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
30 #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC
31 # include <langinfo.h>
33 #if defined HAVE_LOCALE_H || defined _LIBC
36 #if defined HAVE_WCHAR_H || defined _LIBC
38 #endif /* HAVE_WCHAR_H || _LIBC */
39 #if defined HAVE_WCTYPE_H || defined _LIBC
41 #endif /* HAVE_WCTYPE_H || _LIBC */
43 # include <bits/libc-lock.h>
45 # define __libc_lock_define(CLASS,NAME)
46 # define __libc_lock_init(NAME) do { } while (0)
47 # define __libc_lock_lock(NAME) do { } while (0)
48 # define __libc_lock_unlock(NAME) do { } while (0)
51 /* In case that the system doesn't have isblank(). */
52 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
53 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
57 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
58 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
59 # include <locale/localeinfo.h>
60 # include <locale/elem-hash.h>
61 # include <locale/coll-lookup.h>
65 /* This is for other GNU distributions with internationalized messages. */
66 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
70 # define gettext(msgid) \
71 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
74 # define gettext(msgid) (msgid)
78 /* This define is so xgettext can find the internationalizable
80 # define gettext_noop(String) String
83 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
84 # define RE_ENABLE_I18N
88 # define BE(expr, val) __builtin_expect (expr, val)
90 # define BE(expr, val) (expr)
94 /* Number of single byte character. */
97 #define COLL_ELEM_LEN_MAX 8
99 /* The character which represents newline. */
100 #define NEWLINE_CHAR '\n'
101 #define WIDE_NEWLINE_CHAR L'\n'
103 /* Rename to standard API for using out of glibc. */
105 # define __wctype wctype
106 # define __iswctype iswctype
107 # define __btowc btowc
109 # define __mempcpy mempcpy
111 # define __wcrtomb wcrtomb
112 # define __regfree regfree
113 # define attribute_hidden
114 #endif /* not _LIBC */
117 # define __attribute(arg) __attribute__ (arg)
119 # define __attribute(arg)
122 extern const char __re_error_msgid[] attribute_hidden;
123 extern const size_t __re_error_msgid_idx[] attribute_hidden;
125 typedef __re_idx_t Idx;
127 /* Special return value for failure to match. */
128 #define REG_MISSING ((Idx) -1)
130 /* Special return value for internal error. */
131 #define REG_ERROR ((Idx) -2)
133 /* Test whether N is a valid index, and is not one of the above. */
134 #ifdef _REGEX_LARGE_OFFSETS
135 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
137 # define REG_VALID_INDEX(n) (0 <= (n))
140 /* Test whether N is a valid nonzero index. */
141 #ifdef _REGEX_LARGE_OFFSETS
142 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
144 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
147 /* A hash value, suitable for computing hash tables. */
148 typedef __re_size_t re_hashval_t;
150 /* Number of bits in an unsinged int. */
151 #define UINT_BITS (sizeof (unsigned int) * CHAR_BIT)
152 /* Number of unsigned int in an bit_set. */
153 #define BITSET_UINTS ((SBC_MAX + UINT_BITS - 1) / UINT_BITS)
154 typedef unsigned int bitset[BITSET_UINTS];
155 typedef unsigned int *re_bitset_ptr_t;
156 typedef const unsigned int *re_const_bitset_ptr_t;
158 #define bitset_set(set,i) (set[i / UINT_BITS] |= 1u << i % UINT_BITS)
159 #define bitset_clear(set,i) (set[i / UINT_BITS] &= ~(1u << i % UINT_BITS))
160 #define bitset_contain(set,i) (set[i / UINT_BITS] & (1u << i % UINT_BITS))
161 #define bitset_empty(set) memset (set, 0, sizeof (unsigned int) * BITSET_UINTS)
162 #define bitset_set_all(set) \
163 memset (set, 255, sizeof (unsigned int) * BITSET_UINTS)
164 #define bitset_copy(dest,src) \
165 memcpy (dest, src, sizeof (unsigned int) * BITSET_UINTS)
167 #define PREV_WORD_CONSTRAINT 0x0001
168 #define PREV_NOTWORD_CONSTRAINT 0x0002
169 #define NEXT_WORD_CONSTRAINT 0x0004
170 #define NEXT_NOTWORD_CONSTRAINT 0x0008
171 #define PREV_NEWLINE_CONSTRAINT 0x0010
172 #define NEXT_NEWLINE_CONSTRAINT 0x0020
173 #define PREV_BEGBUF_CONSTRAINT 0x0040
174 #define NEXT_ENDBUF_CONSTRAINT 0x0080
175 #define WORD_DELIM_CONSTRAINT 0x0100
176 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
180 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
