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) 1993-2011
6 Free Software Foundation, Inc.
8 This program is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 3, or (at your option)
13 This program is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301,
24 - structure the opcode space into opcode+flag.
25 - merge with glibc's regex.[ch].
26 - replace (succeed_n + jump_n + set_number_at) with something that doesn't
27 need to modify the compiled regexp so that re_match can be reentrant.
28 - get rid of on_failure_jump_smart by doing the optimization in re_comp
29 rather than at run-time, so that re_match can be reentrant.
32 /* AIX requires this to be the first thing in the file. */
33 #if defined _AIX && !defined REGEX_MALLOC
41 #if defined STDC_HEADERS && !defined emacs
44 /* We need this for `regex.h', and perhaps for the Emacs include files. */
45 # include <sys/types.h>
48 /* Whether to use ISO C Amendment 1 wide char functions.
49 Those should not be used for Emacs since it uses its own. */
51 #define WIDE_CHAR_SUPPORT 1
53 #define WIDE_CHAR_SUPPORT \
54 (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC && !emacs)
57 /* For platform which support the ISO C amendement 1 functionality we
58 support user defined character classes. */
60 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
66 /* We have to keep the namespace clean. */
67 # define regfree(preg) __regfree (preg)
68 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
69 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
70 # define regerror(err_code, preg, errbuf, errbuf_size) \
71 __regerror(err_code, preg, errbuf, errbuf_size)
72 # define re_set_registers(bu, re, nu, st, en) \
73 __re_set_registers (bu, re, nu, st, en)
74 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
75 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
76 # define re_match(bufp, string, size, pos, regs) \
77 __re_match (bufp, string, size, pos, regs)
78 # define re_search(bufp, string, size, startpos, range, regs) \
79 __re_search (bufp, string, size, startpos, range, regs)
80 # define re_compile_pattern(pattern, length, bufp) \
81 __re_compile_pattern (pattern, length, bufp)
82 # define re_set_syntax(syntax) __re_set_syntax (syntax)
83 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
84 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
85 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
87 /* Make sure we call libc's function even if the user overrides them. */
88 # define btowc __btowc
89 # define iswctype __iswctype
90 # define wctype __wctype
92 # define WEAK_ALIAS(a,b) weak_alias (a, b)
94 /* We are also using some library internals. */
95 # include <locale/localeinfo.h>
96 # include <locale/elem-hash.h>
97 # include <langinfo.h>
99 # define WEAK_ALIAS(a,b)
102 /* This is for other GNU distributions with internationalized messages. */
103 #if HAVE_LIBINTL_H || defined _LIBC
104 # include <libintl.h>
106 # define gettext(msgid) (msgid)
110 /* This define is so xgettext can find the internationalizable
112 # define gettext_noop(String) String
115 /* The `emacs' switch turns on certain matching commands
116 that make sense only in Emacs. */
123 /* Make syntax table lookup grant data in gl_state. */
124 # define SYNTAX_ENTRY_VIA_PROPERTY
127 # include "character.h"
128 # include "category.h"
133 # define malloc xmalloc
137 # define realloc xrealloc
143 /* Converts the pointer to the char to BEG-based offset from the start. */
144 # define PTR_TO_OFFSET(d) POS_AS_IN_BUFFER (POINTER_TO_OFFSET (d))
145 # define POS_AS_IN_BUFFER(p) ((p) + (NILP (re_match_object) || BUFFERP (re_match_object)))
147 # define RE_MULTIBYTE_P(bufp) ((bufp)->multibyte)
148 # define RE_TARGET_MULTIBYTE_P(bufp) ((bufp)->target_multibyte)
149 # define RE_STRING_CHAR(p, multibyte) \
150 (multibyte ? (STRING_CHAR (p)) : (*(p)))
151 # define RE_STRING_CHAR_AND_LENGTH(p, len, multibyte) \
152 (multibyte ? (STRING_CHAR_AND_LENGTH (p, len)) : ((len) = 1, *(p)))
154 # define RE_CHAR_TO_MULTIBYTE(c) UNIBYTE_TO_CHAR (c)
156 # define RE_CHAR_TO_UNIBYTE(c) CHAR_TO_BYTE_SAFE (c)
158 /* Set C a (possibly converted to multibyte) character before P. P
159 points into a string which is the virtual concatenation of STR1
160 (which ends at END1) or STR2 (which ends at END2). */
161 # define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
163 if (target_multibyte) \
165 re_char *dtemp = (p) == (str2) ? (end1) : (p); \
166 re_char *dlimit = ((p) > (str2) && (p) <= (end2)) ? (str2) : (str1); \
167 while (dtemp-- > dlimit && !CHAR_HEAD_P (*dtemp)); \
168 c = STRING_CHAR (dtemp); \
172 (c = ((p) == (str2) ? (end1) : (p))[-1]); \
173 (c) = RE_CHAR_TO_MULTIBYTE (c); \
177 /* Set C a (possibly converted to multibyte) character at P, and set
178 LEN to the byte length of that character. */
179 # define GET_CHAR_AFTER(c, p, len) \
181 if (target_multibyte) \
182 (c) = STRING_CHAR_AND_LENGTH (p, len); \
187 (c) = RE_CHAR_TO_MULTIBYTE (c); \
191 #else /* not emacs */
193 /* If we are not linking with Emacs proper,
194 we can't use the relocating allocator
195 even if config.h says that we can. */
200 /* When used in Emacs's lib-src, we need xmalloc and xrealloc. */
203 xmalloc (size_t size
)
206 val
= (void *) malloc (size
);
209 write (2, "virtual memory exhausted\n", 25);
216 xrealloc (void *block
, size_t size
)
219 /* We must call malloc explicitly when BLOCK is 0, since some
220 reallocs don't do this. */
222 val
= (void *) malloc (size
);
224 val
= (void *) realloc (block
, size
);
227 write (2, "virtual memory exhausted\n", 25);
236 # define malloc xmalloc
240 # define realloc xrealloc
242 /* This is the normal way of making sure we have memcpy, memcmp and memset. */
243 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
246 # include <strings.h>
248 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
251 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
255 /* Define the syntax stuff for \<, \>, etc. */
257 /* Sword must be nonzero for the wordchar pattern commands in re_match_2. */
258 enum syntaxcode
{ Swhitespace
= 0, Sword
= 1, Ssymbol
= 2 };
260 # define SWITCH_ENUM_CAST(x) (x)
262 /* Dummy macros for non-Emacs environments. */
263 # define CHAR_CHARSET(c) 0
264 # define CHARSET_LEADING_CODE_BASE(c) 0
265 # define MAX_MULTIBYTE_LENGTH 1
266 # define RE_MULTIBYTE_P(x) 0
267 # define RE_TARGET_MULTIBYTE_P(x) 0
268 # define WORD_BOUNDARY_P(c1, c2) (0)
269 # define CHAR_HEAD_P(p) (1)
270 # define SINGLE_BYTE_CHAR_P(c) (1)
271 # define SAME_CHARSET_P(c1, c2) (1)
272 # define BYTES_BY_CHAR_HEAD(p) (1)
273 # define PREV_CHAR_BOUNDARY(p, limit) ((p)--)
274 # define STRING_CHAR(p) (*(p))
275 # define RE_STRING_CHAR(p, multibyte) STRING_CHAR (p)
276 # define CHAR_STRING(c, s) (*(s) = (c), 1)
277 # define STRING_CHAR_AND_LENGTH(p, actual_len) ((actual_len) = 1, *(p))
278 # define RE_STRING_CHAR_AND_LENGTH(p, len, multibyte) STRING_CHAR_AND_LENGTH (p, len)
279 # define RE_CHAR_TO_MULTIBYTE(c) (c)
280 # define RE_CHAR_TO_UNIBYTE(c) (c)
281 # define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
282 (c = ((p) == (str2) ? *((end1) - 1) : *((p) - 1)))
283 # define GET_CHAR_AFTER(c, p, len) \
285 # define MAKE_CHAR(charset, c1, c2) (c1)
286 # define BYTE8_TO_CHAR(c) (c)
287 # define CHAR_BYTE8_P(c) (0)
288 # define CHAR_LEADING_CODE(c) (c)
290 #endif /* not emacs */
293 # define RE_TRANSLATE(TBL, C) ((unsigned char)(TBL)[C])
294 # define RE_TRANSLATE_P(TBL) (TBL)
297 /* Get the interface, including the syntax bits. */
300 /* isalpha etc. are used for the character classes. */
305 /* 1 if C is an ASCII character. */
306 # define IS_REAL_ASCII(c) ((c) < 0200)
308 /* 1 if C is a unibyte character. */
309 # define ISUNIBYTE(c) (SINGLE_BYTE_CHAR_P ((c)))
311 /* The Emacs definitions should not be directly affected by locales. */
313 /* In Emacs, these are only used for single-byte characters. */
314 # define ISDIGIT(c) ((c) >= '0' && (c) <= '9')
315 # define ISCNTRL(c) ((c) < ' ')
316 # define ISXDIGIT(c) (((c) >= '0' && (c) <= '9') \
317 || ((c) >= 'a' && (c) <= 'f') \
318 || ((c) >= 'A' && (c) <= 'F'))
320 /* This is only used for single-byte characters. */
321 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
323 /* The rest must handle multibyte characters. */
325 # define ISGRAPH(c) (SINGLE_BYTE_CHAR_P (c) \
326 ? (c) > ' ' && !((c) >= 0177 && (c) <= 0237) \
329 # define ISPRINT(c) (SINGLE_BYTE_CHAR_P (c) \
330 ? (c) >= ' ' && !((c) >= 0177 && (c) <= 0237) \
333 # define ISALNUM(c) (IS_REAL_ASCII (c) \
334 ? (((c) >= 'a' && (c) <= 'z') \
335 || ((c) >= 'A' && (c) <= 'Z') \
336 || ((c) >= '0' && (c) <= '9')) \
337 : SYNTAX (c) == Sword)
339 # define ISALPHA(c) (IS_REAL_ASCII (c) \
340 ? (((c) >= 'a' && (c) <= 'z') \
341 || ((c) >= 'A' && (c) <= 'Z')) \
342 : SYNTAX (c) == Sword)
344 # define ISLOWER(c) (LOWERCASEP (c))
346 # define ISPUNCT(c) (IS_REAL_ASCII (c) \
347 ? ((c) > ' ' && (c) < 0177 \
348 && !(((c) >= 'a' && (c) <= 'z') \
349 || ((c) >= 'A' && (c) <= 'Z') \
350 || ((c) >= '0' && (c) <= '9'))) \
351 : SYNTAX (c) != Sword)
353 # define ISSPACE(c) (SYNTAX (c) == Swhitespace)
355 # define ISUPPER(c) (UPPERCASEP (c))
357 # define ISWORD(c) (SYNTAX (c) == Sword)
359 #else /* not emacs */
361 /* Jim Meyering writes:
363 "... Some ctype macros are valid only for character codes that
364 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
365 using /bin/cc or gcc but without giving an ansi option). So, all
366 ctype uses should be through macros like ISPRINT... If
367 STDC_HEADERS is defined, then autoconf has verified that the ctype
368 macros don't need to be guarded with references to isascii. ...
369 Defining isascii to 1 should let any compiler worth its salt
370 eliminate the && through constant folding."
371 Solaris defines some of these symbols so we must undefine them first. */
374 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
375 # define ISASCII(c) 1
377 # define ISASCII(c) isascii(c)
380 /* 1 if C is an ASCII character. */
381 # define IS_REAL_ASCII(c) ((c) < 0200)
383 /* This distinction is not meaningful, except in Emacs. */
384 # define ISUNIBYTE(c) 1
387 # define ISBLANK(c) (ISASCII (c) && isblank (c))
389 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
392 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
394 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
398 # define ISPRINT(c) (ISASCII (c) && isprint (c))
399 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
400 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
401 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
402 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
403 # define ISLOWER(c) (ISASCII (c) && islower (c))
404 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
405 # define ISSPACE(c) (ISASCII (c) && isspace (c))
406 # define ISUPPER(c) (ISASCII (c) && isupper (c))
407 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
409 # define ISWORD(c) ISALPHA(c)
412 # define TOLOWER(c) _tolower(c)
414 # define TOLOWER(c) tolower(c)
417 /* How many characters in the character set. */
418 # define CHAR_SET_SIZE 256
422 extern char *re_syntax_table
;
424 # else /* not SYNTAX_TABLE */
426 static char re_syntax_table
[CHAR_SET_SIZE
];
429 init_syntax_once (void)
437 memset (re_syntax_table
, 0, sizeof re_syntax_table
);
439 for (c
= 0; c
< CHAR_SET_SIZE
; ++c
)
441 re_syntax_table
[c
] = Sword
;
443 re_syntax_table
['_'] = Ssymbol
;
448 # endif /* not SYNTAX_TABLE */
450 # define SYNTAX(c) re_syntax_table[(c)]
452 #endif /* not emacs */
455 # define NULL (void *)0
458 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
459 since ours (we hope) works properly with all combinations of
460 machines, compilers, `char' and `unsigned char' argument types.
461 (Per Bothner suggested the basic approach.) */
462 #undef SIGN_EXTEND_CHAR
464 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
465 #else /* not __STDC__ */
466 /* As in Harbison and Steele. */
467 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
470 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
471 use `alloca' instead of `malloc'. This is because using malloc in
472 re_search* or re_match* could cause memory leaks when C-g is used in
473 Emacs; also, malloc is slower and causes storage fragmentation. On
474 the other hand, malloc is more portable, and easier to debug.
476 Because we sometimes use alloca, some routines have to be macros,
477 not functions -- `alloca'-allocated space disappears at the end of the
478 function it is called in. */
482 # define REGEX_ALLOCATE malloc
483 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
484 # define REGEX_FREE free
486 #else /* not REGEX_MALLOC */
488 /* Emacs already defines alloca, sometimes. */
491 /* Make alloca work the best possible way. */
493 # define alloca __builtin_alloca
494 # else /* not __GNUC__ */
495 # ifdef HAVE_ALLOCA_H
497 # endif /* HAVE_ALLOCA_H */
498 # endif /* not __GNUC__ */
500 # endif /* not alloca */
502 # define REGEX_ALLOCATE alloca
504 /* Assumes a `char *destination' variable. */
505 # define REGEX_REALLOCATE(source, osize, nsize) \
506 (destination = (char *) alloca (nsize), \
507 memcpy (destination, source, osize))
509 /* No need to do anything to free, after alloca. */
510 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
512 #endif /* not REGEX_MALLOC */
514 /* Define how to allocate the failure stack. */
516 #if defined REL_ALLOC && defined REGEX_MALLOC
518 # define REGEX_ALLOCATE_STACK(size) \
519 r_alloc (&failure_stack_ptr, (size))
520 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
521 r_re_alloc (&failure_stack_ptr, (nsize))
522 # define REGEX_FREE_STACK(ptr) \
523 r_alloc_free (&failure_stack_ptr)
525 #else /* not using relocating allocator */
529 # define REGEX_ALLOCATE_STACK malloc
530 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
531 # define REGEX_FREE_STACK free
533 # else /* not REGEX_MALLOC */
535 # define REGEX_ALLOCATE_STACK alloca
537 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
538 REGEX_REALLOCATE (source, osize, nsize)
539 /* No need to explicitly free anything. */
540 # define REGEX_FREE_STACK(arg) ((void)0)
542 # endif /* not REGEX_MALLOC */
543 #endif /* not using relocating allocator */
546 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
547 `string1' or just past its end. This works if PTR is NULL, which is
549 #define FIRST_STRING_P(ptr) \
550 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
552 /* (Re)Allocate N items of type T using malloc, or fail. */
553 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
554 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
555 #define RETALLOC_IF(addr, n, t) \
556 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
557 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
559 #define BYTEWIDTH 8 /* In bits. */
561 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
565 #define MAX(a, b) ((a) > (b) ? (a) : (b))
566 #define MIN(a, b) ((a) < (b) ? (a) : (b))
568 /* Type of source-pattern and string chars. */
569 typedef const unsigned char re_char
;
571 typedef char boolean
;
575 static int re_match_2_internal
_RE_ARGS ((struct re_pattern_buffer
*bufp
,
576 re_char
*string1
, int size1
,
577 re_char
*string2
, int size2
,
579 struct re_registers
*regs
,
582 /* These are the command codes that appear in compiled regular
583 expressions. Some opcodes are followed by argument bytes. A
584 command code can specify any interpretation whatsoever for its
585 arguments. Zero bytes may appear in the compiled regular expression. */
591 /* Succeed right away--no more backtracking. */
594 /* Followed by one byte giving n, then by n literal bytes. */
597 /* Matches any (more or less) character. */
600 /* Matches any one char belonging to specified set. First
601 following byte is number of bitmap bytes. Then come bytes
602 for a bitmap saying which chars are in. Bits in each byte
603 are ordered low-bit-first. A character is in the set if its
604 bit is 1. A character too large to have a bit in the map is
605 automatically not in the set.
607 If the length byte has the 0x80 bit set, then that stuff
608 is followed by a range table:
609 2 bytes of flags for character sets (low 8 bits, high 8 bits)
610 See RANGE_TABLE_WORK_BITS below.
611 2 bytes, the number of pairs that follow (upto 32767)
612 pairs, each 2 multibyte characters,
613 each multibyte character represented as 3 bytes. */
616 /* Same parameters as charset, but match any character that is
617 not one of those specified. */
620 /* Start remembering the text that is matched, for storing in a
621 register. Followed by one byte with the register number, in
622 the range 0 to one less than the pattern buffer's re_nsub
626 /* Stop remembering the text that is matched and store it in a
627 memory register. Followed by one byte with the register
628 number, in the range 0 to one less than `re_nsub' in the
632 /* Match a duplicate of something remembered. Followed by one
633 byte containing the register number. */
636 /* Fail unless at beginning of line. */
639 /* Fail unless at end of line. */
642 /* Succeeds if at beginning of buffer (if emacs) or at beginning
643 of string to be matched (if not). */
646 /* Analogously, for end of buffer/string. */
649 /* Followed by two byte relative address to which to jump. */
652 /* Followed by two-byte relative address of place to resume at
653 in case of failure. */
656 /* Like on_failure_jump, but pushes a placeholder instead of the
657 current string position when executed. */
658 on_failure_keep_string_jump
,
660 /* Just like `on_failure_jump', except that it checks that we
661 don't get stuck in an infinite loop (matching an empty string
663 on_failure_jump_loop
,
665 /* Just like `on_failure_jump_loop', except that it checks for
666 a different kind of loop (the kind that shows up with non-greedy
667 operators). This operation has to be immediately preceded
669 on_failure_jump_nastyloop
,
671 /* A smart `on_failure_jump' used for greedy * and + operators.
672 It analyses the loop before which it is put and if the
673 loop does not require backtracking, it changes itself to
674 `on_failure_keep_string_jump' and short-circuits the loop,
675 else it just defaults to changing itself into `on_failure_jump'.
676 It assumes that it is pointing to just past a `jump'. */
677 on_failure_jump_smart
,
679 /* Followed by two-byte relative address and two-byte number n.
680 After matching N times, jump to the address upon failure.
681 Does not work if N starts at 0: use on_failure_jump_loop
685 /* Followed by two-byte relative address, and two-byte number n.
686 Jump to the address N times, then fail. */
689 /* Set the following two-byte relative address to the
690 subsequent two-byte number. The address *includes* the two
694 wordbeg
, /* Succeeds if at word beginning. */
695 wordend
, /* Succeeds if at word end. */
697 wordbound
, /* Succeeds if at a word boundary. */
698 notwordbound
, /* Succeeds if not at a word boundary. */
700 symbeg
, /* Succeeds if at symbol beginning. */
701 symend
, /* Succeeds if at symbol end. */
703 /* Matches any character whose syntax is specified. Followed by
704 a byte which contains a syntax code, e.g., Sword. */
707 /* Matches any character whose syntax is not that specified. */
711 ,before_dot
, /* Succeeds if before point. */
712 at_dot
, /* Succeeds if at point. */
713 after_dot
, /* Succeeds if after point. */
715 /* Matches any character whose category-set contains the specified
716 category. The operator is followed by a byte which contains a
717 category code (mnemonic ASCII character). */
720 /* Matches any character whose category-set does not contain the
721 specified category. The operator is followed by a byte which
722 contains the category code (mnemonic ASCII character). */
727 /* Common operations on the compiled pattern. */
729 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
731 #define STORE_NUMBER(destination, number) \
733 (destination)[0] = (number) & 0377; \
734 (destination)[1] = (number) >> 8; \
737 /* Same as STORE_NUMBER, except increment DESTINATION to
738 the byte after where the number is stored. Therefore, DESTINATION
739 must be an lvalue. */
741 #define STORE_NUMBER_AND_INCR(destination, number) \
743 STORE_NUMBER (destination, number); \
744 (destination) += 2; \
747 /* Put into DESTINATION a number stored in two contiguous bytes starting
750 #define EXTRACT_NUMBER(destination, source) \
752 (destination) = *(source) & 0377; \
753 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
757 static void extract_number
_RE_ARGS ((int *dest
, re_char
*source
));
759 extract_number (dest
, source
)
763 int temp
= SIGN_EXTEND_CHAR (*(source
+ 1));
764 *dest
= *source
& 0377;
768 # ifndef EXTRACT_MACROS /* To debug the macros. */
769 # undef EXTRACT_NUMBER
770 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
771 # endif /* not EXTRACT_MACROS */
775 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
776 SOURCE must be an lvalue. */
778 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
780 EXTRACT_NUMBER (destination, source); \
785 static void extract_number_and_incr
_RE_ARGS ((int *destination
,
788 extract_number_and_incr (destination
, source
)
792 extract_number (destination
, *source
);
796 # ifndef EXTRACT_MACROS
797 # undef EXTRACT_NUMBER_AND_INCR
798 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
799 extract_number_and_incr (&dest, &src)
800 # endif /* not EXTRACT_MACROS */
804 /* Store a multibyte character in three contiguous bytes starting
805 DESTINATION, and increment DESTINATION to the byte after where the
806 character is stored. Therefore, DESTINATION must be an lvalue. */
808 #define STORE_CHARACTER_AND_INCR(destination, character) \
810 (destination)[0] = (character) & 0377; \
811 (destination)[1] = ((character) >> 8) & 0377; \
812 (destination)[2] = (character) >> 16; \
813 (destination) += 3; \
816 /* Put into DESTINATION a character stored in three contiguous bytes
817 starting at SOURCE. */
819 #define EXTRACT_CHARACTER(destination, source) \
821 (destination) = ((source)[0] \
822 | ((source)[1] << 8) \
823 | ((source)[2] << 16)); \
827 /* Macros for charset. */
829 /* Size of bitmap of charset P in bytes. P is a start of charset,
830 i.e. *P is (re_opcode_t) charset or (re_opcode_t) charset_not. */
831 #define CHARSET_BITMAP_SIZE(p) ((p)[1] & 0x7F)
833 /* Nonzero if charset P has range table. */
834 #define CHARSET_RANGE_TABLE_EXISTS_P(p) ((p)[1] & 0x80)
836 /* Return the address of range table of charset P. But not the start
837 of table itself, but the before where the number of ranges is
838 stored. `2 +' means to skip re_opcode_t and size of bitmap,
839 and the 2 bytes of flags at the start of the range table. */
840 #define CHARSET_RANGE_TABLE(p) (&(p)[4 + CHARSET_BITMAP_SIZE (p)])
842 /* Extract the bit flags that start a range table. */
843 #define CHARSET_RANGE_TABLE_BITS(p) \
844 ((p)[2 + CHARSET_BITMAP_SIZE (p)] \
845 + (p)[3 + CHARSET_BITMAP_SIZE (p)] * 0x100)
847 /* Test if C is listed in the bitmap of charset P. */
848 #define CHARSET_LOOKUP_BITMAP(p, c) \
849 ((c) < CHARSET_BITMAP_SIZE (p) * BYTEWIDTH \
850 && (p)[2 + (c) / BYTEWIDTH] & (1 << ((c) % BYTEWIDTH)))
852 /* Return the address of end of RANGE_TABLE. COUNT is number of
853 ranges (which is a pair of (start, end)) in the RANGE_TABLE. `* 2'
854 is start of range and end of range. `* 3' is size of each start
856 #define CHARSET_RANGE_TABLE_END(range_table, count) \
857 ((range_table) + (count) * 2 * 3)
859 /* Test if C is in RANGE_TABLE. A flag NOT is negated if C is in.
