]>
Commit | Line | Data |
---|---|---|
db3b92b4 CS |
1 | /**********************************************************************\r |
2 | regparse.c - Oniguruma (regular expression library)\r | |
3 | **********************************************************************/\r | |
4 | /*-\r | |
5 | * Copyright (c) 2002-2008 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>\r | |
6 | * All rights reserved.\r | |
7 | *\r | |
8 | * Copyright (c) 2015, Hewlett Packard Enterprise Development, L.P.<BR>\r | |
9 | *\r | |
10 | * Redistribution and use in source and binary forms, with or without\r | |
11 | * modification, are permitted provided that the following conditions\r | |
12 | * are met:\r | |
13 | * 1. Redistributions of source code must retain the above copyright\r | |
14 | * notice, this list of conditions and the following disclaimer.\r | |
15 | * 2. Redistributions in binary form must reproduce the above copyright\r | |
16 | * notice, this list of conditions and the following disclaimer in the\r | |
17 | * documentation and/or other materials provided with the distribution.\r | |
18 | *\r | |
19 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND\r | |
20 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE\r | |
21 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE\r | |
22 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE\r | |
23 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL\r | |
24 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS\r | |
25 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)\r | |
26 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT\r | |
27 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\r | |
28 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\r | |
29 | * SUCH DAMAGE.\r | |
30 | */\r | |
31 | \r | |
32 | #include "regparse.h"\r | |
33 | #include "st.h"\r | |
34 | \r | |
35 | #define WARN_BUFSIZE 256\r | |
36 | \r | |
37 | #define CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS\r | |
38 | \r | |
39 | \r | |
40 | OnigSyntaxType OnigSyntaxRuby = {\r | |
41 | (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY |\r | |
42 | ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 |\r | |
43 | ONIG_SYN_OP_ESC_X_BRACE_HEX8 | ONIG_SYN_OP_ESC_CONTROL_CHARS |\r | |
44 | ONIG_SYN_OP_ESC_C_CONTROL )\r | |
45 | & ~ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END )\r | |
46 | , ( ONIG_SYN_OP2_QMARK_GROUP_EFFECT |\r | |
47 | ONIG_SYN_OP2_OPTION_RUBY |\r | |
48 | ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP | ONIG_SYN_OP2_ESC_K_NAMED_BACKREF |\r | |
49 | ONIG_SYN_OP2_ESC_G_SUBEXP_CALL |\r | |
50 | ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY |\r | |
51 | ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT |\r | |
52 | ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT |\r | |
53 | ONIG_SYN_OP2_CCLASS_SET_OP | ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL |\r | |
54 | ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META | ONIG_SYN_OP2_ESC_V_VTAB |\r | |
55 | ONIG_SYN_OP2_ESC_H_XDIGIT )\r | |
56 | , ( SYN_GNU_REGEX_BV | \r | |
57 | ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV |\r | |
58 | ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND |\r | |
59 | ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP |\r | |
60 | ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME |\r | |
61 | ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY |\r | |
62 | ONIG_SYN_WARN_CC_OP_NOT_ESCAPED |\r | |
63 | ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT )\r | |
64 | , ONIG_OPTION_NONE\r | |
65 | ,\r | |
66 | {\r | |
67 | (OnigCodePoint )'\\' /* esc */\r | |
68 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar '.' */\r | |
69 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anytime '*' */\r | |
70 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* zero or one time '?' */\r | |
71 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* one or more time '+' */\r | |
72 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar anytime */\r | |
73 | }\r | |
74 | };\r | |
75 | \r | |
76 | OnigSyntaxType* OnigDefaultSyntax = ONIG_SYNTAX_RUBY;\r | |
77 | \r | |
78 | extern void onig_null_warn(const char* s ARG_UNUSED) { }\r | |
79 | \r | |
80 | #ifdef DEFAULT_WARN_FUNCTION\r | |
81 | static OnigWarnFunc onig_warn = (OnigWarnFunc )DEFAULT_WARN_FUNCTION;\r | |
82 | #else\r | |
83 | static OnigWarnFunc onig_warn = onig_null_warn;\r | |
84 | #endif\r | |
85 | \r | |
86 | #ifdef DEFAULT_VERB_WARN_FUNCTION\r | |
87 | static OnigWarnFunc onig_verb_warn = (OnigWarnFunc )DEFAULT_VERB_WARN_FUNCTION;\r | |
88 | #else\r | |
89 | static OnigWarnFunc onig_verb_warn = onig_null_warn;\r | |
90 | #endif\r | |
91 | \r | |
92 | extern void onig_set_warn_func(OnigWarnFunc f)\r | |
93 | {\r | |
94 | onig_warn = f;\r | |
95 | }\r | |
96 | \r | |
97 | extern void onig_set_verb_warn_func(OnigWarnFunc f)\r | |
98 | {\r | |
99 | onig_verb_warn = f;\r | |
100 | }\r | |
101 | \r | |
102 | static void\r | |
103 | bbuf_free(BBuf* bbuf)\r | |
104 | {\r | |
105 | if (IS_NOT_NULL(bbuf)) {\r | |
106 | if (IS_NOT_NULL(bbuf->p)) xfree(bbuf->p);\r | |
107 | xfree(bbuf);\r | |
108 | }\r | |
109 | }\r | |
110 | \r | |
111 | static int\r | |
112 | bbuf_clone(BBuf** rto, BBuf* from)\r | |
113 | {\r | |
114 | int r;\r | |
115 | BBuf *to;\r | |
116 | \r | |
117 | *rto = to = (BBuf* )xmalloc(sizeof(BBuf));\r | |
118 | CHECK_NULL_RETURN_MEMERR(to);\r | |
119 | r = BBUF_INIT(to, from->alloc);\r | |
120 | if (r != 0) return r;\r | |
121 | to->used = from->used;\r | |
122 | xmemcpy(to->p, from->p, from->used);\r | |
123 | return 0;\r | |
124 | }\r | |
125 | \r | |
126 | #define BACKREF_REL_TO_ABS(rel_no, env) \\r | |
127 | ((env)->num_mem + 1 + (rel_no))\r | |
128 | \r | |
129 | #define ONOFF(v,f,negative) (negative) ? ((v) &= ~(f)) : ((v) |= (f))\r | |
130 | \r | |
131 | #define MBCODE_START_POS(enc) \\r | |
132 | (OnigCodePoint )(ONIGENC_MBC_MINLEN(enc) > 1 ? 0 : 0x80)\r | |
133 | \r | |
134 | #define SET_ALL_MULTI_BYTE_RANGE(enc, pbuf) \\r | |
135 | add_code_range_to_buf(pbuf, MBCODE_START_POS(enc), ~((OnigCodePoint )0))\r | |
136 | \r | |
137 | #define ADD_ALL_MULTI_BYTE_RANGE(enc, mbuf) do {\\r | |
138 | if (! ONIGENC_IS_SINGLEBYTE(enc)) {\\r | |
139 | r = SET_ALL_MULTI_BYTE_RANGE(enc, &(mbuf));\\r | |
140 | if (r) return r;\\r | |
141 | }\\r | |
142 | } while (0)\r | |
143 | \r | |
144 | \r | |
145 | #define BITSET_IS_EMPTY(bs,empty) do {\\r | |
146 | int i;\\r | |
147 | empty = 1;\\r | |
148 | for (i = 0; i < (int )BITSET_SIZE; i++) {\\r | |
149 | if ((bs)[i] != 0) {\\r | |
150 | empty = 0; break;\\r | |
151 | }\\r | |
152 | }\\r | |
153 | } while (0)\r | |
154 | \r | |
155 | static void\r | |
156 | bitset_set_range(BitSetRef bs, int from, int to)\r | |
157 | {\r | |
158 | int i;\r | |
159 | for (i = from; i <= to && i < SINGLE_BYTE_SIZE; i++) {\r | |
160 | BITSET_SET_BIT(bs, i);\r | |
161 | }\r | |
162 | }\r | |
163 | \r | |
164 | #if 0\r | |
165 | static void\r | |
166 | bitset_set_all(BitSetRef bs)\r | |
167 | {\r | |
168 | int i;\r | |
169 | for (i = 0; i < BITSET_SIZE; i++) { bs[i] = ~((Bits )0); }\r | |
170 | }\r | |
171 | #endif\r | |
172 | \r | |
173 | static void\r | |
174 | bitset_invert(BitSetRef bs)\r | |
175 | {\r | |
176 | int i;\r | |
177 | for (i = 0; i < (int )BITSET_SIZE; i++) { bs[i] = ~(bs[i]); }\r | |
178 | }\r | |
179 | \r | |
180 | static void\r | |
181 | bitset_invert_to(BitSetRef from, BitSetRef to)\r | |
182 | {\r | |
183 | int i;\r | |
184 | for (i = 0; i < (int )BITSET_SIZE; i++) { to[i] = ~(from[i]); }\r | |
185 | }\r | |
186 | \r | |
187 | static void\r | |
188 | bitset_and(BitSetRef dest, BitSetRef bs)\r | |
189 | {\r | |
190 | int i;\r | |
191 | for (i = 0; i < (int )BITSET_SIZE; i++) { dest[i] &= bs[i]; }\r | |
192 | }\r | |
193 | \r | |
194 | static void\r | |
195 | bitset_or(BitSetRef dest, BitSetRef bs)\r | |
196 | {\r | |
197 | int i;\r | |
198 | for (i = 0; i < (int )BITSET_SIZE; i++) { dest[i] |= bs[i]; }\r | |
199 | }\r | |
200 | \r | |
201 | static void\r | |
202 | bitset_copy(BitSetRef dest, BitSetRef bs)\r | |
203 | {\r | |
204 | int i;\r | |
205 | for (i = 0; i < (int )BITSET_SIZE; i++) { dest[i] = bs[i]; }\r | |
206 | }\r | |
207 | \r | |
208 | extern int\r | |
209 | onig_strncmp(const UChar* s1, const UChar* s2, int n)\r | |
210 | {\r | |
211 | int x;\r | |
212 | \r | |
213 | while (n-- > 0) {\r | |
214 | x = *s2++ - *s1++;\r | |
215 | if (x) return x;\r | |
216 | }\r | |
217 | return 0;\r | |
218 | }\r | |
219 | \r | |
220 | extern void\r | |
221 | onig_strcpy(UChar* dest, const UChar* src, const UChar* end)\r | |
222 | {\r | |
223 | int len = (int)(end - src);\r | |
224 | if (len > 0) {\r | |
225 | xmemcpy(dest, src, len);\r | |
226 | dest[len] = (UChar )0;\r | |
227 | }\r | |
228 | }\r | |
229 | \r | |
230 | #ifdef USE_NAMED_GROUP\r | |
231 | static UChar*\r | |
232 | strdup_with_null(OnigEncoding enc, UChar* s, UChar* end)\r | |
233 | {\r | |
234 | int slen, term_len, i;\r | |
235 | UChar *r;\r | |
236 | \r | |
237 | slen = (int)(end - s);\r | |
238 | term_len = ONIGENC_MBC_MINLEN(enc);\r | |
239 | \r | |
240 | r = (UChar* )xmalloc(slen + term_len);\r | |
241 | CHECK_NULL_RETURN(r);\r | |
242 | xmemcpy(r, s, slen);\r | |
243 | \r | |
244 | for (i = 0; i < term_len; i++)\r | |
245 | r[slen + i] = (UChar )0;\r | |
246 | \r | |
247 | return r;\r | |
248 | }\r | |
249 | #endif\r | |
250 | \r | |
251 | /* scan pattern methods */\r | |
252 | #define PEND_VALUE 0\r | |
253 | \r | |
254 | #define PFETCH_READY UChar* pfetch_prev\r | |
255 | #define PEND (p < end ? 0 : 1)\r | |
256 | #define PUNFETCH p = pfetch_prev\r | |
257 | #define PINC do { \\r | |
258 | pfetch_prev = p; \\r | |
259 | p += ONIGENC_MBC_ENC_LEN(enc, p); \\r | |
260 | } while (0)\r | |
261 | #define PFETCH(c) do { \\r | |
262 | c = ONIGENC_MBC_TO_CODE(enc, p, end); \\r | |
263 | pfetch_prev = p; \\r | |
264 | p += ONIGENC_MBC_ENC_LEN(enc, p); \\r | |
265 | } while (0)\r | |
266 | \r | |
267 | #define PINC_S do { \\r | |
268 | p += ONIGENC_MBC_ENC_LEN(enc, p); \\r | |
269 | } while (0)\r | |
270 | #define PFETCH_S(c) do { \\r | |
271 | c = ONIGENC_MBC_TO_CODE(enc, p, end); \\r | |
272 | p += ONIGENC_MBC_ENC_LEN(enc, p); \\r | |
273 | } while (0)\r | |
274 | \r | |
275 | #define PPEEK (p < end ? ONIGENC_MBC_TO_CODE(enc, p, end) : PEND_VALUE)\r | |
276 | #define PPEEK_IS(c) (PPEEK == (OnigCodePoint )c)\r | |
277 | \r | |
278 | static UChar*\r | |
279 | strcat_capa(UChar* dest, UChar* dest_end, const UChar* src, const UChar* src_end,\r | |
280 | int capa, int oldCapa)\r | |
281 | {\r | |
282 | UChar* r;\r | |
283 | \r | |
284 | if (dest)\r | |
285 | r = (UChar* )xrealloc(dest, capa + 1, oldCapa);\r | |
286 | else\r | |
287 | r = (UChar* )xmalloc(capa + 1);\r | |
288 | \r | |
289 | CHECK_NULL_RETURN(r);\r | |
290 | onig_strcpy(r + (dest_end - dest), src, src_end);\r | |
291 | return r;\r | |
292 | }\r | |
293 | \r | |
294 | /* dest on static area */\r | |
295 | static UChar*\r | |
296 | strcat_capa_from_static(UChar* dest, UChar* dest_end,\r | |
297 | const UChar* src, const UChar* src_end, int capa)\r | |
298 | {\r | |
299 | UChar* r;\r | |
300 | \r | |
301 | r = (UChar* )xmalloc(capa + 1);\r | |
302 | CHECK_NULL_RETURN(r);\r | |
303 | onig_strcpy(r, dest, dest_end);\r | |
304 | onig_strcpy(r + (dest_end - dest), src, src_end);\r | |
305 | return r;\r | |
306 | }\r | |
307 | \r | |
308 | \r | |
309 | #ifdef USE_ST_LIBRARY\r | |
310 | \r | |
311 | typedef struct {\r | |
312 | UChar* s;\r | |
313 | UChar* end;\r | |
314 | } st_str_end_key;\r | |
315 | \r | |
316 | static int\r | |
317 | str_end_cmp(st_str_end_key* x, st_str_end_key* y)\r | |
318 | {\r | |
319 | UChar *p, *q;\r | |
320 | int c;\r | |
321 | \r | |
322 | if ((x->end - x->s) != (y->end - y->s))\r | |
323 | return 1;\r | |
324 | \r | |
325 | p = x->s;\r | |
326 | q = y->s;\r | |
327 | while (p < x->end) {\r | |
328 | c = (int )*p - (int )*q;\r | |
329 | if (c != 0) return c;\r | |
330 | \r | |
331 | p++; q++;\r | |
332 | }\r | |
333 | \r | |
334 | return 0;\r | |
335 | }\r | |
336 | \r | |
337 | static int\r | |
338 | str_end_hash(st_str_end_key* x)\r | |
339 | {\r | |
340 | UChar *p;\r | |
341 | int val = 0;\r | |
342 | \r | |
343 | p = x->s;\r | |
344 | while (p < x->end) {\r | |
345 | val = val * 997 + (int )*p++;\r | |
346 | }\r | |
347 | \r | |
348 | return val + (val >> 5);\r | |
349 | }\r | |
350 | \r | |
351 | extern hash_table_type*\r | |
352 | onig_st_init_strend_table_with_size(int size)\r | |
353 | {\r | |
354 | static struct st_hash_type hashType = {\r | |
355 | str_end_cmp,\r | |
356 | str_end_hash,\r | |
357 | };\r | |
358 | \r | |
359 | return (hash_table_type* )\r | |
360 | onig_st_init_table_with_size(&hashType, size);\r | |
361 | }\r | |
362 | \r | |
363 | extern int\r | |
364 | onig_st_lookup_strend(hash_table_type* table, const UChar* str_key,\r | |
365 | const UChar* end_key, hash_data_type *value)\r | |
366 | {\r | |
367 | st_str_end_key key;\r | |
368 | \r | |
369 | key.s = (UChar* )str_key;\r | |
370 | key.end = (UChar* )end_key;\r | |
371 | \r | |
372 | return onig_st_lookup(table, (st_data_t )(&key), value);\r | |
373 | }\r | |
374 | \r | |
375 | extern int\r | |
376 | onig_st_insert_strend(hash_table_type* table, const UChar* str_key,\r | |
377 | const UChar* end_key, hash_data_type value)\r | |
378 | {\r | |
379 | st_str_end_key* key;\r | |
380 | int result;\r | |
381 | \r | |
382 | key = (st_str_end_key* )xmalloc(sizeof(st_str_end_key));\r | |
383 | key->s = (UChar* )str_key;\r | |
384 | key->end = (UChar* )end_key;\r | |
385 | result = onig_st_insert(table, (st_data_t )key, value);\r | |
386 | if (result) {\r | |
387 | xfree(key);\r | |
388 | }\r | |
389 | return result;\r | |
390 | }\r | |
391 | \r | |
392 | #endif /* USE_ST_LIBRARY */\r | |
393 | \r | |
394 | \r | |
395 | #ifdef USE_NAMED_GROUP\r | |
396 | \r | |
397 | #define INIT_NAME_BACKREFS_ALLOC_NUM 8\r | |
398 | \r | |
399 | typedef struct {\r | |
400 | UChar* name;\r | |
401 | int name_len; /* byte length */\r | |
402 | int back_num; /* number of backrefs */\r | |
403 | int back_alloc;\r | |
404 | int back_ref1;\r | |
405 | int* back_refs;\r | |
406 | } NameEntry;\r | |
407 | \r | |
408 | #ifdef USE_ST_LIBRARY\r | |
409 | \r | |
410 | typedef st_table NameTable;\r | |
411 | typedef st_data_t HashDataType; /* 1.6 st.h doesn't define st_data_t type */\r | |
412 | \r | |
413 | #define NAMEBUF_SIZE 24\r | |
414 | #define NAMEBUF_SIZE_1 25\r | |
415 | \r | |
416 | #ifdef ONIG_DEBUG\r | |
417 | static int\r | |
418 | i_print_name_entry(UChar* key, NameEntry* e, void* arg)\r | |
419 | {\r | |
420 | int i;\r | |
421 | FILE* fp = (FILE* )arg;\r | |
422 | \r | |
423 | fprintf(fp, "%s: ", e->name);\r | |
424 | if (e->back_num == 0)\r | |
425 | fputs("-", fp);\r | |
426 | else if (e->back_num == 1)\r | |
427 | fprintf(fp, "%d", e->back_ref1);\r | |
428 | else {\r | |
429 | for (i = 0; i < e->back_num; i++) {\r | |
430 | if (i > 0) fprintf(fp, ", ");\r | |
431 | fprintf(fp, "%d", e->back_refs[i]);\r | |
432 | }\r | |
433 | }\r | |
434 | fputs("\n", fp);\r | |
435 | return ST_CONTINUE;\r | |
436 | }\r | |
437 | \r | |
438 | extern int\r | |
439 | onig_print_names(FILE* fp, regex_t* reg)\r | |
440 | {\r | |
441 | NameTable* t = (NameTable* )reg->name_table;\r | |
442 | \r | |
443 | if (IS_NOT_NULL(t)) {\r | |
444 | fprintf(fp, "name table\n");\r | |
445 | onig_st_foreach(t, i_print_name_entry, (HashDataType )fp);\r | |
446 | fputs("\n", fp);\r | |
447 | }\r | |
448 | return 0;\r | |
449 | }\r | |
450 | #endif /* ONIG_DEBUG */\r | |
451 | \r | |
452 | static int\r | |
453 | i_free_name_entry(UChar* key, NameEntry* e, void* arg ARG_UNUSED)\r | |
454 | {\r | |
455 | xfree(e->name);\r | |
456 | if (IS_NOT_NULL(e->back_refs)) xfree(e->back_refs);\r | |
457 | xfree(key);\r | |
458 | xfree(e);\r | |
459 | return ST_DELETE;\r | |
460 | }\r | |
461 | \r | |
462 | static int\r | |
463 | names_clear(regex_t* reg)\r | |
464 | {\r | |
465 | NameTable* t = (NameTable* )reg->name_table;\r | |
466 | \r | |
467 | if (IS_NOT_NULL(t)) {\r | |
468 | onig_st_foreach(t, i_free_name_entry, 0);\r | |
469 | }\r | |
470 | return 0;\r | |
471 | }\r | |
472 | \r | |
473 | extern int\r | |
474 | onig_names_free(regex_t* reg)\r | |
475 | {\r | |
476 | int r;\r | |
477 | NameTable* t;\r | |
478 | \r | |
479 | r = names_clear(reg);\r | |
480 | if (r) return r;\r | |
481 | \r | |
482 | t = (NameTable* )reg->name_table;\r | |
483 | if (IS_NOT_NULL(t)) onig_st_free_table(t);\r | |
484 | reg->name_table = (void* )NULL;\r | |
485 | return 0;\r | |
486 | }\r | |
487 | \r | |
488 | static NameEntry*\r | |
489 | name_find(regex_t* reg, const UChar* name, const UChar* name_end)\r | |
490 | {\r | |
491 | NameEntry* e;\r | |
492 | NameTable* t = (NameTable* )reg->name_table;\r | |
493 | \r | |
494 | e = (NameEntry* )NULL;\r | |
495 | if (IS_NOT_NULL(t)) {\r | |
496 | onig_st_lookup_strend(t, name, name_end, (HashDataType* )((void* )(&e)));\r | |
497 | }\r | |
498 | return e;\r | |
499 | }\r | |
500 | \r | |
501 | typedef struct {\r | |
502 | int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*);\r | |
503 | regex_t* reg;\r | |
504 | void* arg;\r | |
505 | int ret;\r | |
506 | OnigEncoding enc;\r | |
507 | } INamesArg;\r | |
508 | \r | |
509 | static int\r | |
510 | i_names(UChar* key ARG_UNUSED, NameEntry* e, INamesArg* arg)\r | |
511 | {\r | |
512 | int r = (*(arg->func))(e->name,\r | |
513 | e->name + e->name_len,\r | |
514 | e->back_num,\r | |
515 | (e->back_num > 1 ? e->back_refs : &(e->back_ref1)),\r | |
516 | arg->reg, arg->arg);\r | |
517 | if (r != 0) {\r | |
518 | arg->ret = r;\r | |
519 | return ST_STOP;\r | |
520 | }\r | |
521 | return ST_CONTINUE;\r | |
522 | }\r | |
523 | \r | |
524 | extern int\r | |
525 | onig_foreach_name(regex_t* reg,\r | |
526 | int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)\r | |
527 | {\r | |
528 | INamesArg narg;\r | |
529 | NameTable* t = (NameTable* )reg->name_table;\r | |
530 | \r | |
531 | narg.ret = 0;\r | |
532 | if (IS_NOT_NULL(t)) {\r | |
533 | narg.func = func;\r | |
534 | narg.reg = reg;\r | |
535 | narg.arg = arg;\r | |
536 | narg.enc = reg->enc; /* should be pattern encoding. */\r | |
537 | onig_st_foreach(t, i_names, (HashDataType )&narg);\r | |
538 | }\r | |
539 | return narg.ret;\r | |
540 | }\r | |
541 | \r | |
542 | static int\r | |
543 | i_renumber_name(UChar* key ARG_UNUSED, NameEntry* e, GroupNumRemap* map)\r | |
544 | {\r | |
545 | int i;\r | |
546 | \r | |
547 | if (e->back_num > 1) {\r | |
548 | for (i = 0; i < e->back_num; i++) {\r | |
549 | e->back_refs[i] = map[e->back_refs[i]].new_val;\r | |
550 | }\r | |
551 | }\r | |
552 | else if (e->back_num == 1) {\r | |
553 | e->back_ref1 = map[e->back_ref1].new_val;\r | |
554 | }\r | |
555 | \r | |
556 | return ST_CONTINUE;\r | |
557 | }\r | |
558 | \r | |
559 | extern int\r | |
560 | onig_renumber_name_table(regex_t* reg, GroupNumRemap* map)\r | |
561 | {\r | |
562 | NameTable* t = (NameTable* )reg->name_table;\r | |
563 | \r | |
564 | if (IS_NOT_NULL(t)) {\r | |
565 | onig_st_foreach(t, i_renumber_name, (HashDataType )map);\r | |
566 | }\r | |
567 | return 0;\r | |
568 | }\r | |
569 | \r | |
570 | \r | |
571 | extern int\r | |
572 | onig_number_of_names(regex_t* reg)\r | |
573 | {\r | |
574 | NameTable* t = (NameTable* )reg->name_table;\r | |
575 | \r | |
576 | if (IS_NOT_NULL(t))\r | |
577 | return t->num_entries;\r | |
578 | else\r | |
579 | return 0;\r | |
580 | }\r | |
581 | \r | |
