byte of buffer space and then add C to it. */
1687 #define BUF_PUSH(c) \
1689 GET_BUFFER_SPACE (1); \
1690 *b++ = (unsigned char) (c); \
1694 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1695 #define BUF_PUSH_2(c1, c2) \
1697 GET_BUFFER_SPACE (2); \
1698 *b++ = (unsigned char) (c1); \
1699 *b++ = (unsigned char) (c2); \
1703 /* As with BUF_PUSH_2, except for three bytes. */
1704 #define BUF_PUSH_3(c1, c2, c3) \
1706 GET_BUFFER_SPACE (3); \
1707 *b++ = (unsigned char) (c1); \
1708 *b++ = (unsigned char) (c2); \
1709 *b++ = (unsigned char) (c3); \
1713 /* Store a jump with opcode OP at LOC to location TO. We store a
1714 relative address offset by the three bytes the jump itself occupies. */
1715 #define STORE_JUMP(op, loc, to) \
1716 store_op1 (op, loc, (to) - (loc) - 3)
1718 /* Likewise, for a two-argument jump. */
1719 #define STORE_JUMP2(op, loc, to, arg) \
1720 store_op2 (op, loc, (to) - (loc) - 3, arg)
1722 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1723 #define INSERT_JUMP(op, loc, to) \
1724 insert_op1 (op, loc, (to) - (loc) - 3, b)
1726 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1727 #define INSERT_JUMP2(op, loc, to, arg) \
1728 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1731 /* This is not an arbitrary limit: the arguments which represent offsets
1732 into the pattern are two bytes long. So if 2^16 bytes turns out to
1733 be too small, many things would have to change. */
1734 /* Any other compiler which, like MSC, has allocation limit below 2^16
1735 bytes will have to use approach similar to what was done below for
1736 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1737 reallocating to 0 bytes. Such thing is not going to work too well.
1738 You have been warned!! */
1739 #if defined _MSC_VER && !defined WIN32
1740 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes. */
1741 # define MAX_BUF_SIZE 65500L
1743 # define MAX_BUF_SIZE (1L << 16)
1746 /* Extend the buffer by twice its current size via realloc and
1747 reset the pointers that pointed into the old block to point to the
1748 correct places in the new one. If extending the buffer results in it
1749 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1750 #if __BOUNDED_POINTERS__
1751 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
1752 # define MOVE_BUFFER_POINTER(P) \
1753 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
1754 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
1757 SET_HIGH_BOUND (b); \
1758 SET_HIGH_BOUND (begalt); \
1759 if (fixup_alt_jump) \
1760 SET_HIGH_BOUND (fixup_alt_jump); \
1762 SET_HIGH_BOUND (laststart); \
1763 if (pending_exact) \
1764 SET_HIGH_BOUND (pending_exact); \
1767 # define MOVE_BUFFER_POINTER(P) (P) += incr
1768 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
1770 #define EXTEND_BUFFER() \
1772 re_char *old_buffer = bufp->buffer; \
1773 if (bufp->allocated == MAX_BUF_SIZE) \
1775 bufp->allocated <<= 1; \
1776 if (bufp->allocated > MAX_BUF_SIZE) \
1777 bufp->allocated = MAX_BUF_SIZE; \
1778 RETALLOC (bufp->buffer, bufp->allocated, unsigned char); \
1779 if (bufp->buffer == NULL) \
1780 return REG_ESPACE; \
1781 /* If the buffer moved, move all the pointers into it. */ \
1782 if (old_buffer != bufp->buffer) \
1784 int incr = bufp->buffer - old_buffer; \
1785 MOVE_BUFFER_POINTER (b); \
1786 MOVE_BUFFER_POINTER (begalt); \
1787 if (fixup_alt_jump) \
1788 MOVE_BUFFER_POINTER (fixup_alt_jump); \
1790 MOVE_BUFFER_POINTER (laststart); \
1791 if (pending_exact) \
1792 MOVE_BUFFER_POINTER (pending_exact); \
1794 ELSE_EXTEND_BUFFER_HIGH_BOUND \
1798 /* Since we have one byte reserved for the register number argument to
1799 {start,stop}_memory, the maximum number of groups we can report
1800 things about is what fits in that byte. */
1801 #define MAX_REGNUM 255
1803 /* But patterns can have more than `MAX_REGNUM' registers. We just
1804 ignore the excess. */
1805 typedef unsigned regnum_t;
1808 /* Macros for the compile stack. */
1810 /* Since offsets can go either forwards or backwards, this type needs to
1811 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1812 /* int may be not enough when sizeof(int) == 2. */
1813 typedef long pattern_offset_t;
1817 pattern_offset_t begalt_offset;
1818 pattern_offset_t fixup_alt_jump;
1819 pattern_offset_t laststart_offset;
1821 } compile_stack_elt_t;
1826 compile_stack_elt_t *stack;
1828 unsigned avail; /* Offset of next open position. */
1829 } compile_stack_type;
1832 #define INIT_COMPILE_STACK_SIZE 32
1834 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1835 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1837 /* The next available element. */
1838 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1841 /* Structure to manage work area for range table. */
1842 struct range_table_work_area
1844 int *table; /* actual work area. */
1845 int allocated; /* allocated size for work area in bytes. */
1846 int used; /* actually used size in words. */
1847 int bits; /* flag to record character classes */
1850 /* Make sure that WORK_AREA can hold more N multibyte characters. */
1851 #define EXTEND_RANGE_TABLE_WORK_AREA(work_area, n) \
1853 if (((work_area).used + (n)) * sizeof (int) > (work_area).allocated) \
1855 (work_area).allocated += 16 * sizeof (int); \
1856 if ((work_area).table) \
1858 = (int *) realloc ((work_area).table, (work_area).allocated); \
1861 = (int *) malloc ((work_area).allocated); \
1862 if ((work_area).table == 0) \
1863 FREE_STACK_RETURN (REG_ESPACE); \
1867 #define SET_RANGE_TABLE_WORK_AREA_BIT(work_area, bit) \
1868 (work_area).bits |= (bit)
1870 /* Bits used to implement the multibyte-part of the various character classes
1871 such as [:alnum:] in a charset's range table. */
1872 #define BIT_WORD 0x1
1873 #define BIT_LOWER 0x2
1874 #define BIT_PUNCT 0x4
1875 #define BIT_SPACE 0x8
1876 #define BIT_UPPER 0x10
1877 #define BIT_MULTIBYTE 0x20
1879 /* Set a range (RANGE_START, RANGE_END) to WORK_AREA. */
1880 #define SET_RANGE_TABLE_WORK_AREA(work_area, range_start, range_end) \
1882 EXTEND_RANGE_TABLE_WORK_AREA ((work_area), 2); \
1883 (work_area).table[(work_area).used++] = (range_start); \
1884 (work_area).table[(work_area).used++] = (range_end); \
1887 /* Free allocated memory for WORK_AREA. */
1888 #define FREE_RANGE_TABLE_WORK_AREA(work_area) \
1890 if ((work_area).table) \
1891 free ((work_area).table); \
1894 #define CLEAR_RANGE_TABLE_WORK_USED(work_area) ((work_area).used = 0, (work_area).bits = 0)
1895 #define RANGE_TABLE_WORK_USED(work_area) ((work_area).used)
1896 #define RANGE_TABLE_WORK_BITS(work_area) ((work_area).bits)
1897 #define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i])
1900 /* Set the bit for character C in a list. */
1901 #define SET_LIST_BIT(c) (b[((c)) / BYTEWIDTH] |= 1 << ((c) % BYTEWIDTH))
1904 /* Get the next unsigned number in the uncompiled pattern. */
1905 #define GET_UNSIGNED_NUMBER(num) \
1906 do { if (p != pend) \
1909 while ('0' <= c && c <= '9') \
1913 num = num * 10 + c - '0'; \
1921 #if WIDE_CHAR_SUPPORT
1922 /* The GNU C library provides support for user-defined character classes
1923 and the functions from ISO C amendement 1. */
1924 # ifdef CHARCLASS_NAME_MAX
1925 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1927 /* This shouldn't happen but some implementation might still have this
1928 problem. Use a reasonable default value. */
1929 # define CHAR_CLASS_MAX_LENGTH 256
1931 typedef wctype_t re_wctype_t;
1932 typedef wchar_t re_wchar_t;
1933 # define re_wctype wctype
1934 # define re_iswctype iswctype
1935 # define re_wctype_to_bit(cc) 0
1937 # define CHAR_CLASS_MAX_LENGTH 9 /* Namely, `multibyte'. */
1940 /* Character classes. */
1941 typedef enum { RECC_ERROR = 0,
1942 RECC_ALNUM, RECC_ALPHA, RECC_WORD,
1943 RECC_GRAPH, RECC_PRINT,
1944 RECC_LOWER, RECC_UPPER,
1945 RECC_PUNCT, RECC_CNTRL,
1946 RECC_DIGIT, RECC_XDIGIT,
1947 RECC_BLANK, RECC_SPACE,
1948 RECC_MULTIBYTE, RECC_NONASCII,
1949 RECC_ASCII, RECC_UNIBYTE
1952 typedef int re_wchar_t;
1954 /* Map a string to the char class it names (if any). */
1959 if (STREQ (string, "alnum")) return RECC_ALNUM;
1960 else if (STREQ (string, "alpha")) return RECC_ALPHA;
1961 else if (STREQ (string, "word")) return RECC_WORD;
1962 else if (STREQ (string, "ascii")) return RECC_ASCII;
1963 else if (STREQ (string, "nonascii")) return RECC_NONASCII;
1964 else if (STREQ (string, "graph")) return RECC_GRAPH;
1965 else if (STREQ (string, "lower")) return RECC_LOWER;
1966 else if (STREQ (string, "print")) return RECC_PRINT;
1967 else if (STREQ (string, "punct")) return RECC_PUNCT;
1968 else if (STREQ (string, "space")) return RECC_SPACE;
1969 else if (STREQ (string, "upper")) return RECC_UPPER;
1970 else if (STREQ (string, "unibyte")) return RECC_UNIBYTE;
1971 else if (STREQ (string, "multibyte")) return RECC_MULTIBYTE;
1972 else if (STREQ (string, "digit")) return RECC_DIGIT;
1973 else if (STREQ (string, "xdigit")) return RECC_XDIGIT;
1974 else if (STREQ (string, "cntrl")) return RECC_CNTRL;
1975 else if (STREQ (string, "blank")) return RECC_BLANK;
1979 /* True iff CH is in the char class CC. */
1981 re_iswctype (ch, cc)
1985 boolean ret = false;
1989 case RECC_ALNUM: ret = ISALNUM (ch);
1990 case RECC_ALPHA: ret = ISALPHA (ch);
1991 case RECC_BLANK: ret = ISBLANK (ch);
1992 case RECC_CNTRL: ret = ISCNTRL (ch);
1993 case RECC_DIGIT: ret = ISDIGIT (ch);
1994 case RECC_GRAPH: ret = ISGRAPH (ch);
1995 case RECC_LOWER: ret = ISLOWER (ch);
1996 case RECC_PRINT: ret = ISPRINT (ch);
1997 case RECC_PUNCT: ret = ISPUNCT (ch);
1998 case RECC_SPACE: ret = ISSPACE (ch);
1999 case RECC_UPPER: ret = ISUPPER (ch);
2000 case RECC_XDIGIT: ret = ISXDIGIT (ch);
2001 case RECC_ASCII: ret = IS_REAL_ASCII (ch);
2002 case RECC_NONASCII: ret = !IS_REAL_ASCII (ch);
2003 case RECC_UNIBYTE: ret = ISUNIBYTE (ch);
2004 case RECC_MULTIBYTE: ret = !ISUNIBYTE (ch);
2005 case RECC_WORD: ret = ISWORD (ch);
2006 case RECC_ERROR: ret = false;
2011 /* Return a bit-pattern to use in the range-table bits to match multibyte
2012 chars of class CC. */
2014 re_wctype_to_bit (cc)
2021 case RECC_NONASCII: case RECC_PRINT: case RECC_GRAPH:
2022 case RECC_MULTIBYTE: ret = BIT_MULTIBYTE;
2023 case RECC_ALPHA: case RECC_ALNUM: case RECC_WORD: ret = BIT_WORD;
2024 case RECC_LOWER: ret = BIT_LOWER;
2025 case RECC_UPPER: ret = BIT_UPPER;
2026 case RECC_PUNCT: ret = BIT_PUNCT;
2027 case RECC_SPACE: ret = BIT_SPACE;
2028 case RECC_ASCII: case RECC_DIGIT: case RECC_XDIGIT: case RECC_CNTRL:
2029 case RECC_BLANK: case RECC_UNIBYTE: case RECC_ERROR: ret = 0;
2035 /* Explicit quit checking is only used on NTemacs. */
2036 #if defined WINDOWSNT && defined emacs && defined QUIT
2037 extern int immediate_quit;
2038 # define IMMEDIATE_QUIT_CHECK \
2040 if (immediate_quit) QUIT; \
2043 # define IMMEDIATE_QUIT_CHECK ((void)0)
2046 #ifndef MATCH_MAY_ALLOCATE
2048 /* If we cannot allocate large objects within re_match_2_internal,
2049 we make the fail stack and register vectors global.
2050 The fail stack, we grow to the maximum size when a regexp
2052 The register vectors, we adjust in size each time we
2053 compile a regexp, according to the number of registers it needs. */
2055 static fail_stack_type fail_stack;
2057 /* Size with which the following vectors are currently allocated.