860 COUNT is number of ranges in RANGE_TABLE. */
861 #define CHARSET_LOOKUP_RANGE_TABLE_RAW(not, c, range_table, count) \
864 re_wchar_t range_start, range_end; \
866 re_char *range_table_end \
867 = CHARSET_RANGE_TABLE_END ((range_table), (count)); \
869 for (p = (range_table); p < range_table_end; p += 2 * 3) \
871 EXTRACT_CHARACTER (range_start, p); \
872 EXTRACT_CHARACTER (range_end, p + 3); \
874 if (range_start <= (c) && (c) <= range_end) \
883 /* Test if C is in range table of CHARSET. The flag NOT is negated if
884 C is listed in it. */
885 #define CHARSET_LOOKUP_RANGE_TABLE(not, c, charset) \
888 /* Number of ranges in range table. */ \
890 re_char *range_table = CHARSET_RANGE_TABLE (charset); \
892 EXTRACT_NUMBER_AND_INCR (count, range_table); \
893 CHARSET_LOOKUP_RANGE_TABLE_RAW ((not), (c), range_table, count); \
897 /* If DEBUG is defined, Regex prints many voluminous messages about what
898 it is doing (if the variable `debug' is nonzero). If linked with the
899 main program in `iregex.c', you can enter patterns and strings
900 interactively. And if linked with the main program in `main.c' and
901 the other test files, you can run the already-written tests. */
905 /* We use standard I/O for debugging. */
908 /* It is useful to test things that ``must'' be true when debugging. */
911 static int debug
= -100000;
913 # define DEBUG_STATEMENT(e) e
914 # define DEBUG_PRINT1(x) if (debug > 0) printf (x)
915 # define DEBUG_PRINT2(x1, x2) if (debug > 0) printf (x1, x2)
916 # define DEBUG_PRINT3(x1, x2, x3) if (debug > 0) printf (x1, x2, x3)
917 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug > 0) printf (x1, x2, x3, x4)
918 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
919 if (debug > 0) print_partial_compiled_pattern (s, e)
920 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
921 if (debug > 0) print_double_string (w, s1, sz1, s2, sz2)
924 /* Print the fastmap in human-readable form. */
927 print_fastmap (fastmap
)
930 unsigned was_a_range
= 0;
933 while (i
< (1 << BYTEWIDTH
))
939 while (i
< (1 << BYTEWIDTH
) && fastmap
[i
])
955 /* Print a compiled pattern string in human-readable form, starting at
956 the START pointer into it and ending just before the pointer END. */
959 print_partial_compiled_pattern (start
, end
)
969 fprintf (stderr
, "(null)\n");
973 /* Loop over pattern commands. */
976 fprintf (stderr
, "%d:\t", p
- start
);
978 switch ((re_opcode_t
) *p
++)
981 fprintf (stderr
, "/no_op");
985 fprintf (stderr
, "/succeed");
990 fprintf (stderr
, "/exactn/%d", mcnt
);
993 fprintf (stderr
, "/%c", *p
++);
999 fprintf (stderr
, "/start_memory/%d", *p
++);
1003 fprintf (stderr
, "/stop_memory/%d", *p
++);
1007 fprintf (stderr
, "/duplicate/%d", *p
++);
1011 fprintf (stderr
, "/anychar");
1017 register int c
, last
= -100;
1018 register int in_range
= 0;
1019 int length
= CHARSET_BITMAP_SIZE (p
- 1);
1020 int has_range_table
= CHARSET_RANGE_TABLE_EXISTS_P (p
- 1);
1022 fprintf (stderr
, "/charset [%s",
1023 (re_opcode_t
) *(p
- 1) == charset_not
? "^" : "");
1026 fprintf (stderr
, " !extends past end of pattern! ");
1028 for (c
= 0; c
< 256; c
++)
1030 && (p
[1 + (c
/8)] & (1 << (c
% 8))))
1032 /* Are we starting a range? */
1033 if (last
+ 1 == c
&& ! in_range
)
1035 fprintf (stderr
, "-");
1038 /* Have we broken a range? */
1039 else if (last
+ 1 != c
&& in_range
)
1041 fprintf (stderr
, "%c", last
);
1046 fprintf (stderr
, "%c", c
);
1052 fprintf (stderr
, "%c", last
);
1054 fprintf (stderr
, "]");
1058 if (has_range_table
)
1061 fprintf (stderr
, "has-range-table");
1063 /* ??? Should print the range table; for now, just skip it. */
1064 p
+= 2; /* skip range table bits */
1065 EXTRACT_NUMBER_AND_INCR (count
, p
);
1066 p
= CHARSET_RANGE_TABLE_END (p
, count
);
1072 fprintf (stderr
, "/begline");
1076 fprintf (stderr
, "/endline");
1079 case on_failure_jump
:
1080 extract_number_and_incr (&mcnt
, &p
);
1081 fprintf (stderr
, "/on_failure_jump to %d", p
+ mcnt
- start
);
1084 case on_failure_keep_string_jump
:
1085 extract_number_and_incr (&mcnt
, &p
);
1086 fprintf (stderr
, "/on_failure_keep_string_jump to %d", p
+ mcnt
- start
);
1089 case on_failure_jump_nastyloop
:
1090 extract_number_and_incr (&mcnt
, &p
);
1091 fprintf (stderr
, "/on_failure_jump_nastyloop to %d", p
+ mcnt
- start
);
1094 case on_failure_jump_loop
:
1095 extract_number_and_incr (&mcnt
, &p
);
1096 fprintf (stderr
, "/on_failure_jump_loop to %d", p
+ mcnt
- start
);
1099 case on_failure_jump_smart
:
1100 extract_number_and_incr (&mcnt
, &p
);
1101 fprintf (stderr
, "/on_failure_jump_smart to %d", p
+ mcnt
- start
);
1105 extract_number_and_incr (&mcnt
, &p
);
1106 fprintf (stderr
, "/jump to %d", p
+ mcnt
- start
);
1110 extract_number_and_incr (&mcnt
, &p
);
1111 extract_number_and_incr (&mcnt2
, &p
);
1112 fprintf (stderr
, "/succeed_n to %d, %d times", p
- 2 + mcnt
- start
, mcnt2
);
1116 extract_number_and_incr (&mcnt
, &p
);
1117 extract_number_and_incr (&mcnt2
, &p
);
1118 fprintf (stderr
, "/jump_n to %d, %d times", p
- 2 + mcnt
- start
, mcnt2
);
1122 extract_number_and_incr (&mcnt
, &p
);
1123 extract_number_and_incr (&mcnt2
, &p
);
1124 fprintf (stderr
, "/set_number_at location %d to %d", p
- 2 + mcnt
- start
, mcnt2
);
1128 fprintf (stderr
, "/wordbound");
1132 fprintf (stderr
, "/notwordbound");
1136 fprintf (stderr
, "/wordbeg");
1140 fprintf (stderr
, "/wordend");
1144 fprintf (stderr
, "/symbeg");
1148 fprintf (stderr
, "/symend");
1152 fprintf (stderr
, "/syntaxspec");
1154 fprintf (stderr
, "/%d", mcnt
);
1158 fprintf (stderr
, "/notsyntaxspec");
1160 fprintf (stderr
, "/%d", mcnt
);
1165 fprintf (stderr
, "/before_dot");
1169 fprintf (stderr
, "/at_dot");
1173 fprintf (stderr
, "/after_dot");
1177 fprintf (stderr
, "/categoryspec");
1179 fprintf (stderr
, "/%d", mcnt
);
1182 case notcategoryspec
:
1183 fprintf (stderr
, "/notcategoryspec");
1185 fprintf (stderr
, "/%d", mcnt
);
1190 fprintf (stderr
, "/begbuf");
1194 fprintf (stderr
, "/endbuf");
1198 fprintf (stderr
, "?%d", *(p
-1));
1201 fprintf (stderr
, "\n");
1204 fprintf (stderr
, "%d:\tend of pattern.\n", p
- start
);
1209 print_compiled_pattern (bufp
)
1210 struct re_pattern_buffer
*bufp
;
1212 re_char
*buffer
= bufp
->buffer
;
1214 print_partial_compiled_pattern (buffer
, buffer
+ bufp
->used
);
1215 printf ("%ld bytes used/%ld bytes allocated.\n",
1216 bufp
->used
, bufp
->allocated
);
1218 if (bufp
->fastmap_accurate
&& bufp
->fastmap
)
1220 printf ("fastmap: ");
1221 print_fastmap (bufp
->fastmap
);
1224 printf ("re_nsub: %d\t", bufp
->re_nsub
);
1225 printf ("regs_alloc: %d\t", bufp
->regs_allocated
);
1226 printf ("can_be_null: %d\t", bufp
->can_be_null
);
1227 printf ("no_sub: %d\t", bufp
->no_sub
);
1228 printf ("not_bol: %d\t", bufp
->not_bol
);
1229 printf ("not_eol: %d\t", bufp
->not_eol
);
1230 printf ("syntax: %lx\n", bufp
->syntax
);
1232 /* Perhaps we should print the translate table? */
1237 print_double_string (where
, string1
, size1
, string2
, size2
)
1250 if (FIRST_STRING_P (where
))
1252 for (this_char
= where
- string1
; this_char
< size1
; this_char
++)
1253 putchar (string1
[this_char
]);
1258 for (this_char
= where
- string2
; this_char
< size2
; this_char
++)
1259 putchar (string2
[this_char
]);
1263 #else /* not DEBUG */
1268 # define DEBUG_STATEMENT(e)
1269 # define DEBUG_PRINT1(x)
1270 # define DEBUG_PRINT2(x1, x2)
1271 # define DEBUG_PRINT3(x1, x2, x3)
1272 # define DEBUG_PRINT4(x1, x2, x3, x4)
1273 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1274 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1276 #endif /* not DEBUG */
1278 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1279 also be assigned to arbitrarily: each pattern buffer stores its own
1280 syntax, so it can be changed between regex compilations. */
1281 /* This has no initializer because initialized variables in Emacs
1282 become read-only after dumping. */
1283 reg_syntax_t re_syntax_options
;
1286 /* Specify the precise syntax of regexps for compilation. This provides
1287 for compatibility for various utilities which historically have
1288 different, incompatible syntaxes.
1290 The argument SYNTAX is a bit mask comprised of the various bits
1291 defined in regex.h. We return the old syntax. */
1294 re_set_syntax (reg_syntax_t syntax
)
1296 reg_syntax_t ret
= re_syntax_options
;
1298 re_syntax_options
= syntax
;
1301 WEAK_ALIAS (__re_set_syntax
, re_set_syntax
)
1303 /* Regexp to use to replace spaces, or NULL meaning don't. */
1304 static re_char
*whitespace_regexp
;
1307 re_set_whitespace_regexp (const char *regexp
)
1309 whitespace_regexp
= (re_char
*) regexp
;
1311 WEAK_ALIAS (__re_set_syntax
, re_set_syntax
)
1313 /* This table gives an error message for each of the error codes listed
1314 in regex.h. Obviously the order here has to be same as there.
1315 POSIX doesn't require that we do anything for REG_NOERROR,
1316 but why not be nice? */
1318 static const char *re_error_msgid
[] =
1320 gettext_noop ("Success"), /* REG_NOERROR */
1321 gettext_noop ("No match"), /* REG_NOMATCH */
1322 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
1323 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
1324 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
1325 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
1326 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1327 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1328 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1329 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1330 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1331 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1332 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1333 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1334 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1335 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1336 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1337 gettext_noop ("Range striding over charsets") /* REG_ERANGEX */
1340 /* Avoiding alloca during matching, to placate r_alloc. */
1342 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1343 searching and matching functions should not call alloca. On some
1344 systems, alloca is implemented in terms of malloc, and if we're
1345 using the relocating allocator routines, then malloc could cause a
1346 relocation, which might (if the strings being searched are in the
1347 ralloc heap) shift the data out from underneath the regexp
1350 Here's another reason to avoid allocation: Emacs
1351 processes input from X in a signal handler; processing X input may
1352 call malloc; if input arrives while a matching routine is calling
1353 malloc, then we're scrod. But Emacs can't just block input while
1354 calling matching routines; then we don't notice interrupts when
1355 they come in. So, Emacs blocks input around all regexp calls
1356 except the matching calls, which it leaves unprotected, in the
1357 faith that they will not malloc. */
1359 /* Normally, this is fine. */
1360 #define MATCH_MAY_ALLOCATE
1362 /* The match routines may not allocate if (1) they would do it with malloc
1363 and (2) it's not safe for them to use malloc.
1364 Note that if REL_ALLOC is defined, matching would not use malloc for the
1365 failure stack, but we would still use it for the register vectors;
1366 so REL_ALLOC should not affect this. */
1367 #if defined REGEX_MALLOC && defined emacs
1368 # undef MATCH_MAY_ALLOCATE
1372 /* Failure stack declarations and macros; both re_compile_fastmap and
1373 re_match_2 use a failure stack. These have to be macros because of
1374 REGEX_ALLOCATE_STACK. */
1377 /* Approximate number of failure points for which to initially allocate space
1378 when matching. If this number is exceeded, we allocate more
1379 space, so it is not a hard limit. */
1380 #ifndef INIT_FAILURE_ALLOC
1381 # define INIT_FAILURE_ALLOC 20
1384 /* Roughly the maximum number of failure points on the stack. Would be
1385 exactly that if always used TYPICAL_FAILURE_SIZE items each time we failed.
1386 This is a variable only so users of regex can assign to it; we never
1387 change it ourselves. We always multiply it by TYPICAL_FAILURE_SIZE
1388 before using it, so it should probably be a byte-count instead. */
1389 # if defined MATCH_MAY_ALLOCATE
1390 /* Note that 4400 was enough to cause a crash on Alpha OSF/1,
1391 whose default stack limit is 2mb. In order for a larger
1392 value to work reliably, you have to try to make it accord
1393 with the process stack limit. */
1394 size_t re_max_failures
= 40000;
1396 size_t re_max_failures
= 4000;
1399 union fail_stack_elt
1402 /* This should be the biggest `int' that's no bigger than a pointer. */
1406 typedef union fail_stack_elt fail_stack_elt_t
;
1410 fail_stack_elt_t
*stack
;
1412 size_t avail
; /* Offset of next open position. */
1413 size_t frame
; /* Offset of the cur constructed frame. */
1416 #define FAIL_STACK_EMPTY() (fail_stack.frame == 0)
1417 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1420 /* Define macros to initialize and free the failure stack.
1421 Do `return -2' if the alloc fails. */
1423 #ifdef MATCH_MAY_ALLOCATE
1424 # define INIT_FAIL_STACK() \
1426 fail_stack.stack = (fail_stack_elt_t *) \
1427 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * TYPICAL_FAILURE_SIZE \
1428 * sizeof (fail_stack_elt_t)); \
1430 if (fail_stack.stack == NULL) \
1433 fail_stack.size = INIT_FAILURE_ALLOC; \
1434 fail_stack.avail = 0; \
1435 fail_stack.frame = 0; \
1438 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1440 # define INIT_FAIL_STACK() \
1442 fail_stack.avail = 0; \
1443 fail_stack.frame = 0; \
1446 # define RESET_FAIL_STACK() ((void)0)
1450 /* Double the size of FAIL_STACK, up to a limit
1451 which allows approximately `re_max_failures' items.
1453 Return 1 if succeeds, and 0 if either ran out of memory
1454 allocating space for it or it was already too large.
1456 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1458 /* Factor to increase the failure stack size by
1459 when we increase it.
1460 This used to be 2, but 2 was too wasteful
1461 because the old discarded stacks added up to as much space
1462 were as ultimate, maximum-size stack. */
1463 #define FAIL_STACK_GROWTH_FACTOR 4
1465 #define GROW_FAIL_STACK(fail_stack) \
1466 (((fail_stack).size * sizeof (fail_stack_elt_t) \
1467 >= re_max_failures * TYPICAL_FAILURE_SIZE) \
1469 : ((fail_stack).stack \
1470 = (fail_stack_elt_t *) \
1471 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1472 (fail_stack).size * sizeof (fail_stack_elt_t), \
1473 MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
1474 ((fail_stack).size * sizeof (fail_stack_elt_t) \
1475 * FAIL_STACK_GROWTH_FACTOR))), \
1477 (fail_stack).stack == NULL \
1479 : ((fail_stack).size \
1480 = (MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
1481 ((fail_stack).size * sizeof (fail_stack_elt_t) \
1482 * FAIL_STACK_GROWTH_FACTOR)) \
1483 / sizeof (fail_stack_elt_t)), \
1487 /* Push a pointer value onto the failure stack.
1488 Assumes the variable `fail_stack'. Probably should only
1489 be called from within `PUSH_FAILURE_POINT'. */
1490 #define PUSH_FAILURE_POINTER(item) \
1491 fail_stack.stack[fail_stack.avail++].pointer = (item)
1493 /* This pushes an integer-valued item onto the failure stack.
1494 Assumes the variable `fail_stack'. Probably should only
1495 be called from within `PUSH_FAILURE_POINT'. */
1496 #define PUSH_FAILURE_INT(item) \
1497 fail_stack.stack[fail_stack.avail++].integer = (item)
1499 /* Push a fail_stack_elt_t value onto the failure stack.
1500 Assumes the variable `fail_stack'. Probably should only
1501 be called from within `PUSH_FAILURE_POINT'. */
1502 #define PUSH_FAILURE_ELT(item) \
1503 fail_stack.stack[fail_stack.avail++] = (item)
1505 /* These three POP... operations complement the three PUSH... operations.
1506 All assume that `fail_stack' is nonempty. */
1507 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1508 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1509 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1511 /* Individual items aside from the registers. */
1512 #define NUM_NONREG_ITEMS 3
1514 /* Used to examine the stack (to detect infinite loops). */
1515 #define FAILURE_PAT(h) fail_stack.stack[(h) - 1].pointer
1516 #define FAILURE_STR(h) (fail_stack.stack[(h) - 2].pointer)
1517 #define NEXT_FAILURE_HANDLE(h) fail_stack.stack[(h) - 3].integer
1518 #define TOP_FAILURE_HANDLE() fail_stack.frame
1521 #define ENSURE_FAIL_STACK(space) \
1522 while (REMAINING_AVAIL_SLOTS <= space) { \
1523 if (!GROW_FAIL_STACK (fail_stack)) \
1525 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", (fail_stack).size);\
1526 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1529 /* Push register NUM onto the stack. */
1530 #define PUSH_FAILURE_REG(num) \
1532 char *destination; \
1533 ENSURE_FAIL_STACK(3); \
1534 DEBUG_PRINT4 (" Push reg %d (spanning %p -> %p)\n", \
1535 num, regstart[num], regend[num]); \
1536 PUSH_FAILURE_POINTER (regstart[num]); \
1537 PUSH_FAILURE_POINTER (regend[num]); \
1538 PUSH_FAILURE_INT (num); \
1541 /* Change the counter's value to VAL, but make sure that it will
1542 be reset when backtracking. */
1543 #define PUSH_NUMBER(ptr,val) \
1545 char *destination; \
1547 ENSURE_FAIL_STACK(3); \
1548 EXTRACT_NUMBER (c, ptr); \
1549 DEBUG_PRINT4 (" Push number %p = %d -> %d\n", ptr, c, val); \
1550 PUSH_FAILURE_INT (c); \
1551 PUSH_FAILURE_POINTER (ptr); \
1552 PUSH_FAILURE_INT (-1); \
1553 STORE_NUMBER (ptr, val); \
1556 /* Pop a saved register off the stack. */
1557 #define POP_FAILURE_REG_OR_COUNT() \
1559 int reg = POP_FAILURE_INT (); \
1562 /* It's a counter. */ \
1563 /* Here, we discard `const', making re_match non-reentrant. */ \
1564 unsigned char *ptr = (unsigned char*) POP_FAILURE_POINTER (); \
1565 reg = POP_FAILURE_INT (); \
1566 STORE_NUMBER (ptr, reg); \
1567 DEBUG_PRINT3 (" Pop counter %p = %d\n", ptr, reg); \
1571 regend[reg] = POP_FAILURE_POINTER (); \
1572 regstart[reg] = POP_FAILURE_POINTER (); \
1573 DEBUG_PRINT4 (" Pop reg %d (spanning %p -> %p)\n", \
1574 reg, regstart[reg], regend[reg]); \
1578 /* Check that we are not stuck in an infinite loop. */
1579 #define CHECK_INFINITE_LOOP(pat_cur, string_place) \
1581 int failure = TOP_FAILURE_HANDLE (); \
1582 /* Check for infinite matching loops */ \
1583 while (failure > 0 \
1584 && (FAILURE_STR (failure) == string_place \
1585 || FAILURE_STR (failure) == NULL)) \
1587 assert (FAILURE_PAT (failure) >= bufp->buffer \
1588 && FAILURE_PAT (failure) <= bufp->buffer + bufp->used); \
1589 if (FAILURE_PAT (failure) == pat_cur) \
1594 DEBUG_PRINT2 (" Other pattern: %p\n", FAILURE_PAT (failure)); \
1595 failure = NEXT_FAILURE_HANDLE(failure); \
1597 DEBUG_PRINT2 (" Other string: %p\n", FAILURE_STR (failure)); \
1600 /* Push the information about the state we will need
1601 if we ever fail back to it.
1603 Requires variables fail_stack, regstart, regend and
1604 num_regs be declared. GROW_FAIL_STACK requires `destination' be
1607 Does `return FAILURE_CODE' if runs out of memory. */
1609 #define PUSH_FAILURE_POINT(pattern, string_place) \
1611 char *destination; \
1612 /* Must be int, so when we don't save any registers, the arithmetic \
1613 of 0 + -1 isn't done as unsigned. */ \
1615 DEBUG_STATEMENT (nfailure_points_pushed++); \
1616 DEBUG_PRINT1 ("\nPUSH_FAILURE_POINT:\n"); \
1617 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail); \
1618 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1620 ENSURE_FAIL_STACK (NUM_NONREG_ITEMS); \
1622 DEBUG_PRINT1 ("\n"); \
1624 DEBUG_PRINT2 (" Push frame index: %d\n", fail_stack.frame); \
1625 PUSH_FAILURE_INT (fail_stack.frame); \
1627 DEBUG_PRINT2 (" Push string %p: `", string_place); \
1628 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, size2);\
1629 DEBUG_PRINT1 ("'\n"); \
1630 PUSH_FAILURE_POINTER (string_place); \
1632 DEBUG_PRINT2 (" Push pattern %p: ", pattern); \
1633 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern, pend); \
1634 PUSH_FAILURE_POINTER (pattern); \
1636 /* Close the frame by moving the frame pointer past it. */ \
1637 fail_stack.frame = fail_stack.avail; \
1640 /* Estimate the size of data pushed by a typical failure stack entry.
1641 An estimate is all we need, because all we use this for
1642 is to choose a limit for how big to make the failure stack. */
1643 /* BEWARE, the value `20' is hard-coded in emacs.c:main(). */
1644 #define TYPICAL_FAILURE_SIZE 20
1646 /* How many items can still be added to the stack without overflowing it. */
1647 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1650 /* Pops what PUSH_FAIL_STACK pushes.
