1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002, 2003, 2004, 2005 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2, or (at your option)
11 This program 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
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License along
17 with this program; if not, write to the Free Software Foundation,
18 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
20 #ifndef _REGEX_INTERNAL_H
21 #define _REGEX_INTERNAL_H 1
29 #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC
30 # include <langinfo.h>
32 #if defined HAVE_LOCALE_H || defined _LIBC
35 #if defined HAVE_WCHAR_H || defined _LIBC
37 #endif /* HAVE_WCHAR_H || _LIBC */
38 #if defined HAVE_WCTYPE_H || defined _LIBC
40 #endif /* HAVE_WCTYPE_H || _LIBC */
42 # include <bits/libc-lock.h>
44 # define __libc_lock_define(CLASS,NAME)
45 # define __libc_lock_init(NAME) do { } while (0)
46 # define __libc_lock_lock(NAME) do { } while (0)
47 # define __libc_lock_unlock(NAME) do { } while (0)
50 /* In case that the system doesn't have isblank(). */
51 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
52 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
56 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
57 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
58 # include <locale/localeinfo.h>
59 # include <locale/elem-hash.h>
60 # include <locale/coll-lookup.h>
64 /* This is for other GNU distributions with internationalized messages. */
65 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
69 # define gettext(msgid) \
70 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
73 # define gettext(msgid) (msgid)
77 /* This define is so xgettext can find the internationalizable
79 # define gettext_noop(String) String
82 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
83 # define RE_ENABLE_I18N
87 # define BE(expr, val) __builtin_expect (expr, val)
89 # define BE(expr, val) (expr)
93 /* Number of single byte character. */
96 #define COLL_ELEM_LEN_MAX 8
98 /* The character which represents newline. */
99 #define NEWLINE_CHAR '\n'
100 #define WIDE_NEWLINE_CHAR L'\n'
102 /* Rename to standard API for using out of glibc. */
104 # define __wctype wctype
105 # define __iswctype iswctype
106 # define __btowc btowc
108 # define __mempcpy mempcpy
110 # define __wcrtomb wcrtomb
111 # define __regfree regfree
112 # define attribute_hidden
113 #endif /* not _LIBC */
116 # define __attribute(arg) __attribute__ (arg)
118 # define __attribute(arg)
121 extern const char __re_error_msgid[] attribute_hidden;
122 extern const size_t __re_error_msgid_idx[] attribute_hidden;
124 /* Number of bits in an unsinged int. */
125 #define UINT_BITS (sizeof (unsigned int) * CHAR_BIT)
126 /* Number of unsigned int in an bit_set. */
127 #define BITSET_UINTS ((SBC_MAX + UINT_BITS - 1) / UINT_BITS)
128 typedef unsigned int bitset[BITSET_UINTS];
129 typedef unsigned int *re_bitset_ptr_t;
130 typedef const unsigned int *re_const_bitset_ptr_t;
132 #define bitset_set(set,i) (set[i / UINT_BITS] |= 1u << i % UINT_BITS)
133 #define bitset_clear(set,i) (set[i / UINT_BITS] &= ~(1u << i % UINT_BITS))
134 #define bitset_contain(set,i) (set[i / UINT_BITS] & (1u << i % UINT_BITS))
135 #define bitset_empty(set) memset (set, 0, sizeof (unsigned int) * BITSET_UINTS)
136 #define bitset_set_all(set) \
137 memset (set, 255, sizeof (unsigned int) * BITSET_UINTS)
138 #define bitset_copy(dest,src) \
139 memcpy (dest, src, sizeof (unsigned int) * BITSET_UINTS)
141 #define PREV_WORD_CONSTRAINT 0x0001
142 #define PREV_NOTWORD_CONSTRAINT 0x0002
143 #define NEXT_WORD_CONSTRAINT 0x0004
144 #define NEXT_NOTWORD_CONSTRAINT 0x0008
145 #define PREV_NEWLINE_CONSTRAINT 0x0010
146 #define NEXT_NEWLINE_CONSTRAINT 0x0020
147 #define PREV_BEGBUF_CONSTRAINT 0x0040
148 #define NEXT_ENDBUF_CONSTRAINT 0x0080
149 #define WORD_DELIM_CONSTRAINT 0x0100
150 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
154 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
155 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
156 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
157 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