582 | #else /* USE_ST_LIBRARY */\r | |
583 | \r | |
584 | #define INIT_NAMES_ALLOC_NUM 8\r | |
585 | \r | |
586 | typedef struct {\r | |
587 | NameEntry* e;\r | |
588 | int num;\r | |
589 | int alloc;\r | |
590 | } NameTable;\r | |
591 | \r | |
592 | #ifdef ONIG_DEBUG\r | |
593 | extern int\r | |
594 | onig_print_names(FILE* fp, regex_t* reg)\r | |
595 | {\r | |
596 | int i, j;\r | |
597 | NameEntry* e;\r | |
598 | NameTable* t = (NameTable* )reg->name_table;\r | |
599 | \r | |
600 | if (IS_NOT_NULL(t) && t->num > 0) {\r | |
601 | fprintf(fp, "name table\n");\r | |
602 | for (i = 0; i < t->num; i++) {\r | |
603 | e = &(t->e[i]);\r | |
604 | fprintf(fp, "%s: ", e->name);\r | |
605 | if (e->back_num == 0) {\r | |
606 | fputs("-", fp);\r | |
607 | }\r | |
608 | else if (e->back_num == 1) {\r | |
609 | fprintf(fp, "%d", e->back_ref1);\r | |
610 | }\r | |
611 | else {\r | |
612 | for (j = 0; j < e->back_num; j++) {\r | |
613 | if (j > 0) fprintf(fp, ", ");\r | |
614 | fprintf(fp, "%d", e->back_refs[j]);\r | |
615 | }\r | |
616 | }\r | |
617 | fputs("\n", fp);\r | |
618 | }\r | |
619 | fputs("\n", fp);\r | |
620 | }\r | |
621 | return 0;\r | |
622 | }\r | |
623 | #endif\r | |
624 | \r | |
625 | static int\r | |
626 | names_clear(regex_t* reg)\r | |
627 | {\r | |
628 | int i;\r | |
629 | NameEntry* e;\r | |
630 | NameTable* t = (NameTable* )reg->name_table;\r | |
631 | \r | |
632 | if (IS_NOT_NULL(t)) {\r | |
633 | for (i = 0; i < t->num; i++) {\r | |
634 | e = &(t->e[i]);\r | |
635 | if (IS_NOT_NULL(e->name)) {\r | |
636 | xfree(e->name);\r | |
637 | e->name = NULL;\r | |
638 | e->name_len = 0;\r | |
639 | e->back_num = 0;\r | |
640 | e->back_alloc = 0;\r | |
641 | if (IS_NOT_NULL(e->back_refs)) xfree(e->back_refs);\r | |
642 | e->back_refs = (int* )NULL;\r | |
643 | }\r | |
644 | }\r | |
645 | if (IS_NOT_NULL(t->e)) {\r | |
646 | xfree(t->e);\r | |
647 | t->e = NULL;\r | |
648 | }\r | |
649 | t->num = 0;\r | |
650 | }\r | |
651 | return 0;\r | |
652 | }\r | |
653 | \r | |
654 | extern int\r | |
655 | onig_names_free(regex_t* reg)\r | |
656 | {\r | |
657 | int r;\r | |
658 | NameTable* t;\r | |
659 | \r | |
660 | r = names_clear(reg);\r | |
661 | if (r) return r;\r | |
662 | \r | |
663 | t = (NameTable* )reg->name_table;\r | |
664 | if (IS_NOT_NULL(t)) xfree(t);\r | |
665 | reg->name_table = NULL;\r | |
666 | return 0;\r | |
667 | }\r | |
668 | \r | |
669 | static NameEntry*\r | |
670 | name_find(regex_t* reg, UChar* name, UChar* name_end)\r | |
671 | {\r | |
672 | int i, len;\r | |
673 | NameEntry* e;\r | |
674 | NameTable* t = (NameTable* )reg->name_table;\r | |
675 | \r | |
676 | if (IS_NOT_NULL(t)) {\r | |
677 | len = name_end - name;\r | |
678 | for (i = 0; i < t->num; i++) {\r | |
679 | e = &(t->e[i]);\r | |
680 | if (len == e->name_len && onig_strncmp(name, e->name, len) == 0)\r | |
681 | return e;\r | |
682 | }\r | |
683 | }\r | |
684 | return (NameEntry* )NULL;\r | |
685 | }\r | |
686 | \r | |
687 | extern int\r | |
688 | onig_foreach_name(regex_t* reg,\r | |
689 | int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)\r | |
690 | {\r | |
691 | int i, r;\r | |
692 | NameEntry* e;\r | |
693 | NameTable* t = (NameTable* )reg->name_table;\r | |
694 | \r | |
695 | if (IS_NOT_NULL(t)) {\r | |
696 | for (i = 0; i < t->num; i++) {\r | |
697 | e = &(t->e[i]);\r | |
698 | r = (*func)(e->name, e->name + e->name_len, e->back_num,\r | |
699 | (e->back_num > 1 ? e->back_refs : &(e->back_ref1)),\r | |
700 | reg, arg);\r | |
701 | if (r != 0) return r;\r | |
702 | }\r | |
703 | }\r | |
704 | return 0;\r | |
705 | }\r | |
706 | \r | |
707 | extern int\r | |
708 | onig_number_of_names(regex_t* reg)\r | |
709 | {\r | |
710 | NameTable* t = (NameTable* )reg->name_table;\r | |
711 | \r | |
712 | if (IS_NOT_NULL(t))\r | |
713 | return t->num;\r | |
714 | else\r | |
715 | return 0;\r | |
716 | }\r | |
717 | \r | |
718 | #endif /* else USE_ST_LIBRARY */\r | |
719 | \r | |
720 | static int\r | |
721 | name_add(regex_t* reg, UChar* name, UChar* name_end, int backref, ScanEnv* env)\r | |
722 | {\r | |
723 | int alloc;\r | |
724 | NameEntry* e;\r | |
725 | NameTable* t = (NameTable* )reg->name_table;\r | |
726 | \r | |
727 | if (name_end - name <= 0)\r | |
728 | return ONIGERR_EMPTY_GROUP_NAME;\r | |
729 | \r | |
730 | e = name_find(reg, name, name_end);\r | |
731 | if (IS_NULL(e)) {\r | |
732 | #ifdef USE_ST_LIBRARY\r | |
733 | if (IS_NULL(t)) {\r | |
734 | t = onig_st_init_strend_table_with_size(5);\r | |
735 | reg->name_table = (void* )t;\r | |
736 | }\r | |
737 | e = (NameEntry* )xmalloc(sizeof(NameEntry));\r | |
738 | CHECK_NULL_RETURN_MEMERR(e);\r | |
739 | \r | |
740 | e->name = strdup_with_null(reg->enc, name, name_end);\r | |
741 | if (IS_NULL(e->name)) {\r | |
742 | xfree(e); return ONIGERR_MEMORY;\r | |
743 | }\r | |
744 | onig_st_insert_strend(t, e->name, (e->name + (name_end - name)),\r | |
745 | (HashDataType )e);\r | |
746 | \r | |
747 | e->name_len = (int)(name_end - name);\r | |
748 | e->back_num = 0;\r | |
749 | e->back_alloc = 0;\r | |
750 | e->back_refs = (int* )NULL;\r | |
751 | \r | |
752 | #else\r | |
753 | \r | |
754 | if (IS_NULL(t)) {\r | |
755 | alloc = INIT_NAMES_ALLOC_NUM;\r | |
756 | t = (NameTable* )xmalloc(sizeof(NameTable));\r | |
757 | CHECK_NULL_RETURN_MEMERR(t);\r | |
758 | t->e = NULL;\r | |
759 | t->alloc = 0;\r | |
760 | t->num = 0;\r | |
761 | \r | |
762 | t->e = (NameEntry* )xmalloc(sizeof(NameEntry) * alloc);\r | |
763 | if (IS_NULL(t->e)) {\r | |
764 | xfree(t);\r | |
765 | return ONIGERR_MEMORY;\r | |
766 | }\r | |
767 | t->alloc = alloc;\r | |
768 | reg->name_table = t;\r | |
769 | goto clear;\r | |
770 | }\r | |
771 | else if (t->num == t->alloc) {\r | |
772 | int i;\r | |
773 | \r | |
774 | alloc = t->alloc * 2;\r | |
775 | t->e = (NameEntry* )xrealloc(t->e, sizeof(NameEntry) * alloc);\r | |
776 | CHECK_NULL_RETURN_MEMERR(t->e);\r | |
777 | t->alloc = alloc;\r | |
778 | \r | |
779 | clear:\r | |
780 | for (i = t->num; i < t->alloc; i++) {\r | |
781 | t->e[i].name = NULL;\r | |
782 | t->e[i].name_len = 0;\r | |
783 | t->e[i].back_num = 0;\r | |
784 | t->e[i].back_alloc = 0;\r | |
785 | t->e[i].back_refs = (int* )NULL;\r | |
786 | }\r | |
787 | }\r | |
788 | e = &(t->e[t->num]);\r | |
789 | t->num++;\r | |
790 | e->name = strdup_with_null(reg->enc, name, name_end);\r | |
791 | if (IS_NULL(e->name)) return ONIGERR_MEMORY;\r | |
792 | e->name_len = name_end - name;\r | |
793 | #endif\r | |
794 | }\r | |
795 | \r | |
796 | if (e->back_num >= 1 &&\r | |
797 | ! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME)) {\r | |
798 | onig_scan_env_set_error_string(env, ONIGERR_MULTIPLEX_DEFINED_NAME,\r | |
799 | name, name_end);\r | |
800 | return ONIGERR_MULTIPLEX_DEFINED_NAME;\r | |
801 | }\r | |
802 | \r | |
803 | e->back_num++;\r | |
804 | if (e->back_num == 1) {\r | |
805 | e->back_ref1 = backref;\r | |
806 | }\r | |
807 | else {\r | |
808 | if (e->back_num == 2) {\r | |
809 | alloc = INIT_NAME_BACKREFS_ALLOC_NUM;\r | |
810 | e->back_refs = (int* )xmalloc(sizeof(int) * alloc);\r | |
811 | CHECK_NULL_RETURN_MEMERR(e->back_refs);\r | |
812 | e->back_alloc = alloc;\r | |
813 | e->back_refs[0] = e->back_ref1;\r | |
814 | e->back_refs[1] = backref;\r | |
815 | }\r | |
816 | else {\r | |
817 | if (e->back_num > e->back_alloc) {\r | |
818 | alloc = e->back_alloc * 2;\r | |
819 | e->back_refs = (int* )xrealloc(e->back_refs, sizeof(int) * alloc, sizeof(int) * e->back_alloc);\r | |
820 | CHECK_NULL_RETURN_MEMERR(e->back_refs);\r | |
821 | e->back_alloc = alloc;\r | |
822 | }\r | |
823 | e->back_refs[e->back_num - 1] = backref;\r | |
824 | }\r | |
825 | }\r | |
826 | \r | |
827 | return 0;\r | |
828 | }\r | |
829 | \r | |
830 | extern int\r | |
831 | onig_name_to_group_numbers(regex_t* reg, const UChar* name,\r | |
832 | const UChar* name_end, int** nums)\r | |
833 | {\r | |
834 | NameEntry* e = name_find(reg, name, name_end);\r | |
835 | \r | |
836 | if (IS_NULL(e)) return ONIGERR_UNDEFINED_NAME_REFERENCE;\r | |
837 | \r | |
838 | switch (e->back_num) {\r | |
839 | case 0:\r | |
840 | break;\r | |
841 | case 1:\r | |
842 | *nums = &(e->back_ref1);\r | |
843 | break;\r | |
844 | default:\r | |
845 | *nums = e->back_refs;\r | |
846 | break;\r | |
847 | }\r | |
848 | return e->back_num;\r | |
849 | }\r | |
850 | \r | |
851 | extern int\r | |
852 | onig_name_to_backref_number(regex_t* reg, const UChar* name,\r | |
853 | const UChar* name_end, OnigRegion *region)\r | |
854 | {\r | |
855 | int i, n, *nums;\r | |
856 | \r | |
857 | n = onig_name_to_group_numbers(reg, name, name_end, &nums);\r | |
858 | if (n < 0)\r | |
859 | return n;\r | |
860 | else if (n == 0)\r | |
861 | return ONIGERR_PARSER_BUG;\r | |
862 | else if (n == 1)\r | |
863 | return nums[0];\r | |
864 | else {\r | |
865 | if (IS_NOT_NULL(region)) {\r | |
866 | for (i = n - 1; i >= 0; i--) {\r | |
867 | if (region->beg[nums[i]] != ONIG_REGION_NOTPOS)\r | |
868 | return nums[i];\r | |
869 | }\r | |
870 | }\r | |
871 | return nums[n - 1];\r | |
872 | }\r | |
873 | }\r | |
874 | \r | |
875 | #else /* USE_NAMED_GROUP */\r | |
876 | \r | |
877 | extern int\r | |
878 | onig_name_to_group_numbers(regex_t* reg, const UChar* name,\r | |
879 | const UChar* name_end, int** nums)\r | |
880 | {\r | |
881 | return ONIG_NO_SUPPORT_CONFIG;\r | |
882 | }\r | |
883 | \r | |
884 | extern int\r | |
885 | onig_name_to_backref_number(regex_t* reg, const UChar* name,\r | |
886 | const UChar* name_end, OnigRegion* region)\r | |
887 | {\r | |
888 | return ONIG_NO_SUPPORT_CONFIG;\r | |
889 | }\r | |
890 | \r | |
891 | extern int\r | |
892 | onig_foreach_name(regex_t* reg,\r | |
893 | int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)\r | |
894 | {\r | |
895 | return ONIG_NO_SUPPORT_CONFIG;\r | |
896 | }\r | |
897 | \r | |
898 | extern int\r | |
899 | onig_number_of_names(regex_t* reg)\r | |
900 | {\r | |
901 | return 0;\r | |
902 | }\r | |
903 | #endif /* else USE_NAMED_GROUP */\r | |
904 | \r | |
905 | extern int\r | |
906 | onig_noname_group_capture_is_active(regex_t* reg)\r | |
907 | {\r | |
908 | if (ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_DONT_CAPTURE_GROUP))\r | |
909 | return 0;\r | |
910 | \r | |
911 | #ifdef USE_NAMED_GROUP\r | |
912 | if (onig_number_of_names(reg) > 0 &&\r | |
913 | IS_SYNTAX_BV(reg->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) &&\r | |
914 | !ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_CAPTURE_GROUP)) {\r | |
915 | return 0;\r | |
916 | }\r | |
917 | #endif\r | |
918 | \r | |
919 | return 1;\r | |
920 | }\r | |
921 | \r | |
922 | \r | |
923 | #define INIT_SCANENV_MEMNODES_ALLOC_SIZE 16\r | |
924 | \r | |
925 | static void\r | |
926 | scan_env_clear(ScanEnv* env)\r | |
927 | {\r | |
928 | int i;\r | |
929 | \r | |
930 | BIT_STATUS_CLEAR(env->capture_history);\r | |
931 | BIT_STATUS_CLEAR(env->bt_mem_start);\r | |
932 | BIT_STATUS_CLEAR(env->bt_mem_end);\r | |
933 | BIT_STATUS_CLEAR(env->backrefed_mem);\r | |
934 | env->error = (UChar* )NULL;\r | |
935 | env->error_end = (UChar* )NULL;\r | |
936 | env->num_call = 0;\r | |
937 | env->num_mem = 0;\r | |
938 | #ifdef USE_NAMED_GROUP\r | |
939 | env->num_named = 0;\r | |
940 | #endif\r | |
941 | env->mem_alloc = 0;\r | |
942 | env->mem_nodes_dynamic = (Node** )NULL;\r | |
943 | \r | |
944 | for (i = 0; i < SCANENV_MEMNODES_SIZE; i++)\r | |
945 | env->mem_nodes_static[i] = NULL_NODE;\r | |
946 | \r | |
947 | #ifdef USE_COMBINATION_EXPLOSION_CHECK\r | |
948 | env->num_comb_exp_check = 0;\r | |
949 | env->comb_exp_max_regnum = 0;\r | |
950 | env->curr_max_regnum = 0;\r | |
951 | env->has_recursion = 0;\r | |
952 | #endif\r | |
953 | }\r | |
954 | \r | |
955 | static int\r | |
956 | scan_env_add_mem_entry(ScanEnv* env)\r | |
957 | {\r | |
958 | int i, need, alloc;\r | |
959 | Node** p;\r | |
960 | \r | |
961 | need = env->num_mem + 1;\r | |
962 | if (need >= SCANENV_MEMNODES_SIZE) {\r | |
963 | if (env->mem_alloc <= need) {\r | |
964 | if (IS_NULL(env->mem_nodes_dynamic)) {\r | |
965 | alloc = INIT_SCANENV_MEMNODES_ALLOC_SIZE;\r | |
966 | p = (Node** )xmalloc(sizeof(Node*) * alloc);\r | |
967 | xmemcpy(p, env->mem_nodes_static,\r | |
968 | sizeof(Node*) * SCANENV_MEMNODES_SIZE);\r | |
969 | }\r | |
970 | else {\r | |
971 | alloc = env->mem_alloc * 2;\r | |
972 | p = (Node** )xrealloc(env->mem_nodes_dynamic, sizeof(Node*) * alloc, sizeof(Node*) * env->mem_alloc);\r | |
973 | }\r | |
974 | CHECK_NULL_RETURN_MEMERR(p);\r | |
975 | \r | |
976 | for (i = env->num_mem + 1; i < alloc; i++)\r | |
977 | p[i] = NULL_NODE;\r | |
978 | \r | |
979 | env->mem_nodes_dynamic = p;\r | |
980 | env->mem_alloc = alloc;\r | |
981 | }\r | |
982 | }\r | |
983 | \r | |
984 | env->num_mem++;\r | |
985 | return env->num_mem;\r | |
986 | }\r | |
987 | \r | |
988 | static int\r | |
989 | scan_env_set_mem_node(ScanEnv* env, int num, Node* node)\r | |
990 | {\r | |
991 | if (env->num_mem >= num)\r | |
992 | SCANENV_MEM_NODES(env)[num] = node;\r | |
993 | else\r | |
994 | return ONIGERR_PARSER_BUG;\r | |
995 | return 0;\r | |
996 | }\r | |
997 | \r | |
998 | \r | |
999 | #ifdef USE_PARSE_TREE_NODE_RECYCLE\r | |
1000 | typedef struct _FreeNode {\r | |
1001 | struct _FreeNode* next;\r | |
1002 | } FreeNode;\r | |
1003 | \r | |
1004 | static FreeNode* FreeNodeList = (FreeNode* )NULL;\r | |
1005 | #endif\r | |
1006 | \r | |
1007 | extern void\r | |
1008 | onig_node_free(Node* node)\r | |
1009 | {\r | |
1010 | start:\r | |
1011 | if (IS_NULL(node)) return ;\r | |
1012 | \r | |
1013 | switch (NTYPE(node)) {\r | |
1014 | case NT_STR:\r | |
1015 | if (NSTR(node)->capa != 0 &&\r | |
1016 | IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) {\r | |
1017 | xfree(NSTR(node)->s);\r | |
1018 | }\r | |
1019 | break;\r | |
1020 | \r | |
1021 | case NT_LIST:\r | |
1022 | case NT_ALT:\r | |
1023 | onig_node_free(NCAR(node));\r | |
1024 | {\r | |
1025 | Node* next_node = NCDR(node);\r | |
1026 | \r | |
1027 | #ifdef USE_PARSE_TREE_NODE_RECYCLE\r | |
1028 | {\r | |
1029 | FreeNode* n = (FreeNode* )node;\r | |
1030 | \r | |
1031 | THREAD_ATOMIC_START;\r | |
1032 | n->next = FreeNodeList;\r | |
1033 | FreeNodeList = n;\r | |
1034 | THREAD_ATOMIC_END;\r | |
1035 | }\r | |
1036 | #else\r | |
1037 | xfree(node);\r | |
1038 | #endif\r | |
1039 | node = next_node;\r | |
1040 | goto start;\r | |
1041 | }\r | |
1042 | break;\r | |
1043 | \r | |
1044 | case NT_CCLASS:\r | |
1045 | {\r | |
1046 | CClassNode* cc = NCCLASS(node);\r | |
1047 | \r | |
1048 | if (IS_NCCLASS_SHARE(cc)) return ;\r | |
1049 | if (cc->mbuf)\r | |
1050 | bbuf_free(cc->mbuf);\r | |
1051 | }\r | |
1052 | break;\r | |
1053 | \r | |
1054 | case NT_QTFR:\r | |
1055 | if (NQTFR(node)->target)\r | |
1056 | onig_node_free(NQTFR(node)->target);\r | |
1057 | break;\r | |
1058 | \r | |
1059 | case NT_ENCLOSE:\r | |
1060 | if (NENCLOSE(node)->target)\r | |
1061 | onig_node_free(NENCLOSE(node)->target);\r | |
1062 | break;\r | |
1063 | \r | |
1064 | case NT_BREF:\r | |
1065 | if (IS_NOT_NULL(NBREF(node)->back_dynamic))\r | |
1066 | xfree(NBREF(node)->back_dynamic);\r | |
1067 | break;\r | |
1068 | \r | |
1069 | case NT_ANCHOR:\r | |
1070 | if (NANCHOR(node)->target)\r | |
1071 | onig_node_free(NANCHOR(node)->target);\r | |
1072 | break;\r | |
1073 | }\r | |
1074 | \r | |
1075 | #ifdef USE_PARSE_TREE_NODE_RECYCLE\r | |
1076 | {\r | |
1077 | FreeNode* n = (FreeNode* )node;\r | |
1078 | \r | |
1079 | THREAD_ATOMIC_START;\r | |
1080 | n->next = FreeNodeList;\r | |
1081 | FreeNodeList = n;\r | |
1082 | THREAD_ATOMIC_END;\r | |
1083 | }\r | |
1084 | #else\r | |
1085 | xfree(node);\r | |
1086 | #endif\r | |
1087 | }\r | |
1088 | \r | |
1089 | #ifdef USE_PARSE_TREE_NODE_RECYCLE\r | |
1090 | extern int\r | |
1091 | onig_free_node_list(void)\r | |
1092 | {\r | |
1093 | FreeNode* n;\r | |
1094 | \r | |
1095 | /* THREAD_ATOMIC_START; */\r | |
1096 | while (IS_NOT_NULL(FreeNodeList)) {\r | |
1097 | n = FreeNodeList;\r | |
1098 | FreeNodeList = FreeNodeList->next;\r | |
1099 | xfree(n);\r | |
1100 | }\r | |
1101 | /* THREAD_ATOMIC_END; */\r | |
1102 | return 0;\r | |
1103 | }\r | |
1104 | #endif\r | |
1105 | \r | |
1106 | static Node*\r | |
1107 | node_new(void)\r | |
1108 | {\r | |
1109 | Node* node;\r | |
1110 | \r | |
1111 | #ifdef USE_PARSE_TREE_NODE_RECYCLE\r | |
1112 | THREAD_ATOMIC_START;\r | |
1113 | if (IS_NOT_NULL(FreeNodeList)) {\r | |
1114 | node = (Node* )FreeNodeList;\r | |
1115 | FreeNodeList = FreeNodeList->next;\r | |
1116 | THREAD_ATOMIC_END;\r | |
1117 | return node;\r | |
1118 | }\r | |
1119 | THREAD_ATOMIC_END;\r | |
1120 | #endif\r | |
1121 | \r | |
1122 | node = (Node* )xmalloc(sizeof(Node));\r | |
1123 | /* xmemset(node, 0, sizeof(Node)); */\r | |
1124 | return node;\r | |
1125 | }\r | |
1126 | \r | |
1127 | \r | |
1128 | static void\r | |
1129 | initialize_cclass(CClassNode* cc)\r | |
1130 | {\r | |
1131 | BITSET_CLEAR(cc->bs);\r | |
1132 | /* cc->base.flags = 0; */\r | |
1133 | cc->flags = 0;\r | |
1134 | cc->mbuf = NULL;\r | |
1135 | }\r | |
1136 | \r | |
1137 | static Node*\r | |
1138 | node_new_cclass(void)\r | |
1139 | {\r | |
1140 | Node* node = node_new();\r | |
1141 | CHECK_NULL_RETURN(node);\r | |
1142 | \r | |
1143 | SET_NTYPE(node, NT_CCLASS);\r | |
1144 | initialize_cclass(NCCLASS(node));\r | |
1145 | return node;\r | |
1146 | }\r | |
1147 | \r | |
1148 | static Node*\r | |
1149 | node_new_cclass_by_codepoint_range(int not, OnigCodePoint sb_out,\r | |
1150 | const OnigCodePoint ranges[])\r | |
1151 | {\r | |
1152 | int n, i;\r | |
1153 | CClassNode* cc;\r | |
1154 | OnigCodePoint j;\r | |
1155 | \r | |
1156 | Node* node = node_new_cclass();\r | |
1157 | CHECK_NULL_RETURN(node);\r | |
1158 | \r | |
1159 | cc = NCCLASS(node);\r | |
1160 | if (not != 0) NCCLASS_SET_NOT(cc);\r | |
1161 | \r | |
1162 | BITSET_CLEAR(cc->bs);\r | |
1163 | if (sb_out > 0 && IS_NOT_NULL(ranges)) {\r | |
1164 | n = ONIGENC_CODE_RANGE_NUM(ranges);\r | |
1165 | for (i = 0; i < n; i++) {\r | |
1166 | for (j = ONIGENC_CODE_RANGE_FROM(ranges, i);\r | |
1167 | j <= (OnigCodePoint )ONIGENC_CODE_RANGE_TO(ranges, i); j++) {\r | |
1168 | if (j >= sb_out) goto sb_end;\r | |
1169 | \r | |
1170 | BITSET_SET_BIT(cc->bs, j);\r | |
1171 | }\r | |
1172 | }\r | |
1173 | }\r | |
1174 | \r | |
1175 | sb_end:\r | |
1176 | if (IS_NULL(ranges)) {\r | |
1177 | is_null:\r | |
1178 | cc->mbuf = NULL;\r | |
1179 | }\r | |
1180 | else {\r | |
1181 | BBuf* bbuf;\r | |
1182 | \r | |
1183 | n = ONIGENC_CODE_RANGE_NUM(ranges);\r | |
1184 | if (n == 0) goto is_null;\r | |
1185 | \r | |
1186 | bbuf = (BBuf* )xmalloc(sizeof(BBuf));\r | |
1187 | CHECK_NULL_RETURN(bbuf);\r | |
1188 | bbuf->alloc = n + 1;\r | |
1189 | bbuf->used = n + 1;\r | |
1190 | bbuf->p = (UChar* )((void* )ranges);\r | |
1191 | \r | |
1192 | cc->mbuf = bbuf;\r | |
1193 | }\r | |
1194 | \r | |
1195 | return node;\r | |
1196 | }\r | |
1197 | \r | |
1198 | static Node*\r | |
1199 | node_new_ctype(int type, int not)\r | |
1200 | {\r | |
1201 | Node* node = node_new();\r | |
1202 | CHECK_NULL_RETURN(node);\r | |
1203 | \r | |
1204 | SET_NTYPE(node, NT_CTYPE);\r | |
1205 | NCTYPE(node)->ctype = type;\r | |
1206 | NCTYPE(node)->not = not;\r | |
1207 | return node;\r | |
1208 | }\r | |
1209 | \r | |
1210 | static Node*\r | |
1211 | node_new_anychar(void)\r | |
1212 | {\r | |
1213 | Node* node = node_new();\r | |
1214 | CHECK_NULL_RETURN(node);\r | |
1215 | \r | |
1216 | SET_NTYPE(node, NT_CANY);\r | |
1217 | return node;\r | |
1218 | }\r | |
1219 | \r | |
1220 | static Node*\r | |
1221 | node_new_list(Node* left, Node* right)\r | |