1652 We restore into the parameters, all of which should be lvalues:
1653 STR -- the saved data position.
1654 PAT -- the saved pattern position.
1655 REGSTART, REGEND -- arrays of string positions.
1657 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1658 `pend', `string1', `size1', `string2', and `size2'. */
1660 #define POP_FAILURE_POINT(str, pat) \
1662 assert (!FAIL_STACK_EMPTY ()); \
1664 /* Remove failure points and point to how many regs pushed. */ \
1665 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1666 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1667 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1669 /* Pop the saved registers. */ \
1670 while (fail_stack.frame < fail_stack.avail) \
1671 POP_FAILURE_REG_OR_COUNT (); \
1673 pat = POP_FAILURE_POINTER (); \
1674 DEBUG_PRINT2 (" Popping pattern %p: ", pat); \
1675 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1677 /* If the saved string location is NULL, it came from an \
1678 on_failure_keep_string_jump opcode, and we want to throw away the \
1679 saved NULL, thus retaining our current position in the string. */ \
1680 str = POP_FAILURE_POINTER (); \
1681 DEBUG_PRINT2 (" Popping string %p: `", str); \
1682 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1683 DEBUG_PRINT1 ("'\n"); \
1685 fail_stack.frame = POP_FAILURE_INT (); \
1686 DEBUG_PRINT2 (" Popping frame index: %d\n", fail_stack.frame); \
1688 assert (fail_stack.avail >= 0); \
1689 assert (fail_stack.frame <= fail_stack.avail); \
1691 DEBUG_STATEMENT (nfailure_points_popped++); \
1692 } while (0) /* POP_FAILURE_POINT */
1696 /* Registers are set to a sentinel when they haven't yet matched. */
1697 #define REG_UNSET(e) ((e) == NULL)
1699 /* Subroutine declarations and macros for regex_compile. */
1701 static reg_errcode_t regex_compile
_RE_ARGS ((re_char
*pattern
, size_t size
,
1702 reg_syntax_t syntax
,
1703 struct re_pattern_buffer
*bufp
));
1704 static void store_op1
_RE_ARGS ((re_opcode_t op
, unsigned char *loc
, int arg
));
1705 static void store_op2
_RE_ARGS ((re_opcode_t op
, unsigned char *loc
,
1706 int arg1
, int arg2
));
1707 static void insert_op1
_RE_ARGS ((re_opcode_t op
, unsigned char *loc
,
1708 int arg
, unsigned char *end
));
1709 static void insert_op2
_RE_ARGS ((re_opcode_t op
, unsigned char *loc
,
1710 int arg1
, int arg2
, unsigned char *end
));
1711 static boolean at_begline_loc_p
_RE_ARGS ((re_char
*pattern
,
1713 reg_syntax_t syntax
));
1714 static boolean at_endline_loc_p
_RE_ARGS ((re_char
*p
,
1716 reg_syntax_t syntax
));
1717 static re_char
*skip_one_char
_RE_ARGS ((re_char
*p
));
1718 static int analyse_first
_RE_ARGS ((re_char
*p
, re_char
*pend
,
1719 char *fastmap
, const int multibyte
));
1721 /* Fetch the next character in the uncompiled pattern, with no
1723 #define PATFETCH(c) \
1726 if (p == pend) return REG_EEND; \
1727 c = RE_STRING_CHAR_AND_LENGTH (p, len, multibyte); \
1732 /* If `translate' is non-null, return translate[D], else just D. We
1733 cast the subscript to translate because some data is declared as
1734 `char *', to avoid warnings when a string constant is passed. But
1735 when we use a character as a subscript we must make it unsigned. */
1737 # define TRANSLATE(d) \
1738 (RE_TRANSLATE_P (translate) ? RE_TRANSLATE (translate, (d)) : (d))
1742 /* Macros for outputting the compiled pattern into `buffer'. */
1744 /* If the buffer isn't allocated when it comes in, use this. */
1745 #define INIT_BUF_SIZE 32
1747 /* Make sure we have at least N more bytes of space in buffer. */
1748 #define GET_BUFFER_SPACE(n) \
1749 while ((size_t) (b - bufp->buffer + (n)) > bufp->allocated) \
1752 /* Make sure we have one more byte of buffer space and then add C to it. */
1753 #define BUF_PUSH(c) \
1755 GET_BUFFER_SPACE (1); \
1756 *b++ = (unsigned char) (c); \
1760 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1761 #define BUF_PUSH_2(c1, c2) \
1763 GET_BUFFER_SPACE (2); \
1764 *b++ = (unsigned char) (c1); \
1765 *b++ = (unsigned char) (c2); \
1769 /* As with BUF_PUSH_2, except for three bytes. */
1770 #define BUF_PUSH_3(c1, c2, c3) \
1772 GET_BUFFER_SPACE (3); \
1773 *b++ = (unsigned char) (c1); \
1774 *b++ = (unsigned char) (c2); \
1775 *b++ = (unsigned char) (c3); \
1779 /* Store a jump with opcode OP at LOC to location TO. We store a
1780 relative address offset by the three bytes the jump itself occupies. */
1781 #define STORE_JUMP(op, loc, to) \
1782 store_op1 (op, loc, (to) - (loc) - 3)
1784 /* Likewise, for a two-argument jump. */
1785 #define STORE_JUMP2(op, loc, to, arg) \
1786 store_op2 (op, loc, (to) - (loc) - 3, arg)
1788 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1789 #define INSERT_JUMP(op, loc, to) \
1790 insert_op1 (op, loc, (to) - (loc) - 3, b)
1792 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1793 #define INSERT_JUMP2(op, loc, to, arg) \
1794 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1797 /* This is not an arbitrary limit: the arguments which represent offsets
1798 into the pattern are two bytes long. So if 2^15 bytes turns out to
1799 be too small, many things would have to change. */
1800 # define MAX_BUF_SIZE (1L << 15)
1802 #if 0 /* This is when we thought it could be 2^16 bytes. */
1803 /* Any other compiler which, like MSC, has allocation limit below 2^16
1804 bytes will have to use approach similar to what was done below for
1805 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1806 reallocating to 0 bytes. Such thing is not going to work too well.
1807 You have been warned!! */
1808 #if defined _MSC_VER && !defined WIN32
1809 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes. */
1810 # define MAX_BUF_SIZE 65500L
1812 # define MAX_BUF_SIZE (1L << 16)
1816 /* Extend the buffer by twice its current size via realloc and
1817 reset the pointers that pointed into the old block to point to the
1818 correct places in the new one. If extending the buffer results in it
1819 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1820 #if __BOUNDED_POINTERS__
1821 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
1822 # define MOVE_BUFFER_POINTER(P) \
1823 (__ptrlow (P) = new_buffer + (__ptrlow (P) - old_buffer), \
1824 SET_HIGH_BOUND (P), \
1825 __ptrvalue (P) = new_buffer + (__ptrvalue (P) - old_buffer))
1826 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
1829 SET_HIGH_BOUND (b); \
1830 SET_HIGH_BOUND (begalt); \
1831 if (fixup_alt_jump) \
1832 SET_HIGH_BOUND (fixup_alt_jump); \
1834 SET_HIGH_BOUND (laststart); \
1835 if (pending_exact) \
1836 SET_HIGH_BOUND (pending_exact); \
1839 # define MOVE_BUFFER_POINTER(P) ((P) = new_buffer + ((P) - old_buffer))
1840 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
1842 #define EXTEND_BUFFER() \
1844 unsigned char *old_buffer = bufp->buffer; \
1845 if (bufp->allocated == MAX_BUF_SIZE) \
1847 bufp->allocated <<= 1; \
1848 if (bufp->allocated > MAX_BUF_SIZE) \
1849 bufp->allocated = MAX_BUF_SIZE; \
1850 RETALLOC (bufp->buffer, bufp->allocated, unsigned char); \
1851 if (bufp->buffer == NULL) \
1852 return REG_ESPACE; \
1853 /* If the buffer moved, move all the pointers into it. */ \
1854 if (old_buffer != bufp->buffer) \
1856 unsigned char *new_buffer = bufp->buffer; \
1857 MOVE_BUFFER_POINTER (b); \
1858 MOVE_BUFFER_POINTER (begalt); \
1859 if (fixup_alt_jump) \
1860 MOVE_BUFFER_POINTER (fixup_alt_jump); \
1862 MOVE_BUFFER_POINTER (laststart); \
1863 if (pending_exact) \
1864 MOVE_BUFFER_POINTER (pending_exact); \
1866 ELSE_EXTEND_BUFFER_HIGH_BOUND \
1870 /* Since we have one byte reserved for the register number argument to
1871 {start,stop}_memory, the maximum number of groups we can report
1872 things about is what fits in that byte. */
1873 #define MAX_REGNUM 255
1875 /* But patterns can have more than `MAX_REGNUM' registers. We just
1876 ignore the excess. */
1877 typedef int regnum_t
;
1880 /* Macros for the compile stack. */
1882 /* Since offsets can go either forwards or backwards, this type needs to
1883 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1884 /* int may be not enough when sizeof(int) == 2. */
1885 typedef long pattern_offset_t
;
1889 pattern_offset_t begalt_offset
;
1890 pattern_offset_t fixup_alt_jump
;
1891 pattern_offset_t laststart_offset
;
1893 } compile_stack_elt_t
;
1898 compile_stack_elt_t
*stack
;
1900 unsigned avail
; /* Offset of next open position. */
1901 } compile_stack_type
;
1904 #define INIT_COMPILE_STACK_SIZE 32
1906 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1907 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1909 /* The next available element. */
1910 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1912 /* Explicit quit checking is only used on NTemacs and whenever we
1913 use polling to process input events. */
1914 #if defined emacs && (defined WINDOWSNT || defined SYNC_INPUT) && defined QUIT
1915 extern int immediate_quit
;
1916 # define IMMEDIATE_QUIT_CHECK \
1918 if (immediate_quit) QUIT; \
1921 # define IMMEDIATE_QUIT_CHECK ((void)0)
1924 /* Structure to manage work area for range table. */
1925 struct range_table_work_area
1927 int *table
; /* actual work area. */
1928 int allocated
; /* allocated size for work area in bytes. */
1929 int used
; /* actually used size in words. */
1930 int bits
; /* flag to record character classes */
1933 /* Make sure that WORK_AREA can hold more N multibyte characters.
1934 This is used only in set_image_of_range and set_image_of_range_1.
1935 It expects WORK_AREA to be a pointer.
1936 If it can't get the space, it returns from the surrounding function. */
1938 #define EXTEND_RANGE_TABLE(work_area, n) \
1940 if (((work_area).used + (n)) * sizeof (int) > (work_area).allocated) \
1942 extend_range_table_work_area (&work_area); \
1943 if ((work_area).table == 0) \
1944 return (REG_ESPACE); \
1948 #define SET_RANGE_TABLE_WORK_AREA_BIT(work_area, bit) \
1949 (work_area).bits |= (bit)
1951 /* Bits used to implement the multibyte-part of the various character classes
1952 such as [:alnum:] in a charset's range table. */
1953 #define BIT_WORD 0x1
1954 #define BIT_LOWER 0x2
1955 #define BIT_PUNCT 0x4
1956 #define BIT_SPACE 0x8
1957 #define BIT_UPPER 0x10
1958 #define BIT_MULTIBYTE 0x20
1960 /* Set a range (RANGE_START, RANGE_END) to WORK_AREA. */
1961 #define SET_RANGE_TABLE_WORK_AREA(work_area, range_start, range_end) \
1963 EXTEND_RANGE_TABLE ((work_area), 2); \
1964 (work_area).table[(work_area).used++] = (range_start); \
1965 (work_area).table[(work_area).used++] = (range_end); \
1968 /* Free allocated memory for WORK_AREA. */
1969 #define FREE_RANGE_TABLE_WORK_AREA(work_area) \
1971 if ((work_area).table) \
1972 free ((work_area).table); \
1975 #define CLEAR_RANGE_TABLE_WORK_USED(work_area) ((work_area).used = 0, (work_area).bits = 0)
1976 #define RANGE_TABLE_WORK_USED(work_area) ((work_area).used)
1977 #define RANGE_TABLE_WORK_BITS(work_area) ((work_area).bits)
1978 #define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i])
1981 /* Set the bit for character C in a list. */
1982 #define SET_LIST_BIT(c) (b[((c)) / BYTEWIDTH] |= 1 << ((c) % BYTEWIDTH))
1987 /* Store characters in the range FROM to TO in the bitmap at B (for
1988 ASCII and unibyte characters) and WORK_AREA (for multibyte
1989 characters) while translating them and paying attention to the
1990 continuity of translated characters.
1992 Implementation note: It is better to implement these fairly big
1993 macros by a function, but it's not that easy because macros called
1994 in this macro assume various local variables already declared. */
1996 /* Both FROM and TO are ASCII characters. */
1998 #define SETUP_ASCII_RANGE(work_area, FROM, TO) \
2002 for (C0 = (FROM); C0 <= (TO); C0++) \
2004 C1 = TRANSLATE (C0); \
2005 if (! ASCII_CHAR_P (C1)) \
2007 SET_RANGE_TABLE_WORK_AREA ((work_area), C1, C1); \
2008 if ((C1 = RE_CHAR_TO_UNIBYTE (C1)) < 0) \
2011 SET_LIST_BIT (C1); \
2016 /* Both FROM and TO are unibyte characters (0x80..0xFF). */
2018 #define SETUP_UNIBYTE_RANGE(work_area, FROM, TO) \
2020 int C0, C1, C2, I; \
2021 int USED = RANGE_TABLE_WORK_USED (work_area); \
2023 for (C0 = (FROM); C0 <= (TO); C0++) \
2025 C1 = RE_CHAR_TO_MULTIBYTE (C0); \
2026 if (CHAR_BYTE8_P (C1)) \
2027 SET_LIST_BIT (C0); \
2030 C2 = TRANSLATE (C1); \
2032 || (C1 = RE_CHAR_TO_UNIBYTE (C2)) < 0) \
2034 SET_LIST_BIT (C1); \
2035 for (I = RANGE_TABLE_WORK_USED (work_area) - 2; I >= USED; I -= 2) \
2037 int from = RANGE_TABLE_WORK_ELT (work_area, I); \
2038 int to = RANGE_TABLE_WORK_ELT (work_area, I + 1); \
2040 if (C2 >= from - 1 && C2 <= to + 1) \
2042 if (C2 == from - 1) \
2043 RANGE_TABLE_WORK_ELT (work_area, I)--; \
2044 else if (C2 == to + 1) \
2045 RANGE_TABLE_WORK_ELT (work_area, I + 1)++; \
2050 SET_RANGE_TABLE_WORK_AREA ((work_area), C2, C2); \
2056 /* Both FROM and TO are multibyte characters. */
2058 #define SETUP_MULTIBYTE_RANGE(work_area, FROM, TO) \
2060 int C0, C1, C2, I, USED = RANGE_TABLE_WORK_USED (work_area); \
2062 SET_RANGE_TABLE_WORK_AREA ((work_area), (FROM), (TO)); \
2063 for (C0 = (FROM); C0 <= (TO); C0++) \
2065 C1 = TRANSLATE (C0); \
2066 if ((C2 = RE_CHAR_TO_UNIBYTE (C1)) >= 0 \
2067 || (C1 != C0 && (C2 = RE_CHAR_TO_UNIBYTE (C0)) >= 0)) \
2068 SET_LIST_BIT (C2); \
2069 if (C1 >= (FROM) && C1 <= (TO)) \
2071 for (I = RANGE_TABLE_WORK_USED (work_area) - 2; I >= USED; I -= 2) \
2073 int from = RANGE_TABLE_WORK_ELT (work_area, I); \
2074 int to = RANGE_TABLE_WORK_ELT (work_area, I + 1); \
2076 if (C1 >= from - 1 && C1 <= to + 1) \
2078 if (C1 == from - 1) \
2079 RANGE_TABLE_WORK_ELT (work_area, I)--; \
2080 else if (C1 == to + 1) \
2081 RANGE_TABLE_WORK_ELT (work_area, I + 1)++; \
2086 SET_RANGE_TABLE_WORK_AREA ((work_area), C1, C1); \
2092 /* Get the next unsigned number in the uncompiled pattern. */
2093 #define GET_UNSIGNED_NUMBER(num) \
2096 FREE_STACK_RETURN (REG_EBRACE); \
2100 while ('0' <= c && c <= '9') \
2106 num = num * 10 + c - '0'; \
2107 if (num / 10 != prev) \
2108 FREE_STACK_RETURN (REG_BADBR); \
2110 FREE_STACK_RETURN (REG_EBRACE); \
2116 #if ! WIDE_CHAR_SUPPORT
2118 /* Map a string to the char class it names (if any). */
2120 re_wctype (const re_char
*str
)
2122 const char *string
= str
;
2123 if (STREQ (string
, "alnum")) return RECC_ALNUM
;
2124 else if (STREQ (string
, "alpha")) return RECC_ALPHA
;
2125 else if (STREQ (string
, "word")) return RECC_WORD
;
2126 else if (STREQ (string
, "ascii")) return RECC_ASCII
;
2127 else if (STREQ (string
, "nonascii")) return RECC_NONASCII
;
2128 else if (STREQ (string
, "graph")) return RECC_GRAPH
;
2129 else if (STREQ (string
, "lower")) return RECC_LOWER
;
2130 else if (STREQ (string
, "print")) return RECC_PRINT
;
2131 else if (STREQ (string
, "punct")) return RECC_PUNCT
;
2132 else if (STREQ (string
, "space")) return RECC_SPACE
;
2133 else if (STREQ (string
, "upper")) return RECC_UPPER
;
2134 else if (STREQ (string
, "unibyte")) return RECC_UNIBYTE
;
2135 else if (STREQ (string
, "multibyte")) return RECC_MULTIBYTE
;
2136 else if (STREQ (string
, "digit")) return RECC_DIGIT
;
2137 else if (STREQ (string
, "xdigit")) return RECC_XDIGIT
;
2138 else if (STREQ (string
, "cntrl")) return RECC_CNTRL
;
2139 else if (STREQ (string
, "blank")) return RECC_BLANK
;
2143 /* True if CH is in the char class CC. */
2145 re_iswctype (int ch
, re_wctype_t cc
)
2149 case RECC_ALNUM
: return ISALNUM (ch
);
2150 case RECC_ALPHA
: return ISALPHA (ch
);
2151 case RECC_BLANK
: return ISBLANK (ch
);
2152 case RECC_CNTRL
: return ISCNTRL (ch
);
2153 case RECC_DIGIT
: return ISDIGIT (ch
);
2154 case RECC_GRAPH
: return ISGRAPH (ch
);
2155 case RECC_LOWER
: return ISLOWER (ch
);
2156 case RECC_PRINT
: return ISPRINT (ch
);
2157 case RECC_PUNCT
: return ISPUNCT (ch
);
2158 case RECC_SPACE
: return ISSPACE (ch
);
2159 case RECC_UPPER
: return ISUPPER (ch
);
2160 case RECC_XDIGIT
: return ISXDIGIT (ch
);
2161 case RECC_ASCII
: return IS_REAL_ASCII (ch
);
2162 case RECC_NONASCII
: return !IS_REAL_ASCII (ch
);
2163 case RECC_UNIBYTE
: return ISUNIBYTE (ch
);
2164 case RECC_MULTIBYTE
: return !ISUNIBYTE (ch
);
2165 case RECC_WORD
: return ISWORD (ch
);
2166 case RECC_ERROR
: return false;
2172 /* Return a bit-pattern to use in the range-table bits to match multibyte
2173 chars of class CC. */
2175 re_wctype_to_bit (re_wctype_t cc
)
2179 case RECC_NONASCII
: case RECC_PRINT
: case RECC_GRAPH
:
2180 case RECC_MULTIBYTE
: return BIT_MULTIBYTE
;
2181 case RECC_ALPHA
: case RECC_ALNUM
: case RECC_WORD
: return BIT_WORD
;
2182 case RECC_LOWER
: return BIT_LOWER
;
2183 case RECC_UPPER
: return BIT_UPPER
;
2184 case RECC_PUNCT
: return BIT_PUNCT
;
2185 case RECC_SPACE
: return BIT_SPACE
;
2186 case RECC_ASCII
: case RECC_DIGIT
: case RECC_XDIGIT
: case RECC_CNTRL
:
2187 case RECC_BLANK
: case RECC_UNIBYTE
: case RECC_ERROR
: return 0;
2194 /* Filling in the work area of a range. */
2196 /* Actually extend the space in WORK_AREA. */
2199 extend_range_table_work_area (struct range_table_work_area
*work_area
)
2201 work_area
->allocated
+= 16 * sizeof (int);
2202 if (work_area
->table
)
2204 = (int *) realloc (work_area
->table
, work_area
->allocated
);
2207 = (int *) malloc (work_area
->allocated
);
2213 /* Carefully find the ranges of codes that are equivalent
2214 under case conversion to the range start..end when passed through
2215 TRANSLATE. Handle the case where non-letters can come in between
2216 two upper-case letters (which happens in Latin-1).
2217 Also handle the case of groups of more than 2 case-equivalent chars.
2219 The basic method is to look at consecutive characters and see
2220 if they can form a run that can be handled as one.
2222 Returns -1 if successful, REG_ESPACE if ran out of space. */
2225 set_image_of_range_1 (work_area
, start
, end
, translate
)
2226 RE_TRANSLATE_TYPE translate
;
2227 struct range_table_work_area
*work_area
;
2228 re_wchar_t start
, end
;
2230 /* `one_case' indicates a character, or a run of characters,
2231 each of which is an isolate (no case-equivalents).
2232 This includes all ASCII non-letters.
2234 `two_case' indicates a character, or a run of characters,
2235 each of which has two case-equivalent forms.
2236 This includes all ASCII letters.
2238 `strange' indicates a character that has more than one
2241 enum case_type
{one_case
, two_case
, strange
};
2243 /* Describe the run that is in progress,
2244 which the next character can try to extend.
2245 If run_type is strange, that means there really is no run.
2246 If run_type is one_case, then run_start...run_end is the run.
2247 If run_type is two_case, then the run is run_start...run_end,
2248 and the case-equivalents end at run_eqv_end. */
2250 enum case_type run_type
= strange
;
2251 int run_start
, run_end
, run_eqv_end
;
2253 Lisp_Object eqv_table
;
2255 if (!RE_TRANSLATE_P (translate
))
2257 EXTEND_RANGE_TABLE (work_area
, 2);
2258 work_area
->table
[work_area
->used
++] = (start
);
2259 work_area
->table
[work_area
->used
++] = (end
);
2263 eqv_table
= XCHAR_TABLE (translate
)->extras
[2];
2265 for (; start
<= end
; start
++)
2267 enum case_type this_type
;
2268 int eqv
= RE_TRANSLATE (eqv_table
, start
);
2269 int minchar
, maxchar
;
2271 /* Classify this character */
2273 this_type
= one_case
;
2274 else if (RE_TRANSLATE (eqv_table
, eqv
) == start
)
2275 this_type
= two_case
;
2277 this_type
= strange
;
2280 minchar
= start
, maxchar
= eqv
;
2282 minchar
= eqv
, maxchar
= start
;
2284 /* Can this character extend the run in progress? */
2285 if (this_type
== strange
|| this_type
!= run_type
2286 || !(minchar
== run_end
+ 1
2287 && (run_type
== two_case
2288 ? maxchar
== run_eqv_end
+ 1 : 1)))
2291 Record each of its equivalent ranges. */
2292 if (run_type
== one_case
)
2294 EXTEND_RANGE_TABLE (work_area
, 2);
2295 work_area
->table
[work_area
->used
++] = run_start
;
2296 work_area
->table
[work_area
->used
++] = run_end
;
2298 else if (run_type
== two_case
)
2300 EXTEND_RANGE_TABLE (work_area
, 4);
2301 work_area
->table
[work_area
->used
++] = run_start
;
2302 work_area
->table
[work_area
->used
++] = run_end
;
2303 work_area
->table
[work_area
->used
++]
2304 = RE_TRANSLATE (eqv_table
, run_start
);
2305 work_area
->table
[work_area
->used
++]
2306 = RE_TRANSLATE (eqv_table
, run_end
);
2311 if (this_type
== strange
)
2313 /* For a strange character, add each of its equivalents, one
2314 by one. Don't start a range. */
2317 EXTEND_RANGE_TABLE (work_area
, 2);
2318 work_area
->table
[work_area
->used
++] = eqv
;
2319 work_area
->table
[work_area
->used
++] = eqv
;
2320 eqv
= RE_TRANSLATE (eqv_table
, eqv
);
2322 while (eqv
!= start
);
2325 /* Add this char to the run, or start a new run. */
2326 else if (run_type
== strange
)
2328 /* Initialize a new range. */
2329 run_type
= this_type
;
2332 run_eqv_end
= RE_TRANSLATE (eqv_table
, run_end
);
2336 /* Extend a running range. */
2338 run_eqv_end
= RE_TRANSLATE (eqv_table
, run_end
);
2342 /* If a run is still in progress at the end, finish it now
2343 by recording its equivalent ranges. */
2344 if (run_type
== one_case
)
2346 EXTEND_RANGE_TABLE (work_area
, 2);
2347 work_area
->table
[work_area
->used
++] = run_start
;
2348 work_area
->table
[work_area
->used
++] = run_end
;
2350 else if (run_type
== two_case
)
2352 EXTEND_RANGE_TABLE (work_area
, 4);
2353 work_area
->table
[work_area
->used
++] = run_start
;
2354 work_area
->table
[work_area
->used
++] = run_end
;
2355 work_area
->table
[work_area
->used
++]
2356 = RE_TRANSLATE (eqv_table
, run_start
);
2357 work_area
->table
[work_area
->used
++]
2358 = RE_TRANSLATE (eqv_table
, run_end
);
2366 /* Record the image of the range start..end when passed through
2367 TRANSLATE. This is not necessarily TRANSLATE(start)..TRANSLATE(end)
2368 and is not even necessarily contiguous.
