1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002-2005, 2007, 2008, 2010, 2011 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
6 The GNU C Library is free software; you can redistribute it and/or
7 modify it under the terms of the GNU Lesser General Public
8 License as published by the Free Software Foundation; either
9 version 2.1 of the License, or (at your option) any later version.
11 The GNU C Library is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 Lesser General Public License for more details.
16 You should have received a copy of the GNU Lesser General Public
17 License along with the GNU C Library; if not, see
18 <http://www.gnu.org/licenses/>. */
20 #ifndef _REGEX_INTERNAL_H
21 #define _REGEX_INTERNAL_H 1
29 #include "mbsupport.h" /* gawk */
31 #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC
32 # include <langinfo.h>
34 #if defined HAVE_LOCALE_H || defined _LIBC
37 #if MBS_SUPPORT && (defined HAVE_WCHAR_H || defined _LIBC)
39 #endif /* HAVE_WCHAR_H || _LIBC */
40 #if MBS_SUPPORT && (defined HAVE_WCTYPE_H || defined _LIBC)
42 #endif /* HAVE_WCTYPE_H || _LIBC */
43 #if defined HAVE_STDBOOL_H || defined _LIBC
45 #endif /* HAVE_STDBOOL_H || _LIBC */
47 #if defined HAVE_STDINT_H || defined _LIBC
49 #endif /* HAVE_STDINT_H || _LIBC */
52 # include <bits/libc-lock.h>
54 # define __libc_lock_define(CLASS,NAME)
55 # define __libc_lock_init(NAME) do { } while (0)
56 # define __libc_lock_lock(NAME) do { } while (0)
57 # define __libc_lock_unlock(NAME) do { } while (0)
61 /* In case that the system doesn't have isblank(). */
62 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
63 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
67 * This is a freaking mess. On glibc systems you have to define
68 * a magic constant to get isblank() out of <ctype.h>, since it's
69 * a C99 function. To heck with all that and borrow a page from
76 return (c == ' ' || c == '\t');
81 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
82 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
83 # include <locale/localeinfo.h>
84 # include <locale/elem-hash.h>
85 # include <locale/coll-lookup.h>
89 /* This is for other GNU distributions with internationalized messages. */
90 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
94 # define gettext(msgid) \
95 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
98 # define gettext(msgid) (msgid)
102 /* This define is so xgettext can find the internationalizable
104 # define gettext_noop(String) String
107 /* For loser systems without the definition. */
109 # define SIZE_MAX ((size_t) -1)
112 #if MBS_SUPPORT || _LIBC
113 # define RE_ENABLE_I18N
117 # define BE(expr, val) __builtin_expect (expr, val)
119 # define BE(expr, val) (expr)
126 /* Number of single byte character. */
129 #define COLL_ELEM_LEN_MAX 8
131 /* The character which represents newline. */
132 #define NEWLINE_CHAR '\n'
133 #define WIDE_NEWLINE_CHAR L'\n'
135 /* Rename to standard API for using out of glibc. */
140 # define __wctype wctype
144 # define __iswctype iswctype
145 # define __btowc btowc
146 # define __mbrtowc mbrtowc
147 #undef __mempcpy /* GAWK */
148 # define __mempcpy mempcpy
149 # define __wcrtomb wcrtomb
150 # define __regfree regfree
151 # define attribute_hidden
152 #endif /* not _LIBC */
155 # define __attribute(arg) __attribute__ (arg)
157 # define __attribute(arg)
162 * Instead of trying to figure out which GCC version introduced
163 * this symbol, just define it out and be done.