1222 | {\r | |
1223 | Node* node = node_new();\r | |
1224 | CHECK_NULL_RETURN(node);\r | |
1225 | \r | |
1226 | SET_NTYPE(node, NT_LIST);\r | |
1227 | NCAR(node) = left;\r | |
1228 | NCDR(node) = right;\r | |
1229 | return node;\r | |
1230 | }\r | |
1231 | \r | |
1232 | extern Node*\r | |
1233 | onig_node_new_list(Node* left, Node* right)\r | |
1234 | {\r | |
1235 | return node_new_list(left, right);\r | |
1236 | }\r | |
1237 | \r | |
1238 | extern Node*\r | |
1239 | onig_node_list_add(Node* list, Node* x)\r | |
1240 | {\r | |
1241 | Node *n;\r | |
1242 | \r | |
1243 | n = onig_node_new_list(x, NULL);\r | |
1244 | if (IS_NULL(n)) return NULL_NODE;\r | |
1245 | \r | |
1246 | if (IS_NOT_NULL(list)) {\r | |
1247 | while (IS_NOT_NULL(NCDR(list)))\r | |
1248 | list = NCDR(list);\r | |
1249 | \r | |
1250 | NCDR(list) = n;\r | |
1251 | }\r | |
1252 | \r | |
1253 | return n;\r | |
1254 | }\r | |
1255 | \r | |
1256 | extern Node*\r | |
1257 | onig_node_new_alt(Node* left, Node* right)\r | |
1258 | {\r | |
1259 | Node* node = node_new();\r | |
1260 | CHECK_NULL_RETURN(node);\r | |
1261 | \r | |
1262 | SET_NTYPE(node, NT_ALT);\r | |
1263 | NCAR(node) = left;\r | |
1264 | NCDR(node) = right;\r | |
1265 | return node;\r | |
1266 | }\r | |
1267 | \r | |
1268 | extern Node*\r | |
1269 | onig_node_new_anchor(int type)\r | |
1270 | {\r | |
1271 | Node* node = node_new();\r | |
1272 | CHECK_NULL_RETURN(node);\r | |
1273 | \r | |
1274 | SET_NTYPE(node, NT_ANCHOR);\r | |
1275 | NANCHOR(node)->type = type;\r | |
1276 | NANCHOR(node)->target = NULL;\r | |
1277 | NANCHOR(node)->char_len = -1;\r | |
1278 | return node;\r | |
1279 | }\r | |
1280 | \r | |
1281 | static Node*\r | |
1282 | node_new_backref(int back_num, int* backrefs, int by_name,\r | |
1283 | #ifdef USE_BACKREF_WITH_LEVEL\r | |
1284 | int exist_level, int nest_level,\r | |
1285 | #endif\r | |
1286 | ScanEnv* env)\r | |
1287 | {\r | |
1288 | int i;\r | |
1289 | Node* node = node_new();\r | |
1290 | \r | |
1291 | CHECK_NULL_RETURN(node);\r | |
1292 | \r | |
1293 | SET_NTYPE(node, NT_BREF);\r | |
1294 | NBREF(node)->state = 0;\r | |
1295 | NBREF(node)->back_num = back_num;\r | |
1296 | NBREF(node)->back_dynamic = (int* )NULL;\r | |
1297 | if (by_name != 0)\r | |
1298 | NBREF(node)->state |= NST_NAME_REF;\r | |
1299 | \r | |
1300 | #ifdef USE_BACKREF_WITH_LEVEL\r | |
1301 | if (exist_level != 0) {\r | |
1302 | NBREF(node)->state |= NST_NEST_LEVEL;\r | |
1303 | NBREF(node)->nest_level = nest_level;\r | |
1304 | }\r | |
1305 | #endif\r | |
1306 | \r | |
1307 | for (i = 0; i < back_num; i++) {\r | |
1308 | if (backrefs[i] <= env->num_mem &&\r | |
1309 | IS_NULL(SCANENV_MEM_NODES(env)[backrefs[i]])) {\r | |
1310 | NBREF(node)->state |= NST_RECURSION; /* /...(\1).../ */\r | |
1311 | break;\r | |
1312 | }\r | |
1313 | }\r | |
1314 | \r | |
1315 | if (back_num <= NODE_BACKREFS_SIZE) {\r | |
1316 | for (i = 0; i < back_num; i++)\r | |
1317 | NBREF(node)->back_static[i] = backrefs[i];\r | |
1318 | }\r | |
1319 | else {\r | |
1320 | int* p = (int* )xmalloc(sizeof(int) * back_num);\r | |
1321 | if (IS_NULL(p)) {\r | |
1322 | onig_node_free(node);\r | |
1323 | return NULL;\r | |
1324 | }\r | |
1325 | NBREF(node)->back_dynamic = p;\r | |
1326 | for (i = 0; i < back_num; i++)\r | |
1327 | p[i] = backrefs[i];\r | |
1328 | }\r | |
1329 | return node;\r | |
1330 | }\r | |
1331 | \r | |
1332 | #ifdef USE_SUBEXP_CALL\r | |
1333 | static Node*\r | |
1334 | node_new_call(UChar* name, UChar* name_end, int gnum)\r | |
1335 | {\r | |
1336 | Node* node = node_new();\r | |
1337 | CHECK_NULL_RETURN(node);\r | |
1338 | \r | |
1339 | SET_NTYPE(node, NT_CALL);\r | |
1340 | NCALL(node)->state = 0;\r | |
1341 | NCALL(node)->target = NULL_NODE;\r | |
1342 | NCALL(node)->name = name;\r | |
1343 | NCALL(node)->name_end = name_end;\r | |
1344 | NCALL(node)->group_num = gnum; /* call by number if gnum != 0 */\r | |
1345 | return node;\r | |
1346 | }\r | |
1347 | #endif\r | |
1348 | \r | |
1349 | static Node*\r | |
1350 | node_new_quantifier(int lower, int upper, int by_number)\r | |
1351 | {\r | |
1352 | Node* node = node_new();\r | |
1353 | CHECK_NULL_RETURN(node);\r | |
1354 | \r | |
1355 | SET_NTYPE(node, NT_QTFR);\r | |
1356 | NQTFR(node)->state = 0;\r | |
1357 | NQTFR(node)->target = NULL;\r | |
1358 | NQTFR(node)->lower = lower;\r | |
1359 | NQTFR(node)->upper = upper;\r | |
1360 | NQTFR(node)->greedy = 1;\r | |
1361 | NQTFR(node)->target_empty_info = NQ_TARGET_ISNOT_EMPTY;\r | |
1362 | NQTFR(node)->head_exact = NULL_NODE;\r | |
1363 | NQTFR(node)->next_head_exact = NULL_NODE;\r | |
1364 | NQTFR(node)->is_refered = 0;\r | |
1365 | if (by_number != 0)\r | |
1366 | NQTFR(node)->state |= NST_BY_NUMBER;\r | |
1367 | \r | |
1368 | #ifdef USE_COMBINATION_EXPLOSION_CHECK\r | |
1369 | NQTFR(node)->comb_exp_check_num = 0;\r | |
1370 | #endif\r | |
1371 | \r | |
1372 | return node;\r | |
1373 | }\r | |
1374 | \r | |
1375 | static Node*\r | |
1376 | node_new_enclose(int type)\r | |
1377 | {\r | |
1378 | Node* node = node_new();\r | |
1379 | CHECK_NULL_RETURN(node);\r | |
1380 | \r | |
1381 | SET_NTYPE(node, NT_ENCLOSE);\r | |
1382 | NENCLOSE(node)->type = type;\r | |
1383 | NENCLOSE(node)->state = 0;\r | |
1384 | NENCLOSE(node)->regnum = 0;\r | |
1385 | NENCLOSE(node)->option = 0;\r | |
1386 | NENCLOSE(node)->target = NULL;\r | |
1387 | NENCLOSE(node)->call_addr = -1;\r | |
1388 | NENCLOSE(node)->opt_count = 0;\r | |
1389 | return node;\r | |
1390 | }\r | |
1391 | \r | |
1392 | extern Node*\r | |
1393 | onig_node_new_enclose(int type)\r | |
1394 | {\r | |
1395 | return node_new_enclose(type);\r | |
1396 | }\r | |
1397 | \r | |
1398 | static Node*\r | |
1399 | node_new_enclose_memory(OnigOptionType option, int is_named)\r | |
1400 | {\r | |
1401 | Node* node = node_new_enclose(ENCLOSE_MEMORY);\r | |
1402 | CHECK_NULL_RETURN(node);\r | |
1403 | if (is_named != 0)\r | |
1404 | SET_ENCLOSE_STATUS(node, NST_NAMED_GROUP);\r | |
1405 | \r | |
1406 | #ifdef USE_SUBEXP_CALL\r | |
1407 | NENCLOSE(node)->option = option;\r | |
1408 | #endif\r | |
1409 | return node;\r | |
1410 | }\r | |
1411 | \r | |
1412 | static Node*\r | |
1413 | node_new_option(OnigOptionType option)\r | |
1414 | {\r | |
1415 | Node* node = node_new_enclose(ENCLOSE_OPTION);\r | |
1416 | CHECK_NULL_RETURN(node);\r | |
1417 | NENCLOSE(node)->option = option;\r | |
1418 | return node;\r | |
1419 | }\r | |
1420 | \r | |
1421 | extern int\r | |
1422 | onig_node_str_cat(Node* node, const UChar* s, const UChar* end)\r | |
1423 | {\r | |
1424 | int addlen = (int)(end - s);\r | |
1425 | \r | |
1426 | if (addlen > 0) {\r | |
1427 | int len = (int)(NSTR(node)->end - NSTR(node)->s);\r | |
1428 | \r | |
1429 | if (NSTR(node)->capa > 0 || (len + addlen > NODE_STR_BUF_SIZE - 1)) {\r | |
1430 | UChar* p;\r | |
1431 | int capa = len + addlen + NODE_STR_MARGIN;\r | |
1432 | \r | |
1433 | if (capa <= NSTR(node)->capa) {\r | |
1434 | onig_strcpy(NSTR(node)->s + len, s, end);\r | |
1435 | }\r | |
1436 | else {\r | |
1437 | if (NSTR(node)->s == NSTR(node)->buf)\r | |
1438 | p = strcat_capa_from_static(NSTR(node)->s, NSTR(node)->end,\r | |
1439 | s, end, capa);\r | |
1440 | else\r | |
1441 | p = strcat_capa(NSTR(node)->s, NSTR(node)->end, s, end, capa, NSTR(node)->capa);\r | |
1442 | \r | |
1443 | CHECK_NULL_RETURN_MEMERR(p);\r | |
1444 | NSTR(node)->s = p;\r | |
1445 | NSTR(node)->capa = capa;\r | |
1446 | }\r | |
1447 | }\r | |
1448 | else {\r | |
1449 | onig_strcpy(NSTR(node)->s + len, s, end);\r | |
1450 | }\r | |
1451 | NSTR(node)->end = NSTR(node)->s + len + addlen;\r | |
1452 | }\r | |
1453 | \r | |
1454 | return 0;\r | |
1455 | }\r | |
1456 | \r | |
1457 | extern int\r | |
1458 | onig_node_str_set(Node* node, const UChar* s, const UChar* end)\r | |
1459 | {\r | |
1460 | onig_node_str_clear(node);\r | |
1461 | return onig_node_str_cat(node, s, end);\r | |
1462 | }\r | |
1463 | \r | |
1464 | static int\r | |
1465 | node_str_cat_char(Node* node, UChar c)\r | |
1466 | {\r | |
1467 | UChar s[1];\r | |
1468 | \r | |
1469 | s[0] = c;\r | |
1470 | return onig_node_str_cat(node, s, s + 1);\r | |
1471 | }\r | |
1472 | \r | |
1473 | extern void\r | |
1474 | onig_node_conv_to_str_node(Node* node, int flag)\r | |
1475 | {\r | |
1476 | SET_NTYPE(node, NT_STR);\r | |
1477 | NSTR(node)->flag = flag;\r | |
1478 | NSTR(node)->capa = 0;\r | |
1479 | NSTR(node)->s = NSTR(node)->buf;\r | |
1480 | NSTR(node)->end = NSTR(node)->buf;\r | |
1481 | }\r | |
1482 | \r | |
1483 | extern void\r | |
1484 | onig_node_str_clear(Node* node)\r | |
1485 | {\r | |
1486 | if (NSTR(node)->capa != 0 &&\r | |
1487 | IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) {\r | |
1488 | xfree(NSTR(node)->s);\r | |
1489 | }\r | |
1490 | \r | |
1491 | NSTR(node)->capa = 0;\r | |
1492 | NSTR(node)->flag = 0;\r | |
1493 | NSTR(node)->s = NSTR(node)->buf;\r | |
1494 | NSTR(node)->end = NSTR(node)->buf;\r | |
1495 | }\r | |
1496 | \r | |
1497 | static Node*\r | |
1498 | node_new_str(const UChar* s, const UChar* end)\r | |
1499 | {\r | |
1500 | Node* node = node_new();\r | |
1501 | CHECK_NULL_RETURN(node);\r | |
1502 | \r | |
1503 | SET_NTYPE(node, NT_STR);\r | |
1504 | NSTR(node)->capa = 0;\r | |
1505 | NSTR(node)->flag = 0;\r | |
1506 | NSTR(node)->s = NSTR(node)->buf;\r | |
1507 | NSTR(node)->end = NSTR(node)->buf;\r | |
1508 | if (onig_node_str_cat(node, s, end)) {\r | |
1509 | onig_node_free(node);\r | |
1510 | return NULL;\r | |
1511 | }\r | |
1512 | return node;\r | |
1513 | }\r | |
1514 | \r | |
1515 | extern Node*\r | |
1516 | onig_node_new_str(const UChar* s, const UChar* end)\r | |
1517 | {\r | |
1518 | return node_new_str(s, end);\r | |
1519 | }\r | |
1520 | \r | |
1521 | static Node*\r | |
1522 | node_new_str_raw(UChar* s, UChar* end)\r | |
1523 | {\r | |
1524 | Node* node = node_new_str(s, end);\r | |
1525 | NSTRING_SET_RAW(node);\r | |
1526 | return node;\r | |
1527 | }\r | |
1528 | \r | |
1529 | static Node*\r | |
1530 | node_new_empty(void)\r | |
1531 | {\r | |
1532 | return node_new_str(NULL, NULL);\r | |
1533 | }\r | |
1534 | \r | |
1535 | static Node*\r | |
1536 | node_new_str_raw_char(UChar c)\r | |
1537 | {\r | |
1538 | UChar p[1];\r | |
1539 | \r | |
1540 | p[0] = c;\r | |
1541 | return node_new_str_raw(p, p + 1);\r | |
1542 | }\r | |
1543 | \r | |
1544 | static Node*\r | |
1545 | str_node_split_last_char(StrNode* sn, OnigEncoding enc)\r | |
1546 | {\r | |
1547 | const UChar *p;\r | |
1548 | Node* n = NULL_NODE;\r | |
1549 | \r | |
1550 | if (sn->end > sn->s) {\r | |
1551 | p = onigenc_get_prev_char_head(enc, sn->s, sn->end);\r | |
1552 | if (p && p > sn->s) { /* can be splitted. */\r | |
1553 | n = node_new_str(p, sn->end);\r | |
1554 | if ((sn->flag & NSTR_RAW) != 0)\r | |
1555 | NSTRING_SET_RAW(n);\r | |
1556 | sn->end = (UChar* )p;\r | |
1557 | }\r | |
1558 | }\r | |
1559 | return n;\r | |
1560 | }\r | |
1561 | \r | |
1562 | static int\r | |
1563 | str_node_can_be_split(StrNode* sn, OnigEncoding enc)\r | |
1564 | {\r | |
1565 | if (sn->end > sn->s) {\r | |
1566 | return ((enclen(enc, sn->s) < sn->end - sn->s) ? 1 : 0);\r | |
1567 | }\r | |
1568 | return 0;\r | |
1569 | }\r | |
1570 | \r | |
1571 | #ifdef USE_PAD_TO_SHORT_BYTE_CHAR\r | |
1572 | static int\r | |
1573 | node_str_head_pad(StrNode* sn, int num, UChar val)\r | |
1574 | {\r | |
1575 | UChar buf[NODE_STR_BUF_SIZE];\r | |
1576 | int i, len;\r | |
1577 | \r | |
1578 | len = sn->end - sn->s;\r | |
1579 | onig_strcpy(buf, sn->s, sn->end);\r | |
1580 | onig_strcpy(&(sn->s[num]), buf, buf + len);\r | |
1581 | sn->end += num;\r | |
1582 | \r | |
1583 | for (i = 0; i < num; i++) {\r | |
1584 | sn->s[i] = val;\r | |
1585 | }\r | |
1586 | }\r | |
1587 | #endif\r | |
1588 | \r | |
1589 | extern int\r | |
1590 | onig_scan_unsigned_number(UChar** src, const UChar* end, OnigEncoding enc)\r | |
1591 | {\r | |
1592 | unsigned int num, val;\r | |
1593 | OnigCodePoint c;\r | |
1594 | UChar* p = *src;\r | |
1595 | PFETCH_READY;\r | |
1596 | \r | |
1597 | num = 0;\r | |
1598 | while (!PEND) {\r | |
1599 | PFETCH(c);\r | |
1600 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {\r | |
1601 | val = (unsigned int )DIGITVAL(c);\r | |
1602 | if ((INT_MAX_LIMIT - val) / 10UL < num)\r | |
1603 | return -1; /* overflow */\r | |
1604 | \r | |
1605 | num = num * 10 + val;\r | |
1606 | }\r | |
1607 | else {\r | |
1608 | PUNFETCH;\r | |
1609 | break;\r | |
1610 | }\r | |
1611 | }\r | |
1612 | *src = p;\r | |
1613 | return num;\r | |
1614 | }\r | |
1615 | \r | |
1616 | static int\r | |
1617 | scan_unsigned_hexadecimal_number(UChar** src, UChar* end, int maxlen,\r | |
1618 | OnigEncoding enc)\r | |
1619 | {\r | |
1620 | OnigCodePoint c;\r | |
1621 | unsigned int num, val;\r | |
1622 | UChar* p = *src;\r | |
1623 | PFETCH_READY;\r | |
1624 | \r | |
1625 | num = 0;\r | |
1626 | while (!PEND && maxlen-- != 0) {\r | |
1627 | PFETCH(c);\r | |
1628 | if (ONIGENC_IS_CODE_XDIGIT(enc, c)) {\r | |
1629 | val = (unsigned int )XDIGITVAL(enc,c);\r | |
1630 | if ((INT_MAX_LIMIT - val) / 16UL < num)\r | |
1631 | return -1; /* overflow */\r | |
1632 | \r | |
1633 | num = (num << 4) + XDIGITVAL(enc,c);\r | |
1634 | }\r | |
1635 | else {\r | |
1636 | PUNFETCH;\r | |
1637 | break;\r | |
1638 | }\r | |
1639 | }\r | |
1640 | *src = p;\r | |
1641 | return num;\r | |
1642 | }\r | |
1643 | \r | |
1644 | static int\r | |
1645 | scan_unsigned_octal_number(UChar** src, UChar* end, int maxlen,\r | |
1646 | OnigEncoding enc)\r | |
1647 | {\r | |
1648 | OnigCodePoint c;\r | |
1649 | unsigned int num, val;\r | |
1650 | UChar* p = *src;\r | |
1651 | PFETCH_READY;\r | |
1652 | \r | |
1653 | num = 0;\r | |
1654 | while (!PEND && maxlen-- != 0) {\r | |
1655 | PFETCH(c);\r | |
1656 | if (ONIGENC_IS_CODE_DIGIT(enc, c) && c < '8') {\r | |
1657 | val = ODIGITVAL(c);\r | |
1658 | if ((INT_MAX_LIMIT - val) / 8UL < num)\r | |
1659 | return -1; /* overflow */\r | |
1660 | \r | |
1661 | num = (num << 3) + val;\r | |
1662 | }\r | |
1663 | else {\r | |
1664 | PUNFETCH;\r | |
1665 | break;\r | |
1666 | }\r | |
1667 | }\r | |
1668 | *src = p;\r | |
1669 | return num;\r | |
1670 | }\r | |
1671 | \r | |
1672 | \r | |
1673 | #define BBUF_WRITE_CODE_POINT(bbuf,pos,code) \\r | |
1674 | BBUF_WRITE(bbuf, pos, &(code), SIZE_CODE_POINT)\r | |
1675 | \r | |
1676 | /* data format:\r | |
1677 | [n][from-1][to-1][from-2][to-2] ... [from-n][to-n]\r | |
1678 | (all data size is OnigCodePoint)\r | |
1679 | */\r | |
1680 | static int\r | |
1681 | new_code_range(BBuf** pbuf)\r | |
1682 | {\r | |
1683 | #define INIT_MULTI_BYTE_RANGE_SIZE (SIZE_CODE_POINT * 5)\r | |
1684 | int r;\r | |
1685 | OnigCodePoint n;\r | |
1686 | BBuf* bbuf;\r | |
1687 | \r | |
1688 | bbuf = *pbuf = (BBuf* )xmalloc(sizeof(BBuf));\r | |
1689 | CHECK_NULL_RETURN_MEMERR(*pbuf);\r | |
1690 | r = BBUF_INIT(*pbuf, INIT_MULTI_BYTE_RANGE_SIZE);\r | |
1691 | if (r) return r;\r | |
1692 | \r | |
1693 | n = 0;\r | |
1694 | BBUF_WRITE_CODE_POINT(bbuf, 0, n);\r | |
1695 | return 0;\r | |
1696 | }\r | |
1697 | \r | |
1698 | static int\r | |
1699 | add_code_range_to_buf(BBuf** pbuf, OnigCodePoint from, OnigCodePoint to)\r | |
1700 | {\r | |
1701 | int r, inc_n, pos;\r | |
1702 | int low, high, bound, x;\r | |
1703 | OnigCodePoint n, *data;\r | |
1704 | BBuf* bbuf;\r | |
1705 | \r | |
1706 | if (from > to) {\r | |
1707 | n = from; from = to; to = n;\r | |
1708 | }\r | |
1709 | \r | |
1710 | if (IS_NULL(*pbuf)) {\r | |
1711 | r = new_code_range(pbuf);\r | |
1712 | if (r) return r;\r | |
1713 | bbuf = *pbuf;\r | |
1714 | n = 0;\r | |
1715 | }\r | |
1716 | else {\r | |
1717 | bbuf = *pbuf;\r | |
1718 | GET_CODE_POINT(n, bbuf->p);\r | |
1719 | }\r | |
1720 | data = (OnigCodePoint* )(bbuf->p);\r | |
1721 | data++;\r | |
1722 | \r | |
1723 | for (low = 0, bound = n; low < bound; ) {\r | |
1724 | x = (low + bound) >> 1;\r | |
1725 | if (from > data[x*2 + 1])\r | |
1726 | low = x + 1;\r | |
1727 | else\r | |
1728 | bound = x;\r | |
1729 | }\r | |
1730 | \r | |
1731 | for (high = low, bound = n; high < bound; ) {\r | |
1732 | x = (high + bound) >> 1;\r | |
1733 | if (to >= data[x*2] - 1)\r | |
1734 | high = x + 1;\r | |
1735 | else\r | |
1736 | bound = x;\r | |
1737 | }\r | |
1738 | \r | |
1739 | inc_n = low + 1 - high;\r | |
1740 | if (n + inc_n > ONIG_MAX_MULTI_BYTE_RANGES_NUM)\r | |
1741 | return ONIGERR_TOO_MANY_MULTI_BYTE_RANGES;\r | |
1742 | \r | |
1743 | if (inc_n != 1) {\r | |
1744 | if (from > data[low*2])\r | |
1745 | from = data[low*2];\r | |
1746 | if (to < data[(high - 1)*2 + 1])\r | |
1747 | to = data[(high - 1)*2 + 1];\r | |
1748 | }\r | |
1749 | \r | |
1750 | if (inc_n != 0 && (OnigCodePoint )high < n) {\r | |
1751 | int from_pos = SIZE_CODE_POINT * (1 + high * 2);\r | |
1752 | int to_pos = SIZE_CODE_POINT * (1 + (low + 1) * 2);\r | |
1753 | int size = (n - high) * 2 * SIZE_CODE_POINT;\r | |
1754 | \r | |
1755 | if (inc_n > 0) {\r | |
1756 | BBUF_MOVE_RIGHT(bbuf, from_pos, to_pos, size);\r | |
1757 | }\r | |
1758 | else {\r | |
1759 | BBUF_MOVE_LEFT_REDUCE(bbuf, from_pos, to_pos);\r | |
1760 | }\r | |
1761 | }\r | |
1762 | \r | |
1763 | pos = SIZE_CODE_POINT * (1 + low * 2);\r | |
1764 | BBUF_ENSURE_SIZE(bbuf, pos + SIZE_CODE_POINT * 2);\r | |
1765 | BBUF_WRITE_CODE_POINT(bbuf, pos, from);\r | |
1766 | BBUF_WRITE_CODE_POINT(bbuf, pos + SIZE_CODE_POINT, to);\r | |
1767 | n += inc_n;\r | |
1768 | BBUF_WRITE_CODE_POINT(bbuf, 0, n);\r | |
1769 | \r | |
1770 | return 0;\r | |
1771 | }\r | |
1772 | \r | |
1773 | static int\r | |
1774 | add_code_range(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePoint to)\r | |
1775 | {\r | |
1776 | if (from > to) {\r | |
1777 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))\r | |
1778 | return 0;\r | |
1779 | else\r | |
1780 | return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;\r | |
1781 | }\r | |
1782 | \r | |
1783 | return add_code_range_to_buf(pbuf, from, to);\r | |
1784 | }\r | |
1785 | \r | |
1786 | static int\r | |
1787 | not_code_range_buf(OnigEncoding enc, BBuf* bbuf, BBuf** pbuf)\r | |
1788 | {\r | |
1789 | int r, i, n;\r | |
1790 | OnigCodePoint pre, from, *data, to = 0;\r | |
1791 | \r | |
1792 | *pbuf = (BBuf* )NULL;\r | |
1793 | if (IS_NULL(bbuf)) {\r | |
1794 | set_all:\r | |
1795 | return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);\r | |
1796 | }\r | |
1797 | \r | |
1798 | data = (OnigCodePoint* )(bbuf->p);\r | |
1799 | GET_CODE_POINT(n, data);\r | |
1800 | data++;\r | |
1801 | if (n <= 0) goto set_all;\r | |
1802 | \r | |
1803 | r = 0;\r | |
1804 | pre = MBCODE_START_POS(enc);\r | |
1805 | for (i = 0; i < n; i++) {\r | |
1806 | from = data[i*2];\r | |
1807 | to = data[i*2+1];\r | |
1808 | if (pre <= from - 1) {\r | |
1809 | r = add_code_range_to_buf(pbuf, pre, from - 1);\r | |
1810 | if (r != 0) return r;\r | |
1811 | }\r | |
1812 | if (to == ~((OnigCodePoint )0)) break;\r | |
1813 | pre = to + 1;\r | |
1814 | }\r | |
1815 | if (to < ~((OnigCodePoint )0)) {\r | |
1816 | r = add_code_range_to_buf(pbuf, to + 1, ~((OnigCodePoint )0));\r | |
1817 | }\r | |
1818 | return r;\r | |
1819 | }\r | |
1820 | \r | |
1821 | #define SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2) do {\\r | |
1822 | BBuf *tbuf; \\r | |
1823 | int tnot; \\r | |
1824 | tnot = not1; not1 = not2; not2 = tnot; \\r | |
1825 | tbuf = bbuf1; bbuf1 = bbuf2; bbuf2 = tbuf; \\r | |
1826 | } while (0)\r | |
1827 | \r | |
1828 | static int\r | |
1829 | or_code_range_buf(OnigEncoding enc, BBuf* bbuf1, int not1,\r | |
1830 | BBuf* bbuf2, int not2, BBuf** pbuf)\r | |
1831 | {\r | |
1832 | int r;\r | |