2369 Normally we approximate it with the smallest contiguous range that contains
2370 all the chars we need. However, for Latin-1 we go to extra effort
2373 This function is not called for ASCII ranges.
2375 Returns -1 if successful, REG_ESPACE if ran out of space. */
2378 set_image_of_range (work_area
, start
, end
, translate
)
2379 RE_TRANSLATE_TYPE translate
;
2380 struct range_table_work_area
*work_area
;
2381 re_wchar_t start
, end
;
2383 re_wchar_t cmin
, cmax
;
2386 /* For Latin-1 ranges, use set_image_of_range_1
2387 to get proper handling of ranges that include letters and nonletters.
2388 For a range that includes the whole of Latin-1, this is not necessary.
2389 For other character sets, we don't bother to get this right. */
2390 if (RE_TRANSLATE_P (translate
) && start
< 04400
2391 && !(start
< 04200 && end
>= 04377))
2398 tem
= set_image_of_range_1 (work_area
, start
, newend
, translate
);
2408 EXTEND_RANGE_TABLE (work_area
, 2);
2409 work_area
->table
[work_area
->used
++] = (start
);
2410 work_area
->table
[work_area
->used
++] = (end
);
2412 cmin
= -1, cmax
= -1;
2414 if (RE_TRANSLATE_P (translate
))
2418 for (ch
= start
; ch
<= end
; ch
++)
2420 re_wchar_t c
= TRANSLATE (ch
);
2421 if (! (start
<= c
&& c
<= end
))
2427 cmin
= MIN (cmin
, c
);
2428 cmax
= MAX (cmax
, c
);
2435 EXTEND_RANGE_TABLE (work_area
, 2);
2436 work_area
->table
[work_area
->used
++] = (cmin
);
2437 work_area
->table
[work_area
->used
++] = (cmax
);
2445 #ifndef MATCH_MAY_ALLOCATE
2447 /* If we cannot allocate large objects within re_match_2_internal,
2448 we make the fail stack and register vectors global.
2449 The fail stack, we grow to the maximum size when a regexp
2451 The register vectors, we adjust in size each time we
2452 compile a regexp, according to the number of registers it needs. */
2454 static fail_stack_type fail_stack
;
2456 /* Size with which the following vectors are currently allocated.
2457 That is so we can make them bigger as needed,
2458 but never make them smaller. */
2459 static int regs_allocated_size
;
2461 static re_char
** regstart
, ** regend
;
2462 static re_char
**best_regstart
, **best_regend
;
2464 /* Make the register vectors big enough for NUM_REGS registers,
2465 but don't make them smaller. */
2468 regex_grow_registers (num_regs
)
2471 if (num_regs
> regs_allocated_size
)
2473 RETALLOC_IF (regstart
, num_regs
, re_char
*);
2474 RETALLOC_IF (regend
, num_regs
, re_char
*);
2475 RETALLOC_IF (best_regstart
, num_regs
, re_char
*);
2476 RETALLOC_IF (best_regend
, num_regs
, re_char
*);
2478 regs_allocated_size
= num_regs
;
2482 #endif /* not MATCH_MAY_ALLOCATE */
2484 static boolean group_in_compile_stack
_RE_ARGS ((compile_stack_type
2488 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2489 Returns one of error codes defined in `regex.h', or zero for success.
2491 Assumes the `allocated' (and perhaps `buffer') and `translate'
2492 fields are set in BUFP on entry.
2494 If it succeeds, results are put in BUFP (if it returns an error, the
2495 contents of BUFP are undefined):
2496 `buffer' is the compiled pattern;
2497 `syntax' is set to SYNTAX;
2498 `used' is set to the length of the compiled pattern;
2499 `fastmap_accurate' is zero;
2500 `re_nsub' is the number of subexpressions in PATTERN;
2501 `not_bol' and `not_eol' are zero;
2503 The `fastmap' field is neither examined nor set. */
2505 /* Insert the `jump' from the end of last alternative to "here".
2506 The space for the jump has already been allocated. */
2507 #define FIXUP_ALT_JUMP() \
2509 if (fixup_alt_jump) \
2510 STORE_JUMP (jump, fixup_alt_jump, b); \
2514 /* Return, freeing storage we allocated. */
2515 #define FREE_STACK_RETURN(value) \
2517 FREE_RANGE_TABLE_WORK_AREA (range_table_work); \
2518 free (compile_stack.stack); \
2522 static reg_errcode_t
2523 regex_compile (const re_char
*pattern
, size_t size
, reg_syntax_t syntax
, struct re_pattern_buffer
*bufp
)
2525 /* We fetch characters from PATTERN here. */
2526 register re_wchar_t c
, c1
;
2528 /* A random temporary spot in PATTERN. */
2531 /* Points to the end of the buffer, where we should append. */
2532 register unsigned char *b
;
2534 /* Keeps track of unclosed groups. */
2535 compile_stack_type compile_stack
;
2537 /* Points to the current (ending) position in the pattern. */
2539 /* `const' makes AIX compiler fail. */
2540 unsigned char *p
= pattern
;
2542 re_char
*p
= pattern
;
2544 re_char
*pend
= pattern
+ size
;
2546 /* How to translate the characters in the pattern. */
2547 RE_TRANSLATE_TYPE translate
= bufp
->translate
;
2549 /* Address of the count-byte of the most recently inserted `exactn'
2550 command. This makes it possible to tell if a new exact-match
2551 character can be added to that command or if the character requires
2552 a new `exactn' command. */
2553 unsigned char *pending_exact
= 0;
2555 /* Address of start of the most recently finished expression.
2556 This tells, e.g., postfix * where to find the start of its
2557 operand. Reset at the beginning of groups and alternatives. */
2558 unsigned char *laststart
= 0;
2560 /* Address of beginning of regexp, or inside of last group. */
2561 unsigned char *begalt
;
2563 /* Place in the uncompiled pattern (i.e., the {) to
2564 which to go back if the interval is invalid. */
2565 re_char
*beg_interval
;
2567 /* Address of the place where a forward jump should go to the end of
2568 the containing expression. Each alternative of an `or' -- except the
2569 last -- ends with a forward jump of this sort. */
2570 unsigned char *fixup_alt_jump
= 0;
2572 /* Work area for range table of charset. */
2573 struct range_table_work_area range_table_work
;
2575 /* If the object matched can contain multibyte characters. */
2576 const boolean multibyte
= RE_MULTIBYTE_P (bufp
);
2578 /* If a target of matching can contain multibyte characters. */
2579 const boolean target_multibyte
= RE_TARGET_MULTIBYTE_P (bufp
);
2581 /* Nonzero if we have pushed down into a subpattern. */
2582 int in_subpattern
= 0;
2584 /* These hold the values of p, pattern, and pend from the main
2585 pattern when we have pushed into a subpattern. */
2587 re_char
*main_pattern
;
2592 DEBUG_PRINT1 ("\nCompiling pattern: ");
2595 unsigned debug_count
;
2597 for (debug_count
= 0; debug_count
< size
; debug_count
++)
2598 putchar (pattern
[debug_count
]);
2603 /* Initialize the compile stack. */
2604 compile_stack
.stack
= TALLOC (INIT_COMPILE_STACK_SIZE
, compile_stack_elt_t
);
2605 if (compile_stack
.stack
== NULL
)
2608 compile_stack
.size
= INIT_COMPILE_STACK_SIZE
;
2609 compile_stack
.avail
= 0;
2611 range_table_work
.table
= 0;
2612 range_table_work
.allocated
= 0;
2614 /* Initialize the pattern buffer. */
2615 bufp
->syntax
= syntax
;
2616 bufp
->fastmap_accurate
= 0;
2617 bufp
->not_bol
= bufp
->not_eol
= 0;
2618 bufp
->used_syntax
= 0;
2620 /* Set `used' to zero, so that if we return an error, the pattern
2621 printer (for debugging) will think there's no pattern. We reset it
2625 /* Always count groups, whether or not bufp->no_sub is set. */
2628 #if !defined emacs && !defined SYNTAX_TABLE
2629 /* Initialize the syntax table. */
2630 init_syntax_once ();
2633 if (bufp
->allocated
== 0)
2636 { /* If zero allocated, but buffer is non-null, try to realloc
2637 enough space. This loses if buffer's address is bogus, but
2638 that is the user's responsibility. */
2639 RETALLOC (bufp
->buffer
, INIT_BUF_SIZE
, unsigned char);
2642 { /* Caller did not allocate a buffer. Do it for them. */
2643 bufp
->buffer
= TALLOC (INIT_BUF_SIZE
, unsigned char);
2645 if (!bufp
->buffer
) FREE_STACK_RETURN (REG_ESPACE
);
2647 bufp
->allocated
= INIT_BUF_SIZE
;
2650 begalt
= b
= bufp
->buffer
;
2652 /* Loop through the uncompiled pattern until we're at the end. */
2657 /* If this is the end of an included regexp,
2658 pop back to the main regexp and try again. */
2662 pattern
= main_pattern
;
2667 /* If this is the end of the main regexp, we are done. */
2679 /* If there's no special whitespace regexp, treat
2680 spaces normally. And don't try to do this recursively. */
2681 if (!whitespace_regexp
|| in_subpattern
)
2684 /* Peek past following spaces. */
2691 /* If the spaces are followed by a repetition op,
2692 treat them normally. */
2694 && (*p1
== '*' || *p1
== '+' || *p1
== '?'
2695 || (*p1
== '\\' && p1
+ 1 != pend
&& p1
[1] == '{')))
2698 /* Replace the spaces with the whitespace regexp. */
2702 main_pattern
= pattern
;
2703 p
= pattern
= whitespace_regexp
;
2704 pend
= p
+ strlen (p
);
2710 if ( /* If at start of pattern, it's an operator. */
2712 /* If context independent, it's an operator. */
2713 || syntax
& RE_CONTEXT_INDEP_ANCHORS
2714 /* Otherwise, depends on what's come before. */
2715 || at_begline_loc_p (pattern
, p
, syntax
))
2716 BUF_PUSH ((syntax
& RE_NO_NEWLINE_ANCHOR
) ? begbuf
: begline
);
2725 if ( /* If at end of pattern, it's an operator. */
2727 /* If context independent, it's an operator. */
2728 || syntax
& RE_CONTEXT_INDEP_ANCHORS
2729 /* Otherwise, depends on what's next. */
2730 || at_endline_loc_p (p
, pend
, syntax
))
2731 BUF_PUSH ((syntax
& RE_NO_NEWLINE_ANCHOR
) ? endbuf
: endline
);
2740 if ((syntax
& RE_BK_PLUS_QM
)
2741 || (syntax
& RE_LIMITED_OPS
))
2745 /* If there is no previous pattern... */
2748 if (syntax
& RE_CONTEXT_INVALID_OPS
)
2749 FREE_STACK_RETURN (REG_BADRPT
);
2750 else if (!(syntax
& RE_CONTEXT_INDEP_OPS
))
2755 /* 1 means zero (many) matches is allowed. */
2756 boolean zero_times_ok
= 0, many_times_ok
= 0;
2759 /* If there is a sequence of repetition chars, collapse it
2760 down to just one (the right one). We can't combine
2761 interval operators with these because of, e.g., `a{2}*',
2762 which should only match an even number of `a's. */
2766 if ((syntax
& RE_FRUGAL
)
2767 && c
== '?' && (zero_times_ok
|| many_times_ok
))
2771 zero_times_ok
|= c
!= '+';
2772 many_times_ok
|= c
!= '?';
2778 || (!(syntax
& RE_BK_PLUS_QM
)
2779 && (*p
== '+' || *p
== '?')))
2781 else if (syntax
& RE_BK_PLUS_QM
&& *p
== '\\')
2784 FREE_STACK_RETURN (REG_EESCAPE
);
2785 if (p
[1] == '+' || p
[1] == '?')
2786 PATFETCH (c
); /* Gobble up the backslash. */
2792 /* If we get here, we found another repeat character. */
2796 /* Star, etc. applied to an empty pattern is equivalent
2797 to an empty pattern. */
2798 if (!laststart
|| laststart
== b
)
2801 /* Now we know whether or not zero matches is allowed
2802 and also whether or not two or more matches is allowed. */
2807 boolean simple
= skip_one_char (laststart
) == b
;
2808 unsigned int startoffset
= 0;
2810 /* Check if the loop can match the empty string. */
2811 (simple
|| !analyse_first (laststart
, b
, NULL
, 0))
2812 ? on_failure_jump
: on_failure_jump_loop
;
2813 assert (skip_one_char (laststart
) <= b
);
2815 if (!zero_times_ok
&& simple
)
2816 { /* Since simple * loops can be made faster by using
2817 on_failure_keep_string_jump, we turn simple P+
2818 into PP* if P is simple. */
2819 unsigned char *p1
, *p2
;
2820 startoffset
= b
- laststart
;
2821 GET_BUFFER_SPACE (startoffset
);
2822 p1
= b
; p2
= laststart
;
2828 GET_BUFFER_SPACE (6);
2831 STORE_JUMP (ofj
, b
, b
+ 6);
2833 /* Simple * loops can use on_failure_keep_string_jump
2834 depending on what follows. But since we don't know
2835 that yet, we leave the decision up to
2836 on_failure_jump_smart. */
2837 INSERT_JUMP (simple
? on_failure_jump_smart
: ofj
,
2838 laststart
+ startoffset
, b
+ 6);
2840 STORE_JUMP (jump
, b
, laststart
+ startoffset
);
2845 /* A simple ? pattern. */
2846 assert (zero_times_ok
);
2847 GET_BUFFER_SPACE (3);
2848 INSERT_JUMP (on_failure_jump
, laststart
, b
+ 3);
2852 else /* not greedy */
2853 { /* I wish the greedy and non-greedy cases could be merged. */
2855 GET_BUFFER_SPACE (7); /* We might use less. */
2858 boolean emptyp
= analyse_first (laststart
, b
, NULL
, 0);
2860 /* The non-greedy multiple match looks like
2861 a repeat..until: we only need a conditional jump
2862 at the end of the loop. */
2863 if (emptyp
) BUF_PUSH (no_op
);
2864 STORE_JUMP (emptyp
? on_failure_jump_nastyloop
2865 : on_failure_jump
, b
, laststart
);
2869 /* The repeat...until naturally matches one or more.
2870 To also match zero times, we need to first jump to
2871 the end of the loop (its conditional jump). */
2872 INSERT_JUMP (jump
, laststart
, b
);
2878 /* non-greedy a?? */
2879 INSERT_JUMP (jump
, laststart
, b
+ 3);
2881 INSERT_JUMP (on_failure_jump
, laststart
, laststart
+ 6);
2898 CLEAR_RANGE_TABLE_WORK_USED (range_table_work
);
2900 if (p
== pend
) FREE_STACK_RETURN (REG_EBRACK
);
2902 /* Ensure that we have enough space to push a charset: the
2903 opcode, the length count, and the bitset; 34 bytes in all. */
2904 GET_BUFFER_SPACE (34);
2908 /* We test `*p == '^' twice, instead of using an if
2909 statement, so we only need one BUF_PUSH. */
2910 BUF_PUSH (*p
== '^' ? charset_not
: charset
);
2914 /* Remember the first position in the bracket expression. */
2917 /* Push the number of bytes in the bitmap. */
2918 BUF_PUSH ((1 << BYTEWIDTH
) / BYTEWIDTH
);
2920 /* Clear the whole map. */
2921 memset (b
, 0, (1 << BYTEWIDTH
) / BYTEWIDTH
);
2923 /* charset_not matches newline according to a syntax bit. */
2924 if ((re_opcode_t
) b
[-2] == charset_not
2925 && (syntax
& RE_HAT_LISTS_NOT_NEWLINE
))
2926 SET_LIST_BIT ('\n');
2928 /* Read in characters and ranges, setting map bits. */
2931 boolean escaped_char
= false;
2932 const unsigned char *p2
= p
;
2935 if (p
== pend
) FREE_STACK_RETURN (REG_EBRACK
);
2937 /* Don't translate yet. The range TRANSLATE(X..Y) cannot
2938 always be determined from TRANSLATE(X) and TRANSLATE(Y)
2939 So the translation is done later in a loop. Example:
2940 (let ((case-fold-search t)) (string-match "[A-_]" "A")) */
2943 /* \ might escape characters inside [...] and [^...]. */
2944 if ((syntax
& RE_BACKSLASH_ESCAPE_IN_LISTS
) && c
== '\\')
2946 if (p
== pend
) FREE_STACK_RETURN (REG_EESCAPE
);
2949 escaped_char
= true;
2953 /* Could be the end of the bracket expression. If it's
2954 not (i.e., when the bracket expression is `[]' so
2955 far), the ']' character bit gets set way below. */
2956 if (c
== ']' && p2
!= p1
)
2960 /* See if we're at the beginning of a possible character
2963 if (!escaped_char
&&
2964 syntax
& RE_CHAR_CLASSES
&& c
== '[' && *p
== ':')
2966 /* Leave room for the null. */
2967 unsigned char str
[CHAR_CLASS_MAX_LENGTH
+ 1];
2968 const unsigned char *class_beg
;
2974 /* If pattern is `[[:'. */
2975 if (p
== pend
) FREE_STACK_RETURN (REG_EBRACK
);
2980 if ((c
== ':' && *p
== ']') || p
== pend
)
2982 if (c1
< CHAR_CLASS_MAX_LENGTH
)
2985 /* This is in any case an invalid class name. */
2990 /* If isn't a word bracketed by `[:' and `:]':
2991 undo the ending character, the letters, and
2992 leave the leading `:' and `[' (but set bits for
2994 if (c
== ':' && *p
== ']')
2999 cc
= re_wctype (str
);
3002 FREE_STACK_RETURN (REG_ECTYPE
);
3004 /* Throw away the ] at the end of the character
3008 if (p
== pend
) FREE_STACK_RETURN (REG_EBRACK
);
3011 for (ch
= 0; ch
< (1 << BYTEWIDTH
); ++ch
)
3012 if (re_iswctype (btowc (ch
), cc
))
3015 if (c
< (1 << BYTEWIDTH
))
3019 /* Most character classes in a multibyte match
3020 just set a flag. Exceptions are is_blank,
3021 is_digit, is_cntrl, and is_xdigit, since
3022 they can only match ASCII characters. We
3023 don't need to handle them for multibyte.
3024 They are distinguished by a negative wctype. */
3026 /* Setup the gl_state object to its buffer-defined
3027 value. This hardcodes the buffer-global
3028 syntax-table for ASCII chars, while the other chars
3029 will obey syntax-table properties. It's not ideal,
3030 but it's the way it's been done until now. */
3031 SETUP_BUFFER_SYNTAX_TABLE ();
3033 for (ch
= 0; ch
< 256; ++ch
)
3035 c
= RE_CHAR_TO_MULTIBYTE (ch
);
3036 if (! CHAR_BYTE8_P (c
)
3037 && re_iswctype (c
, cc
))
3043 if (ASCII_CHAR_P (c1
))
3045 else if ((c1
= RE_CHAR_TO_UNIBYTE (c1
)) >= 0)
3049 SET_RANGE_TABLE_WORK_AREA_BIT
3050 (range_table_work
, re_wctype_to_bit (cc
));
3052 /* In most cases the matching rule for char classes
3053 only uses the syntax table for multibyte chars,
3054 so that the content of the syntax-table it is not
3055 hardcoded in the range_table. SPACE and WORD are
3056 the two exceptions. */
3057 if ((1 << cc
) & ((1 << RECC_SPACE
) | (1 << RECC_WORD
)))
3058 bufp
->used_syntax
= 1;
3060 /* Repeat the loop. */
3065 /* Go back to right after the "[:". */
3069 /* Because the `:' may starts the range, we
3070 can't simply set bit and repeat the loop.
3071 Instead, just set it to C and handle below. */
3076 if (p
< pend
&& p
[0] == '-' && p
[1] != ']')
3079 /* Discard the `-'. */
3082 /* Fetch the character which ends the range. */
3085 if (CHAR_BYTE8_P (c1
)
3086 && ! ASCII_CHAR_P (c
) && ! CHAR_BYTE8_P (c
))
3087 /* Treat the range from a multibyte character to
3088 raw-byte character as empty. */
3093 /* Range from C to C. */
3098 if (syntax
& RE_NO_EMPTY_RANGES
)
3099 FREE_STACK_RETURN (REG_ERANGEX
);
3100 /* Else, repeat the loop. */
3105 /* Set the range into bitmap */
3106 for (; c
<= c1
; c
++)
3109 if (ch
< (1 << BYTEWIDTH
))
3116 SETUP_ASCII_RANGE (range_table_work
, c
, ch
);
3118 if (CHAR_BYTE8_P (c1
))
3119 c
= BYTE8_TO_CHAR (128);
3123 if (CHAR_BYTE8_P (c
))
3125 c
= CHAR_TO_BYTE8 (c
);
3126 c1
= CHAR_TO_BYTE8 (c1
);
3127 for (; c
<= c1
; c
++)
3132 SETUP_MULTIBYTE_RANGE (range_table_work
, c
, c1
);
3136 SETUP_UNIBYTE_RANGE (range_table_work
, c
, c1
);
3143 /* Discard any (non)matching list bytes that are all 0 at the
3144 end of the map. Decrease the map-length byte too. */
3145 while ((int) b
[-1] > 0 && b
[b
[-1] - 1] == 0)
3149 /* Build real range table from work area. */
3150 if (RANGE_TABLE_WORK_USED (range_table_work
)
3151 || RANGE_TABLE_WORK_BITS (range_table_work
))
3154 int used
= RANGE_TABLE_WORK_USED (range_table_work
);
3156 /* Allocate space for COUNT + RANGE_TABLE. Needs two
3157 bytes for flags, two for COUNT, and three bytes for
3159 GET_BUFFER_SPACE (4 + used
* 3);
3161 /* Indicate the existence of range table. */
3162 laststart
[1] |= 0x80;
3164 /* Store the character class flag bits into the range table.
3165 If not in emacs, these flag bits are always 0. */
3166 *b
++ = RANGE_TABLE_WORK_BITS (range_table_work
) & 0xff;
3167 *b
++ = RANGE_TABLE_WORK_BITS (range_table_work
) >> 8;
3169 STORE_NUMBER_AND_INCR (b
, used
/ 2);
3170 for (i
= 0; i
< used
; i
++)
3171 STORE_CHARACTER_AND_INCR
3172 (b
, RANGE_TABLE_WORK_ELT (range_table_work
, i
));
3179 if (syntax
& RE_NO_BK_PARENS
)
3186 if (syntax
& RE_NO_BK_PARENS
)
3193 if (syntax
& RE_NEWLINE_ALT
)
3200 if (syntax
& RE_NO_BK_VBAR
)
3207 if (syntax
& RE_INTERVALS
&& syntax
& RE_NO_BK_BRACES
)
3208 goto handle_interval
;
3214 if (p
== pend
) FREE_STACK_RETURN (REG_EESCAPE
);
3216 /* Do not translate the character after the \, so that we can
3217 distinguish, e.g., \B from \b, even if we normally would
3218 translate, e.g., B to b. */
3224 if (syntax
& RE_NO_BK_PARENS
)
3225 goto normal_backslash
;
3230 regnum_t regnum
= 0;
3233 /* Look for a special (?...) construct */
3234 if ((syntax
& RE_SHY_GROUPS
) && *p
== '?')