2058 That is so we can make them bigger as needed,
2059 but never make them smaller. */
2060 static int regs_allocated_size;
2062 static re_char ** regstart, ** regend;
2063 static re_char **best_regstart, **best_regend;
2065 /* Make the register vectors big enough for NUM_REGS registers,
2066 but don't make them smaller. */
2069 regex_grow_registers (num_regs)
2072 if (num_regs > regs_allocated_size)
2074 RETALLOC_IF (regstart, num_regs, re_char *);
2075 RETALLOC_IF (regend, num_regs, re_char *);
2076 RETALLOC_IF (best_regstart, num_regs, re_char *);
2077 RETALLOC_IF (best_regend, num_regs, re_char *);
2079 regs_allocated_size = num_regs;
2083 #endif /* not MATCH_MAY_ALLOCATE */
2085 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
2089 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2090 Returns one of error codes defined in `regex.h', or zero for success.
2092 Assumes the `allocated' (and perhaps `buffer') and `translate'
2093 fields are set in BUFP on entry.
2095 If it succeeds, results are put in BUFP (if it returns an error, the
2096 contents of BUFP are undefined):
2097 `buffer' is the compiled pattern;
2098 `syntax' is set to SYNTAX;
2099 `used' is set to the length of the compiled pattern;
2100 `fastmap_accurate' is zero;
2101 `re_nsub' is the number of subexpressions in PATTERN;
2102 `not_bol' and `not_eol' are zero;
2104 The `fastmap' field is neither examined nor set. */
2106 /* Insert the `jump' from the end of last alternative to "here".
2107 The space for the jump has already been allocated. */
2108 #define FIXUP_ALT_JUMP() \
2110 if (fixup_alt_jump) \
2111 STORE_JUMP (jump, fixup_alt_jump, b); \
2115 /* Return, freeing storage we allocated. */
2116 #define FREE_STACK_RETURN(value) \
2118 FREE_RANGE_TABLE_WORK_AREA (range_table_work); \
2119 free (compile_stack.stack); \
2123 static reg_errcode_t
2124 regex_compile (pattern, size, syntax, bufp)
2127 reg_syntax_t syntax;
2128 struct re_pattern_buffer *bufp;
2130 /* We fetch characters from PATTERN here. */
2131 register re_wchar_t c, c1;
2133 /* A random temporary spot in PATTERN. */
2136 /* Points to the end of the buffer, where we should append. */
2137 register unsigned char *b;
2139 /* Keeps track of unclosed groups. */
2140 compile_stack_type compile_stack;
2142 /* Points to the current (ending) position in the pattern. */
2144 /* `const' makes AIX compiler fail. */
2145 unsigned char *p = pattern;
2147 re_char *p = pattern;
2149 re_char *pend = pattern + size;
2151 /* How to translate the characters in the pattern. */
2152 RE_TRANSLATE_TYPE translate = bufp->translate;
2154 /* Address of the count-byte of the most recently inserted `exactn'
2155 command. This makes it possible to tell if a new exact-match
2156 character can be added to that command or if the character requires
2157 a new `exactn' command. */
2158 unsigned char *pending_exact = 0;
2160 /* Address of start of the most recently finished expression.
2161 This tells, e.g., postfix * where to find the start of its
2162 operand. Reset at the beginning of groups and alternatives. */
2163 unsigned char *laststart = 0;
2165 /* Address of beginning of regexp, or inside of last group. */
2166 unsigned char *begalt;
2168 /* Place in the uncompiled pattern (i.e., the {) to
2169 which to go back if the interval is invalid. */
2170 re_char *beg_interval;
2172 /* Address of the place where a forward jump should go to the end of
2173 the containing expression. Each alternative of an `or' -- except the
2174 last -- ends with a forward jump of this sort. */
2175 unsigned char *fixup_alt_jump = 0;
2177 /* Counts open-groups as they are encountered. Remembered for the
2178 matching close-group on the compile stack, so the same register
2179 number is put in the stop_memory as the start_memory. */
2180 regnum_t regnum = 0;
2182 /* Work area for range table of charset. */
2183 struct range_table_work_area range_table_work;
2185 /* If the object matched can contain multibyte characters. */
2186 const boolean multibyte = RE_MULTIBYTE_P (bufp);
2190 DEBUG_PRINT1 ("\nCompiling pattern: ");
2193 unsigned debug_count;
2195 for (debug_count = 0; debug_count < size; debug_count++)
2196 putchar (pattern[debug_count]);
2201 /* Initialize the compile stack. */
2202 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2203 if (compile_stack.stack == NULL)
2206 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2207 compile_stack.avail = 0;
2209 range_table_work.table = 0;
2210 range_table_work.allocated = 0;
2212 /* Initialize the pattern buffer. */
2213 bufp->syntax = syntax;
2214 bufp->fastmap_accurate = 0;
2215 bufp->not_bol = bufp->not_eol = 0;
2217 /* Set `used' to zero, so that if we return an error, the pattern
2218 printer (for debugging) will think there's no pattern. We reset it
2222 /* Always count groups, whether or not bufp->no_sub is set. */
2225 #if !defined emacs && !defined SYNTAX_TABLE
2226 /* Initialize the syntax table. */
2227 init_syntax_once ();
2230 if (bufp->allocated == 0)
2233 { /* If zero allocated, but buffer is non-null, try to realloc
2234 enough space. This loses if buffer's address is bogus, but
2235 that is the user's responsibility. */
2236 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
2239 { /* Caller did not allocate a buffer. Do it for them. */
2240 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
2242 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
2244 bufp->allocated = INIT_BUF_SIZE;
2247 begalt = b = bufp->buffer;
2249 /* Loop through the uncompiled pattern until we're at the end. */
2258 if ( /* If at start of pattern, it's an operator. */
2260 /* If context independent, it's an operator. */
2261 || syntax & RE_CONTEXT_INDEP_ANCHORS
2262 /* Otherwise, depends on what's come before. */
2263 || at_begline_loc_p (pattern, p, syntax))
2264 BUF_PUSH ((syntax & RE_NO_NEWLINE_ANCHOR) ? begbuf : begline);
2273 if ( /* If at end of pattern, it's an operator. */
2275 /* If context independent, it's an operator. */
2276 || syntax & RE_CONTEXT_INDEP_ANCHORS
2277 /* Otherwise, depends on what's next. */
2278 || at_endline_loc_p (p, pend, syntax))
2279 BUF_PUSH ((syntax & RE_NO_NEWLINE_ANCHOR) ? endbuf : endline);
2288 if ((syntax & RE_BK_PLUS_QM)
2289 || (syntax & RE_LIMITED_OPS))
2293 /* If there is no previous pattern... */
2296 if (syntax & RE_CONTEXT_INVALID_OPS)
2297 FREE_STACK_RETURN (REG_BADRPT);
2298 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2303 /* 1 means zero (many) matches is allowed. */
2304 boolean zero_times_ok = 0, many_times_ok = 0;
2307 /* If there is a sequence of repetition chars, collapse it
2308 down to just one (the right one). We can't combine
2309 interval operators with these because of, e.g., `a{2}*',
2310 which should only match an even number of `a's. */
2314 if ((syntax & RE_FRUGAL)
2315 && c == '?' && (zero_times_ok || many_times_ok))
2319 zero_times_ok |= c != '+';
2320 many_times_ok |= c != '?';
2326 || (!(syntax & RE_BK_PLUS_QM)
2327 && (*p == '+' || *p == '?')))
2329 else if (syntax & RE_BK_PLUS_QM && *p == '\\')
2332 FREE_STACK_RETURN (REG_EESCAPE);
2333 if (p[1] == '+' || p[1] == '?')
2334 PATFETCH (c); /* Gobble up the backslash. */
2340 /* If we get here, we found another repeat character. */
2344 /* Star, etc. applied to an empty pattern is equivalent
2345 to an empty pattern. */
2346 if (!laststart || laststart == b)
2349 /* Now we know whether or not zero matches is allowed
2350 and also whether or not two or more matches is allowed. */
2355 boolean simple = skip_one_char (laststart) == b;
2356 unsigned int startoffset = 0;
2358 /* Check if the loop can match the empty string. */
2359 (simple || !analyse_first (laststart, b, NULL, 0)) ?
2360 on_failure_jump : on_failure_jump_loop;
2361 assert (skip_one_char (laststart) <= b);
2363 if (!zero_times_ok && simple)
2364 { /* Since simple * loops can be made faster by using
2365 on_failure_keep_string_jump, we turn simple P+
2366 into PP* if P is simple. */
2367 unsigned char *p1, *p2;
2368 startoffset = b - laststart;
2369 GET_BUFFER_SPACE (startoffset);
2370 p1 = b; p2 = laststart;
2376 GET_BUFFER_SPACE (6);
2379 STORE_JUMP (ofj, b, b + 6);
2381 /* Simple * loops can use on_failure_keep_string_jump
2382 depending on what follows. But since we don't know
2383 that yet, we leave the decision up to
2384 on_failure_jump_smart. */
2385 INSERT_JUMP (simple ? on_failure_jump_smart : ofj,
2386 laststart + startoffset, b + 6);
2388 STORE_JUMP (jump, b, laststart + startoffset);
2393 /* A simple ? pattern. */
2394 assert (zero_times_ok);
2395 GET_BUFFER_SPACE (3);
2396 INSERT_JUMP (on_failure_jump, laststart, b + 3);
2400 else /* not greedy */
2401 { /* I wish the greedy and non-greedy cases could be merged. */
2403 GET_BUFFER_SPACE (7); /* We might use less. */
2406 boolean emptyp = analyse_first (laststart, b, NULL, 0);
2408 /* The non-greedy multiple match looks like a repeat..until:
2409 we only need a conditional jump at the end of the loop */
2410 if (emptyp) BUF_PUSH (no_op);
2411 STORE_JUMP (emptyp ? on_failure_jump_nastyloop
2412 : on_failure_jump, b, laststart);
2416 /* The repeat...until naturally matches one or more.
2417 To also match zero times, we need to first jump to
2418 the end of the loop (its conditional jump). */
2419 INSERT_JUMP (jump, laststart, b);
2425 /* non-greedy a?? */
2426 INSERT_JUMP (jump, laststart, b + 3);
2428 INSERT_JUMP (on_failure_jump, laststart, laststart + 6);
2445 CLEAR_RANGE_TABLE_WORK_USED (range_table_work);
2447 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2449 /* Ensure that we have enough space to push a charset: the
2450 opcode, the length count, and the bitset; 34 bytes in all. */
2451 GET_BUFFER_SPACE (34);
2455 /* We test `*p == '^' twice, instead of using an if
2456 statement, so we only need one BUF_PUSH. */
2457 BUF_PUSH (*p == '^' ? charset_not : charset);
2461 /* Remember the first position in the bracket expression. */
2464 /* Push the number of bytes in the bitmap. */
2465 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2467 /* Clear the whole map. */
2468 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2470 /* charset_not matches newline according to a syntax bit. */
2471 if ((re_opcode_t) b[-2] == charset_not
2472 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2473 SET_LIST_BIT ('\n');
2475 /* Read in characters and ranges, setting map bits. */
2478 boolean escaped_char = false;
2479 const unsigned char *p2 = p;
2481 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2485 /* \ might escape characters inside [...] and [^...]. */
2486 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2488 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2491 escaped_char = true;
2495 /* Could be the end of the bracket expression. If it's
2496 not (i.e., when the bracket expression is `[]' so
2497 far), the ']' character bit gets set way below. */
2498 if (c == ']' && p2 != p1)
2502 /* What should we do for the character which is
2503 greater than 0x7F, but not BASE_LEADING_CODE_P?
2506 /* See if we're at the beginning of a possible character
2509 if (!escaped_char &&
2510 syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2512 /* Leave room for the null. */
2513 unsigned char str[CHAR_CLASS_MAX_LENGTH + 1];
2514 const unsigned char *class_beg;
2520 /* If pattern is `[[:'. */
2521 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2526 if ((c == ':' && *p == ']') || p == pend)
2528 if (c1 < CHAR_CLASS_MAX_LENGTH)
2531 /* This is in any case an invalid class name. */
2536 /* If isn't a word bracketed by `[:' and `:]':
2537 undo the ending character, the letters, and
2538 leave the leading `:' and `[' (but set bits for
2540 if (c == ':' && *p == ']')
2545 cc = re_wctype (str);
2548 FREE_STACK_RETURN (REG_ECTYPE);
2550 /* Throw away the ] at the end of the character
2554 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2556 /* Most character classes in a multibyte match
2557 just set a flag. Exceptions are is_blank,
2558 is_digit, is_cntrl, and is_xdigit, since
2559 they can only match ASCII characters. We
2560 don't need to handle them for multibyte.
2561 They are distinguished by a negative wctype. */
2564 SET_RANGE_TABLE_WORK_AREA_BIT (range_table_work,
2565 re_wctype_to_bit (cc));
2567 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2569 int translated = TRANSLATE (ch);
2570 if (re_iswctype (btowc (ch), cc))
2571 SET_LIST_BIT (translated);
2574 /* Repeat the loop. */
2579 /* Go back to right after the "[:". */
2583 /* Because the `:' may starts the range, we
2584 can't simply set bit and repeat the loop.
2585 Instead, just set it to C and handle below. */
2590 if (p < pend && p[0] == '-' && p[1] != ']')
2593 /* Discard the `-'. */
2596 /* Fetch the character which ends the range. */
2599 if (SINGLE_BYTE_CHAR_P (c))
2601 if (! SINGLE_BYTE_CHAR_P (c1))
2603 /* Handle a range starting with a
2604 character of less than 256, and ending
2605 with a character of not less than 256.