1833 | OnigCodePoint i, n1, *data1;\r | |
1834 | OnigCodePoint from, to;\r | |
1835 | \r | |
1836 | *pbuf = (BBuf* )NULL;\r | |
1837 | if (IS_NULL(bbuf1) && IS_NULL(bbuf2)) {\r | |
1838 | if (not1 != 0 || not2 != 0)\r | |
1839 | return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);\r | |
1840 | return 0;\r | |
1841 | }\r | |
1842 | \r | |
1843 | r = 0;\r | |
1844 | if (IS_NULL(bbuf2))\r | |
1845 | SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);\r | |
1846 | \r | |
1847 | if (IS_NULL(bbuf1)) {\r | |
1848 | if (not1 != 0) {\r | |
1849 | return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);\r | |
1850 | }\r | |
1851 | else {\r | |
1852 | if (not2 == 0) {\r | |
1853 | return bbuf_clone(pbuf, bbuf2);\r | |
1854 | }\r | |
1855 | else {\r | |
1856 | return not_code_range_buf(enc, bbuf2, pbuf);\r | |
1857 | }\r | |
1858 | }\r | |
1859 | }\r | |
1860 | \r | |
1861 | if (not1 != 0)\r | |
1862 | SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);\r | |
1863 | \r | |
1864 | data1 = (OnigCodePoint* )(bbuf1->p);\r | |
1865 | GET_CODE_POINT(n1, data1);\r | |
1866 | data1++;\r | |
1867 | \r | |
1868 | if (not2 == 0 && not1 == 0) { /* 1 OR 2 */\r | |
1869 | r = bbuf_clone(pbuf, bbuf2);\r | |
1870 | }\r | |
1871 | else if (not1 == 0) { /* 1 OR (not 2) */\r | |
1872 | r = not_code_range_buf(enc, bbuf2, pbuf);\r | |
1873 | }\r | |
1874 | if (r != 0) return r;\r | |
1875 | \r | |
1876 | for (i = 0; i < n1; i++) {\r | |
1877 | from = data1[i*2];\r | |
1878 | to = data1[i*2+1];\r | |
1879 | r = add_code_range_to_buf(pbuf, from, to);\r | |
1880 | if (r != 0) return r;\r | |
1881 | }\r | |
1882 | return 0;\r | |
1883 | }\r | |
1884 | \r | |
1885 | static int\r | |
1886 | and_code_range1(BBuf** pbuf, OnigCodePoint from1, OnigCodePoint to1,\r | |
1887 | OnigCodePoint* data, int n)\r | |
1888 | {\r | |
1889 | int i, r;\r | |
1890 | OnigCodePoint from2, to2;\r | |
1891 | \r | |
1892 | for (i = 0; i < n; i++) {\r | |
1893 | from2 = data[i*2];\r | |
1894 | to2 = data[i*2+1];\r | |
1895 | if (from2 < from1) {\r | |
1896 | if (to2 < from1) continue;\r | |
1897 | else {\r | |
1898 | from1 = to2 + 1;\r | |
1899 | }\r | |
1900 | }\r | |
1901 | else if (from2 <= to1) {\r | |
1902 | if (to2 < to1) {\r | |
1903 | if (from1 <= from2 - 1) {\r | |
1904 | r = add_code_range_to_buf(pbuf, from1, from2-1);\r | |
1905 | if (r != 0) return r;\r | |
1906 | }\r | |
1907 | from1 = to2 + 1;\r | |
1908 | }\r | |
1909 | else {\r | |
1910 | to1 = from2 - 1;\r | |
1911 | }\r | |
1912 | }\r | |
1913 | else {\r | |
1914 | from1 = from2;\r | |
1915 | }\r | |
1916 | if (from1 > to1) break;\r | |
1917 | }\r | |
1918 | if (from1 <= to1) {\r | |
1919 | r = add_code_range_to_buf(pbuf, from1, to1);\r | |
1920 | if (r != 0) return r;\r | |
1921 | }\r | |
1922 | return 0;\r | |
1923 | }\r | |
1924 | \r | |
1925 | static int\r | |
1926 | and_code_range_buf(BBuf* bbuf1, int not1, BBuf* bbuf2, int not2, BBuf** pbuf)\r | |
1927 | {\r | |
1928 | int r;\r | |
1929 | OnigCodePoint i, j, n1, n2, *data1, *data2;\r | |
1930 | OnigCodePoint from, to, from1, to1, from2, to2;\r | |
1931 | \r | |
1932 | *pbuf = (BBuf* )NULL;\r | |
1933 | if (IS_NULL(bbuf1)) {\r | |
1934 | if (not1 != 0 && IS_NOT_NULL(bbuf2)) /* not1 != 0 -> not2 == 0 */\r | |
1935 | return bbuf_clone(pbuf, bbuf2);\r | |
1936 | return 0;\r | |
1937 | }\r | |
1938 | else if (IS_NULL(bbuf2)) {\r | |
1939 | if (not2 != 0)\r | |
1940 | return bbuf_clone(pbuf, bbuf1);\r | |
1941 | return 0;\r | |
1942 | }\r | |
1943 | \r | |
1944 | if (not1 != 0)\r | |
1945 | SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);\r | |
1946 | \r | |
1947 | data1 = (OnigCodePoint* )(bbuf1->p);\r | |
1948 | data2 = (OnigCodePoint* )(bbuf2->p);\r | |
1949 | GET_CODE_POINT(n1, data1);\r | |
1950 | GET_CODE_POINT(n2, data2);\r | |
1951 | data1++;\r | |
1952 | data2++;\r | |
1953 | \r | |
1954 | if (not2 == 0 && not1 == 0) { /* 1 AND 2 */\r | |
1955 | for (i = 0; i < n1; i++) {\r | |
1956 | from1 = data1[i*2];\r | |
1957 | to1 = data1[i*2+1];\r | |
1958 | for (j = 0; j < n2; j++) {\r | |
1959 | from2 = data2[j*2];\r | |
1960 | to2 = data2[j*2+1];\r | |
1961 | if (from2 > to1) break;\r | |
1962 | if (to2 < from1) continue;\r | |
1963 | from = MAX(from1, from2);\r | |
1964 | to = MIN(to1, to2);\r | |
1965 | r = add_code_range_to_buf(pbuf, from, to);\r | |
1966 | if (r != 0) return r;\r | |
1967 | }\r | |
1968 | }\r | |
1969 | }\r | |
1970 | else if (not1 == 0) { /* 1 AND (not 2) */\r | |
1971 | for (i = 0; i < n1; i++) {\r | |
1972 | from1 = data1[i*2];\r | |
1973 | to1 = data1[i*2+1];\r | |
1974 | r = and_code_range1(pbuf, from1, to1, data2, n2);\r | |
1975 | if (r != 0) return r;\r | |
1976 | }\r | |
1977 | }\r | |
1978 | \r | |
1979 | return 0;\r | |
1980 | }\r | |
1981 | \r | |
1982 | static int\r | |
1983 | and_cclass(CClassNode* dest, CClassNode* cc, OnigEncoding enc)\r | |
1984 | {\r | |
1985 | int r, not1, not2;\r | |
1986 | BBuf *buf1, *buf2, *pbuf;\r | |
1987 | BitSetRef bsr1, bsr2;\r | |
1988 | BitSet bs1, bs2;\r | |
1989 | \r | |
1990 | not1 = IS_NCCLASS_NOT(dest);\r | |
1991 | bsr1 = dest->bs;\r | |
1992 | buf1 = dest->mbuf;\r | |
1993 | not2 = IS_NCCLASS_NOT(cc);\r | |
1994 | bsr2 = cc->bs;\r | |
1995 | buf2 = cc->mbuf;\r | |
1996 | \r | |
1997 | if (not1 != 0) {\r | |
1998 | bitset_invert_to(bsr1, bs1);\r | |
1999 | bsr1 = bs1;\r | |
2000 | }\r | |
2001 | if (not2 != 0) {\r | |
2002 | bitset_invert_to(bsr2, bs2);\r | |
2003 | bsr2 = bs2;\r | |
2004 | }\r | |
2005 | bitset_and(bsr1, bsr2);\r | |
2006 | if (bsr1 != dest->bs) {\r | |
2007 | bitset_copy(dest->bs, bsr1);\r | |
2008 | bsr1 = dest->bs;\r | |
2009 | }\r | |
2010 | if (not1 != 0) {\r | |
2011 | bitset_invert(dest->bs);\r | |
2012 | }\r | |
2013 | \r | |
2014 | if (! ONIGENC_IS_SINGLEBYTE(enc)) {\r | |
2015 | if (not1 != 0 && not2 != 0) {\r | |
2016 | r = or_code_range_buf(enc, buf1, 0, buf2, 0, &pbuf);\r | |
2017 | }\r | |
2018 | else {\r | |
2019 | r = and_code_range_buf(buf1, not1, buf2, not2, &pbuf);\r | |
2020 | if (r == 0 && not1 != 0) {\r | |
2021 | BBuf *tbuf;\r | |
2022 | r = not_code_range_buf(enc, pbuf, &tbuf);\r | |
2023 | if (r != 0) {\r | |
2024 | bbuf_free(pbuf);\r | |
2025 | return r;\r | |
2026 | }\r | |
2027 | bbuf_free(pbuf);\r | |
2028 | pbuf = tbuf;\r | |
2029 | }\r | |
2030 | }\r | |
2031 | if (r != 0) return r;\r | |
2032 | \r | |
2033 | dest->mbuf = pbuf;\r | |
2034 | bbuf_free(buf1);\r | |
2035 | return r;\r | |
2036 | }\r | |
2037 | return 0;\r | |
2038 | }\r | |
2039 | \r | |
2040 | static int\r | |
2041 | or_cclass(CClassNode* dest, CClassNode* cc, OnigEncoding enc)\r | |
2042 | {\r | |
2043 | int r, not1, not2;\r | |
2044 | BBuf *buf1, *buf2, *pbuf;\r | |
2045 | BitSetRef bsr1, bsr2;\r | |
2046 | BitSet bs1, bs2;\r | |
2047 | \r | |
2048 | not1 = IS_NCCLASS_NOT(dest);\r | |
2049 | bsr1 = dest->bs;\r | |
2050 | buf1 = dest->mbuf;\r | |
2051 | not2 = IS_NCCLASS_NOT(cc);\r | |
2052 | bsr2 = cc->bs;\r | |
2053 | buf2 = cc->mbuf;\r | |
2054 | \r | |
2055 | if (not1 != 0) {\r | |
2056 | bitset_invert_to(bsr1, bs1);\r | |
2057 | bsr1 = bs1;\r | |
2058 | }\r | |
2059 | if (not2 != 0) {\r | |
2060 | bitset_invert_to(bsr2, bs2);\r | |
2061 | bsr2 = bs2;\r | |
2062 | }\r | |
2063 | bitset_or(bsr1, bsr2);\r | |
2064 | if (bsr1 != dest->bs) {\r | |
2065 | bitset_copy(dest->bs, bsr1);\r | |
2066 | bsr1 = dest->bs;\r | |
2067 | }\r | |
2068 | if (not1 != 0) {\r | |
2069 | bitset_invert(dest->bs);\r | |
2070 | }\r | |
2071 | \r | |
2072 | if (! ONIGENC_IS_SINGLEBYTE(enc)) {\r | |
2073 | if (not1 != 0 && not2 != 0) {\r | |
2074 | r = and_code_range_buf(buf1, 0, buf2, 0, &pbuf);\r | |
2075 | }\r | |
2076 | else {\r | |
2077 | r = or_code_range_buf(enc, buf1, not1, buf2, not2, &pbuf);\r | |
2078 | if (r == 0 && not1 != 0) {\r | |
2079 | BBuf *tbuf;\r | |
2080 | r = not_code_range_buf(enc, pbuf, &tbuf);\r | |
2081 | if (r != 0) {\r | |
2082 | bbuf_free(pbuf);\r | |
2083 | return r;\r | |
2084 | }\r | |
2085 | bbuf_free(pbuf);\r | |
2086 | pbuf = tbuf;\r | |
2087 | }\r | |
2088 | }\r | |
2089 | if (r != 0) return r;\r | |
2090 | \r | |
2091 | dest->mbuf = pbuf;\r | |
2092 | bbuf_free(buf1);\r | |
2093 | return r;\r | |
2094 | }\r | |
2095 | else\r | |
2096 | return 0;\r | |
2097 | }\r | |
2098 | \r | |
2099 | static int\r | |
2100 | conv_backslash_value(int c, ScanEnv* env)\r | |
2101 | {\r | |
2102 | if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_CONTROL_CHARS)) {\r | |
2103 | switch (c) {\r | |
2104 | case 'n': return '\n';\r | |
2105 | case 't': return '\t';\r | |
2106 | case 'r': return '\r';\r | |
2107 | case 'f': return '\f';\r | |
2108 | case 'a': return '\007';\r | |
2109 | case 'b': return '\010';\r | |
2110 | case 'e': return '\033';\r | |
2111 | case 'v':\r | |
2112 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_V_VTAB))\r | |
2113 | return '\v';\r | |
2114 | break;\r | |
2115 | \r | |
2116 | default:\r | |
2117 | break;\r | |
2118 | }\r | |
2119 | }\r | |
2120 | return c;\r | |
2121 | }\r | |
2122 | \r | |
2123 | static int\r | |
2124 | is_invalid_quantifier_target(Node* node)\r | |
2125 | {\r | |
2126 | switch (NTYPE(node)) {\r | |
2127 | case NT_ANCHOR:\r | |
2128 | return 1;\r | |
2129 | break;\r | |
2130 | \r | |
2131 | case NT_ENCLOSE:\r | |
2132 | /* allow enclosed elements */\r | |
2133 | /* return is_invalid_quantifier_target(NENCLOSE(node)->target); */\r | |
2134 | break;\r | |
2135 | \r | |
2136 | case NT_LIST:\r | |
2137 | do {\r | |
2138 | if (! is_invalid_quantifier_target(NCAR(node))) return 0;\r | |
2139 | } while (IS_NOT_NULL(node = NCDR(node)));\r | |
2140 | return 0;\r | |
2141 | break;\r | |
2142 | \r | |
2143 | case NT_ALT:\r | |
2144 | do {\r | |
2145 | if (is_invalid_quantifier_target(NCAR(node))) return 1;\r | |
2146 | } while (IS_NOT_NULL(node = NCDR(node)));\r | |
2147 | break;\r | |
2148 | \r | |
2149 | default:\r | |
2150 | break;\r | |
2151 | }\r | |
2152 | return 0;\r | |
2153 | }\r | |
2154 | \r | |
2155 | /* ?:0, *:1, +:2, ??:3, *?:4, +?:5 */\r | |
2156 | static int\r | |
2157 | popular_quantifier_num(QtfrNode* q)\r | |
2158 | {\r | |
2159 | if (q->greedy) {\r | |
2160 | if (q->lower == 0) {\r | |
2161 | if (q->upper == 1) return 0;\r | |
2162 | else if (IS_REPEAT_INFINITE(q->upper)) return 1;\r | |
2163 | }\r | |
2164 | else if (q->lower == 1) {\r | |
2165 | if (IS_REPEAT_INFINITE(q->upper)) return 2;\r | |
2166 | }\r | |
2167 | }\r | |
2168 | else {\r | |
2169 | if (q->lower == 0) {\r | |
2170 | if (q->upper == 1) return 3;\r | |
2171 | else if (IS_REPEAT_INFINITE(q->upper)) return 4;\r | |
2172 | }\r | |
2173 | else if (q->lower == 1) {\r | |
2174 | if (IS_REPEAT_INFINITE(q->upper)) return 5;\r | |
2175 | }\r | |
2176 | }\r | |
2177 | return -1;\r | |
2178 | }\r | |
2179 | \r | |
2180 | \r | |
2181 | enum ReduceType {\r | |
2182 | RQ_ASIS = 0, /* as is */\r | |
2183 | RQ_DEL = 1, /* delete parent */\r | |
2184 | RQ_A, /* to '*' */\r | |
2185 | RQ_AQ, /* to '*?' */\r | |
2186 | RQ_QQ, /* to '??' */\r | |
2187 | RQ_P_QQ, /* to '+)??' */\r | |
2188 | RQ_PQ_Q /* to '+?)?' */\r | |
2189 | };\r | |
2190 | \r | |
2191 | static enum ReduceType ReduceTypeTable[6][6] = {\r | |
2192 | {RQ_DEL, RQ_A, RQ_A, RQ_QQ, RQ_AQ, RQ_ASIS}, /* '?' */\r | |
2193 | {RQ_DEL, RQ_DEL, RQ_DEL, RQ_P_QQ, RQ_P_QQ, RQ_DEL}, /* '*' */\r | |
2194 | {RQ_A, RQ_A, RQ_DEL, RQ_ASIS, RQ_P_QQ, RQ_DEL}, /* '+' */\r | |
2195 | {RQ_DEL, RQ_AQ, RQ_AQ, RQ_DEL, RQ_AQ, RQ_AQ}, /* '??' */\r | |
2196 | {RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL}, /* '*?' */\r | |
2197 | {RQ_ASIS, RQ_PQ_Q, RQ_DEL, RQ_AQ, RQ_AQ, RQ_DEL} /* '+?' */\r | |
2198 | };\r | |
2199 | \r | |
2200 | extern void\r | |
2201 | onig_reduce_nested_quantifier(Node* pnode, Node* cnode)\r | |
2202 | {\r | |
2203 | int pnum, cnum;\r | |
2204 | QtfrNode *p, *c;\r | |
2205 | \r | |
2206 | p = NQTFR(pnode);\r | |
2207 | c = NQTFR(cnode);\r | |
2208 | pnum = popular_quantifier_num(p);\r | |
2209 | cnum = popular_quantifier_num(c);\r | |
2210 | if (pnum < 0 || cnum < 0) return ;\r | |
2211 | \r | |
2212 | switch(ReduceTypeTable[cnum][pnum]) {\r | |
2213 | case RQ_DEL:\r | |
2214 | *pnode = *cnode;\r | |
2215 | break;\r | |
2216 | case RQ_A:\r | |
2217 | p->target = c->target;\r | |
2218 | p->lower = 0; p->upper = REPEAT_INFINITE; p->greedy = 1;\r | |
2219 | break;\r | |
2220 | case RQ_AQ:\r | |
2221 | p->target = c->target;\r | |
2222 | p->lower = 0; p->upper = REPEAT_INFINITE; p->greedy = 0;\r | |
2223 | break;\r | |
2224 | case RQ_QQ:\r | |
2225 | p->target = c->target;\r | |
2226 | p->lower = 0; p->upper = 1; p->greedy = 0;\r | |
2227 | break;\r | |
2228 | case RQ_P_QQ:\r | |
2229 | p->target = cnode;\r | |
2230 | p->lower = 0; p->upper = 1; p->greedy = 0;\r | |
2231 | c->lower = 1; c->upper = REPEAT_INFINITE; c->greedy = 1;\r | |
2232 | return ;\r | |
2233 | break;\r | |
2234 | case RQ_PQ_Q:\r | |
2235 | p->target = cnode;\r | |
2236 | p->lower = 0; p->upper = 1; p->greedy = 1;\r | |
2237 | c->lower = 1; c->upper = REPEAT_INFINITE; c->greedy = 0;\r | |
2238 | return ;\r | |
2239 | break;\r | |
2240 | case RQ_ASIS:\r | |
2241 | p->target = cnode;\r | |
2242 | return ;\r | |
2243 | break;\r | |
2244 | }\r | |
2245 | \r | |
2246 | c->target = NULL_NODE;\r | |
2247 | onig_node_free(cnode);\r | |
2248 | }\r | |
2249 | \r | |
2250 | \r | |
2251 | enum TokenSyms {\r | |
2252 | TK_EOT = 0, /* end of token */\r | |
2253 | TK_RAW_BYTE = 1,\r | |
2254 | TK_CHAR,\r | |
2255 | TK_STRING,\r | |
2256 | TK_CODE_POINT,\r | |
2257 | TK_ANYCHAR,\r | |
2258 | TK_CHAR_TYPE,\r | |
2259 | TK_BACKREF,\r | |
2260 | TK_CALL,\r | |
2261 | TK_ANCHOR,\r | |
2262 | TK_OP_REPEAT,\r | |
2263 | TK_INTERVAL,\r | |
2264 | TK_ANYCHAR_ANYTIME, /* SQL '%' == .* */\r | |
2265 | TK_ALT,\r | |
2266 | TK_SUBEXP_OPEN,\r | |
2267 | TK_SUBEXP_CLOSE,\r | |
2268 | TK_CC_OPEN,\r | |
2269 | TK_QUOTE_OPEN,\r | |
2270 | TK_CHAR_PROPERTY, /* \p{...}, \P{...} */\r | |
2271 | /* in cc */\r | |
2272 | TK_CC_CLOSE,\r | |
2273 | TK_CC_RANGE,\r | |
2274 | TK_POSIX_BRACKET_OPEN,\r | |
2275 | TK_CC_AND, /* && */\r | |
2276 | TK_CC_CC_OPEN /* [ */\r | |
2277 | };\r | |
2278 | \r | |
2279 | typedef struct {\r | |
2280 | enum TokenSyms type;\r | |
2281 | int escaped;\r | |
2282 | int base; /* is number: 8, 16 (used in [....]) */\r | |
2283 | UChar* backp;\r | |
2284 | union {\r | |
2285 | UChar* s;\r | |
2286 | int c;\r | |
2287 | OnigCodePoint code;\r | |
2288 | int anchor;\r | |
2289 | int subtype;\r | |
2290 | struct {\r | |
2291 | int lower;\r | |
2292 | int upper;\r | |
2293 | int greedy;\r | |
2294 | int possessive;\r | |
2295 | } repeat;\r | |
2296 | struct {\r | |
2297 | int num;\r | |
2298 | int ref1;\r | |
2299 | int* refs;\r | |
2300 | int by_name;\r | |
2301 | #ifdef USE_BACKREF_WITH_LEVEL\r | |
2302 | int exist_level;\r | |
2303 | int level; /* \k<name+n> */\r | |
2304 | #endif\r | |
2305 | } backref;\r | |
2306 | struct {\r | |
2307 | UChar* name;\r | |
2308 | UChar* name_end;\r | |
2309 | int gnum;\r | |
2310 | } call;\r | |
2311 | struct {\r | |
2312 | int ctype;\r | |
2313 | int not;\r | |
2314 | } prop;\r | |
2315 | } u;\r | |
2316 | } OnigToken;\r | |
2317 | \r | |
2318 | \r | |
2319 | static int\r | |
2320 | fetch_range_quantifier(UChar** src, UChar* end, OnigToken* tok, ScanEnv* env)\r | |
2321 | {\r | |
2322 | int low, up, syn_allow, non_low = 0;\r | |
2323 | int r = 0;\r | |
2324 | OnigCodePoint c;\r | |
2325 | OnigEncoding enc = env->enc;\r | |
2326 | UChar* p = *src;\r | |
2327 | PFETCH_READY;\r | |
2328 | \r | |
2329 | syn_allow = IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_INVALID_INTERVAL);\r | |
2330 | \r | |
2331 | if (PEND) {\r | |
2332 | if (syn_allow)\r | |
2333 | return 1; /* "....{" : OK! */\r | |
2334 | else\r | |
2335 | return ONIGERR_END_PATTERN_AT_LEFT_BRACE; /* "....{" syntax error */\r | |
2336 | }\r | |
2337 | \r | |
2338 | if (! syn_allow) {\r | |
2339 | c = PPEEK;\r | |
2340 | if (c == ')' || c == '(' || c == '|') {\r | |
2341 | return ONIGERR_END_PATTERN_AT_LEFT_BRACE;\r | |
2342 | }\r | |
2343 | }\r | |
2344 | \r | |
2345 | low = onig_scan_unsigned_number(&p, end, env->enc);\r | |
2346 | if (low < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;\r | |
2347 | if (low > ONIG_MAX_REPEAT_NUM)\r | |
2348 | return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;\r | |
2349 | \r | |
2350 | if (p == *src) { /* can't read low */\r | |
2351 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV)) {\r | |
2352 | /* allow {,n} as {0,n} */\r | |
2353 | low = 0;\r | |
2354 | non_low = 1;\r | |
2355 | }\r | |
2356 | else\r | |
2357 | goto invalid;\r | |
2358 | }\r | |
2359 | \r | |
2360 | if (PEND) goto invalid;\r | |
2361 | PFETCH(c);\r | |
2362 | if (c == ',') {\r | |
2363 | UChar* prev = p;\r | |
2364 | up = onig_scan_unsigned_number(&p, end, env->enc);\r | |
2365 | if (up < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;\r | |
2366 | if (up > ONIG_MAX_REPEAT_NUM)\r | |
2367 | return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;\r | |
2368 | \r | |
2369 | if (p == prev) {\r | |
2370 | if (non_low != 0)\r | |
2371 | goto invalid;\r | |
2372 | up = REPEAT_INFINITE; /* {n,} : {n,infinite} */\r | |
2373 | }\r | |
2374 | }\r | |
2375 | else {\r | |
2376 | if (non_low != 0)\r | |
2377 | goto invalid;\r | |
2378 | \r | |
2379 | PUNFETCH;\r | |
2380 | up = low; /* {n} : exact n times */\r | |
2381 | r = 2; /* fixed */\r | |
2382 | }\r | |
2383 | \r | |
2384 | if (PEND) goto invalid;\r | |
2385 | PFETCH(c);\r | |
2386 | if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) {\r | |
2387 | if (c != MC_ESC(env->syntax)) goto invalid;\r | |
2388 | PFETCH(c);\r | |
2389 | }\r | |
2390 | if (c != '}') goto invalid;\r | |
2391 | \r | |
2392 | if (!IS_REPEAT_INFINITE(up) && low > up) {\r | |
2393 | return ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE;\r | |
2394 | }\r | |
2395 | \r | |
2396 | tok->type = TK_INTERVAL;\r | |
2397 | tok->u.repeat.lower = low;\r | |
2398 | tok->u.repeat.upper = up;\r | |
2399 | *src = p;\r | |
2400 | return r; /* 0: normal {n,m}, 2: fixed {n} */\r | |
2401 | \r | |
2402 | invalid:\r | |
2403 | if (syn_allow)\r | |
2404 | return 1; /* OK */\r | |
2405 | else\r | |
2406 | return ONIGERR_INVALID_REPEAT_RANGE_PATTERN;\r | |
2407 | }\r | |
2408 | \r | |
2409 | /* \M-, \C-, \c, or \... */\r | |
2410 | static int\r | |
2411 | fetch_escaped_value(UChar** src, UChar* end, ScanEnv* env)\r | |
2412 | {\r | |
2413 | int v;\r | |
2414 | OnigCodePoint c;\r | |
2415 | OnigEncoding enc = env->enc;\r | |
2416 | UChar* p = *src;\r | |
2417 | \r | |
2418 | if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;\r | |
2419 | \r | |
2420 | PFETCH_S(c);\r | |
2421 | switch (c) {\r | |
2422 | case 'M':\r | |
2423 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META)) {\r | |
2424 | if (PEND) return ONIGERR_END_PATTERN_AT_META;\r | |
2425 | PFETCH_S(c);\r | |
2426 | if (c != '-') return ONIGERR_META_CODE_SYNTAX;\r | |
2427 | if (PEND) return ONIGERR_END_PATTERN_AT_META;\r | |
2428 | PFETCH_S(c);\r | |
2429 | if (c == MC_ESC(env->syntax)) {\r | |
2430 | v = fetch_escaped_value(&p, end, env);\r | |
2431 | if (v < 0) return v;\r | |
2432 | c = (OnigCodePoint )v;\r | |
2433 | }\r | |
2434 | c = ((c & 0xff) | 0x80);\r | |
2435 | }\r | |
2436 | else\r | |
2437 | goto backslash;\r | |
2438 | break;\r | |
2439 | \r | |
2440 | case 'C':\r | |
2441 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL)) {\r | |
2442 | if (PEND) return ONIGERR_END_PATTERN_AT_CONTROL;\r | |
2443 | PFETCH_S(c);\r | |
2444 | if (c != '-') return ONIGERR_CONTROL_CODE_SYNTAX;\r | |