3236 PATFETCH (c
); /* Gobble up the '?'. */
3242 case ':': shy
= 1; break;
3244 /* An explicitly specified regnum must start
3247 FREE_STACK_RETURN (REG_BADPAT
);
3248 case '1': case '2': case '3': case '4':
3249 case '5': case '6': case '7': case '8': case '9':
3250 regnum
= 10*regnum
+ (c
- '0'); break;
3252 /* Only (?:...) is supported right now. */
3253 FREE_STACK_RETURN (REG_BADPAT
);
3260 regnum
= ++bufp
->re_nsub
;
3262 { /* It's actually not shy, but explicitly numbered. */
3264 if (regnum
> bufp
->re_nsub
)
3265 bufp
->re_nsub
= regnum
;
3266 else if (regnum
> bufp
->re_nsub
3267 /* Ideally, we'd want to check that the specified
3268 group can't have matched (i.e. all subgroups
3269 using the same regnum are in other branches of
3270 OR patterns), but we don't currently keep track
3271 of enough info to do that easily. */
3272 || group_in_compile_stack (compile_stack
, regnum
))
3273 FREE_STACK_RETURN (REG_BADPAT
);
3276 /* It's really shy. */
3277 regnum
= - bufp
->re_nsub
;
3279 if (COMPILE_STACK_FULL
)
3281 RETALLOC (compile_stack
.stack
, compile_stack
.size
<< 1,
3282 compile_stack_elt_t
);
3283 if (compile_stack
.stack
== NULL
) return REG_ESPACE
;
3285 compile_stack
.size
<<= 1;
3288 /* These are the values to restore when we hit end of this
3289 group. They are all relative offsets, so that if the
3290 whole pattern moves because of realloc, they will still
3292 COMPILE_STACK_TOP
.begalt_offset
= begalt
- bufp
->buffer
;
3293 COMPILE_STACK_TOP
.fixup_alt_jump
3294 = fixup_alt_jump
? fixup_alt_jump
- bufp
->buffer
+ 1 : 0;
3295 COMPILE_STACK_TOP
.laststart_offset
= b
- bufp
->buffer
;
3296 COMPILE_STACK_TOP
.regnum
= regnum
;
3298 /* Do not push a start_memory for groups beyond the last one
3299 we can represent in the compiled pattern. */
3300 if (regnum
<= MAX_REGNUM
&& regnum
> 0)
3301 BUF_PUSH_2 (start_memory
, regnum
);
3303 compile_stack
.avail
++;
3308 /* If we've reached MAX_REGNUM groups, then this open
3309 won't actually generate any code, so we'll have to
3310 clear pending_exact explicitly. */
3316 if (syntax
& RE_NO_BK_PARENS
) goto normal_backslash
;
3318 if (COMPILE_STACK_EMPTY
)
3320 if (syntax
& RE_UNMATCHED_RIGHT_PAREN_ORD
)
3321 goto normal_backslash
;
3323 FREE_STACK_RETURN (REG_ERPAREN
);
3329 /* See similar code for backslashed left paren above. */
3330 if (COMPILE_STACK_EMPTY
)
3332 if (syntax
& RE_UNMATCHED_RIGHT_PAREN_ORD
)
3335 FREE_STACK_RETURN (REG_ERPAREN
);
3338 /* Since we just checked for an empty stack above, this
3339 ``can't happen''. */
3340 assert (compile_stack
.avail
!= 0);
3342 /* We don't just want to restore into `regnum', because
3343 later groups should continue to be numbered higher,
3344 as in `(ab)c(de)' -- the second group is #2. */
3347 compile_stack
.avail
--;
3348 begalt
= bufp
->buffer
+ COMPILE_STACK_TOP
.begalt_offset
;
3350 = COMPILE_STACK_TOP
.fixup_alt_jump
3351 ? bufp
->buffer
+ COMPILE_STACK_TOP
.fixup_alt_jump
- 1
3353 laststart
= bufp
->buffer
+ COMPILE_STACK_TOP
.laststart_offset
;
3354 regnum
= COMPILE_STACK_TOP
.regnum
;
3355 /* If we've reached MAX_REGNUM groups, then this open
3356 won't actually generate any code, so we'll have to
3357 clear pending_exact explicitly. */
3360 /* We're at the end of the group, so now we know how many
3361 groups were inside this one. */
3362 if (regnum
<= MAX_REGNUM
&& regnum
> 0)
3363 BUF_PUSH_2 (stop_memory
, regnum
);
3368 case '|': /* `\|'. */
3369 if (syntax
& RE_LIMITED_OPS
|| syntax
& RE_NO_BK_VBAR
)
3370 goto normal_backslash
;
3372 if (syntax
& RE_LIMITED_OPS
)
3375 /* Insert before the previous alternative a jump which
3376 jumps to this alternative if the former fails. */
3377 GET_BUFFER_SPACE (3);
3378 INSERT_JUMP (on_failure_jump
, begalt
, b
+ 6);
3382 /* The alternative before this one has a jump after it
3383 which gets executed if it gets matched. Adjust that
3384 jump so it will jump to this alternative's analogous
3385 jump (put in below, which in turn will jump to the next
3386 (if any) alternative's such jump, etc.). The last such
3387 jump jumps to the correct final destination. A picture:
3393 If we are at `b', then fixup_alt_jump right now points to a
3394 three-byte space after `a'. We'll put in the jump, set
3395 fixup_alt_jump to right after `b', and leave behind three
3396 bytes which we'll fill in when we get to after `c'. */
3400 /* Mark and leave space for a jump after this alternative,
3401 to be filled in later either by next alternative or
3402 when know we're at the end of a series of alternatives. */
3404 GET_BUFFER_SPACE (3);
3413 /* If \{ is a literal. */
3414 if (!(syntax
& RE_INTERVALS
)
3415 /* If we're at `\{' and it's not the open-interval
3417 || (syntax
& RE_NO_BK_BRACES
))
3418 goto normal_backslash
;
3422 /* If got here, then the syntax allows intervals. */
3424 /* At least (most) this many matches must be made. */
3425 int lower_bound
= 0, upper_bound
= -1;
3429 GET_UNSIGNED_NUMBER (lower_bound
);
3432 GET_UNSIGNED_NUMBER (upper_bound
);
3434 /* Interval such as `{1}' => match exactly once. */
3435 upper_bound
= lower_bound
;
3437 if (lower_bound
< 0 || upper_bound
> RE_DUP_MAX
3438 || (upper_bound
>= 0 && lower_bound
> upper_bound
))
3439 FREE_STACK_RETURN (REG_BADBR
);
3441 if (!(syntax
& RE_NO_BK_BRACES
))
3444 FREE_STACK_RETURN (REG_BADBR
);
3446 FREE_STACK_RETURN (REG_EESCAPE
);
3451 FREE_STACK_RETURN (REG_BADBR
);
3453 /* We just parsed a valid interval. */
3455 /* If it's invalid to have no preceding re. */
3458 if (syntax
& RE_CONTEXT_INVALID_OPS
)
3459 FREE_STACK_RETURN (REG_BADRPT
);
3460 else if (syntax
& RE_CONTEXT_INDEP_OPS
)
3463 goto unfetch_interval
;
3466 if (upper_bound
== 0)
3467 /* If the upper bound is zero, just drop the sub pattern
3470 else if (lower_bound
== 1 && upper_bound
== 1)
3471 /* Just match it once: nothing to do here. */
3474 /* Otherwise, we have a nontrivial interval. When
3475 we're all done, the pattern will look like:
3476 set_number_at <jump count> <upper bound>
3477 set_number_at <succeed_n count> <lower bound>
3478 succeed_n <after jump addr> <succeed_n count>
3480 jump_n <succeed_n addr> <jump count>
3481 (The upper bound and `jump_n' are omitted if
3482 `upper_bound' is 1, though.) */
3484 { /* If the upper bound is > 1, we need to insert
3485 more at the end of the loop. */
3486 unsigned int nbytes
= (upper_bound
< 0 ? 3
3487 : upper_bound
> 1 ? 5 : 0);
3488 unsigned int startoffset
= 0;
3490 GET_BUFFER_SPACE (20); /* We might use less. */
3492 if (lower_bound
== 0)
3494 /* A succeed_n that starts with 0 is really a
3495 a simple on_failure_jump_loop. */
3496 INSERT_JUMP (on_failure_jump_loop
, laststart
,
3502 /* Initialize lower bound of the `succeed_n', even
3503 though it will be set during matching by its
3504 attendant `set_number_at' (inserted next),
3505 because `re_compile_fastmap' needs to know.
3506 Jump to the `jump_n' we might insert below. */
3507 INSERT_JUMP2 (succeed_n
, laststart
,
3512 /* Code to initialize the lower bound. Insert
3513 before the `succeed_n'. The `5' is the last two
3514 bytes of this `set_number_at', plus 3 bytes of
3515 the following `succeed_n'. */
3516 insert_op2 (set_number_at
, laststart
, 5, lower_bound
, b
);
3521 if (upper_bound
< 0)
3523 /* A negative upper bound stands for infinity,
3524 in which case it degenerates to a plain jump. */
3525 STORE_JUMP (jump
, b
, laststart
+ startoffset
);
3528 else if (upper_bound
> 1)
3529 { /* More than one repetition is allowed, so
3530 append a backward jump to the `succeed_n'
3531 that starts this interval.
3533 When we've reached this during matching,
3534 we'll have matched the interval once, so
3535 jump back only `upper_bound - 1' times. */
3536 STORE_JUMP2 (jump_n
, b
, laststart
+ startoffset
,
3540 /* The location we want to set is the second
3541 parameter of the `jump_n'; that is `b-2' as
3542 an absolute address. `laststart' will be
3543 the `set_number_at' we're about to insert;
3544 `laststart+3' the number to set, the source
3545 for the relative address. But we are
3546 inserting into the middle of the pattern --
3547 so everything is getting moved up by 5.
3548 Conclusion: (b - 2) - (laststart + 3) + 5,
3549 i.e., b - laststart.
3551 We insert this at the beginning of the loop
3552 so that if we fail during matching, we'll
3553 reinitialize the bounds. */
3554 insert_op2 (set_number_at
, laststart
, b
- laststart
,
3555 upper_bound
- 1, b
);
3560 beg_interval
= NULL
;
3565 /* If an invalid interval, match the characters as literals. */
3566 assert (beg_interval
);
3568 beg_interval
= NULL
;
3570 /* normal_char and normal_backslash need `c'. */
3573 if (!(syntax
& RE_NO_BK_BRACES
))
3575 assert (p
> pattern
&& p
[-1] == '\\');
3576 goto normal_backslash
;
3582 /* There is no way to specify the before_dot and after_dot
3583 operators. rms says this is ok. --karl */
3591 BUF_PUSH_2 (syntaxspec
, syntax_spec_code
[c
]);
3597 BUF_PUSH_2 (notsyntaxspec
, syntax_spec_code
[c
]);
3603 BUF_PUSH_2 (categoryspec
, c
);
3609 BUF_PUSH_2 (notcategoryspec
, c
);
3615 if (syntax
& RE_NO_GNU_OPS
)
3618 BUF_PUSH_2 (syntaxspec
, Sword
);
3623 if (syntax
& RE_NO_GNU_OPS
)
3626 BUF_PUSH_2 (notsyntaxspec
, Sword
);
3631 if (syntax
& RE_NO_GNU_OPS
)
3637 if (syntax
& RE_NO_GNU_OPS
)
3643 if (syntax
& RE_NO_GNU_OPS
)
3652 FREE_STACK_RETURN (REG_BADPAT
);
3656 if (syntax
& RE_NO_GNU_OPS
)
3658 BUF_PUSH (wordbound
);
3662 if (syntax
& RE_NO_GNU_OPS
)
3664 BUF_PUSH (notwordbound
);
3668 if (syntax
& RE_NO_GNU_OPS
)
3674 if (syntax
& RE_NO_GNU_OPS
)
3679 case '1': case '2': case '3': case '4': case '5':
3680 case '6': case '7': case '8': case '9':
3684 if (syntax
& RE_NO_BK_REFS
)
3685 goto normal_backslash
;
3689 if (reg
> bufp
->re_nsub
|| reg
< 1
3690 /* Can't back reference to a subexp before its end. */
3691 || group_in_compile_stack (compile_stack
, reg
))
3692 FREE_STACK_RETURN (REG_ESUBREG
);
3695 BUF_PUSH_2 (duplicate
, reg
);
3702 if (syntax
& RE_BK_PLUS_QM
)
3705 goto normal_backslash
;
3709 /* You might think it would be useful for \ to mean
3710 not to translate; but if we don't translate it
3711 it will never match anything. */
3718 /* Expects the character in `c'. */
3720 /* If no exactn currently being built. */
3723 /* If last exactn not at current position. */
3724 || pending_exact
+ *pending_exact
+ 1 != b
3726 /* We have only one byte following the exactn for the count. */
3727 || *pending_exact
>= (1 << BYTEWIDTH
) - MAX_MULTIBYTE_LENGTH
3729 /* If followed by a repetition operator. */
3730 || (p
!= pend
&& (*p
== '*' || *p
== '^'))
3731 || ((syntax
& RE_BK_PLUS_QM
)
3732 ? p
+ 1 < pend
&& *p
== '\\' && (p
[1] == '+' || p
[1] == '?')
3733 : p
!= pend
&& (*p
== '+' || *p
== '?'))
3734 || ((syntax
& RE_INTERVALS
)
3735 && ((syntax
& RE_NO_BK_BRACES
)
3736 ? p
!= pend
&& *p
== '{'
3737 : p
+ 1 < pend
&& p
[0] == '\\' && p
[1] == '{')))
3739 /* Start building a new exactn. */
3743 BUF_PUSH_2 (exactn
, 0);
3744 pending_exact
= b
- 1;
3747 GET_BUFFER_SPACE (MAX_MULTIBYTE_LENGTH
);
3754 len
= CHAR_STRING (c
, b
);
3759 c1
= RE_CHAR_TO_MULTIBYTE (c
);
3760 if (! CHAR_BYTE8_P (c1
))
3762 re_wchar_t c2
= TRANSLATE (c1
);
3764 if (c1
!= c2
&& (c1
= RE_CHAR_TO_UNIBYTE (c2
)) >= 0)
3770 (*pending_exact
) += len
;
3775 } /* while p != pend */
3778 /* Through the pattern now. */
3782 if (!COMPILE_STACK_EMPTY
)
3783 FREE_STACK_RETURN (REG_EPAREN
);
3785 /* If we don't want backtracking, force success
3786 the first time we reach the end of the compiled pattern. */
3787 if (syntax
& RE_NO_POSIX_BACKTRACKING
)
3790 /* We have succeeded; set the length of the buffer. */
3791 bufp
->used
= b
- bufp
->buffer
;
3796 re_compile_fastmap (bufp
);
3797 DEBUG_PRINT1 ("\nCompiled pattern: \n");
3798 print_compiled_pattern (bufp
);
3803 #ifndef MATCH_MAY_ALLOCATE
3804 /* Initialize the failure stack to the largest possible stack. This
3805 isn't necessary unless we're trying to avoid calling alloca in
3806 the search and match routines. */
3808 int num_regs
= bufp
->re_nsub
+ 1;
3810 if (fail_stack
.size
< re_max_failures
* TYPICAL_FAILURE_SIZE
)
3812 fail_stack
.size
= re_max_failures
* TYPICAL_FAILURE_SIZE
;
3814 if (! fail_stack
.stack
)
3816 = (fail_stack_elt_t
*) malloc (fail_stack
.size
3817 * sizeof (fail_stack_elt_t
));
3820 = (fail_stack_elt_t
*) realloc (fail_stack
.stack
,
3822 * sizeof (fail_stack_elt_t
)));
3825 regex_grow_registers (num_regs
);
3827 #endif /* not MATCH_MAY_ALLOCATE */
3829 FREE_STACK_RETURN (REG_NOERROR
);
3830 } /* regex_compile */
3832 /* Subroutines for `regex_compile'. */
3834 /* Store OP at LOC followed by two-byte integer parameter ARG. */
3837 store_op1 (re_opcode_t op
, unsigned char *loc
, int arg
)
3839 *loc
= (unsigned char) op
;
3840 STORE_NUMBER (loc
+ 1, arg
);
3844 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
3847 store_op2 (re_opcode_t op
, unsigned char *loc
, int arg1
, int arg2
)
3849 *loc
= (unsigned char) op
;
3850 STORE_NUMBER (loc
+ 1, arg1
);
3851 STORE_NUMBER (loc
+ 3, arg2
);
3855 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
3856 for OP followed by two-byte integer parameter ARG. */
3859 insert_op1 (re_opcode_t op
, unsigned char *loc
, int arg
, unsigned char *end
)
3861 register unsigned char *pfrom
= end
;
3862 register unsigned char *pto
= end
+ 3;
3864 while (pfrom
!= loc
)
3867 store_op1 (op
, loc
, arg
);
3871 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
3874 insert_op2 (re_opcode_t op
, unsigned char *loc
, int arg1
, int arg2
, unsigned char *end
)
3876 register unsigned char *pfrom
= end
;
3877 register unsigned char *pto
= end
+ 5;
3879 while (pfrom
!= loc
)
3882 store_op2 (op
, loc
, arg1
, arg2
);
3886 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
3887 after an alternative or a begin-subexpression. We assume there is at
3888 least one character before the ^. */
3891 at_begline_loc_p (const re_char
*pattern
, const re_char
*p
, reg_syntax_t syntax
)
3893 re_char
*prev
= p
- 2;
3894 boolean prev_prev_backslash
= prev
> pattern
&& prev
[-1] == '\\';
3897 /* After a subexpression? */
3898 (*prev
== '(' && (syntax
& RE_NO_BK_PARENS
|| prev_prev_backslash
))
3899 /* After an alternative? */
3900 || (*prev
== '|' && (syntax
& RE_NO_BK_VBAR
|| prev_prev_backslash
))
3901 /* After a shy subexpression? */
3902 || ((syntax
& RE_SHY_GROUPS
) && prev
- 2 >= pattern
3903 && prev
[-1] == '?' && prev
[-2] == '('
3904 && (syntax
& RE_NO_BK_PARENS
3905 || (prev
- 3 >= pattern
&& prev
[-3] == '\\')));
3909 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3910 at least one character after the $, i.e., `P < PEND'. */
3913 at_endline_loc_p (const re_char
*p
, const re_char
*pend
, reg_syntax_t syntax
)
3916 boolean next_backslash
= *next
== '\\';
3917 re_char
*next_next
= p
+ 1 < pend
? p
+ 1 : 0;
3920 /* Before a subexpression? */
3921 (syntax
& RE_NO_BK_PARENS
? *next
== ')'
3922 : next_backslash
&& next_next
&& *next_next
== ')')
3923 /* Before an alternative? */
3924 || (syntax
& RE_NO_BK_VBAR
? *next
== '|'
3925 : next_backslash
&& next_next
&& *next_next
== '|');
3929 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3930 false if it's not. */
3933 group_in_compile_stack (compile_stack_type compile_stack
, regnum_t regnum
)
3937 for (this_element
= compile_stack
.avail
- 1;
3940 if (compile_stack
.stack
[this_element
].regnum
== regnum
)
3947 If fastmap is non-NULL, go through the pattern and fill fastmap
3948 with all the possible leading chars. If fastmap is NULL, don't
3949 bother filling it up (obviously) and only return whether the
3950 pattern could potentially match the empty string.
3952 Return 1 if p..pend might match the empty string.
3953 Return 0 if p..pend matches at least one char.
3954 Return -1 if fastmap was not updated accurately. */
3957 analyse_first (const re_char
*p
, const re_char
*pend
, char *fastmap
, const int multibyte
)
3962 /* If all elements for base leading-codes in fastmap is set, this
3963 flag is set true. */
3964 boolean match_any_multibyte_characters
= false;
3968 /* The loop below works as follows:
3969 - It has a working-list kept in the PATTERN_STACK and which basically
3970 starts by only containing a pointer to the first operation.
3971 - If the opcode we're looking at is a match against some set of
3972 chars, then we add those chars to the fastmap and go on to the
3973 next work element from the worklist (done via `break').
3974 - If the opcode is a control operator on the other hand, we either
3975 ignore it (if it's meaningless at this point, such as `start_memory')
3976 or execute it (if it's a jump). If the jump has several destinations
3977 (i.e. `on_failure_jump'), then we push the other destination onto the
3979 We guarantee termination by ignoring backward jumps (more or less),
3980 so that `p' is monotonically increasing. More to the point, we
3981 never set `p' (or push) anything `<= p1'. */
3985 /* `p1' is used as a marker of how far back a `on_failure_jump'
3986 can go without being ignored. It is normally equal to `p'
3987 (which prevents any backward `on_failure_jump') except right
3988 after a plain `jump', to allow patterns such as:
3991 10: on_failure_jump 3
3992 as used for the *? operator. */
3995 switch (SWITCH_ENUM_CAST ((re_opcode_t
) *p
++))
4001 /* If the first character has to match a backreference, that means
4002 that the group was empty (since it already matched). Since this
4003 is the only case that interests us here, we can assume that the
4004 backreference must match the empty string. */
4009 /* Following are the cases which match a character. These end
4015 /* If multibyte is nonzero, the first byte of each
4016 character is an ASCII or a leading code. Otherwise,
4017 each byte is a character. Thus, this works in both
4022 /* For the case of matching this unibyte regex
4023 against multibyte, we must set a leading code of
4024 the corresponding multibyte character. */
4025 int c
= RE_CHAR_TO_MULTIBYTE (p
[1]);
4027 fastmap
[CHAR_LEADING_CODE (c
)] = 1;
4034 /* We could put all the chars except for \n (and maybe \0)
4035 but we don't bother since it is generally not worth it. */
4036 if (!fastmap
) break;
4041 if (!fastmap
) break;
4043 /* Chars beyond end of bitmap are possible matches. */
4044 for (j
= CHARSET_BITMAP_SIZE (&p
[-1]) * BYTEWIDTH
;
4045 j
< (1 << BYTEWIDTH
); j
++)
4051 if (!fastmap
) break;
4052 not = (re_opcode_t
) *(p
- 1) == charset_not
;
4053 for (j
= CHARSET_BITMAP_SIZE (&p
[-1]) * BYTEWIDTH
- 1, p
++;
4055 if (!!(p
[j
/ BYTEWIDTH
] & (1 << (j
% BYTEWIDTH
))) ^ not)
4059 if (/* Any leading code can possibly start a character
4060 which doesn't match the specified set of characters. */
4063 /* If we can match a character class, we can match any
4064 multibyte characters. */
4065 (CHARSET_RANGE_TABLE_EXISTS_P (&p
[-2])
4066 && CHARSET_RANGE_TABLE_BITS (&p
[-2]) != 0))
4069 if (match_any_multibyte_characters
== false)
4071 for (j
= MIN_MULTIBYTE_LEADING_CODE
;
4072 j
<= MAX_MULTIBYTE_LEADING_CODE
; j
++)
4074 match_any_multibyte_characters
= true;
4078 else if (!not && CHARSET_RANGE_TABLE_EXISTS_P (&p
[-2])
4079 && match_any_multibyte_characters
== false)
4081 /* Set fastmap[I] to 1 where I is a leading code of each
4082 multibyte character in the range table. */
4084 unsigned char lc1
, lc2
;
4086 /* Make P points the range table. `+ 2' is to skip flag
4087 bits for a character class. */
4088 p
+= CHARSET_BITMAP_SIZE (&p
[-2]) + 2;
4090 /* Extract the number of ranges in range table into COUNT. */
4091 EXTRACT_NUMBER_AND_INCR (count
, p
);
4092 for (; count
> 0; count
--, p
+= 3)
4094 /* Extract the start and end of each range. */
4095 EXTRACT_CHARACTER (c
, p
);
4096 lc1
= CHAR_LEADING_CODE (c
);
4098 EXTRACT_CHARACTER (c
, p
);
4099 lc2
= CHAR_LEADING_CODE (c
);
4100 for (j
= lc1
; j
<= lc2
; j
++)
4109 if (!fastmap
) break;
4111 not = (re_opcode_t
)p
[-1] == notsyntaxspec
;
4113 for (j
= 0; j
< (1 << BYTEWIDTH
); j
++)
4114 if ((SYNTAX (j
) == (enum syntaxcode
) k
) ^ not)
4118 /* This match depends on text properties. These end with
4119 aborting optimizations. */
4123 case notcategoryspec
:
4124 if (!fastmap
) break;
4125 not = (re_opcode_t
)p
[-1] == notcategoryspec
;
4127 for (j
= (1 << BYTEWIDTH
); j
>= 0; j
--)
4128 if ((CHAR_HAS_CATEGORY (j
, k
)) ^ not)
4131 /* Any leading code can possibly start a character which
4132 has or doesn't has the specified category. */
4133 if (match_any_multibyte_characters
== false)
4135 for (j
= MIN_MULTIBYTE_LEADING_CODE
;
4136 j
<= MAX_MULTIBYTE_LEADING_CODE
; j
++)
4138 match_any_multibyte_characters
= true;
4142 /* All cases after this match the empty string. These end with
4164 EXTRACT_NUMBER_AND_INCR (j
, p
);
4166 /* Backward jumps can only go back to code that we've already
4167 visited. `re_compile' should make sure this is true. */
4170 switch (SWITCH_ENUM_CAST ((re_opcode_t
) *p
))
4172 case on_failure_jump
:
4173 case on_failure_keep_string_jump
:
4174 case on_failure_jump_loop
:
4175 case on_failure_jump_nastyloop
:
4176 case on_failure_jump_smart
:
4182 /* Keep `p1' to allow the `on_failure_jump' we are jumping to
4183 to jump back to "just after here". */
4186 case on_failure_jump
:
4187 case on_failure_keep_string_jump
:
4188 case on_failure_jump_nastyloop
:
4189 case on_failure_jump_loop
:
4190 case on_failure_jump_smart
:
4191 EXTRACT_NUMBER_AND_INCR (j
, p
);
4193 ; /* Backward jump to be ignored. */
4195 { /* We have to look down both arms.