2606 Split that into two ranges, the low one
2607 ending at 0377, and the high one
2608 starting at the smallest character in
2609 the charset of C1 and ending at C1. */
2610 int charset = CHAR_CHARSET (c1);
2611 int c2 = MAKE_CHAR (charset, 0, 0);
2613 SET_RANGE_TABLE_WORK_AREA (range_table_work,
2618 else if (!SAME_CHARSET_P (c, c1))
2619 FREE_STACK_RETURN (REG_ERANGE);
2622 /* Range from C to C. */
2625 /* Set the range ... */
2626 if (SINGLE_BYTE_CHAR_P (c))
2627 /* ... into bitmap. */
2629 re_wchar_t this_char;
2630 int range_start = c, range_end = c1;
2632 /* If the start is after the end, the range is empty. */
2633 if (range_start > range_end)
2635 if (syntax & RE_NO_EMPTY_RANGES)
2636 FREE_STACK_RETURN (REG_ERANGE);
2637 /* Else, repeat the loop. */
2641 for (this_char = range_start; this_char <= range_end;
2643 SET_LIST_BIT (TRANSLATE (this_char));
2647 /* ... into range table. */
2648 SET_RANGE_TABLE_WORK_AREA (range_table_work, c, c1);
2651 /* Discard any (non)matching list bytes that are all 0 at the
2652 end of the map. Decrease the map-length byte too. */
2653 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2657 /* Build real range table from work area. */
2658 if (RANGE_TABLE_WORK_USED (range_table_work)
2659 || RANGE_TABLE_WORK_BITS (range_table_work))
2662 int used = RANGE_TABLE_WORK_USED (range_table_work);
2664 /* Allocate space for COUNT + RANGE_TABLE. Needs two
2665 bytes for flags, two for COUNT, and three bytes for
2667 GET_BUFFER_SPACE (4 + used * 3);
2669 /* Indicate the existence of range table. */
2670 laststart[1] |= 0x80;
2672 /* Store the character class flag bits into the range table.
2673 If not in emacs, these flag bits are always 0. */
2674 *b++ = RANGE_TABLE_WORK_BITS (range_table_work) & 0xff;
2675 *b++ = RANGE_TABLE_WORK_BITS (range_table_work) >> 8;
2677 STORE_NUMBER_AND_INCR (b, used / 2);
2678 for (i = 0; i < used; i++)
2679 STORE_CHARACTER_AND_INCR
2680 (b, RANGE_TABLE_WORK_ELT (range_table_work, i));
2687 if (syntax & RE_NO_BK_PARENS)
2694 if (syntax & RE_NO_BK_PARENS)
2701 if (syntax & RE_NEWLINE_ALT)
2708 if (syntax & RE_NO_BK_VBAR)
2715 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2716 goto handle_interval;
2722 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2724 /* Do not translate the character after the \, so that we can
2725 distinguish, e.g., \B from \b, even if we normally would
2726 translate, e.g., B to b. */
2732 if (syntax & RE_NO_BK_PARENS)
2733 goto normal_backslash;
2740 /* Look for a special (?...) construct */
2741 if ((syntax & RE_SHY_GROUPS) && *p == '?')
2743 PATFETCH (c); /* Gobble up the '?'. */
2747 case ':': shy = 1; break;
2749 /* Only (?:...) is supported right now. */
2750 FREE_STACK_RETURN (REG_BADPAT);
2761 if (COMPILE_STACK_FULL)
2763 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2764 compile_stack_elt_t);
2765 if (compile_stack.stack == NULL) return REG_ESPACE;
2767 compile_stack.size <<= 1;
2770 /* These are the values to restore when we hit end of this
2771 group. They are all relative offsets, so that if the
2772 whole pattern moves because of realloc, they will still
2774 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2775 COMPILE_STACK_TOP.fixup_alt_jump
2776 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2777 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2778 COMPILE_STACK_TOP.regnum = shy ? -regnum : regnum;
2781 start_memory for groups beyond the last one we can
2782 represent in the compiled pattern. */
2783 if (regnum <= MAX_REGNUM && !shy)
2784 BUF_PUSH_2 (start_memory, regnum);
2786 compile_stack.avail++;
2791 /* If we've reached MAX_REGNUM groups, then this open
2792 won't actually generate any code, so we'll have to
2793 clear pending_exact explicitly. */
2799 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2801 if (COMPILE_STACK_EMPTY)
2803 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2804 goto normal_backslash;
2806 FREE_STACK_RETURN (REG_ERPAREN);
2812 /* See similar code for backslashed left paren above. */
2813 if (COMPILE_STACK_EMPTY)
2815 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2818 FREE_STACK_RETURN (REG_ERPAREN);
2821 /* Since we just checked for an empty stack above, this
2822 ``can't happen''. */
2823 assert (compile_stack.avail != 0);
2825 /* We don't just want to restore into `regnum', because
2826 later groups should continue to be numbered higher,
2827 as in `(ab)c(de)' -- the second group is #2. */
2828 regnum_t this_group_regnum;
2830 compile_stack.avail--;
2831 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2833 = COMPILE_STACK_TOP.fixup_alt_jump
2834 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2836 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2837 this_group_regnum = COMPILE_STACK_TOP.regnum;
2838 /* If we've reached MAX_REGNUM groups, then this open
2839 won't actually generate any code, so we'll have to
2840 clear pending_exact explicitly. */
2843 /* We're at the end of the group, so now we know how many
2844 groups were inside this one. */
2845 if (this_group_regnum <= MAX_REGNUM && this_group_regnum > 0)
2846 BUF_PUSH_2 (stop_memory, this_group_regnum);
2851 case '|': /* `\|'. */
2852 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2853 goto normal_backslash;
2855 if (syntax & RE_LIMITED_OPS)
2858 /* Insert before the previous alternative a jump which
2859 jumps to this alternative if the former fails. */
2860 GET_BUFFER_SPACE (3);
2861 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2865 /* The alternative before this one has a jump after it
2866 which gets executed if it gets matched. Adjust that
2867 jump so it will jump to this alternative's analogous
2868 jump (put in below, which in turn will jump to the next
2869 (if any) alternative's such jump, etc.). The last such
2870 jump jumps to the correct final destination. A picture:
2876 If we are at `b', then fixup_alt_jump right now points to a
2877 three-byte space after `a'. We'll put in the jump, set
2878 fixup_alt_jump to right after `b', and leave behind three
2879 bytes which we'll fill in when we get to after `c'. */
2883 /* Mark and leave space for a jump after this alternative,
2884 to be filled in later either by next alternative or
2885 when know we're at the end of a series of alternatives. */
2887 GET_BUFFER_SPACE (3);
2896 /* If \{ is a literal. */
2897 if (!(syntax & RE_INTERVALS)
2898 /* If we're at `\{' and it's not the open-interval
2900 || (syntax & RE_NO_BK_BRACES))
2901 goto normal_backslash;
2905 /* If got here, then the syntax allows intervals. */
2907 /* At least (most) this many matches must be made. */
2908 int lower_bound = 0, upper_bound = -1;
2913 FREE_STACK_RETURN (REG_EBRACE);
2915 GET_UNSIGNED_NUMBER (lower_bound);
2918 GET_UNSIGNED_NUMBER (upper_bound);
2920 /* Interval such as `{1}' => match exactly once. */
2921 upper_bound = lower_bound;
2923 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2924 || (upper_bound >= 0 && lower_bound > upper_bound))
2925 FREE_STACK_RETURN (REG_BADBR);
2927 if (!(syntax & RE_NO_BK_BRACES))
2930 FREE_STACK_RETURN (REG_BADBR);
2936 FREE_STACK_RETURN (REG_BADBR);
2938 /* We just parsed a valid interval. */
2940 /* If it's invalid to have no preceding re. */
2943 if (syntax & RE_CONTEXT_INVALID_OPS)
2944 FREE_STACK_RETURN (REG_BADRPT);
2945 else if (syntax & RE_CONTEXT_INDEP_OPS)
2948 goto unfetch_interval;
2951 if (upper_bound == 0)
2952 /* If the upper bound is zero, just drop the sub pattern
2955 else if (lower_bound == 1 && upper_bound == 1)
2956 /* Just match it once: nothing to do here. */
2959 /* Otherwise, we have a nontrivial interval. When
2960 we're all done, the pattern will look like:
2961 set_number_at <jump count> <upper bound>
2962 set_number_at <succeed_n count> <lower bound>
2963 succeed_n <after jump addr> <succeed_n count>
2965 jump_n <succeed_n addr> <jump count>
2966 (The upper bound and `jump_n' are omitted if
2967 `upper_bound' is 1, though.) */
2969 { /* If the upper bound is > 1, we need to insert
2970 more at the end of the loop. */
2971 unsigned int nbytes = (upper_bound < 0 ? 3
2972 : upper_bound > 1 ? 5 : 0);
2973 unsigned int startoffset = 0;
2975 GET_BUFFER_SPACE (20); /* We might use less. */
2977 if (lower_bound == 0)
2979 /* A succeed_n that starts with 0 is really a
2980 a simple on_failure_jump_loop. */
2981 INSERT_JUMP (on_failure_jump_loop, laststart,
2987 /* Initialize lower bound of the `succeed_n', even
2988 though it will be set during matching by its
2989 attendant `set_number_at' (inserted next),
2990 because `re_compile_fastmap' needs to know.
2991 Jump to the `jump_n' we might insert below. */
2992 INSERT_JUMP2 (succeed_n, laststart,
2997 /* Code to initialize the lower bound. Insert
2998 before the `succeed_n'. The `5' is the last two
2999 bytes of this `set_number_at', plus 3 bytes of
3000 the following `succeed_n'. */
3001 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
3006 if (upper_bound < 0)
3008 /* A negative upper bound stands for infinity,
3009 in which case it degenerates to a plain jump. */
3010 STORE_JUMP (jump, b, laststart + startoffset);
3013 else if (upper_bound > 1)
3014 { /* More than one repetition is allowed, so
3015 append a backward jump to the `succeed_n'
3016 that starts this interval.
3018 When we've reached this during matching,
3019 we'll have matched the interval once, so
3020 jump back only `upper_bound - 1' times. */
3021 STORE_JUMP2 (jump_n, b, laststart + startoffset,
3025 /* The location we want to set is the second
3026 parameter of the `jump_n'; that is `b-2' as
3027 an absolute address. `laststart' will be
3028 the `set_number_at' we're about to insert;
3029 `laststart+3' the number to set, the source
3030 for the relative address. But we are
3031 inserting into the middle of the pattern --
3032 so everything is getting moved up by 5.
3033 Conclusion: (b - 2) - (laststart + 3) + 5,
3034 i.e., b - laststart.
3036 We insert this at the beginning of the loop
3037 so that if we fail during matching, we'll
3038 reinitialize the bounds. */
3039 insert_op2 (set_number_at, laststart, b - laststart,
3040 upper_bound - 1, b);
3045 beg_interval = NULL;
3050 /* If an invalid interval, match the characters as literals. */
3051 assert (beg_interval);
3053 beg_interval = NULL;
3055 /* normal_char and normal_backslash need `c'. */
3058 if (!(syntax & RE_NO_BK_BRACES))
3060 assert (p > pattern && p[-1] == '\\');
3061 goto normal_backslash;
3067 /* There is no way to specify the before_dot and after_dot
3068 operators. rms says this is ok. --karl */
3076 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
3082 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
3088 BUF_PUSH_2 (categoryspec, c);
3094 BUF_PUSH_2 (notcategoryspec, c);
3100 if (syntax & RE_NO_GNU_OPS)
3103 BUF_PUSH_2 (syntaxspec, Sword);
3108 if (syntax & RE_NO_GNU_OPS)
3111 BUF_PUSH_2 (notsyntaxspec, Sword);
3116 if (syntax & RE_NO_GNU_OPS)
3122 if (syntax & RE_NO_GNU_OPS)
3128 if (syntax & RE_NO_GNU_OPS)
3130 BUF_PUSH (wordbound);
3134 if (syntax & RE_NO_GNU_OPS)
3136 BUF_PUSH (notwordbound);
3140 if (syntax & RE_NO_GNU_OPS)
3146 if (syntax & RE_NO_GNU_OPS)
3151 case '1': case '2': case '3': case '4': case '5':
3152 case '6': case '7': case '8': case '9':
3153 if (syntax & RE_NO_BK_REFS)
3159 FREE_STACK_RETURN (REG_ESUBREG);
3161 /* Can't back reference to a subexpression if inside of it. */
3162 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
3166 BUF_PUSH_2 (duplicate, c1);
3172 if (syntax & RE_BK_PLUS_QM)
3175 goto normal_backslash;
3179 /* You might think it would be useful for \ to mean
3180 not to translate; but if we don't translate it
3181 it will never match anything. */
3189 /* Expects the character in `c'. */
3191 /* If no exactn currently being built. */
3194 /* If last exactn not at current position. */
3195 || pending_exact + *pending_exact + 1 != b
3197 /* We have only one byte following the exactn for the count. */
3198 || *pending_exact >= (1 << BYTEWIDTH) - MAX_MULTIBYTE_LENGTH
3200 /* If followed by a repetition operator. */
3201 || (p != pend && (*p == '*' || *p == '^'))
3202 || ((syntax & RE_BK_PLUS_QM)
3203 ? p + 1 < pend && *p == '\\' && (p[1] == '+' || p[1] == '?')
3204 : p != pend && (*p == '+' || *p == '?'))