2445 | goto control;\r | |
2446 | }\r | |
2447 | else\r | |
2448 | goto backslash;\r | |
2449 | \r | |
2450 | case 'c':\r | |
2451 | if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_C_CONTROL)) {\r | |
2452 | control:\r | |
2453 | if (PEND) return ONIGERR_END_PATTERN_AT_CONTROL;\r | |
2454 | PFETCH_S(c);\r | |
2455 | if (c == '?') {\r | |
2456 | c = 0177;\r | |
2457 | }\r | |
2458 | else {\r | |
2459 | if (c == MC_ESC(env->syntax)) {\r | |
2460 | v = fetch_escaped_value(&p, end, env);\r | |
2461 | if (v < 0) return v;\r | |
2462 | c = (OnigCodePoint )v;\r | |
2463 | }\r | |
2464 | c &= 0x9f;\r | |
2465 | }\r | |
2466 | break;\r | |
2467 | }\r | |
2468 | /* fall through */\r | |
2469 | \r | |
2470 | default:\r | |
2471 | {\r | |
2472 | backslash:\r | |
2473 | c = conv_backslash_value(c, env);\r | |
2474 | }\r | |
2475 | break;\r | |
2476 | }\r | |
2477 | \r | |
2478 | *src = p;\r | |
2479 | return c;\r | |
2480 | }\r | |
2481 | \r | |
2482 | static int fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env);\r | |
2483 | \r | |
2484 | static OnigCodePoint\r | |
2485 | get_name_end_code_point(OnigCodePoint start)\r | |
2486 | {\r | |
2487 | switch (start) {\r | |
2488 | case '<': return (OnigCodePoint )'>'; break;\r | |
2489 | case '\'': return (OnigCodePoint )'\''; break;\r | |
2490 | default:\r | |
2491 | break;\r | |
2492 | }\r | |
2493 | \r | |
2494 | return (OnigCodePoint )0;\r | |
2495 | }\r | |
2496 | \r | |
2497 | #ifdef USE_NAMED_GROUP\r | |
2498 | #ifdef USE_BACKREF_WITH_LEVEL\r | |
2499 | /*\r | |
2500 | \k<name+n>, \k<name-n>\r | |
2501 | \k<num+n>, \k<num-n>\r | |
2502 | \k<-num+n>, \k<-num-n>\r | |
2503 | */\r | |
2504 | static int\r | |
2505 | fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end,\r | |
2506 | UChar** rname_end, ScanEnv* env,\r | |
2507 | int* rback_num, int* rlevel)\r | |
2508 | {\r | |
2509 | int r, sign, is_num, exist_level;\r | |
2510 | OnigCodePoint end_code;\r | |
2511 | OnigCodePoint c = 0;\r | |
2512 | OnigEncoding enc = env->enc;\r | |
2513 | UChar *name_end;\r | |
2514 | UChar *pnum_head;\r | |
2515 | UChar *p = *src;\r | |
2516 | PFETCH_READY;\r | |
2517 | \r | |
2518 | *rback_num = 0;\r | |
2519 | is_num = exist_level = 0;\r | |
2520 | sign = 1;\r | |
2521 | pnum_head = *src;\r | |
2522 | \r | |
2523 | end_code = get_name_end_code_point(start_code);\r | |
2524 | \r | |
2525 | name_end = end;\r | |
2526 | r = 0;\r | |
2527 | if (PEND) {\r | |
2528 | return ONIGERR_EMPTY_GROUP_NAME;\r | |
2529 | }\r | |
2530 | else {\r | |
2531 | PFETCH(c);\r | |
2532 | if (c == end_code)\r | |
2533 | return ONIGERR_EMPTY_GROUP_NAME;\r | |
2534 | \r | |
2535 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {\r | |
2536 | is_num = 1;\r | |
2537 | }\r | |
2538 | else if (c == '-') {\r | |
2539 | is_num = 2;\r | |
2540 | sign = -1;\r | |
2541 | pnum_head = p;\r | |
2542 | }\r | |
2543 | else if (!ONIGENC_IS_CODE_WORD(enc, c)) {\r | |
2544 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;\r | |
2545 | }\r | |
2546 | }\r | |
2547 | \r | |
2548 | while (!PEND) {\r | |
2549 | name_end = p;\r | |
2550 | PFETCH(c);\r | |
2551 | if (c == end_code || c == ')' || c == '+' || c == '-') {\r | |
2552 | if (is_num == 2) r = ONIGERR_INVALID_GROUP_NAME;\r | |
2553 | break;\r | |
2554 | }\r | |
2555 | \r | |
2556 | if (is_num != 0) {\r | |
2557 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {\r | |
2558 | is_num = 1;\r | |
2559 | }\r | |
2560 | else {\r | |
2561 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2562 | is_num = 0;\r | |
2563 | }\r | |
2564 | }\r | |
2565 | else if (!ONIGENC_IS_CODE_WORD(enc, c)) {\r | |
2566 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;\r | |
2567 | }\r | |
2568 | }\r | |
2569 | \r | |
2570 | if (r == 0 && c != end_code) {\r | |
2571 | if (c == '+' || c == '-') {\r | |
2572 | int level;\r | |
2573 | int flag = (c == '-' ? -1 : 1);\r | |
2574 | \r | |
2575 | PFETCH(c);\r | |
2576 | if (! ONIGENC_IS_CODE_DIGIT(enc, c)) goto err;\r | |
2577 | PUNFETCH;\r | |
2578 | level = onig_scan_unsigned_number(&p, end, enc);\r | |
2579 | if (level < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
2580 | *rlevel = (level * flag);\r | |
2581 | exist_level = 1;\r | |
2582 | \r | |
2583 | PFETCH(c);\r | |
2584 | if (c == end_code)\r | |
2585 | goto end;\r | |
2586 | }\r | |
2587 | \r | |
2588 | err:\r | |
2589 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2590 | name_end = end;\r | |
2591 | }\r | |
2592 | \r | |
2593 | end:\r | |
2594 | if (r == 0) {\r | |
2595 | if (is_num != 0) {\r | |
2596 | *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);\r | |
2597 | if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
2598 | else if (*rback_num == 0) goto err;\r | |
2599 | \r | |
2600 | *rback_num *= sign;\r | |
2601 | }\r | |
2602 | \r | |
2603 | *rname_end = name_end;\r | |
2604 | *src = p;\r | |
2605 | return (exist_level ? 1 : 0);\r | |
2606 | }\r | |
2607 | else {\r | |
2608 | onig_scan_env_set_error_string(env, r, *src, name_end);\r | |
2609 | return r;\r | |
2610 | }\r | |
2611 | }\r | |
2612 | #endif /* USE_BACKREF_WITH_LEVEL */\r | |
2613 | \r | |
2614 | /*\r | |
2615 | def: 0 -> define name (don't allow number name)\r | |
2616 | 1 -> reference name (allow number name)\r | |
2617 | */\r | |
2618 | static int\r | |
2619 | fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,\r | |
2620 | UChar** rname_end, ScanEnv* env, int* rback_num, int ref)\r | |
2621 | {\r | |
2622 | int r, is_num, sign;\r | |
2623 | OnigCodePoint end_code;\r | |
2624 | OnigCodePoint c = 0;\r | |
2625 | OnigEncoding enc = env->enc;\r | |
2626 | UChar *name_end;\r | |
2627 | UChar *pnum_head;\r | |
2628 | UChar *p = *src;\r | |
2629 | \r | |
2630 | *rback_num = 0;\r | |
2631 | \r | |
2632 | end_code = get_name_end_code_point(start_code);\r | |
2633 | \r | |
2634 | name_end = end;\r | |
2635 | pnum_head = *src;\r | |
2636 | r = 0;\r | |
2637 | is_num = 0;\r | |
2638 | sign = 1;\r | |
2639 | if (PEND) {\r | |
2640 | return ONIGERR_EMPTY_GROUP_NAME;\r | |
2641 | }\r | |
2642 | else {\r | |
2643 | PFETCH_S(c);\r | |
2644 | if (c == end_code)\r | |
2645 | return ONIGERR_EMPTY_GROUP_NAME;\r | |
2646 | \r | |
2647 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {\r | |
2648 | if (ref == 1)\r | |
2649 | is_num = 1;\r | |
2650 | else {\r | |
2651 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2652 | is_num = 0;\r | |
2653 | }\r | |
2654 | }\r | |
2655 | else if (c == '-') {\r | |
2656 | if (ref == 1) {\r | |
2657 | is_num = 2;\r | |
2658 | sign = -1;\r | |
2659 | pnum_head = p;\r | |
2660 | }\r | |
2661 | else {\r | |
2662 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2663 | is_num = 0;\r | |
2664 | }\r | |
2665 | }\r | |
2666 | else if (!ONIGENC_IS_CODE_WORD(enc, c)) {\r | |
2667 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;\r | |
2668 | }\r | |
2669 | }\r | |
2670 | \r | |
2671 | if (r == 0) {\r | |
2672 | while (!PEND) {\r | |
2673 | name_end = p;\r | |
2674 | PFETCH_S(c);\r | |
2675 | if (c == end_code || c == ')') {\r | |
2676 | if (is_num == 2) r = ONIGERR_INVALID_GROUP_NAME;\r | |
2677 | break;\r | |
2678 | }\r | |
2679 | \r | |
2680 | if (is_num != 0) {\r | |
2681 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {\r | |
2682 | is_num = 1;\r | |
2683 | }\r | |
2684 | else {\r | |
2685 | if (!ONIGENC_IS_CODE_WORD(enc, c))\r | |
2686 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;\r | |
2687 | else\r | |
2688 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2689 | is_num = 0;\r | |
2690 | }\r | |
2691 | }\r | |
2692 | else {\r | |
2693 | if (!ONIGENC_IS_CODE_WORD(enc, c)) {\r | |
2694 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;\r | |
2695 | }\r | |
2696 | }\r | |
2697 | }\r | |
2698 | \r | |
2699 | if (c != end_code) {\r | |
2700 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2701 | name_end = end;\r | |
2702 | }\r | |
2703 | \r | |
2704 | if (is_num != 0) {\r | |
2705 | *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);\r | |
2706 | if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
2707 | else if (*rback_num == 0) {\r | |
2708 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2709 | goto err;\r | |
2710 | }\r | |
2711 | \r | |
2712 | *rback_num *= sign;\r | |
2713 | }\r | |
2714 | \r | |
2715 | *rname_end = name_end;\r | |
2716 | *src = p;\r | |
2717 | return 0;\r | |
2718 | }\r | |
2719 | else {\r | |
2720 | while (!PEND) {\r | |
2721 | name_end = p;\r | |
2722 | PFETCH_S(c);\r | |
2723 | if (c == end_code || c == ')')\r | |
2724 | break;\r | |
2725 | }\r | |
2726 | if (PEND)\r | |
2727 | name_end = end;\r | |
2728 | \r | |
2729 | err:\r | |
2730 | onig_scan_env_set_error_string(env, r, *src, name_end);\r | |
2731 | return r;\r | |
2732 | }\r | |
2733 | }\r | |
2734 | #else\r | |
2735 | static int\r | |
2736 | fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,\r | |
2737 | UChar** rname_end, ScanEnv* env, int* rback_num, int ref)\r | |
2738 | {\r | |
2739 | int r, is_num, sign;\r | |
2740 | OnigCodePoint end_code;\r | |
2741 | OnigCodePoint c = 0;\r | |
2742 | UChar *name_end;\r | |
2743 | OnigEncoding enc = env->enc;\r | |
2744 | UChar *pnum_head;\r | |
2745 | UChar *p = *src;\r | |
2746 | PFETCH_READY;\r | |
2747 | \r | |
2748 | *rback_num = 0;\r | |
2749 | \r | |
2750 | end_code = get_name_end_code_point(start_code);\r | |
2751 | \r | |
2752 | *rname_end = name_end = end;\r | |
2753 | r = 0;\r | |
2754 | pnum_head = *src;\r | |
2755 | is_num = 0;\r | |
2756 | sign = 1;\r | |
2757 | \r | |
2758 | if (PEND) {\r | |
2759 | return ONIGERR_EMPTY_GROUP_NAME;\r | |
2760 | }\r | |
2761 | else {\r | |
2762 | PFETCH(c);\r | |
2763 | if (c == end_code)\r | |
2764 | return ONIGERR_EMPTY_GROUP_NAME;\r | |
2765 | \r | |
2766 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {\r | |
2767 | is_num = 1;\r | |
2768 | }\r | |
2769 | else if (c == '-') {\r | |
2770 | is_num = 2;\r | |
2771 | sign = -1;\r | |
2772 | pnum_head = p;\r | |
2773 | }\r | |
2774 | else {\r | |
2775 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;\r | |
2776 | }\r | |
2777 | }\r | |
2778 | \r | |
2779 | while (!PEND) {\r | |
2780 | name_end = p;\r | |
2781 | \r | |
2782 | PFETCH(c);\r | |
2783 | if (c == end_code || c == ')') break;\r | |
2784 | if (! ONIGENC_IS_CODE_DIGIT(enc, c))\r | |
2785 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;\r | |
2786 | }\r | |
2787 | if (r == 0 && c != end_code) {\r | |
2788 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2789 | name_end = end;\r | |
2790 | }\r | |
2791 | \r | |
2792 | if (r == 0) {\r | |
2793 | *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);\r | |
2794 | if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
2795 | else if (*rback_num == 0) {\r | |
2796 | r = ONIGERR_INVALID_GROUP_NAME;\r | |
2797 | goto err;\r | |
2798 | }\r | |
2799 | *rback_num *= sign;\r | |
2800 | \r | |
2801 | *rname_end = name_end;\r | |
2802 | *src = p;\r | |
2803 | return 0;\r | |
2804 | }\r | |
2805 | else {\r | |
2806 | err:\r | |
2807 | onig_scan_env_set_error_string(env, r, *src, name_end);\r | |
2808 | return r;\r | |
2809 | }\r | |
2810 | }\r | |
2811 | #endif /* USE_NAMED_GROUP */\r | |
2812 | \r | |
2813 | static void\r | |
2814 | CC_ESC_WARN(ScanEnv* env, UChar *c)\r | |
2815 | {\r | |
2816 | if (onig_warn == onig_null_warn) return ;\r | |
2817 | \r | |
2818 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED) &&\r | |
2819 | IS_SYNTAX_BV(env->syntax, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC)) {\r | |
2820 | UChar buf[WARN_BUFSIZE];\r | |
2821 | onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,\r | |
2822 | env->pattern, env->pattern_end,\r | |
2823 | (UChar* )"character class has '%s' without escape", c);\r | |
2824 | (*onig_warn)((char* )buf);\r | |
2825 | }\r | |
2826 | }\r | |
2827 | \r | |
2828 | static void\r | |
2829 | CLOSE_BRACKET_WITHOUT_ESC_WARN(ScanEnv* env, UChar* c)\r | |
2830 | {\r | |
2831 | if (onig_warn == onig_null_warn) return ;\r | |
2832 | \r | |
2833 | if (IS_SYNTAX_BV((env)->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED)) {\r | |
2834 | UChar buf[WARN_BUFSIZE];\r | |
2835 | onig_snprintf_with_pattern(buf, WARN_BUFSIZE, (env)->enc,\r | |
2836 | (env)->pattern, (env)->pattern_end,\r | |
2837 | (UChar* )"regular expression has '%s' without escape", c);\r | |
2838 | (*onig_warn)((char* )buf);\r | |
2839 | }\r | |
2840 | }\r | |
2841 | \r | |
2842 | static UChar*\r | |
2843 | find_str_position(OnigCodePoint s[], int n, UChar* from, UChar* to,\r | |
2844 | UChar **next, OnigEncoding enc)\r | |
2845 | {\r | |
2846 | int i;\r | |
2847 | OnigCodePoint x;\r | |
2848 | UChar *q;\r | |
2849 | UChar *p = from;\r | |
2850 | \r | |
2851 | while (p < to) {\r | |
2852 | x = ONIGENC_MBC_TO_CODE(enc, p, to);\r | |
2853 | q = p + enclen(enc, p);\r | |
2854 | if (x == s[0]) {\r | |
2855 | for (i = 1; i < n && q < to; i++) {\r | |
2856 | x = ONIGENC_MBC_TO_CODE(enc, q, to);\r | |
2857 | if (x != s[i]) break;\r | |
2858 | q += enclen(enc, q);\r | |
2859 | }\r | |
2860 | if (i >= n) {\r | |
2861 | if (IS_NOT_NULL(next))\r | |
2862 | *next = q;\r | |
2863 | return p;\r | |
2864 | }\r | |
2865 | }\r | |
2866 | p = q;\r | |
2867 | }\r | |
2868 | return NULL_UCHARP;\r | |
2869 | }\r | |
2870 | \r | |
2871 | static int\r | |
2872 | str_exist_check_with_esc(OnigCodePoint s[], int n, UChar* from, UChar* to,\r | |
2873 | OnigCodePoint bad, OnigEncoding enc, OnigSyntaxType* syn)\r | |
2874 | {\r | |
2875 | int i, in_esc;\r | |
2876 | OnigCodePoint x;\r | |
2877 | UChar *q;\r | |
2878 | UChar *p = from;\r | |
2879 | \r | |
2880 | in_esc = 0;\r | |
2881 | while (p < to) {\r | |
2882 | if (in_esc) {\r | |
2883 | in_esc = 0;\r | |
2884 | p += enclen(enc, p);\r | |
2885 | }\r | |
2886 | else {\r | |
2887 | x = ONIGENC_MBC_TO_CODE(enc, p, to);\r | |
2888 | q = p + enclen(enc, p);\r | |
2889 | if (x == s[0]) {\r | |
2890 | for (i = 1; i < n && q < to; i++) {\r | |
2891 | x = ONIGENC_MBC_TO_CODE(enc, q, to);\r | |
2892 | if (x != s[i]) break;\r | |
2893 | q += enclen(enc, q);\r | |
2894 | }\r | |
2895 | if (i >= n) return 1;\r | |
2896 | p += enclen(enc, p);\r | |
2897 | }\r | |
2898 | else {\r | |
2899 | x = ONIGENC_MBC_TO_CODE(enc, p, to);\r | |
2900 | if (x == bad) return 0;\r | |
2901 | else if (x == MC_ESC(syn)) in_esc = 1;\r | |
2902 | p = q;\r | |
2903 | }\r | |
2904 | }\r | |
2905 | }\r | |
2906 | return 0;\r | |
2907 | }\r | |
2908 | \r | |
2909 | static int\r | |
2910 | fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)\r | |
2911 | {\r | |
2912 | int num;\r | |
2913 | OnigCodePoint c, c2;\r | |
2914 | OnigSyntaxType* syn = env->syntax;\r | |
2915 | OnigEncoding enc = env->enc;\r | |
2916 | UChar* prev;\r | |
2917 | UChar* p = *src;\r | |
2918 | PFETCH_READY;\r | |
2919 | \r | |
2920 | if (PEND) {\r | |
2921 | tok->type = TK_EOT;\r | |
2922 | return tok->type;\r | |
2923 | }\r | |
2924 | \r | |
2925 | PFETCH(c);\r | |
2926 | tok->type = TK_CHAR;\r | |
2927 | tok->base = 0;\r | |
2928 | tok->u.c = c;\r | |
2929 | tok->escaped = 0;\r | |
2930 | \r | |
2931 | if (c == ']') {\r | |
2932 | tok->type = TK_CC_CLOSE;\r | |
2933 | }\r | |
2934 | else if (c == '-') {\r | |
2935 | tok->type = TK_CC_RANGE;\r | |
2936 | }\r | |
2937 | else if (c == MC_ESC(syn)) {\r | |
2938 | if (! IS_SYNTAX_BV(syn, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC))\r | |
2939 | goto end;\r | |
2940 | \r | |
2941 | if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;\r | |
2942 | \r | |
2943 | PFETCH(c);\r | |
2944 | tok->escaped = 1;\r | |
2945 | tok->u.c = c;\r | |
2946 | switch (c) {\r | |
2947 | case 'w':\r | |
2948 | tok->type = TK_CHAR_TYPE;\r | |
2949 | tok->u.prop.ctype = ONIGENC_CTYPE_WORD;\r | |
2950 | tok->u.prop.not = 0;\r | |
2951 | break;\r | |
2952 | case 'W':\r | |
2953 | tok->type = TK_CHAR_TYPE;\r | |
2954 | tok->u.prop.ctype = ONIGENC_CTYPE_WORD;\r | |
2955 | tok->u.prop.not = 1;\r | |
2956 | break;\r | |
2957 | case 'd':\r | |
2958 | tok->type = TK_CHAR_TYPE;\r | |
2959 | tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;\r | |
2960 | tok->u.prop.not = 0;\r | |
2961 | break;\r | |
2962 | case 'D':\r | |
2963 | tok->type = TK_CHAR_TYPE;\r | |
2964 | tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;\r | |
2965 | tok->u.prop.not = 1;\r | |
2966 | break;\r | |
2967 | case 's':\r | |
2968 | tok->type = TK_CHAR_TYPE;\r | |
2969 | tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;\r | |
2970 | tok->u.prop.not = 0;\r | |
2971 | break;\r | |
2972 | case 'S':\r | |
2973 | tok->type = TK_CHAR_TYPE;\r | |
2974 | tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;\r | |
2975 | tok->u.prop.not = 1;\r | |
2976 | break;\r | |
2977 | case 'h':\r | |
2978 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;\r | |
2979 | tok->type = TK_CHAR_TYPE;\r | |
2980 | tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;\r | |
2981 | tok->u.prop.not = 0;\r | |
2982 | break;\r | |
2983 | case 'H':\r | |
2984 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;\r | |
2985 | tok->type = TK_CHAR_TYPE;\r | |
2986 | tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;\r | |
2987 | tok->u.prop.not = 1;\r | |
2988 | break;\r | |
2989 | \r | |
2990 | case 'p':\r | |
2991 | case 'P':\r | |
2992 | c2 = PPEEK;\r | |
2993 | if (c2 == '{' &&\r | |
2994 | IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) {\r | |
2995 | PINC;\r | |
2996 | tok->type = TK_CHAR_PROPERTY;\r | |
2997 | tok->u.prop.not = (c == 'P' ? 1 : 0);\r | |
2998 | \r | |
2999 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) {\r | |
3000 | PFETCH(c2);\r | |
3001 | if (c2 == '^') {\r | |
3002 | tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0);\r | |
3003 | }\r | |
3004 | else\r | |
3005 | PUNFETCH;\r | |
3006 | }\r | |
3007 | }\r | |
3008 | break;\r | |
3009 | \r | |
3010 | case 'x':\r | |
3011 | if (PEND) break;\r | |
3012 | \r | |
3013 | prev = p;\r | |
3014 | if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {\r | |
3015 | PINC;\r | |
3016 | num = scan_unsigned_hexadecimal_number(&p, end, 8, enc);\r | |
3017 | if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;\r | |
3018 | if (!PEND) {\r | |
3019 | c2 = PPEEK;\r | |
3020 | if (ONIGENC_IS_CODE_XDIGIT(enc, c2))\r | |
3021 | return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;\r | |
3022 | }\r | |
3023 | \r | |
3024 | if (p > prev + enclen(enc, prev) && !PEND && (PPEEK_IS('}'))) {\r | |
3025 | PINC;\r | |
3026 | tok->type = TK_CODE_POINT;\r | |
3027 | tok->base = 16;\r | |
3028 | tok->u.code = (OnigCodePoint )num;\r | |
3029 | }\r | |
3030 | else {\r | |
3031 | /* can't read nothing or invalid format */\r | |
3032 | p = prev;\r | |
3033 | }\r | |
3034 | }\r | |
3035 | else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {\r | |
3036 | num = scan_unsigned_hexadecimal_number(&p, end, 2, enc);\r | |
3037 | if (num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
3038 | if (p == prev) { /* can't read nothing. */\r | |
3039 | num = 0; /* but, it's not error */\r | |
3040 | }\r | |
3041 | tok->type = TK_RAW_BYTE;\r | |
3042 | tok->base = 16;\r | |
3043 | tok->u.c = num;\r | |
3044 | }\r | |
3045 | break;\r | |
3046 | \r | |
3047 | case 'u':\r | |
3048 | if (PEND) break;\r | |
3049 | \r | |
3050 | prev = p;\r | |
3051 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {\r | |
3052 | num = scan_unsigned_hexadecimal_number(&p, end, 4, enc);\r | |
3053 | if (num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
3054 | if (p == prev) { /* can't read nothing. */\r | |