181 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
182 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
183 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
184 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
185 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
186 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
187 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
188 WORD_DELIM = WORD_DELIM_CONSTRAINT,
189 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
203 /* Node type, These are used by token, node, tree. */
209 #ifdef RE_ENABLE_I18N
212 #endif /* RE_ENABLE_I18N */
214 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
215 when the debugger shows values of this enum type. */
216 #define EPSILON_BIT 8
217 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
218 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
219 OP_ALT = EPSILON_BIT | 2,
220 OP_DUP_ASTERISK = EPSILON_BIT | 3,
221 ANCHOR = EPSILON_BIT | 4,
223 /* Tree type, these are used only by tree. */
227 /* Token type, these are used only by token. */
239 OP_CLOSE_EQUIV_CLASS,
250 #ifdef RE_ENABLE_I18N
253 /* Multibyte characters. */
256 /* Collating symbols. */
261 /* Equivalence classes. */
263 int32_t *equiv_classes;
266 /* Range expressions. */
268 uint32_t *range_starts;
269 uint32_t *range_ends;
270 # else /* not _LIBC */
271 wchar_t *range_starts;
273 # endif /* not _LIBC */
275 /* Character classes. */
276 wctype_t *char_classes;
278 /* If this character set is the non-matching list. */
279 unsigned int non_match : 1;
281 /* # of multibyte characters. */
284 /* # of collating symbols. */
287 /* # of equivalence classes. */
290 /* # of range expressions. */
293 /* # of character classes. */
296 #endif /* RE_ENABLE_I18N */
302 unsigned char c; /* for CHARACTER */
303 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
304 #ifdef RE_ENABLE_I18N
305 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
306 #endif /* RE_ENABLE_I18N */
307 Idx idx; /* for BACK_REF */
308 re_context_type ctx_type; /* for ANCHOR */
311 re_token_type_t type : 8;
313 re_token_type_t type;
315 unsigned int constraint : 10; /* context constraint */
316 unsigned int duplicated : 1;
317 unsigned int opt_subexp : 1;
318 #ifdef RE_ENABLE_I18N
319 unsigned int accept_mb : 1;
320 /* These 2 bits can be moved into the union if needed (e.g. if running out
321 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
322 unsigned int mb_partial : 1;
324 unsigned int word_char : 1;
327 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
331 /* Indicate the raw buffer which is the original string passed as an
332 argument of regexec(), re_search(), etc.. */
333 const unsigned char *raw_mbs;
334 /* Store the multibyte string. In case of "case insensitive mode" like
335 REG_ICASE, upper cases of the string are stored, otherwise MBS points
336 the same address that RAW_MBS points. */
338 #ifdef RE_ENABLE_I18N
339 /* Store the wide character string which is corresponding to MBS. */
344 /* Index in RAW_MBS. Each character mbs[i] corresponds to
345 raw_mbs[raw_mbs_idx + i]. */
347 /* The length of the valid characters in the buffers. */
349 /* The corresponding number of bytes in raw_mbs array. */
351 /* The length of the buffers MBS and WCS. */
353 /* The index in MBS, which is updated by re_string_fetch_byte. */
355 /* length of RAW_MBS array. */
357 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
359 /* End of the buffer may be shorter than its length in the cases such
360 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
363 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
366 /* The context of mbs[0]. We store the context independently, since
367 the context of mbs[0] may be different from raw_mbs[0], which is
368 the beginning of the input string. */
369 unsigned int tip_context;
370 /* The translation passed as a part of an argument of re_compile_pattern. */
371 unsigned REG_TRANSLATE_TYPE trans;
372 /* Copy of re_dfa_t's word_char. */
373 re_const_bitset_ptr_t word_char;
374 /* true if REG_ICASE. */
376 unsigned char is_utf8;
377 unsigned char map_notascii;
378 unsigned char mbs_allocated;
379 unsigned char offsets_needed;
380 unsigned char newline_anchor;
381 unsigned char word_ops_used;