3205 || ((syntax & RE_INTERVALS)
3206 && ((syntax & RE_NO_BK_BRACES)
3207 ? p != pend && *p == '{'
3208 : p + 1 < pend && p[0] == '\\' && p[1] == '{')))
3210 /* Start building a new exactn. */
3214 BUF_PUSH_2 (exactn, 0);
3215 pending_exact = b - 1;
3218 GET_BUFFER_SPACE (MAX_MULTIBYTE_LENGTH);
3223 len = CHAR_STRING (c, b);
3227 (*pending_exact) += len;
3232 } /* while p != pend */
3235 /* Through the pattern now. */
3239 if (!COMPILE_STACK_EMPTY)
3240 FREE_STACK_RETURN (REG_EPAREN);
3242 /* If we don't want backtracking, force success
3243 the first time we reach the end of the compiled pattern. */
3244 if (syntax & RE_NO_POSIX_BACKTRACKING)
3247 free (compile_stack.stack);
3249 /* We have succeeded; set the length of the buffer. */
3250 bufp->used = b - bufp->buffer;
3255 re_compile_fastmap (bufp);
3256 DEBUG_PRINT1 ("\nCompiled pattern: \n");
3257 print_compiled_pattern (bufp);
3262 #ifndef MATCH_MAY_ALLOCATE
3263 /* Initialize the failure stack to the largest possible stack. This
3264 isn't necessary unless we're trying to avoid calling alloca in
3265 the search and match routines. */
3267 int num_regs = bufp->re_nsub + 1;
3269 if (fail_stack.size < re_max_failures * TYPICAL_FAILURE_SIZE)
3271 fail_stack.size = re_max_failures * TYPICAL_FAILURE_SIZE;
3273 if (! fail_stack.stack)
3275 = (fail_stack_elt_t *) malloc (fail_stack.size
3276 * sizeof (fail_stack_elt_t));
3279 = (fail_stack_elt_t *) realloc (fail_stack.stack,
3281 * sizeof (fail_stack_elt_t)));
3284 regex_grow_registers (num_regs);
3286 #endif /* not MATCH_MAY_ALLOCATE */
3289 } /* regex_compile */
3291 /* Subroutines for `regex_compile'. */
3293 /* Store OP at LOC followed by two-byte integer parameter ARG. */
3296 store_op1 (op, loc, arg)
3301 *loc = (unsigned char) op;
3302 STORE_NUMBER (loc + 1, arg);
3306 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
3309 store_op2 (op, loc, arg1, arg2)
3314 *loc = (unsigned char) op;
3315 STORE_NUMBER (loc + 1, arg1);
3316 STORE_NUMBER (loc + 3, arg2);
3320 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
3321 for OP followed by two-byte integer parameter ARG. */
3324 insert_op1 (op, loc, arg, end)
3330 register unsigned char *pfrom = end;
3331 register unsigned char *pto = end + 3;
3333 while (pfrom != loc)
3336 store_op1 (op, loc, arg);
3340 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
3343 insert_op2 (op, loc, arg1, arg2, end)
3349 register unsigned char *pfrom = end;
3350 register unsigned char *pto = end + 5;
3352 while (pfrom != loc)
3355 store_op2 (op, loc, arg1, arg2);
3359 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
3360 after an alternative or a begin-subexpression. We assume there is at
3361 least one character before the ^. */
3364 at_begline_loc_p (pattern, p, syntax)
3365 re_char *pattern, *p;
3366 reg_syntax_t syntax;
3368 re_char *prev = p - 2;
3369 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
3372 /* After a subexpression? */
3373 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
3374 /* After an alternative? */
3375 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash))
3376 /* After a shy subexpression? */
3377 || ((syntax & RE_SHY_GROUPS) && prev - 2 >= pattern
3378 && prev[-1] == '?' && prev[-2] == '('
3379 && (syntax & RE_NO_BK_PARENS
3380 || (prev - 3 >= pattern && prev[-3] == '\\')));
3384 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3385 at least one character after the $, i.e., `P < PEND'. */
3388 at_endline_loc_p (p, pend, syntax)
3390 reg_syntax_t syntax;
3393 boolean next_backslash = *next == '\\';
3394 re_char *next_next = p + 1 < pend ? p + 1 : 0;
3397 /* Before a subexpression? */
3398 (syntax & RE_NO_BK_PARENS ? *next == ')'
3399 : next_backslash && next_next && *next_next == ')')
3400 /* Before an alternative? */
3401 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3402 : next_backslash && next_next && *next_next == '|');
3406 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3407 false if it's not. */
3410 group_in_compile_stack (compile_stack, regnum)
3411 compile_stack_type compile_stack;
3416 for (this_element = compile_stack.avail - 1;
3419 if (compile_stack.stack[this_element].regnum == regnum)
3426 If fastmap is non-NULL, go through the pattern and fill fastmap
3427 with all the possible leading chars. If fastmap is NULL, don't
3428 bother filling it up (obviously) and only return whether the
3429 pattern could potentially match the empty string.
3431 Return 1 if p..pend might match the empty string.
3432 Return 0 if p..pend matches at least one char.
3433 Return -1 if fastmap was not updated accurately. */
3436 analyse_first (p, pend, fastmap, multibyte)
3439 const int multibyte;
3444 /* If all elements for base leading-codes in fastmap is set, this
3445 flag is set true. */
3446 boolean match_any_multibyte_characters = false;
3450 /* The loop below works as follows:
3451 - It has a working-list kept in the PATTERN_STACK and which basically
3452 starts by only containing a pointer to the first operation.
3453 - If the opcode we're looking at is a match against some set of
3454 chars, then we add those chars to the fastmap and go on to the
3455 next work element from the worklist (done via `break').
3456 - If the opcode is a control operator on the other hand, we either
3457 ignore it (if it's meaningless at this point, such as `start_memory')
3458 or execute it (if it's a jump). If the jump has several destinations
3459 (i.e. `on_failure_jump'), then we push the other destination onto the
3461 We guarantee termination by ignoring backward jumps (more or less),
3462 so that `p' is monotonically increasing. More to the point, we
3463 never set `p' (or push) anything `<= p1'. */
3467 /* `p1' is used as a marker of how far back a `on_failure_jump'
3468 can go without being ignored. It is normally equal to `p'
3469 (which prevents any backward `on_failure_jump') except right
3470 after a plain `jump', to allow patterns such as:
3473 10: on_failure_jump 3
3474 as used for the *? operator. */
3477 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3484 /* If the first character has to match a backreference, that means
3485 that the group was empty (since it already matched). Since this
3486 is the only case that interests us here, we can assume that the
3487 backreference must match the empty string. */
3492 /* Following are the cases which match a character. These end
3498 int c = RE_STRING_CHAR (p + 1, pend - p);
3500 if (SINGLE_BYTE_CHAR_P (c))
3509 /* We could put all the chars except for \n (and maybe \0)
3510 but we don't bother since it is generally not worth it. */
3511 if (!fastmap) break;
3516 /* Chars beyond end of bitmap are possible matches.
3517 All the single-byte codes can occur in multibyte buffers.
3518 So any that are not listed in the charset
3519 are possible matches, even in multibyte buffers. */
3520 if (!fastmap) break;
3521 for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH;
3522 j < (1 << BYTEWIDTH); j++)
3526 if (!fastmap) break;
3527 not = (re_opcode_t) *(p - 1) == charset_not;
3528 for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH - 1, p++;
3530 if (!!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) ^ not)
3533 if ((not && multibyte)
3534 /* Any character set can possibly contain a character
3535 which doesn't match the specified set of characters. */
3536 || (CHARSET_RANGE_TABLE_EXISTS_P (&p[-2])
3537 && CHARSET_RANGE_TABLE_BITS (&p[-2]) != 0))
3538 /* If we can match a character class, we can match
3539 any character set. */
3541 set_fastmap_for_multibyte_characters:
3542 if (match_any_multibyte_characters == false)
3544 for (j = 0x80; j < 0xA0; j++) /* XXX */
3545 if (BASE_LEADING_CODE_P (j))
3547 match_any_multibyte_characters = true;
3551 else if (!not && CHARSET_RANGE_TABLE_EXISTS_P (&p[-2])
3552 && match_any_multibyte_characters == false)
3554 /* Set fastmap[I] 1 where I is a base leading code of each
3555 multibyte character in the range table. */
3558 /* Make P points the range table. `+ 2' is to skip flag
3559 bits for a character class. */
3560 p += CHARSET_BITMAP_SIZE (&p[-2]) + 2;
3562 /* Extract the number of ranges in range table into COUNT. */
3563 EXTRACT_NUMBER_AND_INCR (count, p);
3564 for (; count > 0; count--, p += 2 * 3) /* XXX */
3566 /* Extract the start of each range. */
3567 EXTRACT_CHARACTER (c, p);
3568 j = CHAR_CHARSET (c);
3569 fastmap[CHARSET_LEADING_CODE_BASE (j)] = 1;
3576 if (!fastmap) break;
3578 not = (re_opcode_t)p[-1] == notsyntaxspec;
3580 for (j = 0; j < (1 << BYTEWIDTH); j++)
3581 if ((SYNTAX (j) == (enum syntaxcode) k) ^ not)
3585 /* This match depends on text properties. These end with
3586 aborting optimizations. */
3590 case notcategoryspec:
3591 if (!fastmap) break;
3592 not = (re_opcode_t)p[-1] == notcategoryspec;
3594 for (j = 0; j < (1 << BYTEWIDTH); j++)
3595 if ((CHAR_HAS_CATEGORY (j, k)) ^ not)
3599 /* Any character set can possibly contain a character
3600 whose category is K (or not). */
3601 goto set_fastmap_for_multibyte_characters;
3604 /* All cases after this match the empty string. These end with
3624 EXTRACT_NUMBER_AND_INCR (j, p);
3626 /* Backward jumps can only go back to code that we've already
3627 visited. `re_compile' should make sure this is true. */
3630 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p))
3632 case on_failure_jump:
3633 case on_failure_keep_string_jump:
3634 case on_failure_jump_loop:
3635 case on_failure_jump_nastyloop:
3636 case on_failure_jump_smart:
3642 /* Keep `p1' to allow the `on_failure_jump' we are jumping to
3643 to jump back to "just after here". */
3646 case on_failure_jump:
3647 case on_failure_keep_string_jump:
3648 case on_failure_jump_nastyloop:
3649 case on_failure_jump_loop:
3650 case on_failure_jump_smart:
3651 EXTRACT_NUMBER_AND_INCR (j, p);
3653 ; /* Backward jump to be ignored. */
3655 { /* We have to look down both arms.
3656 We first go down the "straight" path so as to minimize
3657 stack usage when going through alternatives. */
3658 int r = analyse_first (p, pend, fastmap, multibyte);
3666 /* This code simply does not properly handle forward jump_n. */
3667 DEBUG_STATEMENT (EXTRACT_NUMBER (j, p); assert (j < 0));
3669 /* jump_n can either jump or fall through. The (backward) jump
3670 case has already been handled, so we only need to look at the
3671 fallthrough case. */
3675 /* If N == 0, it should be an on_failure_jump_loop instead. */
3676 DEBUG_STATEMENT (EXTRACT_NUMBER (j, p + 2); assert (j > 0));
3678 /* We only care about one iteration of the loop, so we don't
3679 need to consider the case where this behaves like an
3696 abort (); /* We have listed all the cases. */
3699 /* Getting here means we have found the possible starting
3700 characters for one path of the pattern -- and that the empty
3701 string does not match. We need not follow this path further. */
3705 /* We reached the end without matching anything. */
3708 } /* analyse_first */
3710 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3711 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3712 characters can start a string that matches the pattern. This fastmap
3713 is used by re_search to skip quickly over impossible starting points.
3715 Character codes above (1 << BYTEWIDTH) are not represented in the
3716 fastmap, but the leading codes are represented. Thus, the fastmap
3717 indicates which character sets could start a match.
3719 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3720 area as BUFP->fastmap.
3722 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3725 Returns 0 if we succeed, -2 if an internal error. */
3728 re_compile_fastmap (bufp)
3729 struct re_pattern_buffer *bufp;
3731 char *fastmap = bufp->fastmap;
3734 assert (fastmap && bufp->buffer);
3736 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3737 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3739 analysis = analyse_first (bufp->buffer, bufp->buffer + bufp->used,
3740 fastmap, RE_MULTIBYTE_P (bufp));
3741 bufp->can_be_null = (analysis != 0);
3743 } /* re_compile_fastmap */
3745 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3746 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3747 this memory for recording register information. STARTS and ENDS
3748 must be allocated using the malloc library routine, and must each
3749 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3751 If NUM_REGS == 0, then subsequent matches should allocate their own
3754 Unless this function is called, the first search or match using
3755 PATTERN_BUFFER will allocate its own register data, without
3756 freeing the old data. */
3759 re_set_registers (bufp, regs, num_regs, starts, ends)
3760 struct re_pattern_buffer *bufp;
3761 struct re_registers *regs;
3763 regoff_t *starts, *ends;
3767 bufp->regs_allocated = REGS_REALLOCATE;
3768 regs->num_regs = num_regs;
3769 regs->start = starts;
3774 bufp->regs_allocated = REGS_UNALLOCATED;
3776 regs->start = regs->end = (regoff_t *) 0;
3779 WEAK_ALIAS (__re_set_registers, re_set_registers)
3781 /* Searching routines. */
3783 /* Like re_search_2, below, but only one string is specified, and
3784 doesn't let you say where to stop matching. */
3787 re_search (bufp, string, size, startpos, range, regs)
3788 struct re_pattern_buffer *bufp;
3790 int size, startpos, range;
3791 struct re_registers *regs;
3793 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3796 WEAK_ALIAS (__re_search, re_search)
3798 /* End address of virtual concatenation of string. */
3799 #define STOP_ADDR_VSTRING(P) \
3800 (((P) >= size1 ? string2 + size2 : string1 + size1))
3802 /* Address of POS in the concatenation of virtual string. */
3803 #define POS_ADDR_VSTRING(POS) \
3804 (((POS) >= size1 ? string2 - size1 : string1) + (POS))
3806 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3807 virtual concatenation of STRING1 and STRING2, starting first at index
3808 STARTPOS, then at STARTPOS + 1, and so on.
3810 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3812 RANGE is how far to scan while trying to match. RANGE = 0 means try
3813 only at STARTPOS; in general, the last start tried is STARTPOS +
3816 In REGS, return the indices of the virtual concatenation of STRING1
3817 and STRING2 that matched the entire BUFP->buffer and its contained
3820 Do not consider matching one past the index STOP in the virtual
3821 concatenation of STRING1 and STRING2.