158 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
159 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
160 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
161 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
162 WORD_DELIM = WORD_DELIM_CONSTRAINT,
163 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
177 /* Node type, These are used by token, node, tree. */
183 #ifdef RE_ENABLE_I18N
186 #endif /* RE_ENABLE_I18N */
188 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
189 when the debugger shows values of this enum type. */
190 #define EPSILON_BIT 8
191 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
192 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
193 OP_ALT = EPSILON_BIT | 2,
194 OP_DUP_ASTERISK = EPSILON_BIT | 3,
195 ANCHOR = EPSILON_BIT | 4,
197 /* Tree type, these are used only by tree. */
201 /* Token type, these are used only by token. */
213 OP_CLOSE_EQUIV_CLASS,
224 #ifdef RE_ENABLE_I18N
227 /* Multibyte characters. */
230 /* Collating symbols. */
235 /* Equivalence classes. */
237 int32_t *equiv_classes;
240 /* Range expressions. */
242 uint32_t *range_starts;
243 uint32_t *range_ends;
244 # else /* not _LIBC */
245 wchar_t *range_starts;
247 # endif /* not _LIBC */
249 /* Character classes. */
250 wctype_t *char_classes;
252 /* If this character set is the non-matching list. */
253 unsigned int non_match : 1;
255 /* # of multibyte characters. */
258 /* # of collating symbols. */
261 /* # of equivalence classes. */
264 /* # of range expressions. */
267 /* # of character classes. */
270 #endif /* RE_ENABLE_I18N */
276 unsigned char c; /* for CHARACTER */
277 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
278 #ifdef RE_ENABLE_I18N
279 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
280 #endif /* RE_ENABLE_I18N */
281 int idx; /* for BACK_REF */
282 re_context_type ctx_type; /* for ANCHOR */
285 re_token_type_t type : 8;
287 re_token_type_t type;
289 unsigned int constraint : 10; /* context constraint */
290 unsigned int duplicated : 1;
291 unsigned int opt_subexp : 1;
292 #ifdef RE_ENABLE_I18N
293 unsigned int accept_mb : 1;
294 /* These 2 bits can be moved into the union if needed (e.g. if running out
295 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
296 unsigned int mb_partial : 1;
298 unsigned int word_char : 1;
301 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
305 /* Indicate the raw buffer which is the original string passed as an
306 argument of regexec(), re_search(), etc.. */
307 const unsigned char *raw_mbs;
308 /* Store the multibyte string. In case of "case insensitive mode" like
309 REG_ICASE, upper cases of the string are stored, otherwise MBS points
310 the same address that RAW_MBS points. */
312 #ifdef RE_ENABLE_I18N
313 /* Store the wide character string which is corresponding to MBS. */
318 /* Index in RAW_MBS. Each character mbs[i] corresponds to
319 raw_mbs[raw_mbs_idx + i]. */
321 /* The length of the valid characters in the buffers. */
323 /* The corresponding number of bytes in raw_mbs array. */
325 /* The length of the buffers MBS and WCS. */
327 /* The index in MBS, which is updated by re_string_fetch_byte. */
329 /* length of RAW_MBS array. */
331 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
333 /* End of the buffer may be shorter than its length in the cases such
334 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
337 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
340 /* The context of mbs[0]. We store the context independently, since
341 the context of mbs[0] may be different from raw_mbs[0], which is
342 the beginning of the input string. */
343 unsigned int tip_context;
344 /* The translation passed as a part of an argument of re_compile_pattern. */
345 unsigned REG_TRANSLATE_TYPE trans;
346 /* Copy of re_dfa_t's word_char. */
347 re_const_bitset_ptr_t word_char;
348 /* 1 if REG_ICASE. */
350 unsigned char is_utf8;
351 unsigned char map_notascii;
352 unsigned char mbs_allocated;
353 unsigned char offsets_needed;
354 unsigned char newline_anchor;
355 unsigned char word_ops_used;
358 typedef struct re_string_t re_string_t;
362 typedef struct re_dfa_t re_dfa_t;