4196 We first go down the "straight" path so as to minimize
4197 stack usage when going through alternatives. */
4198 int r
= analyse_first (p
, pend
, fastmap
, multibyte
);
4206 /* This code simply does not properly handle forward jump_n. */
4207 DEBUG_STATEMENT (EXTRACT_NUMBER (j
, p
); assert (j
< 0));
4209 /* jump_n can either jump or fall through. The (backward) jump
4210 case has already been handled, so we only need to look at the
4211 fallthrough case. */
4215 /* If N == 0, it should be an on_failure_jump_loop instead. */
4216 DEBUG_STATEMENT (EXTRACT_NUMBER (j
, p
+ 2); assert (j
> 0));
4218 /* We only care about one iteration of the loop, so we don't
4219 need to consider the case where this behaves like an
4236 abort (); /* We have listed all the cases. */
4239 /* Getting here means we have found the possible starting
4240 characters for one path of the pattern -- and that the empty
4241 string does not match. We need not follow this path further. */
4245 /* We reached the end without matching anything. */
4248 } /* analyse_first */
4250 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4251 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4252 characters can start a string that matches the pattern. This fastmap
4253 is used by re_search to skip quickly over impossible starting points.
4255 Character codes above (1 << BYTEWIDTH) are not represented in the
4256 fastmap, but the leading codes are represented. Thus, the fastmap
4257 indicates which character sets could start a match.
4259 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4260 area as BUFP->fastmap.
4262 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4265 Returns 0 if we succeed, -2 if an internal error. */
4268 re_compile_fastmap (struct re_pattern_buffer
*bufp
)
4270 char *fastmap
= bufp
->fastmap
;
4273 assert (fastmap
&& bufp
->buffer
);
4275 memset (fastmap
, 0, 1 << BYTEWIDTH
); /* Assume nothing's valid. */
4276 bufp
->fastmap_accurate
= 1; /* It will be when we're done. */
4278 analysis
= analyse_first (bufp
->buffer
, bufp
->buffer
+ bufp
->used
,
4279 fastmap
, RE_MULTIBYTE_P (bufp
));
4280 bufp
->can_be_null
= (analysis
!= 0);
4282 } /* re_compile_fastmap */
4284 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4285 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4286 this memory for recording register information. STARTS and ENDS
4287 must be allocated using the malloc library routine, and must each
4288 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4290 If NUM_REGS == 0, then subsequent matches should allocate their own
4293 Unless this function is called, the first search or match using
4294 PATTERN_BUFFER will allocate its own register data, without
4295 freeing the old data. */
4298 re_set_registers (struct re_pattern_buffer
*bufp
, struct re_registers
*regs
, unsigned int num_regs
, regoff_t
*starts
, regoff_t
*ends
)
4302 bufp
->regs_allocated
= REGS_REALLOCATE
;
4303 regs
->num_regs
= num_regs
;
4304 regs
->start
= starts
;
4309 bufp
->regs_allocated
= REGS_UNALLOCATED
;
4311 regs
->start
= regs
->end
= (regoff_t
*) 0;
4314 WEAK_ALIAS (__re_set_registers
, re_set_registers
)
4316 /* Searching routines. */
4318 /* Like re_search_2, below, but only one string is specified, and
4319 doesn't let you say where to stop matching. */
4322 re_search (struct re_pattern_buffer
*bufp
, const char *string
, int size
, int startpos
, int range
, struct re_registers
*regs
)
4324 return re_search_2 (bufp
, NULL
, 0, string
, size
, startpos
, range
,
4327 WEAK_ALIAS (__re_search
, re_search
)
4329 /* Head address of virtual concatenation of string. */
4330 #define HEAD_ADDR_VSTRING(P) \
4331 (((P) >= size1 ? string2 : string1))
4333 /* End address of virtual concatenation of string. */
4334 #define STOP_ADDR_VSTRING(P) \
4335 (((P) >= size1 ? string2 + size2 : string1 + size1))
4337 /* Address of POS in the concatenation of virtual string. */
4338 #define POS_ADDR_VSTRING(POS) \
4339 (((POS) >= size1 ? string2 - size1 : string1) + (POS))
4341 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4342 virtual concatenation of STRING1 and STRING2, starting first at index
4343 STARTPOS, then at STARTPOS + 1, and so on.
4345 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
4347 RANGE is how far to scan while trying to match. RANGE = 0 means try
4348 only at STARTPOS; in general, the last start tried is STARTPOS +
4351 In REGS, return the indices of the virtual concatenation of STRING1
4352 and STRING2 that matched the entire BUFP->buffer and its contained
4355 Do not consider matching one past the index STOP in the virtual
4356 concatenation of STRING1 and STRING2.
4358 We return either the position in the strings at which the match was
4359 found, -1 if no match, or -2 if error (such as failure
4363 re_search_2 (struct re_pattern_buffer
*bufp
, const char *str1
, int size1
, const char *str2
, int size2
, int startpos
, int range
, struct re_registers
*regs
, int stop
)
4366 re_char
*string1
= (re_char
*) str1
;
4367 re_char
*string2
= (re_char
*) str2
;
4368 register char *fastmap
= bufp
->fastmap
;
4369 register RE_TRANSLATE_TYPE translate
= bufp
->translate
;
4370 int total_size
= size1
+ size2
;
4371 int endpos
= startpos
+ range
;
4372 boolean anchored_start
;
4373 /* Nonzero if we are searching multibyte string. */
4374 const boolean multibyte
= RE_TARGET_MULTIBYTE_P (bufp
);
4376 /* Check for out-of-range STARTPOS. */
4377 if (startpos
< 0 || startpos
> total_size
)
4380 /* Fix up RANGE if it might eventually take us outside
4381 the virtual concatenation of STRING1 and STRING2.
4382 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
4384 range
= 0 - startpos
;
4385 else if (endpos
> total_size
)
4386 range
= total_size
- startpos
;
4388 /* If the search isn't to be a backwards one, don't waste time in a
4389 search for a pattern anchored at beginning of buffer. */
4390 if (bufp
->used
> 0 && (re_opcode_t
) bufp
->buffer
[0] == begbuf
&& range
> 0)
4399 /* In a forward search for something that starts with \=.
4400 don't keep searching past point. */
4401 if (bufp
->used
> 0 && (re_opcode_t
) bufp
->buffer
[0] == at_dot
&& range
> 0)
4403 range
= PT_BYTE
- BEGV_BYTE
- startpos
;
4409 /* Update the fastmap now if not correct already. */
4410 if (fastmap
&& !bufp
->fastmap_accurate
)
4411 re_compile_fastmap (bufp
);
4413 /* See whether the pattern is anchored. */
4414 anchored_start
= (bufp
->buffer
[0] == begline
);
4417 gl_state
.object
= re_match_object
; /* Used by SYNTAX_TABLE_BYTE_TO_CHAR. */
4419 int charpos
= SYNTAX_TABLE_BYTE_TO_CHAR (POS_AS_IN_BUFFER (startpos
));
4421 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object
, charpos
, 1);
4425 /* Loop through the string, looking for a place to start matching. */
4428 /* If the pattern is anchored,
4429 skip quickly past places we cannot match.
4430 We don't bother to treat startpos == 0 specially
4431 because that case doesn't repeat. */
4432 if (anchored_start
&& startpos
> 0)
4434 if (! ((startpos
<= size1
? string1
[startpos
- 1]
4435 : string2
[startpos
- size1
- 1])
4440 /* If a fastmap is supplied, skip quickly over characters that
4441 cannot be the start of a match. If the pattern can match the
4442 null string, however, we don't need to skip characters; we want
4443 the first null string. */
4444 if (fastmap
&& startpos
< total_size
&& !bufp
->can_be_null
)
4446 register re_char
*d
;
4447 register re_wchar_t buf_ch
;
4449 d
= POS_ADDR_VSTRING (startpos
);
4451 if (range
> 0) /* Searching forwards. */
4453 register int lim
= 0;
4456 if (startpos
< size1
&& startpos
+ range
>= size1
)
4457 lim
= range
- (size1
- startpos
);
4459 /* Written out as an if-else to avoid testing `translate'
4461 if (RE_TRANSLATE_P (translate
))
4468 buf_ch
= STRING_CHAR_AND_LENGTH (d
, buf_charlen
);
4469 buf_ch
= RE_TRANSLATE (translate
, buf_ch
);
4470 if (fastmap
[CHAR_LEADING_CODE (buf_ch
)])
4473 range
-= buf_charlen
;
4479 register re_wchar_t ch
, translated
;
4482 ch
= RE_CHAR_TO_MULTIBYTE (buf_ch
);
4483 translated
= RE_TRANSLATE (translate
, ch
);
4484 if (translated
!= ch
4485 && (ch
= RE_CHAR_TO_UNIBYTE (translated
)) >= 0)
4487 if (fastmap
[buf_ch
])
4500 buf_ch
= STRING_CHAR_AND_LENGTH (d
, buf_charlen
);
4501 if (fastmap
[CHAR_LEADING_CODE (buf_ch
)])
4503 range
-= buf_charlen
;
4507 while (range
> lim
&& !fastmap
[*d
])
4513 startpos
+= irange
- range
;
4515 else /* Searching backwards. */
4519 buf_ch
= STRING_CHAR (d
);
4520 buf_ch
= TRANSLATE (buf_ch
);
4521 if (! fastmap
[CHAR_LEADING_CODE (buf_ch
)])
4526 register re_wchar_t ch
, translated
;
4529 ch
= RE_CHAR_TO_MULTIBYTE (buf_ch
);
4530 translated
= TRANSLATE (ch
);
4531 if (translated
!= ch
4532 && (ch
= RE_CHAR_TO_UNIBYTE (translated
)) >= 0)
4534 if (! fastmap
[TRANSLATE (buf_ch
)])
4540 /* If can't match the null string, and that's all we have left, fail. */
4541 if (range
>= 0 && startpos
== total_size
&& fastmap
4542 && !bufp
->can_be_null
)
4545 val
= re_match_2_internal (bufp
, string1
, size1
, string2
, size2
,
4546 startpos
, regs
, stop
);
4559 /* Update STARTPOS to the next character boundary. */
4562 re_char
*p
= POS_ADDR_VSTRING (startpos
);
4563 re_char
*pend
= STOP_ADDR_VSTRING (startpos
);
4564 int len
= BYTES_BY_CHAR_HEAD (*p
);
4582 /* Update STARTPOS to the previous character boundary. */
4585 re_char
*p
= POS_ADDR_VSTRING (startpos
) + 1;
4587 re_char
*phead
= HEAD_ADDR_VSTRING (startpos
);
4589 /* Find the head of multibyte form. */
4590 PREV_CHAR_BOUNDARY (p
, phead
);
4591 range
+= p0
- 1 - p
;
4595 startpos
-= p0
- 1 - p
;
4601 WEAK_ALIAS (__re_search_2
, re_search_2
)
4603 /* Declarations and macros for re_match_2. */
4605 static int bcmp_translate
_RE_ARGS((re_char
*s1
, re_char
*s2
,
4607 RE_TRANSLATE_TYPE translate
,
4608 const int multibyte
));
4610 /* This converts PTR, a pointer into one of the search strings `string1'
4611 and `string2' into an offset from the beginning of that string. */
4612 #define POINTER_TO_OFFSET(ptr) \
4613 (FIRST_STRING_P (ptr) \
4614 ? ((regoff_t) ((ptr) - string1)) \
4615 : ((regoff_t) ((ptr) - string2 + size1)))
4617 /* Call before fetching a character with *d. This switches over to
4618 string2 if necessary.
4619 Check re_match_2_internal for a discussion of why end_match_2 might
4620 not be within string2 (but be equal to end_match_1 instead). */
4621 #define PREFETCH() \
4624 /* End of string2 => fail. */ \
4625 if (dend == end_match_2) \
4627 /* End of string1 => advance to string2. */ \
4629 dend = end_match_2; \
4632 /* Call before fetching a char with *d if you already checked other limits.
4633 This is meant for use in lookahead operations like wordend, etc..
4634 where we might need to look at parts of the string that might be
4635 outside of the LIMITs (i.e past `stop'). */
4636 #define PREFETCH_NOLIMIT() \
4640 dend = end_match_2; \
4643 /* Test if at very beginning or at very end of the virtual concatenation
4644 of `string1' and `string2'. If only one string, it's `string2'. */
4645 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
4646 #define AT_STRINGS_END(d) ((d) == end2)
4649 /* Test if D points to a character which is word-constituent. We have
4650 two special cases to check for: if past the end of string1, look at
4651 the first character in string2; and if before the beginning of
4652 string2, look at the last character in string1. */
4653 #define WORDCHAR_P(d) \
4654 (SYNTAX ((d) == end1 ? *string2 \
4655 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
4658 /* Disabled due to a compiler bug -- see comment at case wordbound */
4660 /* The comment at case wordbound is following one, but we don't use
4661 AT_WORD_BOUNDARY anymore to support multibyte form.
4663 The DEC Alpha C compiler 3.x generates incorrect code for the
4664 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4665 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4666 macro and introducing temporary variables works around the bug. */
4669 /* Test if the character before D and the one at D differ with respect
4670 to being word-constituent. */
4671 #define AT_WORD_BOUNDARY(d) \
4672 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
4673 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
4676 /* Free everything we malloc. */
4677 #ifdef MATCH_MAY_ALLOCATE
4678 # define FREE_VAR(var) if (var) { REGEX_FREE (var); var = NULL; } else
4679 # define FREE_VARIABLES() \
4681 REGEX_FREE_STACK (fail_stack.stack); \
4682 FREE_VAR (regstart); \
4683 FREE_VAR (regend); \
4684 FREE_VAR (best_regstart); \
4685 FREE_VAR (best_regend); \
4688 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
4689 #endif /* not MATCH_MAY_ALLOCATE */
4692 /* Optimization routines. */
4694 /* If the operation is a match against one or more chars,
4695 return a pointer to the next operation, else return NULL. */
4697 skip_one_char (const re_char
*p
)
4699 switch (SWITCH_ENUM_CAST (*p
++))
4710 if (CHARSET_RANGE_TABLE_EXISTS_P (p
- 1))
4713 p
= CHARSET_RANGE_TABLE (p
- 1);
4714 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
4715 p
= CHARSET_RANGE_TABLE_END (p
, mcnt
);
4718 p
+= 1 + CHARSET_BITMAP_SIZE (p
- 1);
4725 case notcategoryspec
:
4737 /* Jump over non-matching operations. */
4739 skip_noops (const re_char
*p
, const re_char
*pend
)
4744 switch (SWITCH_ENUM_CAST ((re_opcode_t
) *p
))
4753 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
4764 /* Non-zero if "p1 matches something" implies "p2 fails". */
4766 mutually_exclusive_p (struct re_pattern_buffer
*bufp
, const re_char
*p1
, const re_char
*p2
)
4769 const boolean multibyte
= RE_MULTIBYTE_P (bufp
);
4770 unsigned char *pend
= bufp
->buffer
+ bufp
->used
;
4772 assert (p1
>= bufp
->buffer
&& p1
< pend
4773 && p2
>= bufp
->buffer
&& p2
<= pend
);
4775 /* Skip over open/close-group commands.
4776 If what follows this loop is a ...+ construct,
4777 look at what begins its body, since we will have to
4778 match at least one of that. */
4779 p2
= skip_noops (p2
, pend
);
4780 /* The same skip can be done for p1, except that this function
4781 is only used in the case where p1 is a simple match operator. */
4782 /* p1 = skip_noops (p1, pend); */
4784 assert (p1
>= bufp
->buffer
&& p1
< pend
4785 && p2
>= bufp
->buffer
&& p2
<= pend
);
4787 op2
= p2
== pend
? succeed
: *p2
;
4789 switch (SWITCH_ENUM_CAST (op2
))
4793 /* If we're at the end of the pattern, we can change. */
4794 if (skip_one_char (p1
))
4796 DEBUG_PRINT1 (" End of pattern: fast loop.\n");
4804 register re_wchar_t c
4805 = (re_opcode_t
) *p2
== endline
? '\n'
4806 : RE_STRING_CHAR (p2
+ 2, multibyte
);
4808 if ((re_opcode_t
) *p1
== exactn
)
4810 if (c
!= RE_STRING_CHAR (p1
+ 2, multibyte
))
4812 DEBUG_PRINT3 (" '%c' != '%c' => fast loop.\n", c
, p1
[2]);
4817 else if ((re_opcode_t
) *p1
== charset
4818 || (re_opcode_t
) *p1
== charset_not
)
4820 int not = (re_opcode_t
) *p1
== charset_not
;
4822 /* Test if C is listed in charset (or charset_not)
4824 if (! multibyte
|| IS_REAL_ASCII (c
))
4826 if (c
< CHARSET_BITMAP_SIZE (p1
) * BYTEWIDTH
4827 && p1
[2 + c
/ BYTEWIDTH
] & (1 << (c
% BYTEWIDTH
)))
4830 else if (CHARSET_RANGE_TABLE_EXISTS_P (p1
))
4831 CHARSET_LOOKUP_RANGE_TABLE (not, c
, p1
);
4833 /* `not' is equal to 1 if c would match, which means
4834 that we can't change to pop_failure_jump. */
4837 DEBUG_PRINT1 (" No match => fast loop.\n");
4841 else if ((re_opcode_t
) *p1
== anychar
4844 DEBUG_PRINT1 (" . != \\n => fast loop.\n");
4852 if ((re_opcode_t
) *p1
== exactn
)
4853 /* Reuse the code above. */
4854 return mutually_exclusive_p (bufp
, p2
, p1
);
4856 /* It is hard to list up all the character in charset
4857 P2 if it includes multibyte character. Give up in
4859 else if (!multibyte
|| !CHARSET_RANGE_TABLE_EXISTS_P (p2
))
4861 /* Now, we are sure that P2 has no range table.
4862 So, for the size of bitmap in P2, `p2[1]' is
4863 enough. But P1 may have range table, so the
4864 size of bitmap table of P1 is extracted by
4865 using macro `CHARSET_BITMAP_SIZE'.
4867 In a multibyte case, we know that all the character
4868 listed in P2 is ASCII. In a unibyte case, P1 has only a
4869 bitmap table. So, in both cases, it is enough to test
4870 only the bitmap table of P1. */
4872 if ((re_opcode_t
) *p1
== charset
)
4875 /* We win if the charset inside the loop
4876 has no overlap with the one after the loop. */
4879 && idx
< CHARSET_BITMAP_SIZE (p1
));
4881 if ((p2
[2 + idx
] & p1
[2 + idx
]) != 0)
4885 || idx
== CHARSET_BITMAP_SIZE (p1
))
4887 DEBUG_PRINT1 (" No match => fast loop.\n");
4891 else if ((re_opcode_t
) *p1
== charset_not
)
4894 /* We win if the charset_not inside the loop lists
4895 every character listed in the charset after. */
4896 for (idx
= 0; idx
< (int) p2
[1]; idx
++)
4897 if (! (p2
[2 + idx
] == 0
4898 || (idx
< CHARSET_BITMAP_SIZE (p1
)
4899 && ((p2
[2 + idx
] & ~ p1
[2 + idx
]) == 0))))
4904 DEBUG_PRINT1 (" No match => fast loop.\n");
4913 switch (SWITCH_ENUM_CAST (*p1
))
4917 /* Reuse the code above. */
4918 return mutually_exclusive_p (bufp
, p2
, p1
);
4920 /* When we have two charset_not, it's very unlikely that
4921 they don't overlap. The union of the two sets of excluded
4922 chars should cover all possible chars, which, as a matter of
4923 fact, is virtually impossible in multibyte buffers. */
4929 return ((re_opcode_t
) *p1
== syntaxspec
&& p1
[1] == Sword
);
4931 return ((re_opcode_t
) *p1
== syntaxspec
4932 && (p1
[1] == Ssymbol
|| p1
[1] == Sword
));
4934 return ((re_opcode_t
) *p1
== syntaxspec
&& p1
[1] == p2
[1]);
4937 return ((re_opcode_t
) *p1
== notsyntaxspec
&& p1
[1] == Sword
);
4939 return ((re_opcode_t
) *p1
== notsyntaxspec
4940 && (p1
[1] == Ssymbol
|| p1
[1] == Sword
));
4942 return ((re_opcode_t
) *p1
== notsyntaxspec
&& p1
[1] == p2
[1]);
4945 return (((re_opcode_t
) *p1
== notsyntaxspec
4946 || (re_opcode_t
) *p1
== syntaxspec
)
4951 return ((re_opcode_t
) *p1
== notcategoryspec
&& p1
[1] == p2
[1]);
4952 case notcategoryspec
:
4953 return ((re_opcode_t
) *p1
== categoryspec
&& p1
[1] == p2
[1]);
4965 /* Matching routines. */
4967 #ifndef emacs /* Emacs never uses this. */
4968 /* re_match is like re_match_2 except it takes only a single string. */
4971 re_match (struct re_pattern_buffer
*bufp
, const char *string
,
4972 int size
, int pos
, struct re_registers
*regs
)
4974 int result
= re_match_2_internal (bufp
, NULL
, 0, (re_char
*) string
, size
,
4978 WEAK_ALIAS (__re_match
, re_match
)
4979 #endif /* not emacs */
4982 /* In Emacs, this is the string or buffer in which we
4983 are matching. It is used for looking up syntax properties. */
4984 Lisp_Object re_match_object
;
4987 /* re_match_2 matches the compiled pattern in BUFP against the
4988 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
4989 and SIZE2, respectively). We start matching at POS, and stop
4992 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
4993 store offsets for the substring each group matched in REGS. See the
4994 documentation for exactly how many groups we fill.
4996 We return -1 if no match, -2 if an internal error (such as the
4997 failure stack overflowing). Otherwise, we return the length of the
4998 matched substring. */
5001 re_match_2 (struct re_pattern_buffer
*bufp
, const char *string1
, int size1
, const char *string2
, int size2
, int pos
, struct re_registers
*regs
, int stop
)
5007 gl_state
.object
= re_match_object
; /* Used by SYNTAX_TABLE_BYTE_TO_CHAR. */
5008 charpos
= SYNTAX_TABLE_BYTE_TO_CHAR (POS_AS_IN_BUFFER (pos
));
5009 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object
, charpos
, 1);
5012 result
= re_match_2_internal (bufp
, (re_char
*) string1
, size1
,
5013 (re_char
*) string2
, size2
,
5017 WEAK_ALIAS (__re_match_2
, re_match_2
)
5020 /* This is a separate function so that we can force an alloca cleanup
5023 re_match_2_internal (struct re_pattern_buffer
*bufp
, const re_char
*string1
, int size1
, const re_char
*string2
, int size2
, int pos
, struct re_registers
*regs
, int stop
)
5025 /* General temporaries. */
5030 /* Just past the end of the corresponding string. */
5031 re_char
*end1
, *end2
;
5033 /* Pointers into string1 and string2, just past the last characters in
5034 each to consider matching. */
5035 re_char
*end_match_1
, *end_match_2
;
5037 /* Where we are in the data, and the end of the current string. */
5040 /* Used sometimes to remember where we were before starting matching
5041 an operator so that we can go back in case of failure. This "atomic"
5042 behavior of matching opcodes is indispensable to the correctness
5043 of the on_failure_keep_string_jump optimization. */
5046 /* Where we are in the pattern, and the end of the pattern. */
5047 re_char
*p
= bufp
->buffer
;
5048 re_char
*pend
= p
+ bufp
->used
;
5050 /* We use this to map every character in the string. */
5051 RE_TRANSLATE_TYPE translate
= bufp
->translate
;
5053 /* Nonzero if BUFP is setup from a multibyte regex. */
5054 const boolean multibyte
= RE_MULTIBYTE_P (bufp
);
5056 /* Nonzero if STRING1/STRING2 are multibyte. */
5057 const boolean target_multibyte
= RE_TARGET_MULTIBYTE_P (bufp
);
5059 /* Failure point stack. Each place that can handle a failure further
5060 down the line pushes a failure point on this stack. It consists of
5061 regstart, and regend for all registers corresponding to
5062 the subexpressions we're currently inside, plus the number of such
5063 registers, and, finally, two char *'s. The first char * is where
5064 to resume scanning the pattern; the second one is where to resume
5065 scanning the strings. */
5066 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5067 fail_stack_type fail_stack
;
5070 unsigned nfailure_points_pushed
= 0, nfailure_points_popped
= 0;
5073 #if defined REL_ALLOC && defined REGEX_MALLOC
5074 /* This holds the pointer to the failure stack, when
5075 it is allocated relocatably. */
5076 fail_stack_elt_t
*failure_stack_ptr
;
5079 /* We fill all the registers internally, independent of what we
5080 return, for use in backreferences. The number here includes
5081 an element for register zero. */
5082 size_t num_regs
= bufp
->re_nsub
+ 1;
5084 /* Information on the contents of registers. These are pointers into
5085 the input strings; they record just what was matched (on this
5086 attempt) by a subexpression part of the pattern, that is, the
5087 regnum-th regstart pointer points to where in the pattern we began
5088 matching and the regnum-th regend points to right after where we
5089 stopped matching the regnum-th subexpression. (The zeroth register
5090 keeps track of what the whole pattern matches.) */
5091 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5092 re_char
**regstart
, **regend
;
5095 /* The following record the register info as found in the above
5096 variables when we find a match better than any we've seen before.