384 typedef struct re_string_t re_string_t;
388 typedef struct re_dfa_t re_dfa_t;
392 # define internal_function __attribute ((regparm (3), stdcall))
394 # define internal_function
398 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
401 #ifdef RE_ENABLE_I18N
402 static void build_wcs_buffer (re_string_t *pstr) internal_function;
403 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
405 #endif /* RE_ENABLE_I18N */
406 static void build_upper_buffer (re_string_t *pstr) internal_function;
407 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
408 static unsigned int re_string_context_at (const re_string_t *input,
410 internal_function __attribute ((pure));
412 #define re_string_peek_byte(pstr, offset) \
413 ((pstr)->mbs[(pstr)->cur_idx + offset])
414 #define re_string_fetch_byte(pstr) \
415 ((pstr)->mbs[(pstr)->cur_idx++])
416 #define re_string_first_byte(pstr, idx) \
417 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
418 #define re_string_is_single_byte_char(pstr, idx) \
419 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
420 || (pstr)->wcs[(idx) + 1] != WEOF))
421 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
422 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
423 #define re_string_get_buffer(pstr) ((pstr)->mbs)
424 #define re_string_length(pstr) ((pstr)->len)
425 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
426 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
427 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
433 /* The OS usually guarantees only one guard page at the bottom of the stack,
434 and a page size can be as small as 4096 bytes. So we cannot safely
435 allocate anything larger than 4096 bytes. Also care for the possibility
436 of a few compiler-allocated temporary stack slots. */
437 # define __libc_use_alloca(n) ((n) < 4032)
439 /* alloca is implemented with malloc, so just use malloc. */
440 # define __libc_use_alloca(n) 0
444 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
445 #define re_calloc(t,n) ((t *) calloc (n, sizeof (t)))
446 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
447 #define re_free(p) free (p)
451 struct bin_tree_t *parent;
452 struct bin_tree_t *left;
453 struct bin_tree_t *right;
454 struct bin_tree_t *first;
455 struct bin_tree_t *next;
459 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
460 Otherwise `type' indicate the type of this node. */
463 typedef struct bin_tree_t bin_tree_t;
465 #define BIN_TREE_STORAGE_SIZE \
466 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
468 struct bin_tree_storage_t
470 struct bin_tree_storage_t *next;
471 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
473 typedef struct bin_tree_storage_t bin_tree_storage_t;
475 #define CONTEXT_WORD 1
476 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
477 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
478 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
480 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
481 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
482 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
483 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
484 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
486 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
487 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
488 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
489 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
491 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
492 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
493 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
494 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
495 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
497 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
498 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
499 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
500 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
501 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
507 re_node_set non_eps_nodes;
508 re_node_set inveclosure;
509 re_node_set *entrance_nodes;
510 struct re_dfastate_t **trtable, **word_trtable;
511 unsigned int context : 4;
512 unsigned int halt : 1;
513 /* If this state can accept `multi byte'.
