3 union locale_data_value {
8 extern __thread struct locale_data **_nl_current_LC_CTYPE
9 __attribute__ ((tls_model("initial-exec")));
10 typedef unsigned long int reg_syntax_t;
11 extern reg_syntax_t re_syntax_options;
13 REG_ENOSYS = -1, REG_NOERROR =
14 0, REG_NOMATCH, REG_BADPAT, REG_ECOLLATE, REG_ECTYPE, REG_EESCAPE,
15 REG_ESUBREG, REG_EBRACK, REG_EPAREN, REG_EBRACE, REG_BADBR,
16 REG_ERANGE, REG_ESPACE, REG_BADRPT, REG_EEND, REG_ESIZE, REG_ERPAREN
19 unsigned char *buffer;
22 unsigned char *translate;
24 unsigned fastmap_accurate:1;
28 extern __thread void *__libc_tsd_CTYPE_TOLOWER;
29 inline int **__ctype_tolower_loc(void)
35 u.ptr = (&__libc_tsd_CTYPE_TOLOWER);
38 ((int *)((*_nl_current_LC_CTYPE)->values[0].string) + 128);
41 inline int tolower(int __c)
43 return *__ctype_tolower_loc()[__c];
47 typedef unsigned long int *bitset_t;
48 typedef unsigned long int *re_bitset_ptr_t;
54 NON_TYPE = 0, CHARACTER = 1, END_OF_RE = 2, SIMPLE_BRACKET =
55 3, OP_BACK_REF = 4, OP_PERIOD = 5, COMPLEX_BRACKET =
56 6, OP_UTF8_PERIOD = 7, OP_OPEN_SUBEXP = 8 | 0, OP_CLOSE_SUBEXP =
57 8 | 1, OP_ALT = 8 | 2, OP_DUP_ASTERISK = 8 | 3, ANCHOR =
58 8 | 4, CONCAT = 16, SUBEXP = 17, OP_DUP_PLUS =
59 18, OP_DUP_QUESTION, OP_OPEN_BRACKET, OP_CLOSE_BRACKET,
60 OP_CHARSET_RANGE, OP_OPEN_DUP_NUM, OP_CLOSE_DUP_NUM,
61 OP_NON_MATCH_LIST, OP_OPEN_COLL_ELEM, OP_CLOSE_COLL_ELEM,
62 OP_OPEN_EQUIV_CLASS, OP_CLOSE_EQUIV_CLASS, OP_OPEN_CHAR_CLASS,
63 OP_CLOSE_CHAR_CLASS, OP_WORD, OP_NOTWORD, OP_SPACE, OP_NOTSPACE,
73 re_bitset_ptr_t sbcset;
76 re_token_type_t type:8;
78 typedef struct re_dfa_t re_dfa_t;
79 typedef struct bin_tree_t bin_tree_t;
80 struct re_dfastate_t {
83 typedef struct re_dfastate_t re_dfastate_t;
86 unsigned int nodes_len;
90 re_node_set *eclosures;
91 re_dfastate_t *init_state;
93 re_bitset_ptr_t sb_char;
95 unsigned int has_mb_node:1;
96 unsigned int is_utf8:1;
97 unsigned int map_notascii:1;
101 static reg_errcode_t re_compile_internal(regex_t * preg, const char *pattern,
103 reg_syntax_t syntax);
104 static void re_compile_fastmap_iter(regex_t * bufp,
105 const re_dfastate_t * init_state,
107 static reg_errcode_t analyze(regex_t * preg);
108 static reg_errcode_t optimize_subexps(void *extra, bin_tree_t * node);
109 const char *__re_compile_pattern(pattern, length, bufp)
113 ret = re_compile_internal(bufp, pattern, length, re_syntax_options);
116 int __re_compile_fastmap(regex_t * bufp)
118 re_dfa_t *dfa = (re_dfa_t *) bufp->buffer;
119 char *fastmap = bufp->fastmap;
120 re_compile_fastmap_iter(bufp, dfa->init_state, fastmap);
122 static inline void re_set_fastmap(char *fastmap, int icase, int ch)
126 if (sizeof(ch) > 1) {
134 static void re_compile_fastmap_iter(regex_t * bufp,
135 const re_dfastate_t * init_state,
138 re_dfa_t *dfa = (re_dfa_t *) bufp->buffer;
140 int icase = (dfa->mb_cur_max == 1 && (bufp->syntax & 0x40000));
141 for (node_cnt = 0; node_cnt < init_state->nodes.nelem; ++node_cnt) {
142 int node = init_state->nodes.elems[node_cnt];
143 re_token_type_t type = dfa->nodes[node].type;
144 if (type == CHARACTER) {
145 if ((bufp->syntax & 0x40000) && dfa->mb_cur_max > 1) {
146 void *buf = alloca(dfa->mb_cur_max);