5097 This happens as we backtrack through the failure points, which in
5098 turn happens only if we have not yet matched the entire string. */
5099 unsigned best_regs_set
= false;
5100 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5101 re_char
**best_regstart
, **best_regend
;
5104 /* Logically, this is `best_regend[0]'. But we don't want to have to
5105 allocate space for that if we're not allocating space for anything
5106 else (see below). Also, we never need info about register 0 for
5107 any of the other register vectors, and it seems rather a kludge to
5108 treat `best_regend' differently than the rest. So we keep track of
5109 the end of the best match so far in a separate variable. We
5110 initialize this to NULL so that when we backtrack the first time
5111 and need to test it, it's not garbage. */
5112 re_char
*match_end
= NULL
;
5115 /* Counts the total number of registers pushed. */
5116 unsigned num_regs_pushed
= 0;
5119 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
5123 #ifdef MATCH_MAY_ALLOCATE
5124 /* Do not bother to initialize all the register variables if there are
5125 no groups in the pattern, as it takes a fair amount of time. If
5126 there are groups, we include space for register 0 (the whole
5127 pattern), even though we never use it, since it simplifies the
5128 array indexing. We should fix this. */
5131 regstart
= REGEX_TALLOC (num_regs
, re_char
*);
5132 regend
= REGEX_TALLOC (num_regs
, re_char
*);
5133 best_regstart
= REGEX_TALLOC (num_regs
, re_char
*);
5134 best_regend
= REGEX_TALLOC (num_regs
, re_char
*);
5136 if (!(regstart
&& regend
&& best_regstart
&& best_regend
))
5144 /* We must initialize all our variables to NULL, so that
5145 `FREE_VARIABLES' doesn't try to free them. */
5146 regstart
= regend
= best_regstart
= best_regend
= NULL
;
5148 #endif /* MATCH_MAY_ALLOCATE */
5150 /* The starting position is bogus. */
5151 if (pos
< 0 || pos
> size1
+ size2
)
5157 /* Initialize subexpression text positions to -1 to mark ones that no
5158 start_memory/stop_memory has been seen for. Also initialize the
5159 register information struct. */
5160 for (reg
= 1; reg
< num_regs
; reg
++)
5161 regstart
[reg
] = regend
[reg
] = NULL
;
5163 /* We move `string1' into `string2' if the latter's empty -- but not if
5164 `string1' is null. */
5165 if (size2
== 0 && string1
!= NULL
)
5172 end1
= string1
+ size1
;
5173 end2
= string2
+ size2
;
5175 /* `p' scans through the pattern as `d' scans through the data.
5176 `dend' is the end of the input string that `d' points within. `d'
5177 is advanced into the following input string whenever necessary, but
5178 this happens before fetching; therefore, at the beginning of the
5179 loop, `d' can be pointing at the end of a string, but it cannot
5183 /* Only match within string2. */
5184 d
= string2
+ pos
- size1
;
5185 dend
= end_match_2
= string2
+ stop
- size1
;
5186 end_match_1
= end1
; /* Just to give it a value. */
5192 /* Only match within string1. */
5193 end_match_1
= string1
+ stop
;
5195 When we reach end_match_1, PREFETCH normally switches to string2.
5196 But in the present case, this means that just doing a PREFETCH
5197 makes us jump from `stop' to `gap' within the string.
5198 What we really want here is for the search to stop as
5199 soon as we hit end_match_1. That's why we set end_match_2
5200 to end_match_1 (since PREFETCH fails as soon as we hit
5202 end_match_2
= end_match_1
;
5205 { /* It's important to use this code when stop == size so that
5206 moving `d' from end1 to string2 will not prevent the d == dend
5207 check from catching the end of string. */
5209 end_match_2
= string2
+ stop
- size1
;
5215 DEBUG_PRINT1 ("The compiled pattern is: ");
5216 DEBUG_PRINT_COMPILED_PATTERN (bufp
, p
, pend
);
5217 DEBUG_PRINT1 ("The string to match is: `");
5218 DEBUG_PRINT_DOUBLE_STRING (d
, string1
, size1
, string2
, size2
);
5219 DEBUG_PRINT1 ("'\n");
5221 /* This loops over pattern commands. It exits by returning from the
5222 function if the match is complete, or it drops through if the match
5223 fails at this starting point in the input data. */
5226 DEBUG_PRINT2 ("\n%p: ", p
);
5229 { /* End of pattern means we might have succeeded. */
5230 DEBUG_PRINT1 ("end of pattern ... ");
5232 /* If we haven't matched the entire string, and we want the
5233 longest match, try backtracking. */
5234 if (d
!= end_match_2
)
5236 /* 1 if this match ends in the same string (string1 or string2)
5237 as the best previous match. */
5238 boolean same_str_p
= (FIRST_STRING_P (match_end
)
5239 == FIRST_STRING_P (d
));
5240 /* 1 if this match is the best seen so far. */
5241 boolean best_match_p
;
5243 /* AIX compiler got confused when this was combined
5244 with the previous declaration. */
5246 best_match_p
= d
> match_end
;
5248 best_match_p
= !FIRST_STRING_P (d
);
5250 DEBUG_PRINT1 ("backtracking.\n");
5252 if (!FAIL_STACK_EMPTY ())
5253 { /* More failure points to try. */
5255 /* If exceeds best match so far, save it. */
5256 if (!best_regs_set
|| best_match_p
)
5258 best_regs_set
= true;
5261 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
5263 for (reg
= 1; reg
< num_regs
; reg
++)
5265 best_regstart
[reg
] = regstart
[reg
];
5266 best_regend
[reg
] = regend
[reg
];
5272 /* If no failure points, don't restore garbage. And if
5273 last match is real best match, don't restore second
5275 else if (best_regs_set
&& !best_match_p
)
5278 /* Restore best match. It may happen that `dend ==
5279 end_match_1' while the restored d is in string2.
5280 For example, the pattern `x.*y.*z' against the
5281 strings `x-' and `y-z-', if the two strings are
5282 not consecutive in memory. */
5283 DEBUG_PRINT1 ("Restoring best registers.\n");
5286 dend
= ((d
>= string1
&& d
<= end1
)
5287 ? end_match_1
: end_match_2
);
5289 for (reg
= 1; reg
< num_regs
; reg
++)
5291 regstart
[reg
] = best_regstart
[reg
];
5292 regend
[reg
] = best_regend
[reg
];
5295 } /* d != end_match_2 */
5298 DEBUG_PRINT1 ("Accepting match.\n");
5300 /* If caller wants register contents data back, do it. */
5301 if (regs
&& !bufp
->no_sub
)
5303 /* Have the register data arrays been allocated? */
5304 if (bufp
->regs_allocated
== REGS_UNALLOCATED
)
5305 { /* No. So allocate them with malloc. We need one
5306 extra element beyond `num_regs' for the `-1' marker
5308 regs
->num_regs
= MAX (RE_NREGS
, num_regs
+ 1);
5309 regs
->start
= TALLOC (regs
->num_regs
, regoff_t
);
5310 regs
->end
= TALLOC (regs
->num_regs
, regoff_t
);
5311 if (regs
->start
== NULL
|| regs
->end
== NULL
)
5316 bufp
->regs_allocated
= REGS_REALLOCATE
;
5318 else if (bufp
->regs_allocated
== REGS_REALLOCATE
)
5319 { /* Yes. If we need more elements than were already
5320 allocated, reallocate them. If we need fewer, just
5322 if (regs
->num_regs
< num_regs
+ 1)
5324 regs
->num_regs
= num_regs
+ 1;
5325 RETALLOC (regs
->start
, regs
->num_regs
, regoff_t
);
5326 RETALLOC (regs
->end
, regs
->num_regs
, regoff_t
);
5327 if (regs
->start
== NULL
|| regs
->end
== NULL
)
5336 /* These braces fend off a "empty body in an else-statement"
5337 warning under GCC when assert expands to nothing. */
5338 assert (bufp
->regs_allocated
== REGS_FIXED
);
5341 /* Convert the pointer data in `regstart' and `regend' to
5342 indices. Register zero has to be set differently,
5343 since we haven't kept track of any info for it. */
5344 if (regs
->num_regs
> 0)
5346 regs
->start
[0] = pos
;
5347 regs
->end
[0] = POINTER_TO_OFFSET (d
);
5350 /* Go through the first `min (num_regs, regs->num_regs)'
5351 registers, since that is all we initialized. */
5352 for (reg
= 1; reg
< MIN (num_regs
, regs
->num_regs
); reg
++)
5354 if (REG_UNSET (regstart
[reg
]) || REG_UNSET (regend
[reg
]))
5355 regs
->start
[reg
] = regs
->end
[reg
] = -1;
5359 = (regoff_t
) POINTER_TO_OFFSET (regstart
[reg
]);
5361 = (regoff_t
) POINTER_TO_OFFSET (regend
[reg
]);
5365 /* If the regs structure we return has more elements than
5366 were in the pattern, set the extra elements to -1. If
5367 we (re)allocated the registers, this is the case,
5368 because we always allocate enough to have at least one
5370 for (reg
= num_regs
; reg
< regs
->num_regs
; reg
++)
5371 regs
->start
[reg
] = regs
->end
[reg
] = -1;
5372 } /* regs && !bufp->no_sub */
5374 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
5375 nfailure_points_pushed
, nfailure_points_popped
,
5376 nfailure_points_pushed
- nfailure_points_popped
);
5377 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed
);
5379 mcnt
= POINTER_TO_OFFSET (d
) - pos
;
5381 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt
);
5387 /* Otherwise match next pattern command. */
5388 switch (SWITCH_ENUM_CAST ((re_opcode_t
) *p
++))
5390 /* Ignore these. Used to ignore the n of succeed_n's which
5391 currently have n == 0. */
5393 DEBUG_PRINT1 ("EXECUTING no_op.\n");
5397 DEBUG_PRINT1 ("EXECUTING succeed.\n");
5400 /* Match the next n pattern characters exactly. The following
5401 byte in the pattern defines n, and the n bytes after that
5402 are the characters to match. */
5405 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt
);
5407 /* Remember the start point to rollback upon failure. */
5411 /* This is written out as an if-else so we don't waste time
5412 testing `translate' inside the loop. */
5413 if (RE_TRANSLATE_P (translate
))
5417 if (RE_TRANSLATE (translate
, *d
) != *p
++)
5437 /* The cost of testing `translate' is comparatively small. */
5438 if (target_multibyte
)
5441 int pat_charlen
, buf_charlen
;
5446 pat_ch
= STRING_CHAR_AND_LENGTH (p
, pat_charlen
);
5449 pat_ch
= RE_CHAR_TO_MULTIBYTE (*p
);
5452 buf_ch
= STRING_CHAR_AND_LENGTH (d
, buf_charlen
);
5454 if (TRANSLATE (buf_ch
) != pat_ch
)
5462 mcnt
-= pat_charlen
;
5468 int pat_charlen
, buf_charlen
;
5474 pat_ch
= STRING_CHAR_AND_LENGTH (p
, pat_charlen
);
5475 pat_ch
= RE_CHAR_TO_UNIBYTE (pat_ch
);
5482 buf_ch
= RE_CHAR_TO_MULTIBYTE (*d
);
5483 if (! CHAR_BYTE8_P (buf_ch
))
5485 buf_ch
= TRANSLATE (buf_ch
);
5486 buf_ch
= RE_CHAR_TO_UNIBYTE (buf_ch
);
5492 if (buf_ch
!= pat_ch
)
5505 /* Match any character except possibly a newline or a null. */
5511 DEBUG_PRINT1 ("EXECUTING anychar.\n");
5514 buf_ch
= RE_STRING_CHAR_AND_LENGTH (d
, buf_charlen
,
5516 buf_ch
= TRANSLATE (buf_ch
);
5518 if ((!(bufp
->syntax
& RE_DOT_NEWLINE
)
5520 || ((bufp
->syntax
& RE_DOT_NOT_NULL
)
5521 && buf_ch
== '\000'))
5524 DEBUG_PRINT2 (" Matched `%d'.\n", *d
);
5533 register unsigned int c
;
5534 boolean
not = (re_opcode_t
) *(p
- 1) == charset_not
;
5537 /* Start of actual range_table, or end of bitmap if there is no
5539 re_char
*range_table
;
5541 /* Nonzero if there is a range table. */
5542 int range_table_exists
;
5544 /* Number of ranges of range table. This is not included
5545 in the initial byte-length of the command. */
5548 /* Whether matching against a unibyte character. */
5549 boolean unibyte_char
= false;
5551 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
5553 range_table_exists
= CHARSET_RANGE_TABLE_EXISTS_P (&p
[-1]);
5555 if (range_table_exists
)
5557 range_table
= CHARSET_RANGE_TABLE (&p
[-1]); /* Past the bitmap. */
5558 EXTRACT_NUMBER_AND_INCR (count
, range_table
);
5562 c
= RE_STRING_CHAR_AND_LENGTH (d
, len
, target_multibyte
);
5563 if (target_multibyte
)
5568 c1
= RE_CHAR_TO_UNIBYTE (c
);
5571 unibyte_char
= true;
5577 int c1
= RE_CHAR_TO_MULTIBYTE (c
);
5579 if (! CHAR_BYTE8_P (c1
))
5581 c1
= TRANSLATE (c1
);
5582 c1
= RE_CHAR_TO_UNIBYTE (c1
);
5585 unibyte_char
= true;
5590 unibyte_char
= true;
5593 if (unibyte_char
&& c
< (1 << BYTEWIDTH
))
5594 { /* Lookup bitmap. */
5595 /* Cast to `unsigned' instead of `unsigned char' in
5596 case the bit list is a full 32 bytes long. */
5597 if (c
< (unsigned) (CHARSET_BITMAP_SIZE (&p
[-1]) * BYTEWIDTH
)
5598 && p
[1 + c
/ BYTEWIDTH
] & (1 << (c
% BYTEWIDTH
)))
5602 else if (range_table_exists
)
5604 int class_bits
= CHARSET_RANGE_TABLE_BITS (&p
[-1]);
5606 if ( (class_bits
& BIT_LOWER
&& ISLOWER (c
))
5607 | (class_bits
& BIT_MULTIBYTE
)
5608 | (class_bits
& BIT_PUNCT
&& ISPUNCT (c
))
5609 | (class_bits
& BIT_SPACE
&& ISSPACE (c
))
5610 | (class_bits
& BIT_UPPER
&& ISUPPER (c
))
5611 | (class_bits
& BIT_WORD
&& ISWORD (c
)))
5614 CHARSET_LOOKUP_RANGE_TABLE_RAW (not, c
, range_table
, count
);
5618 if (range_table_exists
)
5619 p
= CHARSET_RANGE_TABLE_END (range_table
, count
);
5621 p
+= CHARSET_BITMAP_SIZE (&p
[-1]) + 1;
5623 if (!not) goto fail
;
5630 /* The beginning of a group is represented by start_memory.
5631 The argument is the register number. The text
5632 matched within the group is recorded (in the internal
5633 registers data structure) under the register number. */
5635 DEBUG_PRINT2 ("EXECUTING start_memory %d:\n", *p
);
5637 /* In case we need to undo this operation (via backtracking). */
5638 PUSH_FAILURE_REG ((unsigned int)*p
);
5641 regend
[*p
] = NULL
; /* probably unnecessary. -sm */
5642 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart
[*p
]));
5644 /* Move past the register number and inner group count. */
5649 /* The stop_memory opcode represents the end of a group. Its
5650 argument is the same as start_memory's: the register number. */
5652 DEBUG_PRINT2 ("EXECUTING stop_memory %d:\n", *p
);
5654 assert (!REG_UNSET (regstart
[*p
]));
5655 /* Strictly speaking, there should be code such as:
5657 assert (REG_UNSET (regend[*p]));
5658 PUSH_FAILURE_REGSTOP ((unsigned int)*p);
5660 But the only info to be pushed is regend[*p] and it is known to
5661 be UNSET, so there really isn't anything to push.
5662 Not pushing anything, on the other hand deprives us from the
5663 guarantee that regend[*p] is UNSET since undoing this operation
5664 will not reset its value properly. This is not important since
5665 the value will only be read on the next start_memory or at
5666 the very end and both events can only happen if this stop_memory
5670 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend
[*p
]));
5672 /* Move past the register number and the inner group count. */
5677 /* \<digit> has been turned into a `duplicate' command which is
5678 followed by the numeric value of <digit> as the register number. */
5681 register re_char
*d2
, *dend2
;
5682 int regno
= *p
++; /* Get which register to match against. */
5683 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno
);
5685 /* Can't back reference a group which we've never matched. */
5686 if (REG_UNSET (regstart
[regno
]) || REG_UNSET (regend
[regno
]))
5689 /* Where in input to try to start matching. */
5690 d2
= regstart
[regno
];
5692 /* Remember the start point to rollback upon failure. */
5695 /* Where to stop matching; if both the place to start and
5696 the place to stop matching are in the same string, then
5697 set to the place to stop, otherwise, for now have to use
5698 the end of the first string. */
5700 dend2
= ((FIRST_STRING_P (regstart
[regno
])
5701 == FIRST_STRING_P (regend
[regno
]))
5702 ? regend
[regno
] : end_match_1
);
5705 /* If necessary, advance to next segment in register
5709 if (dend2
== end_match_2
) break;
5710 if (dend2
== regend
[regno
]) break;
5712 /* End of string1 => advance to string2. */
5714 dend2
= regend
[regno
];
5716 /* At end of register contents => success */
5717 if (d2
== dend2
) break;
5719 /* If necessary, advance to next segment in data. */
5722 /* How many characters left in this segment to match. */
5725 /* Want how many consecutive characters we can match in
5726 one shot, so, if necessary, adjust the count. */
5727 if (mcnt
> dend2
- d2
)
5730 /* Compare that many; failure if mismatch, else move
5732 if (RE_TRANSLATE_P (translate
)
5733 ? bcmp_translate (d
, d2
, mcnt
, translate
, target_multibyte
)
5734 : memcmp (d
, d2
, mcnt
))
5739 d
+= mcnt
, d2
+= mcnt
;
5745 /* begline matches the empty string at the beginning of the string
5746 (unless `not_bol' is set in `bufp'), and after newlines. */
5748 DEBUG_PRINT1 ("EXECUTING begline.\n");
5750 if (AT_STRINGS_BEG (d
))
5752 if (!bufp
->not_bol
) break;
5757 GET_CHAR_BEFORE_2 (c
, d
, string1
, end1
, string2
, end2
);
5761 /* In all other cases, we fail. */
5765 /* endline is the dual of begline. */
5767 DEBUG_PRINT1 ("EXECUTING endline.\n");
5769 if (AT_STRINGS_END (d
))
5771 if (!bufp
->not_eol
) break;
5775 PREFETCH_NOLIMIT ();
5782 /* Match at the very beginning of the data. */
5784 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
5785 if (AT_STRINGS_BEG (d
))
5790 /* Match at the very end of the data. */
5792 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
5793 if (AT_STRINGS_END (d
))
5798 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
5799 pushes NULL as the value for the string on the stack. Then
5800 `POP_FAILURE_POINT' will keep the current value for the
5801 string, instead of restoring it. To see why, consider
5802 matching `foo\nbar' against `.*\n'. The .* matches the foo;
5803 then the . fails against the \n. But the next thing we want
5804 to do is match the \n against the \n; if we restored the
5805 string value, we would be back at the foo.
5807 Because this is used only in specific cases, we don't need to
5808 check all the things that `on_failure_jump' does, to make
5809 sure the right things get saved on the stack. Hence we don't
5810 share its code. The only reason to push anything on the
5811 stack at all is that otherwise we would have to change
5812 `anychar's code to do something besides goto fail in this
5813 case; that seems worse than this. */
5814 case on_failure_keep_string_jump
:
5815 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
5816 DEBUG_PRINT3 ("EXECUTING on_failure_keep_string_jump %d (to %p):\n",
5819 PUSH_FAILURE_POINT (p
- 3, NULL
);
5822 /* A nasty loop is introduced by the non-greedy *? and +?.
5823 With such loops, the stack only ever contains one failure point
5824 at a time, so that a plain on_failure_jump_loop kind of
5825 cycle detection cannot work. Worse yet, such a detection
5826 can not only fail to detect a cycle, but it can also wrongly
5827 detect a cycle (between different instantiations of the same
5829 So the method used for those nasty loops is a little different:
5830 We use a special cycle-detection-stack-frame which is pushed
5831 when the on_failure_jump_nastyloop failure-point is *popped*.
5832 This special frame thus marks the beginning of one iteration
5833 through the loop and we can hence easily check right here
5834 whether something matched between the beginning and the end of
5836 case on_failure_jump_nastyloop
:
5837 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
5838 DEBUG_PRINT3 ("EXECUTING on_failure_jump_nastyloop %d (to %p):\n",
5841 assert ((re_opcode_t
)p
[-4] == no_op
);
5844 CHECK_INFINITE_LOOP (p
- 4, d
);
5846 /* If there's a cycle, just continue without pushing
5847 this failure point. The failure point is the "try again"
5848 option, which shouldn't be tried.
5849 We want (x?)*?y\1z to match both xxyz and xxyxz. */
5850 PUSH_FAILURE_POINT (p
- 3, d
);
5854 /* Simple loop detecting on_failure_jump: just check on the
5855 failure stack if the same spot was already hit earlier. */
5856 case on_failure_jump_loop
:
5858 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
5859 DEBUG_PRINT3 ("EXECUTING on_failure_jump_loop %d (to %p):\n",
5863 CHECK_INFINITE_LOOP (p
- 3, d
);
5865 /* If there's a cycle, get out of the loop, as if the matching
5866 had failed. We used to just `goto fail' here, but that was
5867 aborting the search a bit too early: we want to keep the
5868 empty-loop-match and keep matching after the loop.
5869 We want (x?)*y\1z to match both xxyz and xxyxz. */
5872 PUSH_FAILURE_POINT (p
- 3, d
);
5877 /* Uses of on_failure_jump:
5879 Each alternative starts with an on_failure_jump that points
5880 to the beginning of the next alternative. Each alternative
5881 except the last ends with a jump that in effect jumps past
5882 the rest of the alternatives. (They really jump to the
5883 ending jump of the following alternative, because tensioning
5884 these jumps is a hassle.)
5886 Repeats start with an on_failure_jump that points past both
5887 the repetition text and either the following jump or
5888 pop_failure_jump back to this on_failure_jump. */
5889 case on_failure_jump
:
5890 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
5891 DEBUG_PRINT3 ("EXECUTING on_failure_jump %d (to %p):\n",
5894 PUSH_FAILURE_POINT (p
-3, d
);
5897 /* This operation is used for greedy *.