366 # define internal_function __attribute ((regparm (3), stdcall))
368 # define internal_function
372 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
375 #ifdef RE_ENABLE_I18N
376 static void build_wcs_buffer (re_string_t *pstr) internal_function;
377 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
379 #endif /* RE_ENABLE_I18N */
380 static void build_upper_buffer (re_string_t *pstr) internal_function;
381 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
382 static unsigned int re_string_context_at (const re_string_t *input, int idx,
384 internal_function __attribute ((pure));
386 #define re_string_peek_byte(pstr, offset) \
387 ((pstr)->mbs[(pstr)->cur_idx + offset])
388 #define re_string_fetch_byte(pstr) \
389 ((pstr)->mbs[(pstr)->cur_idx++])
390 #define re_string_first_byte(pstr, idx) \
391 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
392 #define re_string_is_single_byte_char(pstr, idx) \
393 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
394 || (pstr)->wcs[(idx) + 1] != WEOF))
395 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
396 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
397 #define re_string_get_buffer(pstr) ((pstr)->mbs)
398 #define re_string_length(pstr) ((pstr)->len)
399 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
400 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
401 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
407 /* The OS usually guarantees only one guard page at the bottom of the stack,
408 and a page size can be as small as 4096 bytes. So we cannot safely
409 allocate anything larger than 4096 bytes. Also care for the possibility
410 of a few compiler-allocated temporary stack slots. */
411 # define __libc_use_alloca(n) ((n) < 4032)
413 /* alloca is implemented with malloc, so just use malloc. */
414 # define __libc_use_alloca(n) 0
418 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
419 #define re_calloc(t,n) ((t *) calloc (n, sizeof (t)))
420 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
421 #define re_free(p) free (p)
425 struct bin_tree_t *parent;
426 struct bin_tree_t *left;
427 struct bin_tree_t *right;
428 struct bin_tree_t *first;
429 struct bin_tree_t *next;
433 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
434 Otherwise `type' indicate the type of this node. */
437 typedef struct bin_tree_t bin_tree_t;
439 #define BIN_TREE_STORAGE_SIZE \
440 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
442 struct bin_tree_storage_t
444 struct bin_tree_storage_t *next;
445 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
447 typedef struct bin_tree_storage_t bin_tree_storage_t;
449 #define CONTEXT_WORD 1
450 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
451 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
452 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
454 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
455 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
456 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
457 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
458 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
460 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
461 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
462 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
463 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
465 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
466 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
467 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
468 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
469 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
471 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
472 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
473 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
474 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
475 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
481 re_node_set non_eps_nodes;
482 re_node_set inveclosure;
483 re_node_set *entrance_nodes;
484 struct re_dfastate_t **trtable, **word_trtable;
485 unsigned int context : 4;
486 unsigned int halt : 1;
487 /* If this state can accept `multi byte'.