150 for (; i < (256 / (sizeof(unsigned long int) * 8));) {
152 unsigned long int w =
153 dfa->nodes[node].opr.sbcset[i];
156 re_set_fastmap(fastmap, icase,
159 } else if (type == COMPLEX_BRACKET) {
161 re_charset_t *cset = dfa->nodes[node].opr.mbcset;
163 ; i < cset->nmbchars;
167 if (__wcrtomb(buf, cset->mbchars[i], &state) !=
169 re_set_fastmap(fastmap, icase,
172 if ((bufp->syntax & 0x4000)
173 && dfa->mb_cur_max > 1) {
175 (buf, towlower(cset->mbchars[i]),
177 re_set_fastmap(fastmap, 0,
184 static bitset_t utf8_sb_map;
185 static reg_errcode_t re_compile_internal(regex_t * preg, const char *pattern,
191 preg->fastmap_accurate = 0;
192 preg->not_bol = preg->not_eol = 0;
194 if (dfa->is_utf8 && !(syntax & 0x400000) && preg->translate == 0)
196 if (dfa->mb_cur_max == 6 && __extension__( {
197 unsigned int __s1_len = strlen(((*_nl_current_LC_CTYPE)->values[0].string));
198 unsigned int __s2_len= strlen("UTF-8");
200 (!((int)((((*_nl_current_LC_CTYPE)->values[0].string)) + 1) - (int)(((*_nl_current_LC_CTYPE)->values[0].string)) == 1) || __s1_len >= 4) && (!((int)(("UTF-8") + 1) - (int)("UTF-8") == 1) || __s2_len >= 4)) ? strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8") : (((int)((((*_nl_current_LC_CTYPE)->values[0].string)) + 1) - (int)(((*_nl_current_LC_CTYPE)->values[0].string)) == 1) && (__s1_len = strlen(((*_nl_current_LC_CTYPE)->values[0].string)),
201 __s1_len < 4) ? (((int)(("UTF-8") + 1) - (int)("UTF-8") == 1) ? strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8") : (__extension__( {
219 (((__const unsigned char *)(__const char *)(((*_nl_current_LC_CTYPE)->values[0].string)))[0] - __s2[0]); __result;}
220 ))): (((int)(("UTF-8") + 1) - (int)("UTF-8") == 1) && (__s2_len = strlen("UTF-8"), __s2_len < 4) ? (((int)((((*_nl_current_LC_CTYPE)->values[0].string)) + 1) - (int)(((*_nl_current_LC_CTYPE)->values[0].string)) == 1) ? strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8") : (__extension__( {
233 (((*_nl_current_LC_CTYPE)->values[0].string)); register int __result = __s1[0] - ((__const unsigned char *)(__const char *)("UTF-8"))[0]; __result;}
234 ))): strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8"))));}
239 (*_nl_current_LC_CTYPE)->values[0].word != 0;
240 if (dfa->mb_cur_max > 1) {
242 dfa->sb_char = (re_bitset_ptr_t) utf8_sb_map;
248 (re_bitset_ptr_t) calloc(sizeof(bitset_t), 1);
251 i < (256 / (sizeof(unsigned long int) * 8));)
256 unsigned int wch = __btowc(ch);
257 if (wch != (0xffffffffu))
258 dfa->sb_char[i] |= 1 << j;
266 for (; node < dfa->nodes_len;)
267 switch (dfa->nodes[node].type) {
269 if (dfa->nodes[node].opr.c >= 0x80)
272 i < (256 / (sizeof(unsigned long int) * 8));)
273 if (dfa->nodes[node].opr.sbcset[i])
276 dfa->has_mb_node = dfa->nbackref > 0 || has_period;
278 static reg_errcode_t analyze(regex_t * preg)
280 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
281 if (dfa->nexts == 0 || dfa->org_indices == 0 || dfa->edests == 0
282 || dfa->eclosures == 0)
284 if (dfa->subexp_map != 0) {
286 for (; i < preg->re_nsub;)
287 dfa->subexp_map[i] = i;
288 preorder(dfa->str_tree, optimize_subexps, dfa);
290 ; i < preg->re_nsub; i++)
291 if (dfa->subexp_map[i] != i)