514 Note that we refer to multibyte characters, and multi character
515 collating elements as `multi byte'. */
516 unsigned int accept_mb : 1;
517 /* If this state has backreference node(s). */
518 unsigned int has_backref : 1;
519 unsigned int has_constraint : 1;
521 typedef struct re_dfastate_t re_dfastate_t;
523 struct re_state_table_entry
527 re_dfastate_t **array;
530 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
536 re_dfastate_t **array;
539 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
544 Idx str_idx; /* The position NODE match at. */
546 } re_sub_match_last_t;
548 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
549 And information about the node, whose type is OP_CLOSE_SUBEXP,
550 corresponding to NODE is stored in LASTS. */
557 Idx alasts; /* Allocation size of LASTS. */
558 Idx nlasts; /* The number of LASTS. */
559 re_sub_match_last_t **lasts;
560 } re_sub_match_top_t;
562 struct re_backref_cache_entry
570 unsigned short int eps_reachable_subexps_map;
575 /* The string object corresponding to the input string. */
577 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
582 /* EFLAGS of the argument of regexec. */
584 /* Where the matching ends. */
587 /* The state log used by the matcher. */
588 re_dfastate_t **state_log;
590 /* Back reference cache. */
593 struct re_backref_cache_entry *bkref_ents;
597 re_sub_match_top_t **sub_tops;
598 } re_match_context_t;
602 re_dfastate_t **sifted_states;
603 re_dfastate_t **limited_states;
609 struct re_fail_stack_ent_t
614 re_node_set eps_via_nodes;
617 struct re_fail_stack_t
621 struct re_fail_stack_ent_t *stack;
632 re_node_set *eclosures;
633 re_node_set *inveclosures;
634 struct re_state_table_entry *state_table;
635 re_dfastate_t *init_state;
636 re_dfastate_t *init_state_word;
637 re_dfastate_t *init_state_nl;
638 re_dfastate_t *init_state_begbuf;
639 bin_tree_t *str_tree;
640 bin_tree_storage_t *str_tree_storage;
641 re_bitset_ptr_t sb_char;
642 int str_tree_storage_idx;
644 /* number of subexpressions `re_nsub' is in regex_t. */
645 re_hashval_t state_hash_mask;
647 Idx nbackref; /* The number of backreference in this dfa. */
649 /* Bitmap expressing which backreference is used. */
650 unsigned int used_bkref_map;
651 unsigned int completed_bkref_map;
653 unsigned int has_plural_match : 1;
654 /* If this dfa has "multibyte node", which is a backreference or
655 a node which can accept multibyte character or multi character
656 collating element. */
657 unsigned int has_mb_node : 1;
658 unsigned int is_utf8 : 1;
659 unsigned int map_notascii : 1;
660 unsigned int word_ops_used : 1;
668 __libc_lock_define (, lock)
671 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
672 #define re_node_set_remove(set,id) \
673 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
674 #define re_node_set_empty(p) ((p)->nelem = 0)
675 #define re_node_set_free(set) re_free ((set)->elems)
677 static void free_state (re_dfastate_t *state) internal_function;
691 bracket_elem_type type;
701 /* Inline functions for bitset operation. */
703 bitset_not (bitset set)
706 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
707 set[bitset_i] = ~set[bitset_i];
711 bitset_merge (bitset dest, const bitset src)
714 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
715 dest[bitset_i] |= src[bitset_i];
719 bitset_not_merge (bitset dest, const bitset src)
722 for (i = 0; i < BITSET_UINTS; ++i)
727 bitset_mask (bitset dest, const bitset src)
730 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
731 dest[bitset_i] &= src[bitset_i];
734 #if defined RE_ENABLE_I18N
735 /* Inline functions for re_string. */
737 internal_function __attribute ((pure))
738 re_string_char_size_at (const re_string_t *pstr, Idx idx)
741 if (pstr->mb_cur_max == 1)
743 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
744 if (pstr->wcs[idx + byte_idx] != WEOF)
750 internal_function __attribute ((pure))
751 re_string_wchar_at (const re_string_t *pstr, Idx idx)
753 if (pstr->mb_cur_max == 1)
754 return (wint_t) pstr->mbs[idx];
755 return (wint_t) pstr->wcs[idx];
759 internal_function __attribute ((pure))
760 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
763 const unsigned char *p, *extra;
764 const int32_t *table, *indirect;
766 # include <locale/weight.h>
767 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
771 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
772 extra = (const unsigned char *)
773 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
774 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
775 _NL_COLLATE_INDIRECTMB);
778 return p - pstr->mbs - idx;
784 #endif /* RE_ENABLE_I18N */
786 #endif /* _REGEX_INTERNAL_H */