5898 Compare the beginning of the repeat with what in the
5899 pattern follows its end. If we can establish that there
5900 is nothing that they would both match, i.e., that we
5901 would have to backtrack because of (as in, e.g., `a*a')
5902 then we can use a non-backtracking loop based on
5903 on_failure_keep_string_jump instead of on_failure_jump. */
5904 case on_failure_jump_smart
:
5905 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
5906 DEBUG_PRINT3 ("EXECUTING on_failure_jump_smart %d (to %p).\n",
5909 re_char
*p1
= p
; /* Next operation. */
5910 /* Here, we discard `const', making re_match non-reentrant. */
5911 unsigned char *p2
= (unsigned char*) p
+ mcnt
; /* Jump dest. */
5912 unsigned char *p3
= (unsigned char*) p
- 3; /* opcode location. */
5914 p
-= 3; /* Reset so that we will re-execute the
5915 instruction once it's been changed. */
5917 EXTRACT_NUMBER (mcnt
, p2
- 2);
5919 /* Ensure this is a indeed the trivial kind of loop
5920 we are expecting. */
5921 assert (skip_one_char (p1
) == p2
- 3);
5922 assert ((re_opcode_t
) p2
[-3] == jump
&& p2
+ mcnt
== p
);
5923 DEBUG_STATEMENT (debug
+= 2);
5924 if (mutually_exclusive_p (bufp
, p1
, p2
))
5926 /* Use a fast `on_failure_keep_string_jump' loop. */
5927 DEBUG_PRINT1 (" smart exclusive => fast loop.\n");
5928 *p3
= (unsigned char) on_failure_keep_string_jump
;
5929 STORE_NUMBER (p2
- 2, mcnt
+ 3);
5933 /* Default to a safe `on_failure_jump' loop. */
5934 DEBUG_PRINT1 (" smart default => slow loop.\n");
5935 *p3
= (unsigned char) on_failure_jump
;
5937 DEBUG_STATEMENT (debug
-= 2);
5941 /* Unconditionally jump (without popping any failure points). */
5944 IMMEDIATE_QUIT_CHECK
;
5945 EXTRACT_NUMBER_AND_INCR (mcnt
, p
); /* Get the amount to jump. */
5946 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt
);
5947 p
+= mcnt
; /* Do the jump. */
5948 DEBUG_PRINT2 ("(to %p).\n", p
);
5952 /* Have to succeed matching what follows at least n times.
5953 After that, handle like `on_failure_jump'. */
5955 /* Signedness doesn't matter since we only compare MCNT to 0. */
5956 EXTRACT_NUMBER (mcnt
, p
+ 2);
5957 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt
);
5959 /* Originally, mcnt is how many times we HAVE to succeed. */
5962 /* Here, we discard `const', making re_match non-reentrant. */
5963 unsigned char *p2
= (unsigned char*) p
+ 2; /* counter loc. */
5966 PUSH_NUMBER (p2
, mcnt
);
5969 /* The two bytes encoding mcnt == 0 are two no_op opcodes. */
5974 /* Signedness doesn't matter since we only compare MCNT to 0. */
5975 EXTRACT_NUMBER (mcnt
, p
+ 2);
5976 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt
);
5978 /* Originally, this is how many times we CAN jump. */
5981 /* Here, we discard `const', making re_match non-reentrant. */
5982 unsigned char *p2
= (unsigned char*) p
+ 2; /* counter loc. */
5984 PUSH_NUMBER (p2
, mcnt
);
5985 goto unconditional_jump
;
5987 /* If don't have to jump any more, skip over the rest of command. */
5994 unsigned char *p2
; /* Location of the counter. */
5995 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
5997 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
5998 /* Here, we discard `const', making re_match non-reentrant. */
5999 p2
= (unsigned char*) p
+ mcnt
;
6000 /* Signedness doesn't matter since we only copy MCNT's bits . */
6001 EXTRACT_NUMBER_AND_INCR (mcnt
, p
);
6002 DEBUG_PRINT3 (" Setting %p to %d.\n", p2
, mcnt
);
6003 PUSH_NUMBER (p2
, mcnt
);
6009 not = (re_opcode_t
) *(p
- 1) == notwordbound
;
6010 DEBUG_PRINT2 ("EXECUTING %swordbound.\n", not?"not":"");
6012 /* We SUCCEED (or FAIL) in one of the following cases: */
6014 /* Case 1: D is at the beginning or the end of string. */
6015 if (AT_STRINGS_BEG (d
) || AT_STRINGS_END (d
))
6019 /* C1 is the character before D, S1 is the syntax of C1, C2
6020 is the character at D, and S2 is the syntax of C2. */
6025 int offset
= PTR_TO_OFFSET (d
- 1);
6026 int charpos
= SYNTAX_TABLE_BYTE_TO_CHAR (offset
);
6027 UPDATE_SYNTAX_TABLE (charpos
);
6029 GET_CHAR_BEFORE_2 (c1
, d
, string1
, end1
, string2
, end2
);
6032 UPDATE_SYNTAX_TABLE_FORWARD (charpos
+ 1);
6034 PREFETCH_NOLIMIT ();
6035 GET_CHAR_AFTER (c2
, d
, dummy
);
6038 if (/* Case 2: Only one of S1 and S2 is Sword. */
6039 ((s1
== Sword
) != (s2
== Sword
))
6040 /* Case 3: Both of S1 and S2 are Sword, and macro
6041 WORD_BOUNDARY_P (C1, C2) returns nonzero. */
6042 || ((s1
== Sword
) && WORD_BOUNDARY_P (c1
, c2
)))
6051 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
6053 /* We FAIL in one of the following cases: */
6055 /* Case 1: D is at the end of string. */
6056 if (AT_STRINGS_END (d
))
6060 /* C1 is the character before D, S1 is the syntax of C1, C2
6061 is the character at D, and S2 is the syntax of C2. */
6066 int offset
= PTR_TO_OFFSET (d
);
6067 int charpos
= SYNTAX_TABLE_BYTE_TO_CHAR (offset
);
6068 UPDATE_SYNTAX_TABLE (charpos
);
6071 GET_CHAR_AFTER (c2
, d
, dummy
);
6074 /* Case 2: S2 is not Sword. */
6078 /* Case 3: D is not at the beginning of string ... */
6079 if (!AT_STRINGS_BEG (d
))
6081 GET_CHAR_BEFORE_2 (c1
, d
, string1
, end1
, string2
, end2
);
6083 UPDATE_SYNTAX_TABLE_BACKWARD (charpos
- 1);
6087 /* ... and S1 is Sword, and WORD_BOUNDARY_P (C1, C2)
6089 if ((s1
== Sword
) && !WORD_BOUNDARY_P (c1
, c2
))
6096 DEBUG_PRINT1 ("EXECUTING wordend.\n");
6098 /* We FAIL in one of the following cases: */
6100 /* Case 1: D is at the beginning of string. */
6101 if (AT_STRINGS_BEG (d
))
6105 /* C1 is the character before D, S1 is the syntax of C1, C2
6106 is the character at D, and S2 is the syntax of C2. */
6111 int offset
= PTR_TO_OFFSET (d
) - 1;
6112 int charpos
= SYNTAX_TABLE_BYTE_TO_CHAR (offset
);
6113 UPDATE_SYNTAX_TABLE (charpos
);
6115 GET_CHAR_BEFORE_2 (c1
, d
, string1
, end1
, string2
, end2
);
6118 /* Case 2: S1 is not Sword. */
6122 /* Case 3: D is not at the end of string ... */
6123 if (!AT_STRINGS_END (d
))
6125 PREFETCH_NOLIMIT ();
6126 GET_CHAR_AFTER (c2
, d
, dummy
);
6128 UPDATE_SYNTAX_TABLE_FORWARD (charpos
);
6132 /* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2)
6134 if ((s2
== Sword
) && !WORD_BOUNDARY_P (c1
, c2
))
6141 DEBUG_PRINT1 ("EXECUTING symbeg.\n");
6143 /* We FAIL in one of the following cases: */
6145 /* Case 1: D is at the end of string. */
6146 if (AT_STRINGS_END (d
))
6150 /* C1 is the character before D, S1 is the syntax of C1, C2
6151 is the character at D, and S2 is the syntax of C2. */
6155 int offset
= PTR_TO_OFFSET (d
);
6156 int charpos
= SYNTAX_TABLE_BYTE_TO_CHAR (offset
);
6157 UPDATE_SYNTAX_TABLE (charpos
);
6160 c2
= RE_STRING_CHAR (d
, target_multibyte
);
6163 /* Case 2: S2 is neither Sword nor Ssymbol. */
6164 if (s2
!= Sword
&& s2
!= Ssymbol
)
6167 /* Case 3: D is not at the beginning of string ... */
6168 if (!AT_STRINGS_BEG (d
))
6170 GET_CHAR_BEFORE_2 (c1
, d
, string1
, end1
, string2
, end2
);
6172 UPDATE_SYNTAX_TABLE_BACKWARD (charpos
- 1);
6176 /* ... and S1 is Sword or Ssymbol. */
6177 if (s1
== Sword
|| s1
== Ssymbol
)
6184 DEBUG_PRINT1 ("EXECUTING symend.\n");
6186 /* We FAIL in one of the following cases: */
6188 /* Case 1: D is at the beginning of string. */
6189 if (AT_STRINGS_BEG (d
))
6193 /* C1 is the character before D, S1 is the syntax of C1, C2
6194 is the character at D, and S2 is the syntax of C2. */
6198 int offset
= PTR_TO_OFFSET (d
) - 1;
6199 int charpos
= SYNTAX_TABLE_BYTE_TO_CHAR (offset
);
6200 UPDATE_SYNTAX_TABLE (charpos
);
6202 GET_CHAR_BEFORE_2 (c1
, d
, string1
, end1
, string2
, end2
);
6205 /* Case 2: S1 is neither Ssymbol nor Sword. */
6206 if (s1
!= Sword
&& s1
!= Ssymbol
)
6209 /* Case 3: D is not at the end of string ... */
6210 if (!AT_STRINGS_END (d
))
6212 PREFETCH_NOLIMIT ();
6213 c2
= RE_STRING_CHAR (d
, target_multibyte
);
6215 UPDATE_SYNTAX_TABLE_FORWARD (charpos
+ 1);
6219 /* ... and S2 is Sword or Ssymbol. */
6220 if (s2
== Sword
|| s2
== Ssymbol
)
6228 not = (re_opcode_t
) *(p
- 1) == notsyntaxspec
;
6230 DEBUG_PRINT3 ("EXECUTING %ssyntaxspec %d.\n", not?"not":"", mcnt
);
6234 int offset
= PTR_TO_OFFSET (d
);
6235 int pos1
= SYNTAX_TABLE_BYTE_TO_CHAR (offset
);
6236 UPDATE_SYNTAX_TABLE (pos1
);
6243 GET_CHAR_AFTER (c
, d
, len
);
6244 if ((SYNTAX (c
) != (enum syntaxcode
) mcnt
) ^ not)
6252 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
6253 if (PTR_BYTE_POS (d
) >= PT_BYTE
)
6258 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
6259 if (PTR_BYTE_POS (d
) != PT_BYTE
)
6264 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
6265 if (PTR_BYTE_POS (d
) <= PT_BYTE
)
6270 case notcategoryspec
:
6271 not = (re_opcode_t
) *(p
- 1) == notcategoryspec
;
6273 DEBUG_PRINT3 ("EXECUTING %scategoryspec %d.\n", not?"not":"", mcnt
);
6279 GET_CHAR_AFTER (c
, d
, len
);
6280 if ((!CHAR_HAS_CATEGORY (c
, mcnt
)) ^ not)
6291 continue; /* Successfully executed one pattern command; keep going. */
6294 /* We goto here if a matching operation fails. */
6296 IMMEDIATE_QUIT_CHECK
;
6297 if (!FAIL_STACK_EMPTY ())
6300 /* A restart point is known. Restore to that state. */
6301 DEBUG_PRINT1 ("\nFAIL:\n");
6302 POP_FAILURE_POINT (str
, pat
);
6303 switch (SWITCH_ENUM_CAST ((re_opcode_t
) *pat
++))
6305 case on_failure_keep_string_jump
:
6306 assert (str
== NULL
);
6307 goto continue_failure_jump
;
6309 case on_failure_jump_nastyloop
:
6310 assert ((re_opcode_t
)pat
[-2] == no_op
);
6311 PUSH_FAILURE_POINT (pat
- 2, str
);
6314 case on_failure_jump_loop
:
6315 case on_failure_jump
:
6318 continue_failure_jump
:
6319 EXTRACT_NUMBER_AND_INCR (mcnt
, pat
);
6324 /* A special frame used for nastyloops. */
6331 assert (p
>= bufp
->buffer
&& p
<= pend
);
6333 if (d
>= string1
&& d
<= end1
)
6337 break; /* Matching at this starting point really fails. */
6341 goto restore_best_regs
;
6345 return -1; /* Failure to match. */
6348 /* Subroutine definitions for re_match_2. */
6350 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
6351 bytes; nonzero otherwise. */
6354 bcmp_translate (const re_char
*s1
, const re_char
*s2
, register int len
,
6355 RE_TRANSLATE_TYPE translate
, const int target_multibyte
)
6357 register re_char
*p1
= s1
, *p2
= s2
;
6358 re_char
*p1_end
= s1
+ len
;
6359 re_char
*p2_end
= s2
+ len
;
6361 /* FIXME: Checking both p1 and p2 presumes that the two strings might have
6362 different lengths, but relying on a single `len' would break this. -sm */
6363 while (p1
< p1_end
&& p2
< p2_end
)
6365 int p1_charlen
, p2_charlen
;
6366 re_wchar_t p1_ch
, p2_ch
;
6368 GET_CHAR_AFTER (p1_ch
, p1
, p1_charlen
);
6369 GET_CHAR_AFTER (p2_ch
, p2
, p2_charlen
);
6371 if (RE_TRANSLATE (translate
, p1_ch
)
6372 != RE_TRANSLATE (translate
, p2_ch
))
6375 p1
+= p1_charlen
, p2
+= p2_charlen
;
6378 if (p1
!= p1_end
|| p2
!= p2_end
)
6384 /* Entry points for GNU code. */
6386 /* re_compile_pattern is the GNU regular expression compiler: it
6387 compiles PATTERN (of length SIZE) and puts the result in BUFP.
6388 Returns 0 if the pattern was valid, otherwise an error string.
6390 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
6391 are set in BUFP on entry.
6393 We call regex_compile to do the actual compilation. */
6396 re_compile_pattern (const char *pattern
, size_t length
, struct re_pattern_buffer
*bufp
)
6400 /* GNU code is written to assume at least RE_NREGS registers will be set
6401 (and at least one extra will be -1). */
6402 bufp
->regs_allocated
= REGS_UNALLOCATED
;
6404 /* And GNU code determines whether or not to get register information
6405 by passing null for the REGS argument to re_match, etc., not by
6409 ret
= regex_compile ((re_char
*) pattern
, length
, re_syntax_options
, bufp
);
6413 return gettext (re_error_msgid
[(int) ret
]);
6415 WEAK_ALIAS (__re_compile_pattern
, re_compile_pattern
)
6417 /* Entry points compatible with 4.2 BSD regex library. We don't define
6418 them unless specifically requested. */
6420 #if defined _REGEX_RE_COMP || defined _LIBC
6422 /* BSD has one and only one pattern buffer. */
6423 static struct re_pattern_buffer re_comp_buf
;
6427 /* Make these definitions weak in libc, so POSIX programs can redefine
6428 these names if they don't use our functions, and still use
6429 regcomp/regexec below without link errors. */
6439 if (!re_comp_buf
.buffer
)
6440 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
6441 return (char *) gettext ("No previous regular expression");
6445 if (!re_comp_buf
.buffer
)
6447 re_comp_buf
.buffer
= (unsigned char *) malloc (200);
6448 if (re_comp_buf
.buffer
== NULL
)
6449 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
6450 return (char *) gettext (re_error_msgid
[(int) REG_ESPACE
]);
6451 re_comp_buf
.allocated
= 200;
6453 re_comp_buf
.fastmap
= (char *) malloc (1 << BYTEWIDTH
);
6454 if (re_comp_buf
.fastmap
== NULL
)
6455 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
6456 return (char *) gettext (re_error_msgid
[(int) REG_ESPACE
]);
6459 /* Since `re_exec' always passes NULL for the `regs' argument, we
6460 don't need to initialize the pattern buffer fields which affect it. */
6462 ret
= regex_compile (s
, strlen (s
), re_syntax_options
, &re_comp_buf
);
6467 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
6468 return (char *) gettext (re_error_msgid
[(int) ret
]);
6479 const int len
= strlen (s
);
6481 0 <= re_search (&re_comp_buf
, s
, len
, 0, len
, (struct re_registers
*) 0);
6483 #endif /* _REGEX_RE_COMP */
6485 /* POSIX.2 functions. Don't define these for Emacs. */
6489 /* regcomp takes a regular expression as a string and compiles it.
6491 PREG is a regex_t *. We do not expect any fields to be initialized,
6492 since POSIX says we shouldn't. Thus, we set
6494 `buffer' to the compiled pattern;
6495 `used' to the length of the compiled pattern;
6496 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
6497 REG_EXTENDED bit in CFLAGS is set; otherwise, to
6498 RE_SYNTAX_POSIX_BASIC;
6499 `fastmap' to an allocated space for the fastmap;
6500 `fastmap_accurate' to zero;
6501 `re_nsub' to the number of subexpressions in PATTERN.
6503 PATTERN is the address of the pattern string.
6505 CFLAGS is a series of bits which affect compilation.
6507 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
6508 use POSIX basic syntax.
6510 If REG_NEWLINE is set, then . and [^...] don't match newline.
6511 Also, regexec will try a match beginning after every newline.
6513 If REG_ICASE is set, then we considers upper- and lowercase
6514 versions of letters to be equivalent when matching.
6516 If REG_NOSUB is set, then when PREG is passed to regexec, that
6517 routine will report only success or failure, and nothing about the
6520 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
6521 the return codes and their meanings.) */
6524 regcomp (regex_t
*__restrict preg
, const char *__restrict pattern
,
6529 = (cflags
& REG_EXTENDED
) ?
6530 RE_SYNTAX_POSIX_EXTENDED
: RE_SYNTAX_POSIX_BASIC
;
6532 /* regex_compile will allocate the space for the compiled pattern. */
6534 preg
->allocated
= 0;
6537 /* Try to allocate space for the fastmap. */
6538 preg
->fastmap
= (char *) malloc (1 << BYTEWIDTH
);
6540 if (cflags
& REG_ICASE
)
6545 = (RE_TRANSLATE_TYPE
) malloc (CHAR_SET_SIZE
6546 * sizeof (*(RE_TRANSLATE_TYPE
)0));
6547 if (preg
->translate
== NULL
)
6548 return (int) REG_ESPACE
;
6550 /* Map uppercase characters to corresponding lowercase ones. */
6551 for (i
= 0; i
< CHAR_SET_SIZE
; i
++)
6552 preg
->translate
[i
] = ISUPPER (i
) ? TOLOWER (i
) : i
;
6555 preg
->translate
= NULL
;
6557 /* If REG_NEWLINE is set, newlines are treated differently. */
6558 if (cflags
& REG_NEWLINE
)
6559 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
6560 syntax
&= ~RE_DOT_NEWLINE
;
6561 syntax
|= RE_HAT_LISTS_NOT_NEWLINE
;
6564 syntax
|= RE_NO_NEWLINE_ANCHOR
;
6566 preg
->no_sub
= !!(cflags
& REG_NOSUB
);
6568 /* POSIX says a null character in the pattern terminates it, so we
6569 can use strlen here in compiling the pattern. */
6570 ret
= regex_compile ((re_char
*) pattern
, strlen (pattern
), syntax
, preg
);
6572 /* POSIX doesn't distinguish between an unmatched open-group and an
6573 unmatched close-group: both are REG_EPAREN. */
6574 if (ret
== REG_ERPAREN
)
6577 if (ret
== REG_NOERROR
&& preg
->fastmap
)
6578 { /* Compute the fastmap now, since regexec cannot modify the pattern
6580 re_compile_fastmap (preg
);
6581 if (preg
->can_be_null
)
6582 { /* The fastmap can't be used anyway. */
6583 free (preg
->fastmap
);
6584 preg
->fastmap
= NULL
;
6589 WEAK_ALIAS (__regcomp
, regcomp
)
6592 /* regexec searches for a given pattern, specified by PREG, in the
6595 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
6596 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
6597 least NMATCH elements, and we set them to the offsets of the
6598 corresponding matched substrings.
6600 EFLAGS specifies `execution flags' which affect matching: if
6601 REG_NOTBOL is set, then ^ does not match at the beginning of the
6602 string; if REG_NOTEOL is set, then $ does not match at the end.
6604 We return 0 if we find a match and REG_NOMATCH if not. */
6607 regexec (const regex_t
*__restrict preg
, const char *__restrict string
,
6608 size_t nmatch
, regmatch_t pmatch
[__restrict_arr
], int eflags
)
6611 struct re_registers regs
;
6612 regex_t private_preg
;
6613 int len
= strlen (string
);
6614 boolean want_reg_info
= !preg
->no_sub
&& nmatch
> 0 && pmatch
;
6616 private_preg
= *preg
;
6618 private_preg
.not_bol
= !!(eflags
& REG_NOTBOL
);
6619 private_preg
.not_eol
= !!(eflags
& REG_NOTEOL
);
6621 /* The user has told us exactly how many registers to return
6622 information about, via `nmatch'. We have to pass that on to the
6623 matching routines. */
6624 private_preg
.regs_allocated
= REGS_FIXED
;
6628 regs
.num_regs
= nmatch
;
6629 regs
.start
= TALLOC (nmatch
* 2, regoff_t
);
6630 if (regs
.start
== NULL
)
6631 return (int) REG_NOMATCH
;
6632 regs
.end
= regs
.start
+ nmatch
;
6635 /* Instead of using not_eol to implement REG_NOTEOL, we could simply
6636 pass (&private_preg, string, len + 1, 0, len, ...) pretending the string
6637 was a little bit longer but still only matching the real part.
6638 This works because the `endline' will check for a '\n' and will find a
6639 '\0', correctly deciding that this is not the end of a line.
6640 But it doesn't work out so nicely for REG_NOTBOL, since we don't have
6641 a convenient '\0' there. For all we know, the string could be preceded
6642 by '\n' which would throw things off. */
6644 /* Perform the searching operation. */
6645 ret
= re_search (&private_preg
, string
, len
,
6646 /* start: */ 0, /* range: */ len
,
6647 want_reg_info
? ®s
: (struct re_registers
*) 0);
6649 /* Copy the register information to the POSIX structure. */
6656 for (r
= 0; r
< nmatch
; r
++)
6658 pmatch
[r
].rm_so
= regs
.start
[r
];
6659 pmatch
[r
].rm_eo
= regs
.end
[r
];
6663 /* If we needed the temporary register info, free the space now. */
6667 /* We want zero return to mean success, unlike `re_search'. */
6668 return ret
>= 0 ? (int) REG_NOERROR
: (int) REG_NOMATCH
;
6670 WEAK_ALIAS (__regexec
, regexec
)
6673 /* Returns a message corresponding to an error code, ERR_CODE, returned
6674 from either regcomp or regexec. We don't use PREG here.
6676 ERR_CODE was previously called ERRCODE, but that name causes an
6677 error with msvc8 compiler. */
6680 regerror (int err_code
, const regex_t
*preg
, char *errbuf
, size_t errbuf_size
)
6686 || err_code
>= (sizeof (re_error_msgid
) / sizeof (re_error_msgid
[0])))
6687 /* Only error codes returned by the rest of the code should be passed
6688 to this routine. If we are given anything else, or if other regex
6689 code generates an invalid error code, then the program has a bug.
6690 Dump core so we can fix it. */
6693 msg
= gettext (re_error_msgid
[err_code
]);
6695 msg_size
= strlen (msg
) + 1; /* Includes the null. */
6697 if (errbuf_size
!= 0)
6699 if (msg_size
> errbuf_size
)
6701 strncpy (errbuf
, msg
, errbuf_size
- 1);
6702 errbuf
[errbuf_size
- 1] = 0;
6705 strcpy (errbuf
, msg
);
6710 WEAK_ALIAS (__regerror
, regerror
)
6713 /* Free dynamically allocated space used by PREG. */
6716 regfree (regex_t
*preg
)
6718 free (preg
->buffer
);
6719 preg
->buffer
= NULL
;
6721 preg
->allocated
= 0;
6724 free (preg
->fastmap
);
6725 preg
->fastmap
= NULL
;
6726 preg
->fastmap_accurate
= 0;
6728 free (preg
->translate
);
6729 preg
->translate
= NULL
;
6731 WEAK_ALIAS (__regfree
, regfree
)
6733 #endif /* not emacs */