3823 We return either the position in the strings at which the match was
3824 found, -1 if no match, or -2 if error (such as failure
3828 re_search_2 (bufp, str1, size1, str2, size2, startpos, range, regs, stop)
3829 struct re_pattern_buffer *bufp;
3830 const char *str1, *str2;
3834 struct re_registers *regs;
3838 re_char *string1 = (re_char*) str1;
3839 re_char *string2 = (re_char*) str2;
3840 register char *fastmap = bufp->fastmap;
3841 register RE_TRANSLATE_TYPE translate = bufp->translate;
3842 int total_size = size1 + size2;
3843 int endpos = startpos + range;
3844 boolean anchored_start;
3846 /* Nonzero if we have to concern multibyte character. */
3847 const boolean multibyte = RE_MULTIBYTE_P (bufp);
3849 /* Check for out-of-range STARTPOS. */
3850 if (startpos < 0 || startpos > total_size)
3853 /* Fix up RANGE if it might eventually take us outside
3854 the virtual concatenation of STRING1 and STRING2.
3855 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3857 range = 0 - startpos;
3858 else if (endpos > total_size)
3859 range = total_size - startpos;
3861 /* If the search isn't to be a backwards one, don't waste time in a
3862 search for a pattern anchored at beginning of buffer. */
3863 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3872 /* In a forward search for something that starts with \=.
3873 don't keep searching past point. */
3874 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3876 range = PT_BYTE - BEGV_BYTE - startpos;
3882 /* Update the fastmap now if not correct already. */
3883 if (fastmap && !bufp->fastmap_accurate)
3884 re_compile_fastmap (bufp);
3886 /* See whether the pattern is anchored. */
3887 anchored_start = (bufp->buffer[0] == begline);
3890 gl_state.object = re_match_object;
3892 int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (POS_AS_IN_BUFFER (startpos));
3894 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1);
3898 /* Loop through the string, looking for a place to start matching. */
3901 /* If the pattern is anchored,
3902 skip quickly past places we cannot match.
3903 We don't bother to treat startpos == 0 specially
3904 because that case doesn't repeat. */
3905 if (anchored_start && startpos > 0)
3907 if (! ((startpos <= size1 ? string1[startpos - 1]
3908 : string2[startpos - size1 - 1])
3913 /* If a fastmap is supplied, skip quickly over characters that
3914 cannot be the start of a match. If the pattern can match the
3915 null string, however, we don't need to skip characters; we want
3916 the first null string. */
3917 if (fastmap && startpos < total_size && !bufp->can_be_null)
3919 register re_char *d;
3920 register re_wchar_t buf_ch;
3922 d = POS_ADDR_VSTRING (startpos);
3924 if (range > 0) /* Searching forwards. */
3926 register int lim = 0;
3929 if (startpos < size1 && startpos + range >= size1)
3930 lim = range - (size1 - startpos);
3932 /* Written out as an if-else to avoid testing `translate'
3934 if (RE_TRANSLATE_P (translate))
3941 buf_ch = STRING_CHAR_AND_LENGTH (d, range - lim,
3944 buf_ch = RE_TRANSLATE (translate, buf_ch);
3949 range -= buf_charlen;
3954 && !fastmap[RE_TRANSLATE (translate, *d)])
3961 while (range > lim && !fastmap[*d])
3967 startpos += irange - range;
3969 else /* Searching backwards. */
3971 int room = (startpos >= size1
3972 ? size2 + size1 - startpos
3973 : size1 - startpos);
3974 buf_ch = RE_STRING_CHAR (d, room);
3975 buf_ch = TRANSLATE (buf_ch);
3977 if (! (buf_ch >= 0400
3978 || fastmap[buf_ch]))
3983 /* If can't match the null string, and that's all we have left, fail. */
3984 if (range >= 0 && startpos == total_size && fastmap
3985 && !bufp->can_be_null)
3988 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3989 startpos, regs, stop);
3990 #ifndef REGEX_MALLOC
4007 /* Update STARTPOS to the next character boundary. */
4010 re_char *p = POS_ADDR_VSTRING (startpos);
4011 re_char *pend = STOP_ADDR_VSTRING (startpos);
4012 int len = MULTIBYTE_FORM_LENGTH (p, pend - p);
4030 /* Update STARTPOS to the previous character boundary. */
4033 re_char *p = POS_ADDR_VSTRING (startpos);
4036 /* Find the head of multibyte form. */
4037 while (!CHAR_HEAD_P (*p))
4042 if (MULTIBYTE_FORM_LENGTH (p, len + 1) != (len + 1))
4058 WEAK_ALIAS (__re_search_2, re_search_2)
4060 /* Declarations and macros for re_match_2. */
4062 static int bcmp_translate _RE_ARGS((re_char *s1, re_char *s2,
4064 RE_TRANSLATE_TYPE translate,
4065 const int multibyte));
4067 /* This converts PTR, a pointer into one of the search strings `string1'
4068 and `string2' into an offset from the beginning of that string. */
4069 #define POINTER_TO_OFFSET(ptr) \
4070 (FIRST_STRING_P (ptr) \
4071 ? ((regoff_t) ((ptr) - string1)) \
4072 : ((regoff_t) ((ptr) - string2 + size1)))
4074 /* Call before fetching a character with *d. This switches over to
4075 string2 if necessary.
4076 Check re_match_2_internal for a discussion of why end_match_2 might
4077 not be within string2 (but be equal to end_match_1 instead). */
4078 #define PREFETCH() \
4081 /* End of string2 => fail. */ \
4082 if (dend == end_match_2) \
4084 /* End of string1 => advance to string2. */ \
4086 dend = end_match_2; \
4089 /* Call before fetching a char with *d if you already checked other limits.
4090 This is meant for use in lookahead operations like wordend, etc..
4091 where we might need to look at parts of the string that might be
4092 outside of the LIMITs (i.e past `stop'). */
4093 #define PREFETCH_NOLIMIT() \
4097 dend = end_match_2; \
4100 /* Test if at very beginning or at very end of the virtual concatenation
4101 of `string1' and `string2'. If only one string, it's `string2'. */
4102 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
4103 #define AT_STRINGS_END(d) ((d) == end2)
4106 /* Test if D points to a character which is word-constituent. We have
4107 two special cases to check for: if past the end of string1, look at
4108 the first character in string2; and if before the beginning of
4109 string2, look at the last character in string1. */
4110 #define WORDCHAR_P(d) \
4111 (SYNTAX ((d) == end1 ? *string2 \
4112 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
4115 /* Disabled due to a compiler bug -- see comment at case wordbound */
4117 /* The comment at case wordbound is following one, but we don't use
4118 AT_WORD_BOUNDARY anymore to support multibyte form.
4120 The DEC Alpha C compiler 3.x generates incorrect code for the
4121 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4122 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4123 macro and introducing temporary variables works around the bug. */
4126 /* Test if the character before D and the one at D differ with respect
4127 to being word-constituent. */
4128 #define AT_WORD_BOUNDARY(d) \
4129 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
4130 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
4133 /* Free everything we malloc. */
4134 #ifdef MATCH_MAY_ALLOCATE
4135 # define FREE_VAR(var) if (var) { REGEX_FREE (var); var = NULL; } else
4136 # define FREE_VARIABLES() \
4138 REGEX_FREE_STACK (fail_stack.stack); \
4139 FREE_VAR (regstart); \
4140 FREE_VAR (regend); \
4141 FREE_VAR (best_regstart); \
4142 FREE_VAR (best_regend); \
4145 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
4146 #endif /* not MATCH_MAY_ALLOCATE */
4149 /* Optimization routines. */
4151 /* If the operation is a match against one or more chars,
4152 return a pointer to the next operation, else return NULL. */
4157 switch (SWITCH_ENUM_CAST (*p++))
4168 if (CHARSET_RANGE_TABLE_EXISTS_P (p - 1))
4171 p = CHARSET_RANGE_TABLE (p - 1);
4172 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4173 p = CHARSET_RANGE_TABLE_END (p, mcnt);
4176 p += 1 + CHARSET_BITMAP_SIZE (p - 1);
4183 case notcategoryspec:
4195 /* Jump over non-matching operations. */
4196 static unsigned char *
4197 skip_noops (p, pend)
4198 unsigned char *p, *pend;
4203 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p))
4212 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4223 /* Non-zero if "p1 matches something" implies "p2 fails". */
4225 mutually_exclusive_p (bufp, p1, p2)
4226 struct re_pattern_buffer *bufp;
4227 unsigned char *p1, *p2;
4230 const boolean multibyte = RE_MULTIBYTE_P (bufp);
4231 unsigned char *pend = bufp->buffer + bufp->used;
4233 assert (p1 >= bufp->buffer && p1 < pend
4234 && p2 >= bufp->buffer && p2 <= pend);
4236 /* Skip over open/close-group commands.
4237 If what follows this loop is a ...+ construct,
4238 look at what begins its body, since we will have to
4239 match at least one of that. */
4240 p2 = skip_noops (p2, pend);
4241 /* The same skip can be done for p1, except that this function
4242 is only used in the case where p1 is a simple match operator. */
4243 /* p1 = skip_noops (p1, pend); */
4245 assert (p1 >= bufp->buffer && p1 < pend
4246 && p2 >= bufp->buffer && p2 <= pend);
4248 op2 = p2 == pend ? succeed : *p2;
4250 switch (SWITCH_ENUM_CAST (op2))
4254 /* If we're at the end of the pattern, we can change. */
4255 if (skip_one_char (p1))
4257 DEBUG_PRINT1 (" End of pattern: fast loop.\n");
4265 register re_wchar_t c
4266 = (re_opcode_t) *p2 == endline ? '\n'
4267 : RE_STRING_CHAR(p2 + 2, pend - p2 - 2);
4269 if ((re_opcode_t) *p1 == exactn)
4271 if (c != RE_STRING_CHAR (p1 + 2, pend - p1 - 2))
4273 DEBUG_PRINT3 (" '%c' != '%c' => fast loop.\n", c, p1[2]);
4278 else if ((re_opcode_t) *p1 == charset
4279 || (re_opcode_t) *p1 == charset_not)
4281 int not = (re_opcode_t) *p1 == charset_not;
4283 /* Test if C is listed in charset (or charset_not)
4285 if (SINGLE_BYTE_CHAR_P (c))
4287 if (c < CHARSET_BITMAP_SIZE (p1) * BYTEWIDTH
4288 && p1[2 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4291 else if (CHARSET_RANGE_TABLE_EXISTS_P (p1))
4292 CHARSET_LOOKUP_RANGE_TABLE (not, c, p1);
4294 /* `not' is equal to 1 if c would match, which means
4295 that we can't change to pop_failure_jump. */
4298 DEBUG_PRINT1 (" No match => fast loop.\n");
4302 else if ((re_opcode_t) *p1 == anychar
4305 DEBUG_PRINT1 (" . != \\n => fast loop.\n");
4314 if ((re_opcode_t) *p1 == exactn)
4315 /* Reuse the code above. */
4316 return mutually_exclusive_p (bufp, p2, p1);
4319 /* It is hard to list up all the character in charset
4320 P2 if it includes multibyte character. Give up in
4322 else if (!multibyte || !CHARSET_RANGE_TABLE_EXISTS_P (p2))
4324 /* Now, we are sure that P2 has no range table.
4325 So, for the size of bitmap in P2, `p2[1]' is
4326 enough. But P1 may have range table, so the
4327 size of bitmap table of P1 is extracted by
4328 using macro `CHARSET_BITMAP_SIZE'.
4330 Since we know that all the character listed in
4331 P2 is ASCII, it is enough to test only bitmap
4337 /* We win if the charset inside the loop
4338 has no overlap with the one after the loop. */
4341 && idx < CHARSET_BITMAP_SIZE (p1));
4343 if ((p2[2 + idx] & p1[2 + idx]) != 0)
4347 || idx == CHARSET_BITMAP_SIZE (p1))
4349 DEBUG_PRINT1 (" No match => fast loop.\n");
4353 else if ((re_opcode_t) *p1 == charset
4354 || (re_opcode_t) *p1 == charset_not)
4357 /* We win if the charset_not inside the loop lists
4358 every character listed in the charset after. */
4359 for (idx = 0; idx < (int) p2[1]; idx++)
4360 if (! (p2[2 + idx] == 0
4361 || (idx < CHARSET_BITMAP_SIZE (p1)
4362 && ((p2[2 + idx] & ~ p1[2 + idx]) == 0))))
4367 DEBUG_PRINT1 (" No match => fast loop.\n");
4376 return ((re_opcode_t) *p1 == syntaxspec
4377 && p1[1] == (op2 == wordend ? Sword : p2[1]));
4381 return ((re_opcode_t) *p1 == notsyntaxspec
4382 && p1[1] == (op2 == wordend ? Sword : p2[1]));
4385 return (((re_opcode_t) *p1 == notsyntaxspec
4386 || (re_opcode_t) *p1 == syntaxspec)
4391 return ((re_opcode_t) *p1 == notcategoryspec && p1[1] == p2[1]);
4392 case notcategoryspec:
4393 return ((re_opcode_t) *p1 == categoryspec && p1[1] == p2[1]);
4405 /* Matching routines. */
4407 #ifndef emacs /* Emacs never uses this. */
4408 /* re_match is like re_match_2 except it takes only a single string. */
4411 re_match (bufp, string, size, pos, regs)
4412 struct re_pattern_buffer *bufp;
4415 struct re_registers *regs;
4417 int result = re_match_2_internal (bufp, NULL, 0, (re_char*) string, size,
4419 # if defined C_ALLOCA && !defined REGEX_MALLOC
4424 WEAK_ALIAS (__re_match, re_match)
4425 #endif /* not emacs */
4428 /* In Emacs, this is the string or buffer in which we
4429 are matching. It is used for looking up syntax properties. */
4430 Lisp_Object re_match_object;
4433 /* re_match_2 matches the compiled pattern in BUFP against the
4434 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
4435 and SIZE2, respectively). We start matching at POS, and stop
4438 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
4439 store offsets for the substring each group matched in REGS. See the
4440 documentation for exactly how many groups we fill.