165 # undef __attribute_warn_unused_result__
166 # define __attribute_warn_unused_result__
169 extern const char __re_error_msgid[] attribute_hidden;
170 extern const size_t __re_error_msgid_idx[] attribute_hidden;
172 /* An integer used to represent a set of bits. It must be unsigned,
173 and must be at least as wide as unsigned int. */
174 typedef unsigned long int bitset_word_t;
175 /* All bits set in a bitset_word_t. */
176 #define BITSET_WORD_MAX ULONG_MAX
177 /* Number of bits in a bitset_word_t. */
178 #define BITSET_WORD_BITS (sizeof (bitset_word_t) * CHAR_BIT)
179 /* Number of bitset_word_t in a bit_set. */
180 #define BITSET_WORDS (SBC_MAX / BITSET_WORD_BITS)
181 typedef bitset_word_t bitset_t[BITSET_WORDS];
182 typedef bitset_word_t *re_bitset_ptr_t;
183 typedef const bitset_word_t *re_const_bitset_ptr_t;
185 #define bitset_set(set,i) \
186 (set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS)
187 #define bitset_clear(set,i) \
188 (set[i / BITSET_WORD_BITS] &= ~((bitset_word_t) 1 << i % BITSET_WORD_BITS))
189 #define bitset_contain(set,i) \
190 (set[i / BITSET_WORD_BITS] & ((bitset_word_t) 1 << i % BITSET_WORD_BITS))
191 #define bitset_empty(set) memset (set, '\0', sizeof (bitset_t))
192 #define bitset_set_all(set) memset (set, '\xff', sizeof (bitset_t))
193 #define bitset_copy(dest,src) memcpy (dest, src, sizeof (bitset_t))
195 #define PREV_WORD_CONSTRAINT 0x0001
196 #define PREV_NOTWORD_CONSTRAINT 0x0002
197 #define NEXT_WORD_CONSTRAINT 0x0004
198 #define NEXT_NOTWORD_CONSTRAINT 0x0008
199 #define PREV_NEWLINE_CONSTRAINT 0x0010
200 #define NEXT_NEWLINE_CONSTRAINT 0x0020
201 #define PREV_BEGBUF_CONSTRAINT 0x0040
202 #define NEXT_ENDBUF_CONSTRAINT 0x0080
203 #define WORD_DELIM_CONSTRAINT 0x0100
204 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
208 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
209 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
210 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
211 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
212 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
213 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
214 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
215 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
216 WORD_DELIM = WORD_DELIM_CONSTRAINT,
217 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
231 /* Node type, These are used by token, node, tree. */
237 #ifdef RE_ENABLE_I18N
240 #endif /* RE_ENABLE_I18N */
242 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
243 when the debugger shows values of this enum type. */
244 #define EPSILON_BIT 8
245 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
246 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
247 OP_ALT = EPSILON_BIT | 2,
248 OP_DUP_ASTERISK = EPSILON_BIT | 3,
249 ANCHOR = EPSILON_BIT | 4,
251 /* Tree type, these are used only by tree. */
255 /* Token type, these are used only by token. */
267 OP_CLOSE_EQUIV_CLASS,
278 #ifdef RE_ENABLE_I18N
281 /* Multibyte characters. */
284 /* Collating symbols. */
289 /* Equivalence classes. */
291 int32_t *equiv_classes;
294 /* Range expressions. */
296 uint32_t *range_starts;
297 uint32_t *range_ends;
298 # else /* not _LIBC */
299 wchar_t *range_starts;
301 # endif /* not _LIBC */
303 /* Character classes. */
304 wctype_t *char_classes;
306 /* If this character set is the non-matching list. */
307 unsigned int non_match : 1;
309 /* # of multibyte characters. */
312 /* # of collating symbols. */
315 /* # of equivalence classes. */
318 /* # of range expressions. */
321 /* # of character classes. */
324 #endif /* RE_ENABLE_I18N */
330 unsigned char c; /* for CHARACTER */
331 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
332 #ifdef RE_ENABLE_I18N
333 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
334 #endif /* RE_ENABLE_I18N */
335 int idx; /* for BACK_REF */
336 re_context_type ctx_type; /* for ANCHOR */
339 re_token_type_t type : 8;
341 re_token_type_t type;
343 unsigned int constraint : 10; /* context constraint */
344 unsigned int duplicated : 1;
345 unsigned int opt_subexp : 1;
346 #ifdef RE_ENABLE_I18N
347 unsigned int accept_mb : 1;
348 /* These 2 bits can be moved into the union if needed (e.g. if running out
349 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
350 unsigned int mb_partial : 1;
352 unsigned int word_char : 1;
355 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
359 /* Indicate the raw buffer which is the original string passed as an
360 argument of regexec(), re_search(), etc.. */
361 const unsigned char *raw_mbs;
362 /* Store the multibyte string. In case of "case insensitive mode" like
363 REG_ICASE, upper cases of the string are stored, otherwise MBS points
364 the same address that RAW_MBS points. */
366 #ifdef RE_ENABLE_I18N
367 /* Store the wide character string which is corresponding to MBS. */
372 /* Index in RAW_MBS. Each character mbs[i] corresponds to
373 raw_mbs[raw_mbs_idx + i]. */
375 /* The length of the valid characters in the buffers. */
377 /* The corresponding number of bytes in raw_mbs array. */
379 /* The length of the buffers MBS and WCS. */