3055 | num = 0; /* but, it's not error */\r | |
3056 | }\r | |
3057 | tok->type = TK_CODE_POINT;\r | |
3058 | tok->base = 16;\r | |
3059 | tok->u.code = (OnigCodePoint )num;\r | |
3060 | }\r | |
3061 | break;\r | |
3062 | \r | |
3063 | case '0':\r | |
3064 | case '1': case '2': case '3': case '4': case '5': case '6': case '7':\r | |
3065 | if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {\r | |
3066 | PUNFETCH;\r | |
3067 | prev = p;\r | |
3068 | num = scan_unsigned_octal_number(&p, end, 3, enc);\r | |
3069 | if (num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
3070 | if (p == prev) { /* can't read nothing. */\r | |
3071 | num = 0; /* but, it's not error */\r | |
3072 | }\r | |
3073 | tok->type = TK_RAW_BYTE;\r | |
3074 | tok->base = 8;\r | |
3075 | tok->u.c = num;\r | |
3076 | }\r | |
3077 | break;\r | |
3078 | \r | |
3079 | default:\r | |
3080 | PUNFETCH;\r | |
3081 | num = fetch_escaped_value(&p, end, env);\r | |
3082 | if (num < 0) return num;\r | |
3083 | if (tok->u.c != num) {\r | |
3084 | tok->u.code = (OnigCodePoint )num;\r | |
3085 | tok->type = TK_CODE_POINT;\r | |
3086 | }\r | |
3087 | break;\r | |
3088 | }\r | |
3089 | }\r | |
3090 | else if (c == '[') {\r | |
3091 | if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_POSIX_BRACKET) && (PPEEK_IS(':'))) {\r | |
3092 | OnigCodePoint send[] = { (OnigCodePoint )':', (OnigCodePoint )']' };\r | |
3093 | tok->backp = p; /* point at '[' is readed */\r | |
3094 | PINC;\r | |
3095 | if (str_exist_check_with_esc(send, 2, p, end,\r | |
3096 | (OnigCodePoint )']', enc, syn)) {\r | |
3097 | tok->type = TK_POSIX_BRACKET_OPEN;\r | |
3098 | }\r | |
3099 | else {\r | |
3100 | PUNFETCH;\r | |
3101 | goto cc_in_cc;\r | |
3102 | }\r | |
3103 | }\r | |
3104 | else {\r | |
3105 | cc_in_cc:\r | |
3106 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP)) {\r | |
3107 | tok->type = TK_CC_CC_OPEN;\r | |
3108 | }\r | |
3109 | else {\r | |
3110 | CC_ESC_WARN(env, (UChar* )"[");\r | |
3111 | }\r | |
3112 | }\r | |
3113 | }\r | |
3114 | else if (c == '&') {\r | |
3115 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP) &&\r | |
3116 | !PEND && (PPEEK_IS('&'))) {\r | |
3117 | PINC;\r | |
3118 | tok->type = TK_CC_AND;\r | |
3119 | }\r | |
3120 | }\r | |
3121 | \r | |
3122 | end:\r | |
3123 | *src = p;\r | |
3124 | return tok->type;\r | |
3125 | }\r | |
3126 | \r | |
3127 | static int\r | |
3128 | fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)\r | |
3129 | {\r | |
3130 | int r, num;\r | |
3131 | OnigCodePoint c;\r | |
3132 | OnigEncoding enc = env->enc;\r | |
3133 | OnigSyntaxType* syn = env->syntax;\r | |
3134 | UChar* prev;\r | |
3135 | UChar* p = *src;\r | |
3136 | PFETCH_READY;\r | |
3137 | \r | |
3138 | start:\r | |
3139 | if (PEND) {\r | |
3140 | tok->type = TK_EOT;\r | |
3141 | return tok->type;\r | |
3142 | }\r | |
3143 | \r | |
3144 | tok->type = TK_STRING;\r | |
3145 | tok->base = 0;\r | |
3146 | tok->backp = p;\r | |
3147 | \r | |
3148 | PFETCH(c);\r | |
3149 | if (IS_MC_ESC_CODE(c, syn)) {\r | |
3150 | if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;\r | |
3151 | \r | |
3152 | tok->backp = p;\r | |
3153 | PFETCH(c);\r | |
3154 | \r | |
3155 | tok->u.c = c;\r | |
3156 | tok->escaped = 1;\r | |
3157 | switch (c) {\r | |
3158 | case '*':\r | |
3159 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF)) break;\r | |
3160 | tok->type = TK_OP_REPEAT;\r | |
3161 | tok->u.repeat.lower = 0;\r | |
3162 | tok->u.repeat.upper = REPEAT_INFINITE;\r | |
3163 | goto greedy_check;\r | |
3164 | break;\r | |
3165 | \r | |
3166 | case '+':\r | |
3167 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_PLUS_ONE_INF)) break;\r | |
3168 | tok->type = TK_OP_REPEAT;\r | |
3169 | tok->u.repeat.lower = 1;\r | |
3170 | tok->u.repeat.upper = REPEAT_INFINITE;\r | |
3171 | goto greedy_check;\r | |
3172 | break;\r | |
3173 | \r | |
3174 | case '?':\r | |
3175 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_QMARK_ZERO_ONE)) break;\r | |
3176 | tok->type = TK_OP_REPEAT;\r | |
3177 | tok->u.repeat.lower = 0;\r | |
3178 | tok->u.repeat.upper = 1;\r | |
3179 | greedy_check:\r | |
3180 | if (!PEND && PPEEK_IS('?') &&\r | |
3181 | IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_NON_GREEDY)) {\r | |
3182 | PFETCH(c);\r | |
3183 | tok->u.repeat.greedy = 0;\r | |
3184 | tok->u.repeat.possessive = 0;\r | |
3185 | }\r | |
3186 | else {\r | |
3187 | possessive_check:\r | |
3188 | if (!PEND && PPEEK_IS('+') &&\r | |
3189 | ((IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT) &&\r | |
3190 | tok->type != TK_INTERVAL) ||\r | |
3191 | (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL) &&\r | |
3192 | tok->type == TK_INTERVAL))) {\r | |
3193 | PFETCH(c);\r | |
3194 | tok->u.repeat.greedy = 1;\r | |
3195 | tok->u.repeat.possessive = 1;\r | |
3196 | }\r | |
3197 | else {\r | |
3198 | tok->u.repeat.greedy = 1;\r | |
3199 | tok->u.repeat.possessive = 0;\r | |
3200 | }\r | |
3201 | }\r | |
3202 | break;\r | |
3203 | \r | |
3204 | case '{':\r | |
3205 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) break;\r | |
3206 | r = fetch_range_quantifier(&p, end, tok, env);\r | |
3207 | if (r < 0) return r; /* error */\r | |
3208 | if (r == 0) goto greedy_check;\r | |
3209 | else if (r == 2) { /* {n} */\r | |
3210 | if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY))\r | |
3211 | goto possessive_check;\r | |
3212 | \r | |
3213 | goto greedy_check;\r | |
3214 | }\r | |
3215 | /* r == 1 : normal char */\r | |
3216 | break;\r | |
3217 | \r | |
3218 | case '|':\r | |
3219 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_VBAR_ALT)) break;\r | |
3220 | tok->type = TK_ALT;\r | |
3221 | break;\r | |
3222 | \r | |
3223 | case '(':\r | |
3224 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LPAREN_SUBEXP)) break;\r | |
3225 | tok->type = TK_SUBEXP_OPEN;\r | |
3226 | break;\r | |
3227 | \r | |
3228 | case ')':\r | |
3229 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LPAREN_SUBEXP)) break;\r | |
3230 | tok->type = TK_SUBEXP_CLOSE;\r | |
3231 | break;\r | |
3232 | \r | |
3233 | case 'w':\r | |
3234 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD)) break;\r | |
3235 | tok->type = TK_CHAR_TYPE;\r | |
3236 | tok->u.prop.ctype = ONIGENC_CTYPE_WORD;\r | |
3237 | tok->u.prop.not = 0;\r | |
3238 | break;\r | |
3239 | \r | |
3240 | case 'W':\r | |
3241 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD)) break;\r | |
3242 | tok->type = TK_CHAR_TYPE;\r | |
3243 | tok->u.prop.ctype = ONIGENC_CTYPE_WORD;\r | |
3244 | tok->u.prop.not = 1;\r | |
3245 | break;\r | |
3246 | \r | |
3247 | case 'b':\r | |
3248 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_B_WORD_BOUND)) break;\r | |
3249 | tok->type = TK_ANCHOR;\r | |
3250 | tok->u.anchor = ANCHOR_WORD_BOUND;\r | |
3251 | break;\r | |
3252 | \r | |
3253 | case 'B':\r | |
3254 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_B_WORD_BOUND)) break;\r | |
3255 | tok->type = TK_ANCHOR;\r | |
3256 | tok->u.anchor = ANCHOR_NOT_WORD_BOUND;\r | |
3257 | break;\r | |
3258 | \r | |
3259 | #ifdef USE_WORD_BEGIN_END\r | |
3260 | case '<':\r | |
3261 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END)) break;\r | |
3262 | tok->type = TK_ANCHOR;\r | |
3263 | tok->u.anchor = ANCHOR_WORD_BEGIN;\r | |
3264 | break;\r | |
3265 | \r | |
3266 | case '>':\r | |
3267 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END)) break;\r | |
3268 | tok->type = TK_ANCHOR;\r | |
3269 | tok->u.anchor = ANCHOR_WORD_END;\r | |
3270 | break;\r | |
3271 | #endif\r | |
3272 | \r | |
3273 | case 's':\r | |
3274 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE)) break;\r | |
3275 | tok->type = TK_CHAR_TYPE;\r | |
3276 | tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;\r | |
3277 | tok->u.prop.not = 0;\r | |
3278 | break;\r | |
3279 | \r | |
3280 | case 'S':\r | |
3281 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE)) break;\r | |
3282 | tok->type = TK_CHAR_TYPE;\r | |
3283 | tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;\r | |
3284 | tok->u.prop.not = 1;\r | |
3285 | break;\r | |
3286 | \r | |
3287 | case 'd':\r | |
3288 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT)) break;\r | |
3289 | tok->type = TK_CHAR_TYPE;\r | |
3290 | tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;\r | |
3291 | tok->u.prop.not = 0;\r | |
3292 | break;\r | |
3293 | \r | |
3294 | case 'D':\r | |
3295 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT)) break;\r | |
3296 | tok->type = TK_CHAR_TYPE;\r | |
3297 | tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;\r | |
3298 | tok->u.prop.not = 1;\r | |
3299 | break;\r | |
3300 | \r | |
3301 | case 'h':\r | |
3302 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;\r | |
3303 | tok->type = TK_CHAR_TYPE;\r | |
3304 | tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;\r | |
3305 | tok->u.prop.not = 0;\r | |
3306 | break;\r | |
3307 | \r | |
3308 | case 'H':\r | |
3309 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;\r | |
3310 | tok->type = TK_CHAR_TYPE;\r | |
3311 | tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;\r | |
3312 | tok->u.prop.not = 1;\r | |
3313 | break;\r | |
3314 | \r | |
3315 | case 'A':\r | |
3316 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;\r | |
3317 | begin_buf:\r | |
3318 | tok->type = TK_ANCHOR;\r | |
3319 | tok->u.subtype = ANCHOR_BEGIN_BUF;\r | |
3320 | break;\r | |
3321 | \r | |
3322 | case 'Z':\r | |
3323 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;\r | |
3324 | tok->type = TK_ANCHOR;\r | |
3325 | tok->u.subtype = ANCHOR_SEMI_END_BUF;\r | |
3326 | break;\r | |
3327 | \r | |
3328 | case 'z':\r | |
3329 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;\r | |
3330 | end_buf:\r | |
3331 | tok->type = TK_ANCHOR;\r | |
3332 | tok->u.subtype = ANCHOR_END_BUF;\r | |
3333 | break;\r | |
3334 | \r | |
3335 | case 'G':\r | |
3336 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR)) break;\r | |
3337 | tok->type = TK_ANCHOR;\r | |
3338 | tok->u.subtype = ANCHOR_BEGIN_POSITION;\r | |
3339 | break;\r | |
3340 | \r | |
3341 | case '`':\r | |
3342 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR)) break;\r | |
3343 | goto begin_buf;\r | |
3344 | break;\r | |
3345 | \r | |
3346 | case '\'':\r | |
3347 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR)) break;\r | |
3348 | goto end_buf;\r | |
3349 | break;\r | |
3350 | \r | |
3351 | case 'x':\r | |
3352 | if (PEND) break;\r | |
3353 | \r | |
3354 | prev = p;\r | |
3355 | if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {\r | |
3356 | PINC;\r | |
3357 | num = scan_unsigned_hexadecimal_number(&p, end, 8, enc);\r | |
3358 | if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;\r | |
3359 | if (!PEND) {\r | |
3360 | if (ONIGENC_IS_CODE_XDIGIT(enc, PPEEK))\r | |
3361 | return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;\r | |
3362 | }\r | |
3363 | \r | |
3364 | if ((p > prev + enclen(enc, prev)) && !PEND && PPEEK_IS('}')) {\r | |
3365 | PINC;\r | |
3366 | tok->type = TK_CODE_POINT;\r | |
3367 | tok->u.code = (OnigCodePoint )num;\r | |
3368 | }\r | |
3369 | else {\r | |
3370 | /* can't read nothing or invalid format */\r | |
3371 | p = prev;\r | |
3372 | }\r | |
3373 | }\r | |
3374 | else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {\r | |
3375 | num = scan_unsigned_hexadecimal_number(&p, end, 2, enc);\r | |
3376 | if (num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
3377 | if (p == prev) { /* can't read nothing. */\r | |
3378 | num = 0; /* but, it's not error */\r | |
3379 | }\r | |
3380 | tok->type = TK_RAW_BYTE;\r | |
3381 | tok->base = 16;\r | |
3382 | tok->u.c = num;\r | |
3383 | }\r | |
3384 | break;\r | |
3385 | \r | |
3386 | case 'u':\r | |
3387 | if (PEND) break;\r | |
3388 | \r | |
3389 | prev = p;\r | |
3390 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {\r | |
3391 | num = scan_unsigned_hexadecimal_number(&p, end, 4, enc);\r | |
3392 | if (num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
3393 | if (p == prev) { /* can't read nothing. */\r | |
3394 | num = 0; /* but, it's not error */\r | |
3395 | }\r | |
3396 | tok->type = TK_CODE_POINT;\r | |
3397 | tok->base = 16;\r | |
3398 | tok->u.code = (OnigCodePoint )num;\r | |
3399 | }\r | |
3400 | break;\r | |
3401 | \r | |
3402 | case '1': case '2': case '3': case '4':\r | |
3403 | case '5': case '6': case '7': case '8': case '9':\r | |
3404 | PUNFETCH;\r | |
3405 | prev = p;\r | |
3406 | num = onig_scan_unsigned_number(&p, end, enc);\r | |
3407 | if (num < 0 || num > ONIG_MAX_BACKREF_NUM) {\r | |
3408 | goto skip_backref;\r | |
3409 | }\r | |
3410 | \r | |
3411 | if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_DECIMAL_BACKREF) && \r | |
3412 | (num <= env->num_mem || num <= 9)) { /* This spec. from GNU regex */\r | |
3413 | if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {\r | |
3414 | if (num > env->num_mem || IS_NULL(SCANENV_MEM_NODES(env)[num]))\r | |
3415 | return ONIGERR_INVALID_BACKREF;\r | |
3416 | }\r | |
3417 | \r | |
3418 | tok->type = TK_BACKREF;\r | |
3419 | tok->u.backref.num = 1;\r | |
3420 | tok->u.backref.ref1 = num;\r | |
3421 | tok->u.backref.by_name = 0;\r | |
3422 | #ifdef USE_BACKREF_WITH_LEVEL\r | |
3423 | tok->u.backref.exist_level = 0;\r | |
3424 | #endif\r | |
3425 | break;\r | |
3426 | }\r | |
3427 | \r | |
3428 | skip_backref:\r | |
3429 | if (c == '8' || c == '9') {\r | |
3430 | /* normal char */\r | |
3431 | p = prev; PINC;\r | |
3432 | break;\r | |
3433 | }\r | |
3434 | \r | |
3435 | p = prev;\r | |
3436 | /* fall through */\r | |
3437 | case '0':\r | |
3438 | if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {\r | |
3439 | prev = p;\r | |
3440 | num = scan_unsigned_octal_number(&p, end, (c == '0' ? 2:3), enc);\r | |
3441 | if (num < 0) return ONIGERR_TOO_BIG_NUMBER;\r | |
3442 | if (p == prev) { /* can't read nothing. */\r | |
3443 | num = 0; /* but, it's not error */\r | |
3444 | }\r | |
3445 | tok->type = TK_RAW_BYTE;\r | |
3446 | tok->base = 8;\r | |
3447 | tok->u.c = num;\r | |
3448 | }\r | |
3449 | else if (c != '0') {\r | |
3450 | PINC;\r | |
3451 | }\r | |
3452 | break;\r | |
3453 | \r | |
3454 | #ifdef USE_NAMED_GROUP\r | |
3455 | case 'k':\r | |
3456 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_K_NAMED_BACKREF)) {\r | |
3457 | PFETCH(c);\r | |
3458 | if (c == '<' || c == '\'') {\r | |
3459 | UChar* name_end;\r | |
3460 | int* backs;\r | |
3461 | int back_num;\r | |
3462 | \r | |
3463 | prev = p;\r | |
3464 | \r | |
3465 | #ifdef USE_BACKREF_WITH_LEVEL\r | |
3466 | name_end = NULL_UCHARP; /* no need. escape gcc warning. */\r | |
3467 | r = fetch_name_with_level((OnigCodePoint )c, &p, end, &name_end,\r | |
3468 | env, &back_num, &tok->u.backref.level);\r | |
3469 | if (r == 1) tok->u.backref.exist_level = 1;\r | |
3470 | else tok->u.backref.exist_level = 0;\r | |
3471 | #else\r | |
3472 | r = fetch_name(&p, end, &name_end, env, &back_num, 1);\r | |
3473 | #endif\r | |
3474 | if (r < 0) return r;\r | |
3475 | \r | |
3476 | if (back_num != 0) {\r | |
3477 | if (back_num < 0) {\r | |
3478 | back_num = BACKREF_REL_TO_ABS(back_num, env);\r | |
3479 | if (back_num <= 0)\r | |
3480 | return ONIGERR_INVALID_BACKREF;\r | |
3481 | }\r | |
3482 | \r | |
3483 | if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {\r | |
3484 | if (back_num > env->num_mem ||\r | |
3485 | IS_NULL(SCANENV_MEM_NODES(env)[back_num]))\r | |
3486 | return ONIGERR_INVALID_BACKREF;\r | |
3487 | }\r | |
3488 | tok->type = TK_BACKREF;\r | |
3489 | tok->u.backref.by_name = 0;\r | |
3490 | tok->u.backref.num = 1;\r | |
3491 | tok->u.backref.ref1 = back_num;\r | |
3492 | }\r | |
3493 | else {\r | |
3494 | num = onig_name_to_group_numbers(env->reg, prev, name_end, &backs);\r | |
3495 | if (num <= 0) {\r | |
3496 | onig_scan_env_set_error_string(env,\r | |
3497 | ONIGERR_UNDEFINED_NAME_REFERENCE, prev, name_end);\r | |
3498 | return ONIGERR_UNDEFINED_NAME_REFERENCE;\r | |
3499 | }\r | |
3500 | if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {\r | |
3501 | int i;\r | |
3502 | for (i = 0; i < num; i++) {\r | |
3503 | if (backs[i] > env->num_mem ||\r | |
3504 | IS_NULL(SCANENV_MEM_NODES(env)[backs[i]]))\r | |
3505 | return ONIGERR_INVALID_BACKREF;\r | |
3506 | }\r | |
3507 | }\r | |
3508 | \r | |
3509 | tok->type = TK_BACKREF;\r | |
3510 | tok->u.backref.by_name = 1;\r | |
3511 | if (num == 1) {\r | |
3512 | tok->u.backref.num = 1;\r | |
3513 | tok->u.backref.ref1 = backs[0];\r | |
3514 | }\r | |
3515 | else {\r | |
3516 | tok->u.backref.num = num;\r | |
3517 | tok->u.backref.refs = backs;\r | |
3518 | }\r | |
3519 | }\r | |
3520 | }\r | |
3521 | else\r | |
3522 | PUNFETCH;\r | |
3523 | }\r | |
3524 | break;\r | |
3525 | #endif\r | |
3526 | \r | |
3527 | #ifdef USE_SUBEXP_CALL\r | |
3528 | case 'g':\r | |
3529 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_G_SUBEXP_CALL)) {\r | |
3530 | PFETCH(c);\r | |
3531 | if (c == '<' || c == '\'') {\r | |
3532 | int gnum;\r | |
3533 | UChar* name_end;\r | |
3534 | \r | |
3535 | prev = p;\r | |
3536 | r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &gnum, 1);\r | |
3537 | if (r < 0) return r;\r | |
3538 | \r | |
3539 | tok->type = TK_CALL;\r | |
3540 | tok->u.call.name = prev;\r | |
3541 | tok->u.call.name_end = name_end;\r | |
3542 | tok->u.call.gnum = gnum;\r | |
3543 | }\r | |
3544 | else\r | |
3545 | PUNFETCH;\r | |
3546 | }\r | |
3547 | break;\r | |
3548 | #endif\r | |
3549 | \r | |
3550 | case 'Q':\r | |
3551 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE)) {\r | |
3552 | tok->type = TK_QUOTE_OPEN;\r | |
3553 | }\r | |
3554 | break;\r | |
3555 | \r | |
3556 | case 'p':\r | |
3557 | case 'P':\r | |
3558 | if (PPEEK_IS('{') &&\r | |
3559 | IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) {\r | |
3560 | PINC;\r | |
3561 | tok->type = TK_CHAR_PROPERTY;\r | |
3562 | tok->u.prop.not = (c == 'P' ? 1 : 0);\r | |
3563 | \r | |
3564 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) {\r | |
3565 | PFETCH(c);\r | |
3566 | if (c == '^') {\r | |
3567 | tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0);\r | |
3568 | }\r | |
3569 | else\r | |
3570 | PUNFETCH;\r | |
3571 | }\r | |
3572 | }\r | |
3573 | break;\r | |
3574 | \r | |
3575 | default:\r | |
3576 | PUNFETCH;\r | |
3577 | num = fetch_escaped_value(&p, end, env);\r | |
3578 | if (num < 0) return num;\r | |
3579 | /* set_raw: */\r | |
3580 | if (tok->u.c != num) {\r | |
3581 | tok->type = TK_CODE_POINT;\r | |
3582 | tok->u.code = (OnigCodePoint )num;\r | |
3583 | }\r | |
3584 | else { /* string */\r | |
3585 | p = tok->backp + enclen(enc, tok->backp);\r | |
3586 | }\r | |
3587 | break;\r | |
3588 | }\r | |
3589 | }\r | |
3590 | else {\r | |
3591 | tok->u.c = c;\r | |
3592 | tok->escaped = 0;\r | |
3593 | \r | |
3594 | #ifdef USE_VARIABLE_META_CHARS\r | |
3595 | if ((c != ONIG_INEFFECTIVE_META_CHAR) &&\r | |
3596 | IS_SYNTAX_OP(syn, ONIG_SYN_OP_VARIABLE_META_CHARACTERS)) {\r | |
3597 | if (c == MC_ANYCHAR(syn))\r | |
3598 | goto any_char;\r | |
3599 | else if (c == MC_ANYTIME(syn))\r | |
3600 | goto anytime;\r | |
3601 | else if (c == MC_ZERO_OR_ONE_TIME(syn))\r | |
3602 | goto zero_or_one_time;\r | |
3603 | else if (c == MC_ONE_OR_MORE_TIME(syn))\r | |
3604 | goto one_or_more_time;\r | |
3605 | else if (c == MC_ANYCHAR_ANYTIME(syn)) {\r | |
3606 | tok->type = TK_ANYCHAR_ANYTIME;\r | |
3607 | goto out;\r | |
3608 | }\r | |
3609 | }\r | |
3610 | #endif\r | |
3611 | \r | |
3612 | switch (c) {\r | |
3613 | case '.':\r | |
3614 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_DOT_ANYCHAR)) break;\r | |