488 Note that we refer to multibyte characters, and multi character
489 collating elements as `multi byte'. */
490 unsigned int accept_mb : 1;
491 /* If this state has backreference node(s). */
492 unsigned int has_backref : 1;
493 unsigned int has_constraint : 1;
495 typedef struct re_dfastate_t re_dfastate_t;
497 struct re_state_table_entry
501 re_dfastate_t **array;
504 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
510 re_dfastate_t **array;
513 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
518 int str_idx; /* The position NODE match at. */
520 } re_sub_match_last_t;
522 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
523 And information about the node, whose type is OP_CLOSE_SUBEXP,
524 corresponding to NODE is stored in LASTS. */
531 int alasts; /* Allocation size of LASTS. */
532 int nlasts; /* The number of LASTS. */
533 re_sub_match_last_t **lasts;
534 } re_sub_match_top_t;
536 struct re_backref_cache_entry
544 unsigned short int eps_reachable_subexps_map;
549 /* The string object corresponding to the input string. */
551 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
556 /* EFLAGS of the argument of regexec. */
558 /* Where the matching ends. */
561 /* The state log used by the matcher. */
562 re_dfastate_t **state_log;
564 /* Back reference cache. */
567 struct re_backref_cache_entry *bkref_ents;
571 re_sub_match_top_t **sub_tops;
572 } re_match_context_t;
576 re_dfastate_t **sifted_states;
577 re_dfastate_t **limited_states;
583 struct re_fail_stack_ent_t
588 re_node_set eps_via_nodes;
591 struct re_fail_stack_t
595 struct re_fail_stack_ent_t *stack;
606 re_node_set *eclosures;
607 re_node_set *inveclosures;
608 struct re_state_table_entry *state_table;
609 re_dfastate_t *init_state;
610 re_dfastate_t *init_state_word;
611 re_dfastate_t *init_state_nl;
612 re_dfastate_t *init_state_begbuf;
613 bin_tree_t *str_tree;
614 bin_tree_storage_t *str_tree_storage;
615 re_bitset_ptr_t sb_char;
616 int str_tree_storage_idx;
618 /* number of subexpressions `re_nsub' is in regex_t. */
619 unsigned int state_hash_mask;
621 int nbackref; /* The number of backreference in this dfa. */
623 /* Bitmap expressing which backreference is used. */
624 unsigned int used_bkref_map;
625 unsigned int completed_bkref_map;
627 unsigned int has_plural_match : 1;
628 /* If this dfa has "multibyte node", which is a backreference or
629 a node which can accept multibyte character or multi character
630 collating element. */
631 unsigned int has_mb_node : 1;
632 unsigned int is_utf8 : 1;
633 unsigned int map_notascii : 1;
634 unsigned int word_ops_used : 1;
642 __libc_lock_define (, lock)
645 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
646 #define re_node_set_remove(set,id) \
647 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
648 #define re_node_set_empty(p) ((p)->nelem = 0)
649 #define re_node_set_free(set) re_free ((set)->elems)
651 static void free_state (re_dfastate_t *state) internal_function;
665 bracket_elem_type type;
675 /* Inline functions for bitset operation. */
677 bitset_not (bitset set)
680 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
681 set[bitset_i] = ~set[bitset_i];
685 bitset_merge (bitset dest, const bitset src)
688 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
689 dest[bitset_i] |= src[bitset_i];
693 bitset_not_merge (bitset dest, const bitset src)
696 for (i = 0; i < BITSET_UINTS; ++i)
701 bitset_mask (bitset dest, const bitset src)
704 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
705 dest[bitset_i] &= src[bitset_i];
708 #if defined RE_ENABLE_I18N
709 /* Inline functions for re_string. */
711 internal_function __attribute ((pure))
712 re_string_char_size_at (const re_string_t *pstr, int idx)
715 if (pstr->mb_cur_max == 1)
717 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
718 if (pstr->wcs[idx + byte_idx] != WEOF)
724 internal_function __attribute ((pure))
725 re_string_wchar_at (const re_string_t *pstr, int idx)
727 if (pstr->mb_cur_max == 1)
728 return (wint_t) pstr->mbs[idx];
729 return (wint_t) pstr->wcs[idx];
733 internal_function __attribute ((pure))
734 re_string_elem_size_at (const re_string_t *pstr, int idx)
737 const unsigned char *p, *extra;
738 const int32_t *table, *indirect;
740 # include <locale/weight.h>
741 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
745 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
746 extra = (const unsigned char *)
747 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
748 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
749 _NL_COLLATE_INDIRECTMB);
752 return p - pstr->mbs - idx;
758 #endif /* RE_ENABLE_I18N */
760 #endif /* _REGEX_INTERNAL_H */