381 /* The index in MBS, which is updated by re_string_fetch_byte. */
383 /* length of RAW_MBS array. */
385 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
387 /* End of the buffer may be shorter than its length in the cases such
388 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
391 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
394 /* The context of mbs[0]. We store the context independently, since
395 the context of mbs[0] may be different from raw_mbs[0], which is
396 the beginning of the input string. */
397 unsigned int tip_context;
398 /* The translation passed as a part of an argument of re_compile_pattern. */
399 RE_TRANSLATE_TYPE trans;
400 /* Copy of re_dfa_t's word_char. */
401 re_const_bitset_ptr_t word_char;
402 /* 1 if REG_ICASE. */
404 unsigned char is_utf8;
405 unsigned char map_notascii;
406 unsigned char mbs_allocated;
407 unsigned char offsets_needed;
408 unsigned char newline_anchor;
409 unsigned char word_ops_used;
412 typedef struct re_string_t re_string_t;
416 typedef struct re_dfa_t re_dfa_t;
420 # define internal_function __attribute ((regparm (3), stdcall))
422 # define internal_function
427 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
430 # ifdef RE_ENABLE_I18N
431 static void build_wcs_buffer (re_string_t *pstr) internal_function;
432 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
434 # endif /* RE_ENABLE_I18N */
435 static void build_upper_buffer (re_string_t *pstr) internal_function;
436 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
437 static unsigned int re_string_context_at (const re_string_t *input, int idx,
439 internal_function __attribute ((pure));
441 #define re_string_peek_byte(pstr, offset) \
442 ((pstr)->mbs[(pstr)->cur_idx + offset])
443 #define re_string_fetch_byte(pstr) \
444 ((pstr)->mbs[(pstr)->cur_idx++])
445 #define re_string_first_byte(pstr, idx) \
446 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
447 #define re_string_is_single_byte_char(pstr, idx) \
448 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
449 || (pstr)->wcs[(idx) + 1] != WEOF))
450 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
451 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
452 #define re_string_get_buffer(pstr) ((pstr)->mbs)
453 #define re_string_length(pstr) ((pstr)->len)
454 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
455 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
456 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
461 /* The OS usually guarantees only one guard page at the bottom of the stack,
462 and a page size can be as small as 4096 bytes. So we cannot safely
463 allocate anything larger than 4096 bytes. Also care for the possibility
464 of a few compiler-allocated temporary stack slots. */
465 # define __libc_use_alloca(n) ((n) < 4032)
467 /* alloca is implemented with malloc, so just use malloc. */
468 # define __libc_use_alloca(n) 0
472 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
473 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
474 #define re_free(p) free (p)
478 struct bin_tree_t *parent;
479 struct bin_tree_t *left;
480 struct bin_tree_t *right;
481 struct bin_tree_t *first;
482 struct bin_tree_t *next;
486 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
487 Otherwise `type' indicate the type of this node. */
490 typedef struct bin_tree_t bin_tree_t;
492 #define BIN_TREE_STORAGE_SIZE \
493 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
495 struct bin_tree_storage_t
497 struct bin_tree_storage_t *next;
498 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
500 typedef struct bin_tree_storage_t bin_tree_storage_t;
502 #define CONTEXT_WORD 1
503 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
504 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
505 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
507 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
508 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
509 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
510 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
511 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
513 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
514 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
515 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
516 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
518 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
519 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
520 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
521 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
522 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
524 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
525 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
526 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
527 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
528 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
534 re_node_set non_eps_nodes;
535 re_node_set inveclosure;
536 re_node_set *entrance_nodes;
537 struct re_dfastate_t **trtable, **word_trtable;
538 unsigned int context : 4;
539 unsigned int halt : 1;
540 /* If this state can accept `multi byte'.