4442 We return -1 if no match, -2 if an internal error (such as the
4443 failure stack overflowing). Otherwise, we return the length of the
4444 matched substring. */
4447 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
4448 struct re_pattern_buffer *bufp;
4449 const char *string1, *string2;
4452 struct re_registers *regs;
4459 gl_state.object = re_match_object;
4460 charpos = SYNTAX_TABLE_BYTE_TO_CHAR (POS_AS_IN_BUFFER (pos));
4461 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1);
4464 result = re_match_2_internal (bufp, (re_char*) string1, size1,
4465 (re_char*) string2, size2,
4467 #if defined C_ALLOCA && !defined REGEX_MALLOC
4472 WEAK_ALIAS (__re_match_2, re_match_2)
4474 /* This is a separate function so that we can force an alloca cleanup
4477 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
4478 struct re_pattern_buffer *bufp;
4479 re_char *string1, *string2;
4482 struct re_registers *regs;
4485 /* General temporaries. */
4490 /* Just past the end of the corresponding string. */
4491 re_char *end1, *end2;
4493 /* Pointers into string1 and string2, just past the last characters in
4494 each to consider matching. */
4495 re_char *end_match_1, *end_match_2;
4497 /* Where we are in the data, and the end of the current string. */
4500 /* Used sometimes to remember where we were before starting matching
4501 an operator so that we can go back in case of failure. This "atomic"
4502 behavior of matching opcodes is indispensable to the correctness
4503 of the on_failure_keep_string_jump optimization. */
4506 /* Where we are in the pattern, and the end of the pattern. */
4507 re_char *p = bufp->buffer;
4508 re_char *pend = p + bufp->used;
4510 /* We use this to map every character in the string. */
4511 RE_TRANSLATE_TYPE translate = bufp->translate;
4513 /* Nonzero if we have to concern multibyte character. */
4514 const boolean multibyte = RE_MULTIBYTE_P (bufp);
4516 /* Failure point stack. Each place that can handle a failure further
4517 down the line pushes a failure point on this stack. It consists of
4518 regstart, and regend for all registers corresponding to
4519 the subexpressions we're currently inside, plus the number of such
4520 registers, and, finally, two char *'s. The first char * is where
4521 to resume scanning the pattern; the second one is where to resume
4522 scanning the strings. */
4523 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
4524 fail_stack_type fail_stack;
4527 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
4530 #if defined REL_ALLOC && defined REGEX_MALLOC
4531 /* This holds the pointer to the failure stack, when
4532 it is allocated relocatably. */
4533 fail_stack_elt_t *failure_stack_ptr;
4536 /* We fill all the registers internally, independent of what we
4537 return, for use in backreferences. The number here includes
4538 an element for register zero. */
4539 size_t num_regs = bufp->re_nsub + 1;
4541 /* Information on the contents of registers. These are pointers into
4542 the input strings; they record just what was matched (on this
4543 attempt) by a subexpression part of the pattern, that is, the
4544 regnum-th regstart pointer points to where in the pattern we began
4545 matching and the regnum-th regend points to right after where we
4546 stopped matching the regnum-th subexpression. (The zeroth register
4547 keeps track of what the whole pattern matches.) */
4548 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4549 re_char **regstart, **regend;
4552 /* The following record the register info as found in the above
4553 variables when we find a match better than any we've seen before.
4554 This happens as we backtrack through the failure points, which in
4555 turn happens only if we have not yet matched the entire string. */
4556 unsigned best_regs_set = false;
4557 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4558 re_char **best_regstart, **best_regend;
4561 /* Logically, this is `best_regend[0]'. But we don't want to have to
4562 allocate space for that if we're not allocating space for anything
4563 else (see below). Also, we never need info about register 0 for
4564 any of the other register vectors, and it seems rather a kludge to
4565 treat `best_regend' differently than the rest. So we keep track of
4566 the end of the best match so far in a separate variable. We
4567 initialize this to NULL so that when we backtrack the first time
4568 and need to test it, it's not garbage. */
4569 re_char *match_end = NULL;
4572 /* Counts the total number of registers pushed. */
4573 unsigned num_regs_pushed = 0;
4576 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
4580 #ifdef MATCH_MAY_ALLOCATE
4581 /* Do not bother to initialize all the register variables if there are
4582 no groups in the pattern, as it takes a fair amount of time. If
4583 there are groups, we include space for register 0 (the whole
4584 pattern), even though we never use it, since it simplifies the
4585 array indexing. We should fix this. */
4588 regstart = REGEX_TALLOC (num_regs, re_char *);
4589 regend = REGEX_TALLOC (num_regs, re_char *);
4590 best_regstart = REGEX_TALLOC (num_regs, re_char *);
4591 best_regend = REGEX_TALLOC (num_regs, re_char *);
4593 if (!(regstart && regend && best_regstart && best_regend))
4601 /* We must initialize all our variables to NULL, so that
4602 `FREE_VARIABLES' doesn't try to free them. */
4603 regstart = regend = best_regstart = best_regend = NULL;
4605 #endif /* MATCH_MAY_ALLOCATE */
4607 /* The starting position is bogus. */
4608 if (pos < 0 || pos > size1 + size2)
4614 /* Initialize subexpression text positions to -1 to mark ones that no
4615 start_memory/stop_memory has been seen for. Also initialize the
4616 register information struct. */
4617 for (reg = 1; reg < num_regs; reg++)
4618 regstart[reg] = regend[reg] = NULL;
4620 /* We move `string1' into `string2' if the latter's empty -- but not if
4621 `string1' is null. */
4622 if (size2 == 0 && string1 != NULL)
4629 end1 = string1 + size1;
4630 end2 = string2 + size2;
4632 /* `p' scans through the pattern as `d' scans through the data.
4633 `dend' is the end of the input string that `d' points within. `d'
4634 is advanced into the following input string whenever necessary, but
4635 this happens before fetching; therefore, at the beginning of the
4636 loop, `d' can be pointing at the end of a string, but it cannot
4640 /* Only match within string2. */
4641 d = string2 + pos - size1;
4642 dend = end_match_2 = string2 + stop - size1;
4643 end_match_1 = end1; /* Just to give it a value. */
4649 /* Only match within string1. */
4650 end_match_1 = string1 + stop;
4652 When we reach end_match_1, PREFETCH normally switches to string2.
4653 But in the present case, this means that just doing a PREFETCH
4654 makes us jump from `stop' to `gap' within the string.
4655 What we really want here is for the search to stop as
4656 soon as we hit end_match_1. That's why we set end_match_2
4657 to end_match_1 (since PREFETCH fails as soon as we hit
4659 end_match_2 = end_match_1;
4662 { /* It's important to use this code when stop == size so that
4663 moving `d' from end1 to string2 will not prevent the d == dend
4664 check from catching the end of string. */
4666 end_match_2 = string2 + stop - size1;
4672 DEBUG_PRINT1 ("The compiled pattern is: ");
4673 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
4674 DEBUG_PRINT1 ("The string to match is: `");
4675 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
4676 DEBUG_PRINT1 ("'\n");
4678 /* This loops over pattern commands. It exits by returning from the
4679 function if the match is complete, or it drops through if the match
4680 fails at this starting point in the input data. */
4683 DEBUG_PRINT2 ("\n%p: ", p);
4686 { /* End of pattern means we might have succeeded. */
4687 DEBUG_PRINT1 ("end of pattern ... ");
4689 /* If we haven't matched the entire string, and we want the
4690 longest match, try backtracking. */
4691 if (d != end_match_2)
4693 /* 1 if this match ends in the same string (string1 or string2)
4694 as the best previous match. */
4695 boolean same_str_p = (FIRST_STRING_P (match_end)
4696 == FIRST_STRING_P (d));
4697 /* 1 if this match is the best seen so far. */
4698 boolean best_match_p;
4700 /* AIX compiler got confused when this was combined
4701 with the previous declaration. */
4703 best_match_p = d > match_end;
4705 best_match_p = !FIRST_STRING_P (d);
4707 DEBUG_PRINT1 ("backtracking.\n");
4709 if (!FAIL_STACK_EMPTY ())
4710 { /* More failure points to try. */
4712 /* If exceeds best match so far, save it. */
4713 if (!best_regs_set || best_match_p)
4715 best_regs_set = true;
4718 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4720 for (reg = 1; reg < num_regs; reg++)
4722 best_regstart[reg] = regstart[reg];
4723 best_regend[reg] = regend[reg];
4729 /* If no failure points, don't restore garbage. And if
4730 last match is real best match, don't restore second
4732 else if (best_regs_set && !best_match_p)
4735 /* Restore best match. It may happen that `dend ==
4736 end_match_1' while the restored d is in string2.
4737 For example, the pattern `x.*y.*z' against the
4738 strings `x-' and `y-z-', if the two strings are
4739 not consecutive in memory. */
4740 DEBUG_PRINT1 ("Restoring best registers.\n");
4743 dend = ((d >= string1 && d <= end1)
4744 ? end_match_1 : end_match_2);
4746 for (reg = 1; reg < num_regs; reg++)
4748 regstart[reg] = best_regstart[reg];
4749 regend[reg] = best_regend[reg];
4752 } /* d != end_match_2 */
4755 DEBUG_PRINT1 ("Accepting match.\n");
4757 /* If caller wants register contents data back, do it. */
4758 if (regs && !bufp->no_sub)
4760 /* Have the register data arrays been allocated? */
4761 if (bufp->regs_allocated == REGS_UNALLOCATED)
4762 { /* No. So allocate them with malloc. We need one
4763 extra element beyond `num_regs' for the `-1' marker
4765 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4766 regs->start = TALLOC (regs->num_regs, regoff_t);
4767 regs->end = TALLOC (regs->num_regs, regoff_t);
4768 if (regs->start == NULL || regs->end == NULL)
4773 bufp->regs_allocated = REGS_REALLOCATE;
4775 else if (bufp->regs_allocated == REGS_REALLOCATE)
4776 { /* Yes. If we need more elements than were already
4777 allocated, reallocate them. If we need fewer, just
4779 if (regs->num_regs < num_regs + 1)
4781 regs->num_regs = num_regs + 1;
4782 RETALLOC (regs->start, regs->num_regs, regoff_t);
4783 RETALLOC (regs->end, regs->num_regs, regoff_t);
4784 if (regs->start == NULL || regs->end == NULL)
4793 /* These braces fend off a "empty body in an else-statement"
4794 warning under GCC when assert expands to nothing. */
4795 assert (bufp->regs_allocated == REGS_FIXED);
4798 /* Convert the pointer data in `regstart' and `regend' to
4799 indices. Register zero has to be set differently,
4800 since we haven't kept track of any info for it. */
4801 if (regs->num_regs > 0)
4803 regs->start[0] = pos;
4804 regs->end[0] = POINTER_TO_OFFSET (d);
4807 /* Go through the first `min (num_regs, regs->num_regs)'
4808 registers, since that is all we initialized. */
4809 for (reg = 1; reg < MIN (num_regs, regs->num_regs); reg++)
4811 if (REG_UNSET (regstart[reg]) || REG_UNSET (regend[reg]))
4812 regs->start[reg] = regs->end[reg] = -1;
4816 = (regoff_t) POINTER_TO_OFFSET (regstart[reg]);
4818 = (regoff_t) POINTER_TO_OFFSET (regend[reg]);
4822 /* If the regs structure we return has more elements than
4823 were in the pattern, set the extra elements to -1. If
4824 we (re)allocated the registers, this is the case,
4825 because we always allocate enough to have at least one
4827 for (reg = num_regs; reg < regs->num_regs; reg++)
4828 regs->start[reg] = regs->end[reg] = -1;
4829 } /* regs && !bufp->no_sub */
4831 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4832 nfailure_points_pushed, nfailure_points_popped,
4833 nfailure_points_pushed - nfailure_points_popped);
4834 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4836 mcnt = POINTER_TO_OFFSET (d) - pos;
4838 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4844 /* Otherwise match next pattern command. */
4845 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4847 /* Ignore these. Used to ignore the n of succeed_n's which
4848 currently have n == 0. */
4850 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4854 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4857 /* Match the next n pattern characters exactly. The following
4858 byte in the pattern defines n, and the n bytes after that
4859 are the characters to match. */
4862 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4864 /* Remember the start point to rollback upon failure. */
4867 /* This is written out as an if-else so we don't waste time
4868 testing `translate' inside the loop. */
4869 if (RE_TRANSLATE_P (translate))
4874 int pat_charlen, buf_charlen;
4875 unsigned int pat_ch, buf_ch;
4878 pat_ch = STRING_CHAR_AND_LENGTH (p, pend - p, pat_charlen);
4879 buf_ch = STRING_CHAR_AND_LENGTH (d, dend - d, buf_charlen);
4881 if (RE_TRANSLATE (translate, buf_ch)
4890 mcnt -= pat_charlen;
4897 if (RE_TRANSLATE (translate, *d) != *p++)
4922 /* Match any character except possibly a newline or a null. */
4928 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4931 buf_ch = RE_STRING_CHAR_AND_LENGTH (d, dend - d, buf_charlen);
4932 buf_ch = TRANSLATE (buf_ch);
4934 if ((!(bufp->syntax & RE_DOT_NEWLINE)
4936 || ((bufp->syntax & RE_DOT_NOT_NULL)
4937 && buf_ch == '\000'))
4940 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4949 register unsigned int c;
4950 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4953 /* Start of actual range_table, or end of bitmap if there is no
4955 re_char *range_table;
4957 /* Nonzero if there is a range table. */
4958 int range_table_exists;
4960 /* Number of ranges of range table. This is not included
4961 in the initial byte-length of the command. */
4964 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4966 range_table_exists = CHARSET_RANGE_TABLE_EXISTS_P (&p[-1]);
4968 if (range_table_exists)
4970 range_table = CHARSET_RANGE_TABLE (&p[-1]); /* Past the bitmap. */
4971 EXTRACT_NUMBER_AND_INCR (count, range_table);
4975 c = RE_STRING_CHAR_AND_LENGTH (d, dend - d, len);
4976 c = TRANSLATE (c); /* The character to match. */
4978 if (SINGLE_BYTE_CHAR_P (c))
4979 { /* Lookup bitmap. */
4980 /* Cast to `unsigned' instead of `unsigned char' in
4981 case the bit list is a full 32 bytes long. */
4982 if (c < (unsigned) (CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH)
4983 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4987 else if (range_table_exists)
4989 int class_bits = CHARSET_RANGE_TABLE_BITS (&p[-1]);
4991 if ( (class_bits & BIT_LOWER && ISLOWER (c))
4992 | (class_bits & BIT_MULTIBYTE)
4993 | (class_bits & BIT_PUNCT && ISPUNCT (c))
4994 | (class_bits & BIT_SPACE && ISSPACE (c))
4995 | (class_bits & BIT_UPPER && ISUPPER (c))
4996 | (class_bits & BIT_WORD && ISWORD (c)))
4999 CHARSET_LOOKUP_RANGE_TABLE_RAW (not, c, range_table, count);
5003 if (range_table_exists)
5004 p = CHARSET_RANGE_TABLE_END (range_table, count);
5006 p += CHARSET_BITMAP_SIZE (&p[-1]) + 1;
5008 if (!not) goto fail;
5015 /* The beginning of a group is represented by start_memory.