3615 | #ifdef USE_VARIABLE_META_CHARS\r | |
3616 | any_char:\r | |
3617 | #endif\r | |
3618 | tok->type = TK_ANYCHAR;\r | |
3619 | break;\r | |
3620 | \r | |
3621 | case '*':\r | |
3622 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ASTERISK_ZERO_INF)) break;\r | |
3623 | #ifdef USE_VARIABLE_META_CHARS\r | |
3624 | anytime:\r | |
3625 | #endif\r | |
3626 | tok->type = TK_OP_REPEAT;\r | |
3627 | tok->u.repeat.lower = 0;\r | |
3628 | tok->u.repeat.upper = REPEAT_INFINITE;\r | |
3629 | goto greedy_check;\r | |
3630 | break;\r | |
3631 | \r | |
3632 | case '+':\r | |
3633 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_PLUS_ONE_INF)) break;\r | |
3634 | #ifdef USE_VARIABLE_META_CHARS\r | |
3635 | one_or_more_time:\r | |
3636 | #endif\r | |
3637 | tok->type = TK_OP_REPEAT;\r | |
3638 | tok->u.repeat.lower = 1;\r | |
3639 | tok->u.repeat.upper = REPEAT_INFINITE;\r | |
3640 | goto greedy_check;\r | |
3641 | break;\r | |
3642 | \r | |
3643 | case '?':\r | |
3644 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_ZERO_ONE)) break;\r | |
3645 | #ifdef USE_VARIABLE_META_CHARS\r | |
3646 | zero_or_one_time:\r | |
3647 | #endif\r | |
3648 | tok->type = TK_OP_REPEAT;\r | |
3649 | tok->u.repeat.lower = 0;\r | |
3650 | tok->u.repeat.upper = 1;\r | |
3651 | goto greedy_check;\r | |
3652 | break;\r | |
3653 | \r | |
3654 | case '{':\r | |
3655 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACE_INTERVAL)) break;\r | |
3656 | r = fetch_range_quantifier(&p, end, tok, env);\r | |
3657 | if (r < 0) return r; /* error */\r | |
3658 | if (r == 0) goto greedy_check;\r | |
3659 | else if (r == 2) { /* {n} */\r | |
3660 | if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY))\r | |
3661 | goto possessive_check;\r | |
3662 | \r | |
3663 | goto greedy_check;\r | |
3664 | }\r | |
3665 | /* r == 1 : normal char */\r | |
3666 | break;\r | |
3667 | \r | |
3668 | case '|':\r | |
3669 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_VBAR_ALT)) break;\r | |
3670 | tok->type = TK_ALT;\r | |
3671 | break;\r | |
3672 | \r | |
3673 | case '(':\r | |
3674 | if (PPEEK_IS('?') &&\r | |
3675 | IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) {\r | |
3676 | PINC;\r | |
3677 | if (PPEEK_IS('#')) {\r | |
3678 | PFETCH(c);\r | |
3679 | while (1) {\r | |
3680 | if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;\r | |
3681 | PFETCH(c);\r | |
3682 | if (c == MC_ESC(syn)) {\r | |
3683 | if (!PEND) PFETCH(c);\r | |
3684 | }\r | |
3685 | else {\r | |
3686 | if (c == ')') break;\r | |
3687 | }\r | |
3688 | }\r | |
3689 | goto start;\r | |
3690 | }\r | |
3691 | PUNFETCH;\r | |
3692 | }\r | |
3693 | \r | |
3694 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP)) break;\r | |
3695 | tok->type = TK_SUBEXP_OPEN;\r | |
3696 | break;\r | |
3697 | \r | |
3698 | case ')':\r | |
3699 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP)) break;\r | |
3700 | tok->type = TK_SUBEXP_CLOSE;\r | |
3701 | break;\r | |
3702 | \r | |
3703 | case '^':\r | |
3704 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR)) break;\r | |
3705 | tok->type = TK_ANCHOR;\r | |
3706 | tok->u.subtype = (IS_SINGLELINE(env->option)\r | |
3707 | ? ANCHOR_BEGIN_BUF : ANCHOR_BEGIN_LINE);\r | |
3708 | break;\r | |
3709 | \r | |
3710 | case '$':\r | |
3711 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR)) break;\r | |
3712 | tok->type = TK_ANCHOR;\r | |
3713 | tok->u.subtype = (IS_SINGLELINE(env->option)\r | |
3714 | ? ANCHOR_SEMI_END_BUF : ANCHOR_END_LINE);\r | |
3715 | break;\r | |
3716 | \r | |
3717 | case '[':\r | |
3718 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACKET_CC)) break;\r | |
3719 | tok->type = TK_CC_OPEN;\r | |
3720 | break;\r | |
3721 | \r | |
3722 | case ']':\r | |
3723 | if (*src > env->pattern) /* /].../ is allowed. */\r | |
3724 | CLOSE_BRACKET_WITHOUT_ESC_WARN(env, (UChar* )"]");\r | |
3725 | break;\r | |
3726 | \r | |
3727 | case '#':\r | |
3728 | if (IS_EXTEND(env->option)) {\r | |
3729 | while (!PEND) {\r | |
3730 | PFETCH(c);\r | |
3731 | if (ONIGENC_IS_CODE_NEWLINE(enc, c))\r | |
3732 | break;\r | |
3733 | }\r | |
3734 | goto start;\r | |
3735 | break;\r | |
3736 | }\r | |
3737 | break;\r | |
3738 | \r | |
3739 | case ' ': case '\t': case '\n': case '\r': case '\f':\r | |
3740 | if (IS_EXTEND(env->option))\r | |
3741 | goto start;\r | |
3742 | break;\r | |
3743 | \r | |
3744 | default:\r | |
3745 | /* string */\r | |
3746 | break;\r | |
3747 | }\r | |
3748 | }\r | |
3749 | \r | |
3750 | #ifdef USE_VARIABLE_META_CHARS\r | |
3751 | out:\r | |
3752 | #endif\r | |
3753 | *src = p;\r | |
3754 | return tok->type;\r | |
3755 | }\r | |
3756 | \r | |
3757 | static int\r | |
3758 | add_ctype_to_cc_by_range(CClassNode* cc, int ctype ARG_UNUSED, int not,\r | |
3759 | OnigEncoding enc ARG_UNUSED,\r | |
3760 | OnigCodePoint sb_out, const OnigCodePoint mbr[])\r | |
3761 | {\r | |
3762 | int i, r;\r | |
3763 | OnigCodePoint j;\r | |
3764 | \r | |
3765 | int n = ONIGENC_CODE_RANGE_NUM(mbr);\r | |
3766 | \r | |
3767 | if (not == 0) {\r | |
3768 | for (i = 0; i < n; i++) {\r | |
3769 | for (j = ONIGENC_CODE_RANGE_FROM(mbr, i);\r | |
3770 | j <= ONIGENC_CODE_RANGE_TO(mbr, i); j++) {\r | |
3771 | if (j >= sb_out) {\r | |
3772 | if (j == ONIGENC_CODE_RANGE_TO(mbr, i)) i++;\r | |
3773 | else if (j > ONIGENC_CODE_RANGE_FROM(mbr, i)) {\r | |
3774 | r = add_code_range_to_buf(&(cc->mbuf), j,\r | |
3775 | ONIGENC_CODE_RANGE_TO(mbr, i));\r | |
3776 | if (r != 0) return r;\r | |
3777 | i++;\r | |
3778 | }\r | |
3779 | \r | |
3780 | goto sb_end;\r | |
3781 | }\r | |
3782 | BITSET_SET_BIT(cc->bs, j);\r | |
3783 | }\r | |
3784 | }\r | |
3785 | \r | |
3786 | sb_end:\r | |
3787 | for ( ; i < n; i++) {\r | |
3788 | r = add_code_range_to_buf(&(cc->mbuf),\r | |
3789 | ONIGENC_CODE_RANGE_FROM(mbr, i),\r | |
3790 | ONIGENC_CODE_RANGE_TO(mbr, i));\r | |
3791 | if (r != 0) return r;\r | |
3792 | }\r | |
3793 | }\r | |
3794 | else {\r | |
3795 | OnigCodePoint prev = 0;\r | |
3796 | \r | |
3797 | for (i = 0; i < n; i++) {\r | |
3798 | for (j = prev;\r | |
3799 | j < ONIGENC_CODE_RANGE_FROM(mbr, i); j++) {\r | |
3800 | if (j >= sb_out) {\r | |
3801 | goto sb_end2;\r | |
3802 | }\r | |
3803 | BITSET_SET_BIT(cc->bs, j);\r | |
3804 | }\r | |
3805 | prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1;\r | |
3806 | }\r | |
3807 | for (j = prev; j < sb_out; j++) {\r | |
3808 | BITSET_SET_BIT(cc->bs, j);\r | |
3809 | }\r | |
3810 | \r | |
3811 | sb_end2:\r | |
3812 | prev = sb_out;\r | |
3813 | \r | |
3814 | for (i = 0; i < n; i++) {\r | |
3815 | if (prev < ONIGENC_CODE_RANGE_FROM(mbr, i)) {\r | |
3816 | r = add_code_range_to_buf(&(cc->mbuf), prev,\r | |
3817 | ONIGENC_CODE_RANGE_FROM(mbr, i) - 1);\r | |
3818 | if (r != 0) return r;\r | |
3819 | }\r | |
3820 | prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1;\r | |
3821 | }\r | |
3822 | if (prev < 0x7fffffff) {\r | |
3823 | r = add_code_range_to_buf(&(cc->mbuf), prev, 0x7fffffff);\r | |
3824 | if (r != 0) return r;\r | |
3825 | }\r | |
3826 | }\r | |
3827 | \r | |
3828 | return 0;\r | |
3829 | }\r | |
3830 | \r | |
3831 | static int\r | |
3832 | add_ctype_to_cc(CClassNode* cc, int ctype, int not, ScanEnv* env)\r | |
3833 | {\r | |
3834 | int c, r;\r | |
3835 | const OnigCodePoint *ranges;\r | |
3836 | OnigCodePoint sb_out;\r | |
3837 | OnigEncoding enc = env->enc;\r | |
3838 | \r | |
3839 | r = ONIGENC_GET_CTYPE_CODE_RANGE(enc, ctype, &sb_out, &ranges);\r | |
3840 | if (r == 0) {\r | |
3841 | return add_ctype_to_cc_by_range(cc, ctype, not, env->enc, sb_out, ranges);\r | |
3842 | }\r | |
3843 | else if (r != ONIG_NO_SUPPORT_CONFIG) {\r | |
3844 | return r;\r | |
3845 | }\r | |
3846 | \r | |
3847 | r = 0;\r | |
3848 | switch (ctype) {\r | |
3849 | case ONIGENC_CTYPE_ALPHA:\r | |
3850 | case ONIGENC_CTYPE_BLANK:\r | |
3851 | case ONIGENC_CTYPE_CNTRL:\r | |
3852 | case ONIGENC_CTYPE_DIGIT:\r | |
3853 | case ONIGENC_CTYPE_LOWER:\r | |
3854 | case ONIGENC_CTYPE_PUNCT:\r | |
3855 | case ONIGENC_CTYPE_SPACE:\r | |
3856 | case ONIGENC_CTYPE_UPPER:\r | |
3857 | case ONIGENC_CTYPE_XDIGIT:\r | |
3858 | case ONIGENC_CTYPE_ASCII:\r | |
3859 | case ONIGENC_CTYPE_ALNUM:\r | |
3860 | if (not != 0) {\r | |
3861 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {\r | |
3862 | if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))\r | |
3863 | BITSET_SET_BIT(cc->bs, c);\r | |
3864 | }\r | |
3865 | ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);\r | |
3866 | }\r | |
3867 | else {\r | |
3868 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {\r | |
3869 | if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))\r | |
3870 | BITSET_SET_BIT(cc->bs, c);\r | |
3871 | }\r | |
3872 | }\r | |
3873 | break;\r | |
3874 | \r | |
3875 | case ONIGENC_CTYPE_GRAPH:\r | |
3876 | case ONIGENC_CTYPE_PRINT:\r | |
3877 | if (not != 0) {\r | |
3878 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {\r | |
3879 | if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))\r | |
3880 | BITSET_SET_BIT(cc->bs, c);\r | |
3881 | }\r | |
3882 | }\r | |
3883 | else {\r | |
3884 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {\r | |
3885 | if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))\r | |
3886 | BITSET_SET_BIT(cc->bs, c);\r | |
3887 | }\r | |
3888 | ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);\r | |
3889 | }\r | |
3890 | break;\r | |
3891 | \r | |
3892 | case ONIGENC_CTYPE_WORD:\r | |
3893 | if (not == 0) {\r | |
3894 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {\r | |
3895 | if (IS_CODE_SB_WORD(enc, c)) BITSET_SET_BIT(cc->bs, c);\r | |
3896 | }\r | |
3897 | ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);\r | |
3898 | }\r | |
3899 | else {\r | |
3900 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {\r | |
3901 | if ((ONIGENC_CODE_TO_MBCLEN(enc, c) > 0) /* check invalid code point */\r | |
3902 | && ! ONIGENC_IS_CODE_WORD(enc, c))\r | |
3903 | BITSET_SET_BIT(cc->bs, c);\r | |
3904 | }\r | |
3905 | }\r | |
3906 | break;\r | |
3907 | \r | |
3908 | default:\r | |
3909 | return ONIGERR_PARSER_BUG;\r | |
3910 | break;\r | |
3911 | }\r | |
3912 | \r | |
3913 | return r;\r | |
3914 | }\r | |
3915 | \r | |
3916 | static int\r | |
3917 | parse_posix_bracket(CClassNode* cc, UChar** src, UChar* end, ScanEnv* env)\r | |
3918 | {\r | |
3919 | #define POSIX_BRACKET_CHECK_LIMIT_LENGTH 20\r | |
3920 | #define POSIX_BRACKET_NAME_MIN_LEN 4\r | |
3921 | \r | |
3922 | static PosixBracketEntryType PBS[] = {\r | |
3923 | { (UChar* )"alnum", ONIGENC_CTYPE_ALNUM, 5 },\r | |
3924 | { (UChar* )"alpha", ONIGENC_CTYPE_ALPHA, 5 },\r | |
3925 | { (UChar* )"blank", ONIGENC_CTYPE_BLANK, 5 },\r | |
3926 | { (UChar* )"cntrl", ONIGENC_CTYPE_CNTRL, 5 },\r | |
3927 | { (UChar* )"digit", ONIGENC_CTYPE_DIGIT, 5 },\r | |
3928 | { (UChar* )"graph", ONIGENC_CTYPE_GRAPH, 5 },\r | |
3929 | { (UChar* )"lower", ONIGENC_CTYPE_LOWER, 5 },\r | |
3930 | { (UChar* )"print", ONIGENC_CTYPE_PRINT, 5 },\r | |
3931 | { (UChar* )"punct", ONIGENC_CTYPE_PUNCT, 5 },\r | |
3932 | { (UChar* )"space", ONIGENC_CTYPE_SPACE, 5 },\r | |
3933 | { (UChar* )"upper", ONIGENC_CTYPE_UPPER, 5 },\r | |
3934 | { (UChar* )"xdigit", ONIGENC_CTYPE_XDIGIT, 6 },\r | |
3935 | { (UChar* )"ascii", ONIGENC_CTYPE_ASCII, 5 },\r | |
3936 | { (UChar* )"word", ONIGENC_CTYPE_WORD, 4 },\r | |
3937 | { (UChar* )NULL, -1, 0 }\r | |
3938 | };\r | |
3939 | \r | |
3940 | PosixBracketEntryType *pb;\r | |
3941 | int not, i, r;\r | |
3942 | OnigCodePoint c;\r | |
3943 | OnigEncoding enc = env->enc;\r | |
3944 | UChar *p = *src;\r | |
3945 | \r | |
3946 | if (PPEEK_IS('^')) {\r | |
3947 | PINC_S;\r | |
3948 | not = 1;\r | |
3949 | }\r | |
3950 | else\r | |
3951 | not = 0;\r | |
3952 | \r | |
3953 | if (onigenc_strlen(enc, p, end) < POSIX_BRACKET_NAME_MIN_LEN + 3)\r | |
3954 | goto not_posix_bracket;\r | |
3955 | \r | |
3956 | for (pb = PBS; IS_NOT_NULL(pb->name); pb++) {\r | |
3957 | if (onigenc_with_ascii_strncmp(enc, p, end, pb->name, pb->len) == 0) {\r | |
3958 | p = (UChar* )onigenc_step(enc, p, end, pb->len);\r | |
3959 | if (onigenc_with_ascii_strncmp(enc, p, end, (UChar* )":]", 2) != 0)\r | |
3960 | return ONIGERR_INVALID_POSIX_BRACKET_TYPE;\r | |
3961 | \r | |
3962 | r = add_ctype_to_cc(cc, pb->ctype, not, env);\r | |
3963 | if (r != 0) return r;\r | |
3964 | \r | |
3965 | PINC_S; PINC_S;\r | |
3966 | *src = p;\r | |
3967 | return 0;\r | |
3968 | }\r | |
3969 | }\r | |
3970 | \r | |
3971 | not_posix_bracket:\r | |
3972 | c = 0;\r | |
3973 | i = 0;\r | |
3974 | while (!PEND && ((c = PPEEK) != ':') && c != ']') {\r | |
3975 | PINC_S;\r | |
3976 | if (++i > POSIX_BRACKET_CHECK_LIMIT_LENGTH) break;\r | |
3977 | }\r | |
3978 | if (c == ':' && ! PEND) {\r | |
3979 | PINC_S;\r | |
3980 | if (! PEND) {\r | |
3981 | PFETCH_S(c);\r | |
3982 | if (c == ']')\r | |
3983 | return ONIGERR_INVALID_POSIX_BRACKET_TYPE;\r | |
3984 | }\r | |
3985 | }\r | |
3986 | \r | |
3987 | return 1; /* 1: is not POSIX bracket, but no error. */\r | |
3988 | }\r | |
3989 | \r | |
3990 | static int\r | |
3991 | fetch_char_property_to_ctype(UChar** src, UChar* end, ScanEnv* env)\r | |
3992 | {\r | |
3993 | int r;\r | |
3994 | OnigCodePoint c;\r | |
3995 | OnigEncoding enc = env->enc;\r | |
3996 | UChar *prev, *start, *p = *src;\r | |
3997 | \r | |
3998 | r = 0;\r | |
3999 | start = prev = p;\r | |
4000 | \r | |
4001 | while (!PEND) {\r | |
4002 | prev = p;\r | |
4003 | PFETCH_S(c);\r | |
4004 | if (c == '}') {\r | |
4005 | r = ONIGENC_PROPERTY_NAME_TO_CTYPE(enc, start, prev);\r | |
4006 | if (r < 0) break;\r | |
4007 | \r | |
4008 | *src = p;\r | |
4009 | return r;\r | |
4010 | }\r | |
4011 | else if (c == '(' || c == ')' || c == '{' || c == '|') {\r | |
4012 | r = ONIGERR_INVALID_CHAR_PROPERTY_NAME;\r | |
4013 | break;\r | |
4014 | }\r | |
4015 | }\r | |
4016 | \r | |
4017 | onig_scan_env_set_error_string(env, r, *src, prev);\r | |
4018 | return r;\r | |
4019 | }\r | |
4020 | \r | |
4021 | static int\r | |
4022 | parse_char_property(Node** np, OnigToken* tok, UChar** src, UChar* end,\r | |
4023 | ScanEnv* env)\r | |
4024 | {\r | |
4025 | int r, ctype;\r | |
4026 | CClassNode* cc;\r | |
4027 | \r | |
4028 | ctype = fetch_char_property_to_ctype(src, end, env);\r | |
4029 | if (ctype < 0) return ctype;\r | |
4030 | \r | |
4031 | *np = node_new_cclass();\r | |
4032 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
4033 | cc = NCCLASS(*np);\r | |
4034 | r = add_ctype_to_cc(cc, ctype, 0, env);\r | |
4035 | if (r != 0) return r;\r | |
4036 | if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);\r | |
4037 | \r | |
4038 | return 0;\r | |
4039 | }\r | |
4040 | \r | |
4041 | \r | |
4042 | enum CCSTATE {\r | |
4043 | CCS_VALUE,\r | |
4044 | CCS_RANGE,\r | |
4045 | CCS_COMPLETE,\r | |
4046 | CCS_START\r | |
4047 | };\r | |
4048 | \r | |
4049 | enum CCVALTYPE {\r | |
4050 | CCV_SB,\r | |
4051 | CCV_CODE_POINT,\r | |
4052 | CCV_CLASS\r | |
4053 | };\r | |
4054 | \r | |
4055 | static int\r | |
4056 | next_state_class(CClassNode* cc, OnigCodePoint* vs, enum CCVALTYPE* type,\r | |
4057 | enum CCSTATE* state, ScanEnv* env)\r | |
4058 | {\r | |
4059 | int r;\r | |
4060 | \r | |
4061 | if (*state == CCS_RANGE)\r | |
4062 | return ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE;\r | |
4063 | \r | |
4064 | if (*state == CCS_VALUE && *type != CCV_CLASS) {\r | |
4065 | if (*type == CCV_SB)\r | |
4066 | BITSET_SET_BIT(cc->bs, (int )(*vs));\r | |
4067 | else if (*type == CCV_CODE_POINT) {\r | |
4068 | r = add_code_range(&(cc->mbuf), env, *vs, *vs);\r | |
4069 | if (r < 0) return r;\r | |
4070 | }\r | |
4071 | }\r | |
4072 | \r | |
4073 | *state = CCS_VALUE;\r | |
4074 | *type = CCV_CLASS;\r | |
4075 | return 0;\r | |
4076 | }\r | |
4077 | \r | |
4078 | static int\r | |
4079 | next_state_val(CClassNode* cc, OnigCodePoint *vs, OnigCodePoint v,\r | |
4080 | int* vs_israw, int v_israw,\r | |
4081 | enum CCVALTYPE intype, enum CCVALTYPE* type,\r | |
4082 | enum CCSTATE* state, ScanEnv* env)\r | |
4083 | {\r | |
4084 | int r;\r | |
4085 | \r | |
4086 | switch (*state) {\r | |
4087 | case CCS_VALUE:\r | |
4088 | if (*type == CCV_SB)\r | |
4089 | BITSET_SET_BIT(cc->bs, (int )(*vs));\r | |
4090 | else if (*type == CCV_CODE_POINT) {\r | |
4091 | r = add_code_range(&(cc->mbuf), env, *vs, *vs);\r | |
4092 | if (r < 0) return r;\r | |
4093 | }\r | |
4094 | break;\r | |
4095 | \r | |
4096 | case CCS_RANGE:\r | |
4097 | if (intype == *type) {\r | |
4098 | if (intype == CCV_SB) {\r | |
4099 | if (*vs > 0xff || v > 0xff)\r | |
4100 | return ONIGERR_INVALID_CODE_POINT_VALUE;\r | |
4101 | \r | |
4102 | if (*vs > v) {\r | |
4103 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))\r | |
4104 | goto ccs_range_end;\r | |
4105 | else\r | |
4106 | return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;\r | |
4107 | }\r | |
4108 | bitset_set_range(cc->bs, (int )*vs, (int )v);\r | |
4109 | }\r | |
4110 | else {\r | |
4111 | r = add_code_range(&(cc->mbuf), env, *vs, v);\r | |
4112 | if (r < 0) return r;\r | |
4113 | }\r | |
4114 | }\r | |
4115 | else {\r | |
4116 | #if 0\r | |
4117 | if (intype == CCV_CODE_POINT && *type == CCV_SB) {\r | |
4118 | #endif\r | |
4119 | if (*vs > v) {\r | |
4120 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))\r | |
4121 | goto ccs_range_end;\r | |
4122 | else\r | |
4123 | return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;\r | |
4124 | }\r | |
4125 | bitset_set_range(cc->bs, (int )*vs, (int )(v < 0xff ? v : 0xff));\r | |
4126 | r = add_code_range(&(cc->mbuf), env, (OnigCodePoint )*vs, v);\r | |
4127 | if (r < 0) return r;\r | |
4128 | #if 0\r | |
4129 | }\r | |
4130 | else\r | |
4131 | return ONIGERR_MISMATCH_CODE_LENGTH_IN_CLASS_RANGE;\r | |
4132 | #endif\r | |
4133 | }\r | |
4134 | ccs_range_end:\r | |
4135 | *state = CCS_COMPLETE;\r | |
4136 | break;\r | |
4137 | \r | |
4138 | case CCS_COMPLETE:\r | |
4139 | case CCS_START:\r | |
4140 | *state = CCS_VALUE;\r | |
4141 | break;\r | |
4142 | \r | |
4143 | default:\r | |
4144 | break;\r | |
4145 | }\r | |
4146 | \r | |
4147 | *vs_israw = v_israw;\r | |
4148 | *vs = v;\r | |
4149 | *type = intype;\r | |
4150 | return 0;\r | |
4151 | }\r | |
4152 | \r | |
4153 | static int\r | |
4154 | code_exist_check(OnigCodePoint c, UChar* from, UChar* end, int ignore_escaped,\r | |
4155 | ScanEnv* env)\r | |
4156 | {\r | |
4157 | int in_esc;\r | |
4158 | OnigCodePoint code;\r | |
4159 | OnigEncoding enc = env->enc;\r | |
4160 | UChar* p = from;\r | |
4161 | \r | |
4162 | in_esc = 0;\r | |
4163 | while (! PEND) {\r | |
4164 | if (ignore_escaped && in_esc) {\r | |
4165 | in_esc = 0;\r | |
4166 | }\r | |
4167 | else {\r | |
4168 | PFETCH_S(code);\r | |
4169 | if (code == c) return 1;\r | |
4170 | if (code == MC_ESC(env->syntax)) in_esc = 1;\r | |
4171 | }\r | |
4172 | }\r | |
4173 | return 0;\r | |
4174 | }\r | |
4175 | \r | |
4176 | static int\r | |
4177 | parse_char_class(Node** np, OnigToken* tok, UChar** src, UChar* end,\r | |
4178 | ScanEnv* env)\r | |
4179 | {\r | |
4180 | int r, neg, len, fetched, and_start;\r | |
4181 | OnigCodePoint v, vs;\r | |
4182 | UChar *p;\r | |
4183 | Node* node;\r | |
4184 | CClassNode *cc, *prev_cc;\r | |
4185 | CClassNode work_cc;\r | |
4186 | \r | |
4187 | enum CCSTATE state;\r | |
4188 | enum CCVALTYPE val_type, in_type;\r | |
4189 | int val_israw, in_israw;\r | |
4190 | \r | |
4191 | prev_cc = (CClassNode* )NULL;\r | |
4192 | *np = NULL_NODE;\r | |
4193 | r = fetch_token_in_cc(tok, src, end, env);\r | |
4194 | if (r == TK_CHAR && tok->u.c == '^' && tok->escaped == 0) {\r | |
