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;
16 unsigned char *buffer;
19 unsigned char *translate;
22 extern __thread void *__libc_tsd_CTYPE_TOLOWER;
23 inline int **__ctype_tolower_loc(void)
29 u.ptr = (&__libc_tsd_CTYPE_TOLOWER);
32 ((int *)((*_nl_current_LC_CTYPE)->values[0].string) + 128);
35 inline int tolower(int __c)
37 return *__ctype_tolower_loc()[__c];
41 typedef unsigned long int *bitset_t;
42 typedef unsigned long int *re_bitset_ptr_t;
48 NON_TYPE = 0, CHARACTER = 1, END_OF_RE = 2, SIMPLE_BRACKET =
49 3, OP_BACK_REF = 4, OP_PERIOD = 5, COMPLEX_BRACKET =
50 6, OP_UTF8_PERIOD = 7, OP_OPEN_SUBEXP = 8 | 0, OP_CLOSE_SUBEXP =
51 8 | 1, OP_ALT = 8 | 2, OP_DUP_ASTERISK = 8 | 3, ANCHOR =
52 8 | 4, CONCAT = 16, SUBEXP = 17, OP_DUP_PLUS =
62 re_bitset_ptr_t sbcset;
65 re_token_type_t type:8;
67 typedef struct re_dfa_t re_dfa_t;
68 typedef struct bin_tree_t bin_tree_t;
69 struct re_dfastate_t {
72 typedef struct re_dfastate_t re_dfastate_t;
75 unsigned int nodes_len;
76 re_dfastate_t *init_state;
78 re_bitset_ptr_t sb_char;
80 unsigned int has_mb_node:1;
81 unsigned int is_utf8:1;
82 unsigned int map_notascii:1;
86 static reg_errcode_t re_compile_internal(regex_t * preg, const char *pattern,
89 static void re_compile_fastmap_iter(regex_t * bufp,
90 const re_dfastate_t * init_state,
92 static reg_errcode_t analyze(regex_t * preg);
93 static reg_errcode_t optimize_subexps(void *extra, bin_tree_t * node);
94 const char *__re_compile_pattern(pattern, length, bufp)
98 ret = re_compile_internal(bufp, pattern, length, re_syntax_options);
101 int __re_compile_fastmap(regex_t * bufp)
103 re_dfa_t *dfa = (re_dfa_t *) bufp->buffer;
104 char *fastmap = bufp->fastmap;
105 re_compile_fastmap_iter(bufp, dfa->init_state, fastmap);
107 static inline void re_set_fastmap(char *fastmap, int icase, int ch)
111 if (sizeof(ch) > 1) {
119 static void re_compile_fastmap_iter(regex_t * bufp,
120 const re_dfastate_t * init_state,
123 re_dfa_t *dfa = (re_dfa_t *) bufp->buffer;
125 int icase = (dfa->mb_cur_max == 1 && (bufp->syntax & 0x40000));
126 for (node_cnt = 0; node_cnt < init_state->nodes.nelem; ++node_cnt) {
127 int node = init_state->nodes.elems[node_cnt];
128 re_token_type_t type = dfa->nodes[node].type;
129 if (type == CHARACTER) {
130 if ((bufp->syntax & 0x40000) && dfa->mb_cur_max > 1) {
131 void *buf = alloca(dfa->mb_cur_max);
135 for (; i < (256 / (sizeof(unsigned long int) * 8));) {
137 unsigned long int w =
138 dfa->nodes[node].opr.sbcset[i];
141 re_set_fastmap(fastmap, icase,
144 } else if (type == COMPLEX_BRACKET) {
146 re_charset_t *cset = dfa->nodes[node].opr.mbcset;
147 for (; i < cset->nmbchars;) {
150 if (__wcrtomb(buf, cset->mbchars[i], &state) !=
152 re_set_fastmap(fastmap, icase, *buf);
153 if ((bufp->syntax & 0x4000)
154 && dfa->mb_cur_max > 1) {
160 static bitset_t utf8_sb_map;
161 static reg_errcode_t re_compile_internal(regex_t * preg, const char *pattern,
168 if (dfa->is_utf8 && !(syntax & 0x400000) && preg->translate == 0)
170 if (dfa->mb_cur_max == 6 && __extension__( {
171 unsigned int __s1_len =
172 strlen(((*_nl_current_LC_CTYPE)->values[0].string));
173 unsigned int __s2_len = strlen("UTF-8");
174 (((!((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)), __s1_len < 4) ? (((int)(("UTF-8") + 1) - (int)("UTF-8") == 1) ? strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8") : (
175 ((*_nl_current_LC_CTYPE)->values[0].string[0] - 'U')
178 (((int)(("UTF-8") + 1) - (int)("UTF-8") == 1) && (
179 __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") : (
180 ((*_nl_current_LC_CTYPE)->values[0].string[0] - 'U')
181 )): strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8"))));}
184 dfa->map_notascii = (*_nl_current_LC_CTYPE)->values[0].word != 0;
185 if (dfa->mb_cur_max > 1) {
187 dfa->sb_char = (re_bitset_ptr_t) utf8_sb_map;
193 i < (256 / (sizeof(unsigned long int) * 8));
195 unsigned int wch = __btowc(ch);
196 if (wch != (0xffffffffu))
197 dfa->sb_char[i] |= 1 << j;
205 for (; node < dfa->nodes_len;)
206 switch (dfa->nodes[node].type) {
208 if (dfa->nodes[node].opr.c >= 0x80)
210 for (; i < (256 / (sizeof(unsigned long int) * 8));)
211 if (dfa->nodes[node].opr.sbcset[i])
214 dfa->has_mb_node = dfa->nbackref > 0 || has_period;
216 static reg_errcode_t analyze(regex_t * preg)
218 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
219 if (dfa->subexp_map != 0) {
221 for (; i < preg->re_nsub;)
222 dfa->subexp_map[i] = i;
223 preorder(dfa->str_tree, optimize_subexps, dfa);
224 for (; i < preg->re_nsub; i++)
225 if (dfa->subexp_map[i] != i)