5016 The argument is the register number. The text
5017 matched within the group is recorded (in the internal
5018 registers data structure) under the register number. */
5020 DEBUG_PRINT2 ("EXECUTING start_memory %d:\n", *p);
5022 /* In case we need to undo this operation (via backtracking). */
5023 PUSH_FAILURE_REG ((unsigned int)*p);
5026 regend[*p] = NULL; /* probably unnecessary. -sm */
5027 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
5029 /* Move past the register number and inner group count. */
5034 /* The stop_memory opcode represents the end of a group. Its
5035 argument is the same as start_memory's: the register number. */
5037 DEBUG_PRINT2 ("EXECUTING stop_memory %d:\n", *p);
5039 assert (!REG_UNSET (regstart[*p]));
5040 /* Strictly speaking, there should be code such as:
5042 assert (REG_UNSET (regend[*p]));
5043 PUSH_FAILURE_REGSTOP ((unsigned int)*p);
5045 But the only info to be pushed is regend[*p] and it is known to
5046 be UNSET, so there really isn't anything to push.
5047 Not pushing anything, on the other hand deprives us from the
5048 guarantee that regend[*p] is UNSET since undoing this operation
5049 will not reset its value properly. This is not important since
5050 the value will only be read on the next start_memory or at
5051 the very end and both events can only happen if this stop_memory
5055 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
5057 /* Move past the register number and the inner group count. */
5062 /* \<digit> has been turned into a `duplicate' command which is
5063 followed by the numeric value of <digit> as the register number. */
5066 register re_char *d2, *dend2;
5067 int regno = *p++; /* Get which register to match against. */
5068 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
5070 /* Can't back reference a group which we've never matched. */
5071 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
5074 /* Where in input to try to start matching. */
5075 d2 = regstart[regno];
5077 /* Remember the start point to rollback upon failure. */
5080 /* Where to stop matching; if both the place to start and
5081 the place to stop matching are in the same string, then
5082 set to the place to stop, otherwise, for now have to use
5083 the end of the first string. */
5085 dend2 = ((FIRST_STRING_P (regstart[regno])
5086 == FIRST_STRING_P (regend[regno]))
5087 ? regend[regno] : end_match_1);
5090 /* If necessary, advance to next segment in register
5094 if (dend2 == end_match_2) break;
5095 if (dend2 == regend[regno]) break;
5097 /* End of string1 => advance to string2. */
5099 dend2 = regend[regno];
5101 /* At end of register contents => success */
5102 if (d2 == dend2) break;
5104 /* If necessary, advance to next segment in data. */
5107 /* How many characters left in this segment to match. */
5110 /* Want how many consecutive characters we can match in
5111 one shot, so, if necessary, adjust the count. */
5112 if (mcnt > dend2 - d2)
5115 /* Compare that many; failure if mismatch, else move
5117 if (RE_TRANSLATE_P (translate)
5118 ? bcmp_translate (d, d2, mcnt, translate, multibyte)
5119 : memcmp (d, d2, mcnt))
5124 d += mcnt, d2 += mcnt;
5130 /* begline matches the empty string at the beginning of the string
5131 (unless `not_bol' is set in `bufp'), and after newlines. */
5133 DEBUG_PRINT1 ("EXECUTING begline.\n");
5135 if (AT_STRINGS_BEG (d))
5137 if (!bufp->not_bol) break;
5142 GET_CHAR_BEFORE_2 (c, d, string1, end1, string2, end2);
5146 /* In all other cases, we fail. */
5150 /* endline is the dual of begline. */
5152 DEBUG_PRINT1 ("EXECUTING endline.\n");
5154 if (AT_STRINGS_END (d))
5156 if (!bufp->not_eol) break;
5160 PREFETCH_NOLIMIT ();
5167 /* Match at the very beginning of the data. */
5169 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
5170 if (AT_STRINGS_BEG (d))
5175 /* Match at the very end of the data. */
5177 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
5178 if (AT_STRINGS_END (d))
5183 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
5184 pushes NULL as the value for the string on the stack. Then
5185 `POP_FAILURE_POINT' will keep the current value for the
5186 string, instead of restoring it. To see why, consider
5187 matching `foo\nbar' against `.*\n'. The .* matches the foo;
5188 then the . fails against the \n. But the next thing we want
5189 to do is match the \n against the \n; if we restored the
5190 string value, we would be back at the foo.
5192 Because this is used only in specific cases, we don't need to
5193 check all the things that `on_failure_jump' does, to make
5194 sure the right things get saved on the stack. Hence we don't
5195 share its code. The only reason to push anything on the
5196 stack at all is that otherwise we would have to change
5197 `anychar's code to do something besides goto fail in this
5198 case; that seems worse than this. */
5199 case on_failure_keep_string_jump:
5200 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5201 DEBUG_PRINT3 ("EXECUTING on_failure_keep_string_jump %d (to %p):\n",
5204 PUSH_FAILURE_POINT (p - 3, NULL);
5207 /* A nasty loop is introduced by the non-greedy *? and +?.
5208 With such loops, the stack only ever contains one failure point
5209 at a time, so that a plain on_failure_jump_loop kind of
5210 cycle detection cannot work. Worse yet, such a detection
5211 can not only fail to detect a cycle, but it can also wrongly
5212 detect a cycle (between different instantiations of the same
5214 So the method used for those nasty loops is a little different:
5215 We use a special cycle-detection-stack-frame which is pushed
5216 when the on_failure_jump_nastyloop failure-point is *popped*.
5217 This special frame thus marks the beginning of one iteration
5218 through the loop and we can hence easily check right here
5219 whether something matched between the beginning and the end of
5221 case on_failure_jump_nastyloop:
5222 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5223 DEBUG_PRINT3 ("EXECUTING on_failure_jump_nastyloop %d (to %p):\n",
5226 assert ((re_opcode_t)p[-4] == no_op);
5227 CHECK_INFINITE_LOOP (p - 4, d);
5228 PUSH_FAILURE_POINT (p - 3, d);
5232 /* Simple loop detecting on_failure_jump: just check on the
5233 failure stack if the same spot was already hit earlier. */
5234 case on_failure_jump_loop:
5236 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5237 DEBUG_PRINT3 ("EXECUTING on_failure_jump_loop %d (to %p):\n",
5240 CHECK_INFINITE_LOOP (p - 3, d);
5241 PUSH_FAILURE_POINT (p - 3, d);
5245 /* Uses of on_failure_jump:
5247 Each alternative starts with an on_failure_jump that points
5248 to the beginning of the next alternative. Each alternative
5249 except the last ends with a jump that in effect jumps past
5250 the rest of the alternatives. (They really jump to the
5251 ending jump of the following alternative, because tensioning
5252 these jumps is a hassle.)
5254 Repeats start with an on_failure_jump that points past both
5255 the repetition text and either the following jump or
5256 pop_failure_jump back to this on_failure_jump. */
5257 case on_failure_jump:
5258 IMMEDIATE_QUIT_CHECK;
5259 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5260 DEBUG_PRINT3 ("EXECUTING on_failure_jump %d (to %p):\n",
5263 PUSH_FAILURE_POINT (p -3, d);
5266 /* This operation is used for greedy *.
5267 Compare the beginning of the repeat with what in the
5268 pattern follows its end. If we can establish that there
5269 is nothing that they would both match, i.e., that we
5270 would have to backtrack because of (as in, e.g., `a*a')
5271 then we can use a non-backtracking loop based on
5272 on_failure_keep_string_jump instead of on_failure_jump. */
5273 case on_failure_jump_smart:
5274 IMMEDIATE_QUIT_CHECK;
5275 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5276 DEBUG_PRINT3 ("EXECUTING on_failure_jump_smart %d (to %p).\n",
5279 re_char *p1 = p; /* Next operation. */
5280 /* Here, we discard `const', making re_match non-reentrant. */
5281 unsigned char *p2 = (unsigned char*) p + mcnt; /* Jump dest. */
5282 unsigned char *p3 = (unsigned char*) p - 3; /* opcode location. */
5284 p -= 3; /* Reset so that we will re-execute the
5285 instruction once it's been changed. */
5287 EXTRACT_NUMBER (mcnt, p2 - 2);
5289 /* Ensure this is a indeed the trivial kind of loop
5290 we are expecting. */
5291 assert (skip_one_char (p1) == p2 - 3);
5292 assert ((re_opcode_t) p2[-3] == jump && p2 + mcnt == p);
5293 DEBUG_STATEMENT (debug += 2);
5294 if (mutually_exclusive_p (bufp, p1, p2))
5296 /* Use a fast `on_failure_keep_string_jump' loop. */
5297 DEBUG_PRINT1 (" smart exclusive => fast loop.\n");
5298 *p3 = (unsigned char) on_failure_keep_string_jump;
5299 STORE_NUMBER (p2 - 2, mcnt + 3);
5303 /* Default to a safe `on_failure_jump' loop. */
5304 DEBUG_PRINT1 (" smart default => slow loop.\n");
5305 *p3 = (unsigned char) on_failure_jump;
5307 DEBUG_STATEMENT (debug -= 2);
5311 /* Unconditionally jump (without popping any failure points). */
5314 IMMEDIATE_QUIT_CHECK;
5315 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
5316 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
5317 p += mcnt; /* Do the jump. */
5318 DEBUG_PRINT2 ("(to %p).\n", p);
5322 /* Have to succeed matching what follows at least n times.