541 Note that we refer to multibyte characters, and multi character
542 collating elements as `multi byte'. */
543 unsigned int accept_mb : 1;
544 /* If this state has backreference node(s). */
545 unsigned int has_backref : 1;
546 unsigned int has_constraint : 1;
548 typedef struct re_dfastate_t re_dfastate_t;
550 struct re_state_table_entry
554 re_dfastate_t **array;
557 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
563 re_dfastate_t **array;
566 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
571 int str_idx; /* The position NODE match at. */
573 } re_sub_match_last_t;
575 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
576 And information about the node, whose type is OP_CLOSE_SUBEXP,
577 corresponding to NODE is stored in LASTS. */
584 int alasts; /* Allocation size of LASTS. */
585 int nlasts; /* The number of LASTS. */
586 re_sub_match_last_t **lasts;
587 } re_sub_match_top_t;
589 struct re_backref_cache_entry
597 unsigned short int eps_reachable_subexps_map;
602 /* The string object corresponding to the input string. */
604 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
605 const re_dfa_t *const dfa;
609 /* EFLAGS of the argument of regexec. */
611 /* Where the matching ends. */
614 /* The state log used by the matcher. */
615 re_dfastate_t **state_log;
617 /* Back reference cache. */
620 struct re_backref_cache_entry *bkref_ents;
624 re_sub_match_top_t **sub_tops;
625 } re_match_context_t;
629 re_dfastate_t **sifted_states;
630 re_dfastate_t **limited_states;
636 struct re_fail_stack_ent_t
641 re_node_set eps_via_nodes;
644 struct re_fail_stack_t
648 struct re_fail_stack_ent_t *stack;
659 re_node_set *eclosures;
660 re_node_set *inveclosures;
661 struct re_state_table_entry *state_table;
662 re_dfastate_t *init_state;
663 re_dfastate_t *init_state_word;
664 re_dfastate_t *init_state_nl;
665 re_dfastate_t *init_state_begbuf;
666 bin_tree_t *str_tree;
667 bin_tree_storage_t *str_tree_storage;
668 re_bitset_ptr_t sb_char;
669 int str_tree_storage_idx;
671 /* number of subexpressions `re_nsub' is in regex_t. */
672 unsigned int state_hash_mask;
674 int nbackref; /* The number of backreference in this dfa. */
676 /* Bitmap expressing which backreference is used. */
677 bitset_word_t used_bkref_map;
678 bitset_word_t completed_bkref_map;
680 unsigned int has_plural_match : 1;
681 /* If this dfa has "multibyte node", which is a backreference or
682 a node which can accept multibyte character or multi character
683 collating element. */
684 unsigned int has_mb_node : 1;
685 unsigned int is_utf8 : 1;
686 unsigned int map_notascii : 1;
687 unsigned int word_ops_used : 1;
696 __libc_lock_define (, lock)
700 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
701 #define re_node_set_remove(set,id) \
702 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
703 #define re_node_set_empty(p) ((p)->nelem = 0)
704 #define re_node_set_free(set) re_free ((set)->elems)
718 bracket_elem_type type;
728 /* Inline functions for bitset operation. */
730 bitset_not (bitset_t set)
733 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
734 set[bitset_i] = ~set[bitset_i];
738 bitset_merge (bitset_t dest, const bitset_t src)
741 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
742 dest[bitset_i] |= src[bitset_i];
746 bitset_mask (bitset_t dest, const bitset_t src)
749 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
750 dest[bitset_i] &= src[bitset_i];
753 #ifdef RE_ENABLE_I18N
754 /* Inline functions for re_string. */
756 internal_function __attribute ((pure))
757 re_string_char_size_at (const re_string_t *pstr, int idx)
760 if (pstr->mb_cur_max == 1)
762 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
763 if (pstr->wcs[idx + byte_idx] != WEOF)
769 internal_function __attribute ((pure))
770 re_string_wchar_at (const re_string_t *pstr, int idx)
772 if (pstr->mb_cur_max == 1)
773 return (wint_t) pstr->mbs[idx];
774 return (wint_t) pstr->wcs[idx];
779 internal_function __attribute ((pure))
780 re_string_elem_size_at (const re_string_t *pstr, int idx)
783 const unsigned char *p, *extra;
784 const int32_t *table, *indirect;
785 # include <locale/weight.h>
786 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
790 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
791 extra = (const unsigned char *)
792 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
793 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
794 _NL_COLLATE_INDIRECTMB);
796 findidx (&p, pstr->len - idx);
797 return p - pstr->mbs - idx;
804 #endif /* RE_ENABLE_I18N */
806 #endif /* _REGEX_INTERNAL_H */