5323 After that, handle like `on_failure_jump'. */
5325 /* Signedness doesn't matter since we only compare MCNT to 0. */
5326 EXTRACT_NUMBER (mcnt, p + 2);
5327 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
5329 /* Originally, mcnt is how many times we HAVE to succeed. */
5332 /* Here, we discard `const', making re_match non-reentrant. */
5333 unsigned char *p2 = (unsigned char*) p + 2; /* counter loc. */
5336 PUSH_NUMBER (p2, mcnt);
5339 /* The two bytes encoding mcnt == 0 are two no_op opcodes. */
5344 /* Signedness doesn't matter since we only compare MCNT to 0. */
5345 EXTRACT_NUMBER (mcnt, p + 2);
5346 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
5348 /* Originally, this is how many times we CAN jump. */
5351 /* Here, we discard `const', making re_match non-reentrant. */
5352 unsigned char *p2 = (unsigned char*) p + 2; /* counter loc. */
5354 PUSH_NUMBER (p2, mcnt);
5355 goto unconditional_jump;
5357 /* If don't have to jump any more, skip over the rest of command. */
5364 unsigned char *p2; /* Location of the counter. */
5365 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
5367 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5368 /* Here, we discard `const', making re_match non-reentrant. */
5369 p2 = (unsigned char*) p + mcnt;
5370 /* Signedness doesn't matter since we only copy MCNT's bits . */
5371 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5372 DEBUG_PRINT3 (" Setting %p to %d.\n", p2, mcnt);
5373 PUSH_NUMBER (p2, mcnt);
5379 not = (re_opcode_t) *(p - 1) == notwordbound;
5380 DEBUG_PRINT2 ("EXECUTING %swordbound.\n", not?"not":"");
5382 /* We SUCCEED (or FAIL) in one of the following cases: */
5384 /* Case 1: D is at the beginning or the end of string. */
5385 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5389 /* C1 is the character before D, S1 is the syntax of C1, C2
5390 is the character at D, and S2 is the syntax of C2. */
5394 int offset = PTR_TO_OFFSET (d - 1);
5395 int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
5396 UPDATE_SYNTAX_TABLE (charpos);
5398 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5401 UPDATE_SYNTAX_TABLE_FORWARD (charpos + 1);
5403 PREFETCH_NOLIMIT ();
5404 c2 = RE_STRING_CHAR (d, dend - d);
5407 if (/* Case 2: Only one of S1 and S2 is Sword. */
5408 ((s1 == Sword) != (s2 == Sword))
5409 /* Case 3: Both of S1 and S2 are Sword, and macro
5410 WORD_BOUNDARY_P (C1, C2) returns nonzero. */
5411 || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2)))
5420 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5422 /* We FAIL in one of the following cases: */
5424 /* Case 1: D is at the end of string. */
5425 if (AT_STRINGS_END (d))
5429 /* C1 is the character before D, S1 is the syntax of C1, C2
5430 is the character at D, and S2 is the syntax of C2. */
5434 int offset = PTR_TO_OFFSET (d);
5435 int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
5436 UPDATE_SYNTAX_TABLE (charpos);
5439 c2 = RE_STRING_CHAR (d, dend - d);
5442 /* Case 2: S2 is not Sword. */
5446 /* Case 3: D is not at the beginning of string ... */
5447 if (!AT_STRINGS_BEG (d))
5449 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5451 UPDATE_SYNTAX_TABLE_BACKWARD (charpos - 1);
5455 /* ... and S1 is Sword, and WORD_BOUNDARY_P (C1, C2)
5457 if ((s1 == Sword) && !WORD_BOUNDARY_P (c1, c2))
5464 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5466 /* We FAIL in one of the following cases: */
5468 /* Case 1: D is at the beginning of string. */
5469 if (AT_STRINGS_BEG (d))
5473 /* C1 is the character before D, S1 is the syntax of C1, C2
5474 is the character at D, and S2 is the syntax of C2. */
5478 int offset = PTR_TO_OFFSET (d) - 1;
5479 int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
5480 UPDATE_SYNTAX_TABLE (charpos);
5482 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5485 /* Case 2: S1 is not Sword. */
5489 /* Case 3: D is not at the end of string ... */
5490 if (!AT_STRINGS_END (d))
5492 PREFETCH_NOLIMIT ();
5493 c2 = RE_STRING_CHAR (d, dend - d);
5495 UPDATE_SYNTAX_TABLE_FORWARD (charpos);
5499 /* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2)
5501 if ((s2 == Sword) && !WORD_BOUNDARY_P (c1, c2))
5509 not = (re_opcode_t) *(p - 1) == notsyntaxspec;
5511 DEBUG_PRINT3 ("EXECUTING %ssyntaxspec %d.\n", not?"not":"", mcnt);
5515 int offset = PTR_TO_OFFSET (d);
5516 int pos1 = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
5517 UPDATE_SYNTAX_TABLE (pos1);
5524 c = RE_STRING_CHAR_AND_LENGTH (d, dend - d, len);
5526 if ((SYNTAX (c) != (enum syntaxcode) mcnt) ^ not)
5534 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5535 if (PTR_BYTE_POS (d) >= PT_BYTE)
5540 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5541 if (PTR_BYTE_POS (d) != PT_BYTE)
5546 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5547 if (PTR_BYTE_POS (d) <= PT_BYTE)
5552 case notcategoryspec:
5553 not = (re_opcode_t) *(p - 1) == notcategoryspec;
5555 DEBUG_PRINT3 ("EXECUTING %scategoryspec %d.\n", not?"not":"", mcnt);
5561 c = RE_STRING_CHAR_AND_LENGTH (d, dend - d, len);
5563 if ((!CHAR_HAS_CATEGORY (c, mcnt)) ^ not)
5574 continue; /* Successfully executed one pattern command; keep going. */
5577 /* We goto here if a matching operation fails. */
5579 IMMEDIATE_QUIT_CHECK;
5580 if (!FAIL_STACK_EMPTY ())
5583 /* A restart point is known. Restore to that state. */
5584 DEBUG_PRINT1 ("\nFAIL:\n");
5585 POP_FAILURE_POINT (str, pat);
5586 switch (SWITCH_ENUM_CAST ((re_opcode_t) *pat++))
5588 case on_failure_keep_string_jump:
5589 assert (str == NULL);
5590 goto continue_failure_jump;
5592 case on_failure_jump_nastyloop:
5593 assert ((re_opcode_t)pat[-2] == no_op);
5594 PUSH_FAILURE_POINT (pat - 2, str);
5597 case on_failure_jump_loop:
5598 case on_failure_jump:
5601 continue_failure_jump:
5602 EXTRACT_NUMBER_AND_INCR (mcnt, pat);
5607 /* A special frame used for nastyloops. */
5614 assert (p >= bufp->buffer && p <= pend);
5616 if (d >= string1 && d <= end1)
5620 break; /* Matching at this starting point really fails. */
5624 goto restore_best_regs;
5628 return -1; /* Failure to match. */
5631 /* Subroutine definitions for re_match_2. */
5633 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5634 bytes; nonzero otherwise. */
5637 bcmp_translate (s1, s2, len, translate, multibyte)
5640 RE_TRANSLATE_TYPE translate;
5641 const int multibyte;
5643 register re_char *p1 = s1, *p2 = s2;
5644 re_char *p1_end = s1 + len;
5645 re_char *p2_end = s2 + len;
5647 /* FIXME: Checking both p1 and p2 presumes that the two strings might have
5648 different lengths, but relying on a single `len' would break this. -sm */
5649 while (p1 < p1_end && p2 < p2_end)
5651 int p1_charlen, p2_charlen;
5652 re_wchar_t p1_ch, p2_ch;
5654 p1_ch = RE_STRING_CHAR_AND_LENGTH (p1, p1_end - p1, p1_charlen);
5655 p2_ch = RE_STRING_CHAR_AND_LENGTH (p2, p2_end - p2, p2_charlen);
5657 if (RE_TRANSLATE (translate, p1_ch)
5658 != RE_TRANSLATE (translate, p2_ch))
5661 p1 += p1_charlen, p2 += p2_charlen;
5664 if (p1 != p1_end || p2 != p2_end)
5670 /* Entry points for GNU code. */
5672 /* re_compile_pattern is the GNU regular expression compiler: it
5673 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5674 Returns 0 if the pattern was valid, otherwise an error string.
5676 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5677 are set in BUFP on entry.
5679 We call regex_compile to do the actual compilation. */
5682 re_compile_pattern (pattern, length, bufp)
5683 const char *pattern;
5685 struct re_pattern_buffer *bufp;
5689 /* GNU code is written to assume at least RE_NREGS registers will be set
5690 (and at least one extra will be -1). */
5691 bufp->regs_allocated = REGS_UNALLOCATED;
5693 /* And GNU code determines whether or not to get register information
5694 by passing null for the REGS argument to re_match, etc., not by
5698 ret = regex_compile ((re_char*) pattern, length, re_syntax_options, bufp);
5702 return gettext (re_error_msgid[(int) ret]);
5704 WEAK_ALIAS (__re_compile_pattern, re_compile_pattern)
5706 /* Entry points compatible with 4.2 BSD regex library. We don't define
5707 them unless specifically requested. */
5709 #if defined _REGEX_RE_COMP || defined _LIBC
5711 /* BSD has one and only one pattern buffer. */
5712 static struct re_pattern_buffer re_comp_buf;
5716 /* Make these definitions weak in libc, so POSIX programs can redefine
5717 these names if they don't use our functions, and still use
5718 regcomp/regexec below without link errors. */
5728 if (!re_comp_buf.buffer)
5729 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5730 return (char *) gettext ("No previous regular expression");
5734 if (!re_comp_buf.buffer)
5736 re_comp_buf.buffer = (unsigned char *) malloc (200);
5737 if (re_comp_buf.buffer == NULL)
5738 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5739 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5740 re_comp_buf.allocated = 200;
5742 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5743 if (re_comp_buf.fastmap == NULL)
5744 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5745 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5748 /* Since `re_exec' always passes NULL for the `regs' argument, we
5749 don't need to initialize the pattern buffer fields which affect it. */
5751 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5756 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5757 return (char *) gettext (re_error_msgid[(int) ret]);
5768 const int len = strlen (s);
5770 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5772 #endif /* _REGEX_RE_COMP */
5774 /* POSIX.2 functions. Don't define these for Emacs. */
5778 /* regcomp takes a regular expression as a string and compiles it.
5780 PREG is a regex_t *. We do not expect any fields to be initialized,
5781 since POSIX says we shouldn't. Thus, we set
5783 `buffer' to the compiled pattern;
5784 `used' to the length of the compiled pattern;
5785 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5786 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5787 RE_SYNTAX_POSIX_BASIC;
5788 `fastmap' to an allocated space for the fastmap;
5789 `fastmap_accurate' to zero;
5790 `re_nsub' to the number of subexpressions in PATTERN.
5792 PATTERN is the address of the pattern string.
5794 CFLAGS is a series of bits which affect compilation.
5796 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5797 use POSIX basic syntax.
5799 If REG_NEWLINE is set, then . and [^...] don't match newline.
5800 Also, regexec will try a match beginning after every newline.
5802 If REG_ICASE is set, then we considers upper- and lowercase
5803 versions of letters to be equivalent when matching.
5805 If REG_NOSUB is set, then when PREG is passed to regexec, that
5806 routine will report only success or failure, and nothing about the
5809 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5810 the return codes and their meanings.) */
5813 regcomp (preg, pattern, cflags)
5815 const char *pattern;
5820 = (cflags & REG_EXTENDED) ?
5821 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5823 /* regex_compile will allocate the space for the compiled pattern. */
5825 preg->allocated = 0;
5828 /* Try to allocate space for the fastmap. */
5829 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
5831 if (cflags & REG_ICASE)
5836 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5837 * sizeof (*(RE_TRANSLATE_TYPE)0));
5838 if (preg->translate == NULL)
5839 return (int) REG_ESPACE;
5841 /* Map uppercase characters to corresponding lowercase ones. */
5842 for (i = 0; i < CHAR_SET_SIZE; i++)
5843 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
5846 preg->translate = NULL;
5848 /* If REG_NEWLINE is set, newlines are treated differently. */
5849 if (cflags & REG_NEWLINE)
5850 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5851 syntax &= ~RE_DOT_NEWLINE;
5852 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5855 syntax |= RE_NO_NEWLINE_ANCHOR;
5857 preg->no_sub = !!(cflags & REG_NOSUB);
5859 /* POSIX says a null character in the pattern terminates it, so we
5860 can use strlen here in compiling the pattern. */
5861 ret = regex_compile ((re_char*) pattern, strlen (pattern), syntax, preg);
5863 /* POSIX doesn't distinguish between an unmatched open-group and an
5864 unmatched close-group: both are REG_EPAREN. */
5865 if (ret == REG_ERPAREN)
5868 if (ret == REG_NOERROR && preg->fastmap)
5869 { /* Compute the fastmap now, since regexec cannot modify the pattern
5871 re_compile_fastmap (preg);
5872 if (preg->can_be_null)
5873 { /* The fastmap can't be used anyway. */
5874 free (preg->fastmap);
5875 preg->fastmap = NULL;
5880 WEAK_ALIAS (__regcomp, regcomp)
5883 /* regexec searches for a given pattern, specified by PREG, in the
5886 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5887 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5888 least NMATCH elements, and we set them to the offsets of the
5889 corresponding matched substrings.
5891 EFLAGS specifies `execution flags' which affect matching: if
5892 REG_NOTBOL is set, then ^ does not match at the beginning of the
5893 string; if REG_NOTEOL is set, then $ does not match at the end.
5895 We return 0 if we find a match and REG_NOMATCH if not. */
5898 regexec (preg, string, nmatch, pmatch, eflags)
5899 const regex_t *preg;
5902 regmatch_t pmatch[];
5906 struct re_registers regs;
5907 regex_t private_preg;
5908 int len = strlen (string);
5909 boolean want_reg_info = !preg->no_sub && nmatch > 0 && pmatch;
5911 private_preg = *preg;
5913 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5914 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5916 /* The user has told us exactly how many registers to return
5917 information about, via `nmatch'. We have to pass that on to the
5918 matching routines. */
5919 private_preg.regs_allocated = REGS_FIXED;
5923 regs.num_regs = nmatch;
5924 regs.start = TALLOC (nmatch * 2, regoff_t);
5925 if (regs.start == NULL)
5926 return (int) REG_NOMATCH;
5927 regs.end = regs.start + nmatch;
5930 /* Instead of using not_eol to implement REG_NOTEOL, we could simply
5931 pass (&private_preg, string, len + 1, 0, len, ...) pretending the string
5932 was a little bit longer but still only matching the real part.
5933 This works because the `endline' will check for a '\n' and will find a
5934 '\0', correctly deciding that this is not the end of a line.
5935 But it doesn't work out so nicely for REG_NOTBOL, since we don't have
5936 a convenient '\0' there. For all we know, the string could be preceded
5937 by '\n' which would throw things off. */
5939 /* Perform the searching operation. */
5940 ret = re_search (&private_preg, string, len,
5941 /* start: */ 0, /* range: */ len,
5942 want_reg_info ? ®s : (struct re_registers *) 0);
5944 /* Copy the register information to the POSIX structure. */
5951 for (r = 0; r < nmatch; r++)
5953 pmatch[r].rm_so = regs.start[r];
5954 pmatch[r].rm_eo = regs.end[r];
5958 /* If we needed the temporary register info, free the space now. */
5962 /* We want zero return to mean success, unlike `re_search'. */
5963 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5965 WEAK_ALIAS (__regexec, regexec)
5968 /* Returns a message corresponding to an error code, ERRCODE, returned
5969 from either regcomp or regexec. We don't use PREG here. */
5972 regerror (errcode, preg, errbuf, errbuf_size)
5974 const regex_t *preg;
5982 || errcode >= (sizeof (re_error_msgid) / sizeof (re_error_msgid[0])))
5983 /* Only error codes returned by the rest of the code should be passed
5984 to this routine. If we are given anything else, or if other regex
5985 code generates an invalid error code, then the program has a bug.
5986 Dump core so we can fix it. */
5989 msg = gettext (re_error_msgid[errcode]);
5991 msg_size = strlen (msg) + 1; /* Includes the null. */
5993 if (errbuf_size != 0)
5995 if (msg_size > errbuf_size)
5997 strncpy (errbuf, msg, errbuf_size - 1);
5998 errbuf[errbuf_size - 1] = 0;
6001 strcpy (errbuf, msg);
6006 WEAK_ALIAS (__regerror, regerror)
6009 /* Free dynamically allocated space used by PREG. */
6015 if (preg->buffer != NULL)
6016 free (preg->buffer);
6017 preg->buffer = NULL;
6019 preg->allocated = 0;
6022 if (preg->fastmap != NULL)
6023 free (preg->fastmap);
6024 preg->fastmap = NULL;
6025 preg->fastmap_accurate = 0;
6027 if (preg->translate != NULL)
6028 free (preg->translate);
6029 preg->translate = NULL;
6031 WEAK_ALIAS (__regfree, regfree)
6033 #endif /* not emacs */
projects / gnulib.git / blob