4195 | neg = 1;\r | |
4196 | r = fetch_token_in_cc(tok, src, end, env);\r | |
4197 | }\r | |
4198 | else {\r | |
4199 | neg = 0;\r | |
4200 | }\r | |
4201 | \r | |
4202 | if (r < 0) return r;\r | |
4203 | if (r == TK_CC_CLOSE) {\r | |
4204 | if (! code_exist_check((OnigCodePoint )']',\r | |
4205 | *src, env->pattern_end, 1, env))\r | |
4206 | return ONIGERR_EMPTY_CHAR_CLASS;\r | |
4207 | \r | |
4208 | CC_ESC_WARN(env, (UChar* )"]");\r | |
4209 | r = tok->type = TK_CHAR; /* allow []...] */\r | |
4210 | }\r | |
4211 | \r | |
4212 | *np = node = node_new_cclass();\r | |
4213 | CHECK_NULL_RETURN_MEMERR(node);\r | |
4214 | cc = NCCLASS(node);\r | |
4215 | \r | |
4216 | and_start = 0;\r | |
4217 | state = CCS_START;\r | |
4218 | p = *src;\r | |
4219 | while (r != TK_CC_CLOSE) {\r | |
4220 | fetched = 0;\r | |
4221 | switch (r) {\r | |
4222 | case TK_CHAR:\r | |
4223 | len = ONIGENC_CODE_TO_MBCLEN(env->enc, tok->u.c);\r | |
4224 | if (len > 1) {\r | |
4225 | in_type = CCV_CODE_POINT;\r | |
4226 | }\r | |
4227 | else if (len < 0) {\r | |
4228 | r = len;\r | |
4229 | goto err;\r | |
4230 | }\r | |
4231 | else {\r | |
4232 | sb_char:\r | |
4233 | in_type = CCV_SB;\r | |
4234 | }\r | |
4235 | v = (OnigCodePoint )tok->u.c;\r | |
4236 | in_israw = 0;\r | |
4237 | goto val_entry2;\r | |
4238 | break;\r | |
4239 | \r | |
4240 | case TK_RAW_BYTE:\r | |
4241 | /* tok->base != 0 : octal or hexadec. */\r | |
4242 | if (! ONIGENC_IS_SINGLEBYTE(env->enc) && tok->base != 0) {\r | |
4243 | UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];\r | |
4244 | UChar* bufe = buf + ONIGENC_CODE_TO_MBC_MAXLEN;\r | |
4245 | UChar* psave = p;\r | |
4246 | int i, base = tok->base;\r | |
4247 | \r | |
4248 | buf[0] = (UChar)tok->u.c;\r | |
4249 | for (i = 1; i < ONIGENC_MBC_MAXLEN(env->enc); i++) {\r | |
4250 | r = fetch_token_in_cc(tok, &p, end, env);\r | |
4251 | if (r < 0) goto err;\r | |
4252 | if (r != TK_RAW_BYTE || tok->base != base) {\r | |
4253 | fetched = 1;\r | |
4254 | break;\r | |
4255 | }\r | |
4256 | buf[i] = (UChar)tok->u.c;\r | |
4257 | }\r | |
4258 | \r | |
4259 | if (i < ONIGENC_MBC_MINLEN(env->enc)) {\r | |
4260 | r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;\r | |
4261 | goto err;\r | |
4262 | }\r | |
4263 | \r | |
4264 | len = enclen(env->enc, buf);\r | |
4265 | if (i < len) {\r | |
4266 | r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;\r | |
4267 | goto err;\r | |
4268 | }\r | |
4269 | else if (i > len) { /* fetch back */\r | |
4270 | p = psave;\r | |
4271 | for (i = 1; i < len; i++) {\r | |
4272 | r = fetch_token_in_cc(tok, &p, end, env);\r | |
4273 | }\r | |
4274 | fetched = 0;\r | |
4275 | }\r | |
4276 | \r | |
4277 | if (i == 1) {\r | |
4278 | v = (OnigCodePoint )buf[0];\r | |
4279 | goto raw_single;\r | |
4280 | }\r | |
4281 | else {\r | |
4282 | v = ONIGENC_MBC_TO_CODE(env->enc, buf, bufe);\r | |
4283 | in_type = CCV_CODE_POINT;\r | |
4284 | }\r | |
4285 | }\r | |
4286 | else {\r | |
4287 | v = (OnigCodePoint )tok->u.c;\r | |
4288 | raw_single:\r | |
4289 | in_type = CCV_SB;\r | |
4290 | }\r | |
4291 | in_israw = 1;\r | |
4292 | goto val_entry2;\r | |
4293 | break;\r | |
4294 | \r | |
4295 | case TK_CODE_POINT:\r | |
4296 | v = tok->u.code;\r | |
4297 | in_israw = 1;\r | |
4298 | val_entry:\r | |
4299 | len = ONIGENC_CODE_TO_MBCLEN(env->enc, v);\r | |
4300 | if (len < 0) {\r | |
4301 | r = len;\r | |
4302 | goto err;\r | |
4303 | }\r | |
4304 | in_type = (len == 1 ? CCV_SB : CCV_CODE_POINT);\r | |
4305 | val_entry2:\r | |
4306 | r = next_state_val(cc, &vs, v, &val_israw, in_israw, in_type, &val_type,\r | |
4307 | &state, env);\r | |
4308 | if (r != 0) goto err;\r | |
4309 | break;\r | |
4310 | \r | |
4311 | case TK_POSIX_BRACKET_OPEN:\r | |
4312 | r = parse_posix_bracket(cc, &p, end, env);\r | |
4313 | if (r < 0) goto err;\r | |
4314 | if (r == 1) { /* is not POSIX bracket */\r | |
4315 | CC_ESC_WARN(env, (UChar* )"[");\r | |
4316 | p = tok->backp;\r | |
4317 | v = (OnigCodePoint )tok->u.c;\r | |
4318 | in_israw = 0;\r | |
4319 | goto val_entry;\r | |
4320 | }\r | |
4321 | goto next_class;\r | |
4322 | break;\r | |
4323 | \r | |
4324 | case TK_CHAR_TYPE:\r | |
4325 | r = add_ctype_to_cc(cc, tok->u.prop.ctype, tok->u.prop.not, env);\r | |
4326 | if (r != 0) return r;\r | |
4327 | \r | |
4328 | next_class:\r | |
4329 | r = next_state_class(cc, &vs, &val_type, &state, env);\r | |
4330 | if (r != 0) goto err;\r | |
4331 | break;\r | |
4332 | \r | |
4333 | case TK_CHAR_PROPERTY:\r | |
4334 | {\r | |
4335 | int ctype;\r | |
4336 | \r | |
4337 | ctype = fetch_char_property_to_ctype(&p, end, env);\r | |
4338 | if (ctype < 0) return ctype;\r | |
4339 | r = add_ctype_to_cc(cc, ctype, tok->u.prop.not, env);\r | |
4340 | if (r != 0) return r;\r | |
4341 | goto next_class;\r | |
4342 | }\r | |
4343 | break;\r | |
4344 | \r | |
4345 | case TK_CC_RANGE:\r | |
4346 | if (state == CCS_VALUE) {\r | |
4347 | r = fetch_token_in_cc(tok, &p, end, env);\r | |
4348 | if (r < 0) goto err;\r | |
4349 | fetched = 1;\r | |
4350 | if (r == TK_CC_CLOSE) { /* allow [x-] */\r | |
4351 | range_end_val:\r | |
4352 | v = (OnigCodePoint )'-';\r | |
4353 | in_israw = 0;\r | |
4354 | goto val_entry;\r | |
4355 | }\r | |
4356 | else if (r == TK_CC_AND) {\r | |
4357 | CC_ESC_WARN(env, (UChar* )"-");\r | |
4358 | goto range_end_val;\r | |
4359 | }\r | |
4360 | state = CCS_RANGE;\r | |
4361 | }\r | |
4362 | else if (state == CCS_START) {\r | |
4363 | /* [-xa] is allowed */\r | |
4364 | v = (OnigCodePoint )tok->u.c;\r | |
4365 | in_israw = 0;\r | |
4366 | \r | |
4367 | r = fetch_token_in_cc(tok, &p, end, env);\r | |
4368 | if (r < 0) goto err;\r | |
4369 | fetched = 1;\r | |
4370 | /* [--x] or [a&&-x] is warned. */\r | |
4371 | if (r == TK_CC_RANGE || and_start != 0)\r | |
4372 | CC_ESC_WARN(env, (UChar* )"-");\r | |
4373 | \r | |
4374 | goto val_entry;\r | |
4375 | }\r | |
4376 | else if (state == CCS_RANGE) {\r | |
4377 | CC_ESC_WARN(env, (UChar* )"-");\r | |
4378 | goto sb_char; /* [!--x] is allowed */\r | |
4379 | }\r | |
4380 | else { /* CCS_COMPLETE */\r | |
4381 | r = fetch_token_in_cc(tok, &p, end, env);\r | |
4382 | if (r < 0) goto err;\r | |
4383 | fetched = 1;\r | |
4384 | if (r == TK_CC_CLOSE) goto range_end_val; /* allow [a-b-] */\r | |
4385 | else if (r == TK_CC_AND) {\r | |
4386 | CC_ESC_WARN(env, (UChar* )"-");\r | |
4387 | goto range_end_val;\r | |
4388 | }\r | |
4389 | \r | |
4390 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC)) {\r | |
4391 | CC_ESC_WARN(env, (UChar* )"-");\r | |
4392 | goto sb_char; /* [0-9-a] is allowed as [0-9\-a] */\r | |
4393 | }\r | |
4394 | r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS;\r | |
4395 | goto err;\r | |
4396 | }\r | |
4397 | break;\r | |
4398 | \r | |
4399 | case TK_CC_CC_OPEN: /* [ */\r | |
4400 | {\r | |
4401 | Node *anode;\r | |
4402 | CClassNode* acc;\r | |
4403 | \r | |
4404 | r = parse_char_class(&anode, tok, &p, end, env);\r | |
4405 | if (r != 0) goto cc_open_err;\r | |
4406 | acc = NCCLASS(anode);\r | |
4407 | r = or_cclass(cc, acc, env->enc);\r | |
4408 | \r | |
4409 | onig_node_free(anode);\r | |
4410 | cc_open_err:\r | |
4411 | if (r != 0) goto err;\r | |
4412 | }\r | |
4413 | break;\r | |
4414 | \r | |
4415 | case TK_CC_AND: /* && */\r | |
4416 | {\r | |
4417 | if (state == CCS_VALUE) {\r | |
4418 | r = next_state_val(cc, &vs, 0, &val_israw, 0, val_type,\r | |
4419 | &val_type, &state, env);\r | |
4420 | if (r != 0) goto err;\r | |
4421 | }\r | |
4422 | /* initialize local variables */\r | |
4423 | and_start = 1;\r | |
4424 | state = CCS_START;\r | |
4425 | \r | |
4426 | if (IS_NOT_NULL(prev_cc)) {\r | |
4427 | r = and_cclass(prev_cc, cc, env->enc);\r | |
4428 | if (r != 0) goto err;\r | |
4429 | bbuf_free(cc->mbuf);\r | |
4430 | }\r | |
4431 | else {\r | |
4432 | prev_cc = cc;\r | |
4433 | cc = &work_cc;\r | |
4434 | }\r | |
4435 | initialize_cclass(cc);\r | |
4436 | }\r | |
4437 | break;\r | |
4438 | \r | |
4439 | case TK_EOT:\r | |
4440 | r = ONIGERR_PREMATURE_END_OF_CHAR_CLASS;\r | |
4441 | goto err;\r | |
4442 | break;\r | |
4443 | default:\r | |
4444 | r = ONIGERR_PARSER_BUG;\r | |
4445 | goto err;\r | |
4446 | break;\r | |
4447 | }\r | |
4448 | \r | |
4449 | if (fetched)\r | |
4450 | r = tok->type;\r | |
4451 | else {\r | |
4452 | r = fetch_token_in_cc(tok, &p, end, env);\r | |
4453 | if (r < 0) goto err;\r | |
4454 | }\r | |
4455 | }\r | |
4456 | \r | |
4457 | if (state == CCS_VALUE) {\r | |
4458 | r = next_state_val(cc, &vs, 0, &val_israw, 0, val_type,\r | |
4459 | &val_type, &state, env);\r | |
4460 | if (r != 0) goto err;\r | |
4461 | }\r | |
4462 | \r | |
4463 | if (IS_NOT_NULL(prev_cc)) {\r | |
4464 | r = and_cclass(prev_cc, cc, env->enc);\r | |
4465 | if (r != 0) goto err;\r | |
4466 | bbuf_free(cc->mbuf);\r | |
4467 | cc = prev_cc;\r | |
4468 | }\r | |
4469 | \r | |
4470 | if (neg != 0)\r | |
4471 | NCCLASS_SET_NOT(cc);\r | |
4472 | else\r | |
4473 | NCCLASS_CLEAR_NOT(cc);\r | |
4474 | if (IS_NCCLASS_NOT(cc) &&\r | |
4475 | IS_SYNTAX_BV(env->syntax, ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC)) {\r | |
4476 | int is_empty;\r | |
4477 | \r | |
4478 | is_empty = (IS_NULL(cc->mbuf) ? 1 : 0);\r | |
4479 | if (is_empty != 0)\r | |
4480 | BITSET_IS_EMPTY(cc->bs, is_empty);\r | |
4481 | \r | |
4482 | if (is_empty == 0) {\r | |
4483 | #define NEWLINE_CODE 0x0a\r | |
4484 | \r | |
4485 | if (ONIGENC_IS_CODE_NEWLINE(env->enc, NEWLINE_CODE)) {\r | |
4486 | if (ONIGENC_CODE_TO_MBCLEN(env->enc, NEWLINE_CODE) == 1)\r | |
4487 | BITSET_SET_BIT(cc->bs, NEWLINE_CODE);\r | |
4488 | else\r | |
4489 | add_code_range(&(cc->mbuf), env, NEWLINE_CODE, NEWLINE_CODE);\r | |
4490 | }\r | |
4491 | }\r | |
4492 | }\r | |
4493 | *src = p;\r | |
4494 | return 0;\r | |
4495 | \r | |
4496 | err:\r | |
4497 | if (cc != NCCLASS(*np))\r | |
4498 | bbuf_free(cc->mbuf);\r | |
4499 | onig_node_free(*np);\r | |
4500 | return r;\r | |
4501 | }\r | |
4502 | \r | |
4503 | static int parse_subexp(Node** top, OnigToken* tok, int term,\r | |
4504 | UChar** src, UChar* end, ScanEnv* env);\r | |
4505 | \r | |
4506 | static int\r | |
4507 | parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end,\r | |
4508 | ScanEnv* env)\r | |
4509 | {\r | |
4510 | int r, num;\r | |
4511 | Node *target;\r | |
4512 | OnigOptionType option;\r | |
4513 | OnigCodePoint c;\r | |
4514 | OnigEncoding enc = env->enc;\r | |
4515 | \r | |
4516 | #ifdef USE_NAMED_GROUP\r | |
4517 | int list_capture;\r | |
4518 | #endif\r | |
4519 | \r | |
4520 | UChar* p = *src;\r | |
4521 | PFETCH_READY;\r | |
4522 | \r | |
4523 | *np = NULL;\r | |
4524 | if (PEND) return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;\r | |
4525 | \r | |
4526 | option = env->option;\r | |
4527 | if (PPEEK_IS('?') &&\r | |
4528 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) {\r | |
4529 | PINC;\r | |
4530 | if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;\r | |
4531 | \r | |
4532 | PFETCH(c);\r | |
4533 | switch (c) {\r | |
4534 | case ':': /* (?:...) grouping only */\r | |
4535 | group:\r | |
4536 | r = fetch_token(tok, &p, end, env);\r | |
4537 | if (r < 0) return r;\r | |
4538 | r = parse_subexp(np, tok, term, &p, end, env);\r | |
4539 | if (r < 0) return r;\r | |
4540 | *src = p;\r | |
4541 | return 1; /* group */\r | |
4542 | break;\r | |
4543 | \r | |
4544 | case '=':\r | |
4545 | *np = onig_node_new_anchor(ANCHOR_PREC_READ);\r | |
4546 | break;\r | |
4547 | case '!': /* preceding read */\r | |
4548 | *np = onig_node_new_anchor(ANCHOR_PREC_READ_NOT);\r | |
4549 | break;\r | |
4550 | case '>': /* (?>...) stop backtrack */\r | |
4551 | *np = node_new_enclose(ENCLOSE_STOP_BACKTRACK);\r | |
4552 | break;\r | |
4553 | \r | |
4554 | #ifdef USE_NAMED_GROUP\r | |
4555 | case '\'':\r | |
4556 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {\r | |
4557 | goto named_group1;\r | |
4558 | }\r | |
4559 | else\r | |
4560 | return ONIGERR_UNDEFINED_GROUP_OPTION;\r | |
4561 | break;\r | |
4562 | #endif\r | |
4563 | \r | |
4564 | case '<': /* look behind (?<=...), (?<!...) */\r | |
4565 | PFETCH(c);\r | |
4566 | if (c == '=')\r | |
4567 | *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND);\r | |
4568 | else if (c == '!')\r | |
4569 | *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND_NOT);\r | |
4570 | #ifdef USE_NAMED_GROUP\r | |
4571 | else {\r | |
4572 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {\r | |
4573 | UChar *name;\r | |
4574 | UChar *name_end;\r | |
4575 | \r | |
4576 | PUNFETCH;\r | |
4577 | c = '<';\r | |
4578 | \r | |
4579 | named_group1:\r | |
4580 | list_capture = 0;\r | |
4581 | \r | |
4582 | named_group2:\r | |
4583 | name = p;\r | |
4584 | r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &num, 0);\r | |
4585 | if (r < 0) return r;\r | |
4586 | \r | |
4587 | num = scan_env_add_mem_entry(env);\r | |
4588 | if (num < 0) return num;\r | |
4589 | if (list_capture != 0 && num >= (int )BIT_STATUS_BITS_NUM)\r | |
4590 | return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;\r | |
4591 | \r | |
4592 | r = name_add(env->reg, name, name_end, num, env);\r | |
4593 | if (r != 0) return r;\r | |
4594 | *np = node_new_enclose_memory(env->option, 1);\r | |
4595 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
4596 | NENCLOSE(*np)->regnum = num;\r | |
4597 | if (list_capture != 0)\r | |
4598 | BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num);\r | |
4599 | env->num_named++;\r | |
4600 | }\r | |
4601 | else {\r | |
4602 | return ONIGERR_UNDEFINED_GROUP_OPTION;\r | |
4603 | }\r | |
4604 | }\r | |
4605 | #else\r | |
4606 | else {\r | |
4607 | return ONIGERR_UNDEFINED_GROUP_OPTION;\r | |
4608 | }\r | |
4609 | #endif\r | |
4610 | break;\r | |
4611 | \r | |
4612 | case '@':\r | |
4613 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY)) {\r | |
4614 | #ifdef USE_NAMED_GROUP\r | |
4615 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {\r | |
4616 | PFETCH(c);\r | |
4617 | if (c == '<' || c == '\'') {\r | |
4618 | list_capture = 1;\r | |
4619 | goto named_group2; /* (?@<name>...) */\r | |
4620 | }\r | |
4621 | PUNFETCH;\r | |
4622 | }\r | |
4623 | #endif\r | |
4624 | *np = node_new_enclose_memory(env->option, 0);\r | |
4625 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
4626 | num = scan_env_add_mem_entry(env);\r | |
4627 | if (num < 0) {\r | |
4628 | onig_node_free(*np);\r | |
4629 | return num;\r | |
4630 | }\r | |
4631 | else if (num >= (int )BIT_STATUS_BITS_NUM) {\r | |
4632 | onig_node_free(*np);\r | |
4633 | return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;\r | |
4634 | }\r | |
4635 | NENCLOSE(*np)->regnum = num;\r | |
4636 | BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num);\r | |
4637 | }\r | |
4638 | else {\r | |
4639 | return ONIGERR_UNDEFINED_GROUP_OPTION;\r | |
4640 | }\r | |
4641 | break;\r | |
4642 | \r | |
4643 | #ifdef USE_POSIXLINE_OPTION\r | |
4644 | case 'p':\r | |
4645 | #endif\r | |
4646 | case '-': case 'i': case 'm': case 's': case 'x':\r | |
4647 | {\r | |
4648 | int neg = 0;\r | |
4649 | \r | |
4650 | while (1) {\r | |
4651 | switch (c) {\r | |
4652 | case ':':\r | |
4653 | case ')':\r | |
4654 | break;\r | |
4655 | \r | |
4656 | case '-': neg = 1; break;\r | |
4657 | case 'x': ONOFF(option, ONIG_OPTION_EXTEND, neg); break;\r | |
4658 | case 'i': ONOFF(option, ONIG_OPTION_IGNORECASE, neg); break;\r | |
4659 | case 's':\r | |
4660 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {\r | |
4661 | ONOFF(option, ONIG_OPTION_MULTILINE, neg);\r | |
4662 | }\r | |
4663 | else\r | |
4664 | return ONIGERR_UNDEFINED_GROUP_OPTION;\r | |
4665 | break;\r | |
4666 | \r | |
4667 | case 'm':\r | |
4668 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {\r | |
4669 | ONOFF(option, ONIG_OPTION_SINGLELINE, (neg == 0 ? 1 : 0));\r | |
4670 | }\r | |
4671 | else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) {\r | |
4672 | ONOFF(option, ONIG_OPTION_MULTILINE, neg);\r | |
4673 | }\r | |
4674 | else\r | |
4675 | return ONIGERR_UNDEFINED_GROUP_OPTION;\r | |
4676 | break;\r | |
4677 | #ifdef USE_POSIXLINE_OPTION\r | |
4678 | case 'p':\r | |
4679 | ONOFF(option, ONIG_OPTION_MULTILINE|ONIG_OPTION_SINGLELINE, neg);\r | |
4680 | break;\r | |
4681 | #endif\r | |
4682 | default:\r | |
4683 | return ONIGERR_UNDEFINED_GROUP_OPTION;\r | |
4684 | }\r | |
4685 | \r | |
4686 | if (c == ')') {\r | |
4687 | *np = node_new_option(option);\r | |
4688 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
4689 | *src = p;\r | |
4690 | return 2; /* option only */\r | |
4691 | }\r | |
4692 | else if (c == ':') {\r | |
4693 | OnigOptionType prev = env->option;\r | |
4694 | \r | |
4695 | env->option = option;\r | |
4696 | r = fetch_token(tok, &p, end, env);\r | |
4697 | if (r < 0) return r;\r | |
4698 | r = parse_subexp(&target, tok, term, &p, end, env);\r | |
4699 | env->option = prev;\r | |
4700 | if (r < 0) return r;\r | |
4701 | *np = node_new_option(option);\r | |
4702 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
4703 | NENCLOSE(*np)->target = target;\r | |
4704 | *src = p;\r | |
4705 | return 0;\r | |
4706 | }\r | |
4707 | \r | |
4708 | if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;\r | |
4709 | PFETCH(c);\r | |
4710 | }\r | |
4711 | }\r | |
4712 | break;\r | |
4713 | \r | |
4714 | default:\r | |
4715 | return ONIGERR_UNDEFINED_GROUP_OPTION;\r | |
4716 | }\r | |
4717 | }\r | |
4718 | else {\r | |
4719 | if (ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_DONT_CAPTURE_GROUP))\r | |
4720 | goto group;\r | |
4721 | \r | |
4722 | *np = node_new_enclose_memory(env->option, 0);\r | |
4723 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
4724 | num = scan_env_add_mem_entry(env);\r | |
4725 | if (num < 0) return num;\r | |
4726 | NENCLOSE(*np)->regnum = num;\r | |
4727 | }\r | |
4728 | \r | |
4729 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
4730 | r = fetch_token(tok, &p, end, env);\r | |
4731 | if (r < 0) return r;\r | |
4732 | r = parse_subexp(&target, tok, term, &p, end, env);\r | |
4733 | if (r < 0) return r;\r | |
4734 | \r | |
4735 | if (NTYPE(*np) == NT_ANCHOR)\r | |
4736 | NANCHOR(*np)->target = target;\r | |
4737 | else {\r | |
4738 | NENCLOSE(*np)->target = target;\r | |
4739 | if (NENCLOSE(*np)->type == ENCLOSE_MEMORY) {\r | |
4740 | /* Don't move this to previous of parse_subexp() */\r | |
4741 | r = scan_env_set_mem_node(env, NENCLOSE(*np)->regnum, *np);\r | |
4742 | if (r != 0) return r;\r | |
4743 | }\r | |
4744 | }\r | |
4745 | \r | |
4746 | *src = p;\r | |
4747 | return 0;\r | |
4748 | }\r | |
4749 | \r | |
4750 | static const char* PopularQStr[] = {\r | |
4751 | "?", "*", "+", "??", "*?", "+?"\r | |
4752 | };\r | |
4753 | \r | |
4754 | static const char* ReduceQStr[] = {\r | |
4755 | "", "", "*", "*?", "??", "+ and ??", "+? and ?"\r | |
4756 | };\r | |
4757 | \r | |
4758 | static int\r | |
4759 | set_quantifier(Node* qnode, Node* target, int group, ScanEnv* env)\r | |
4760 | {\r | |
4761 | QtfrNode* qn;\r | |
4762 | \r | |
4763 | qn = NQTFR(qnode);\r | |
4764 | if (qn->lower == 1 && qn->upper == 1) {\r | |
4765 | return 1;\r | |
4766 | }\r | |
4767 | \r | |
4768 | switch (NTYPE(target)) {\r | |
4769 | case NT_STR:\r | |
4770 | if (! group) {\r | |
4771 | StrNode* sn = NSTR(target);\r | |
4772 | if (str_node_can_be_split(sn, env->enc)) {\r | |
4773 | Node* n = str_node_split_last_char(sn, env->enc);\r | |
4774 | if (IS_NOT_NULL(n)) {\r | |
4775 | qn->target = n;\r | |
4776 | return 2;\r | |
4777 | }\r | |
4778 | }\r | |
4779 | }\r | |
4780 | break;\r | |
4781 | \r | |
4782 | case NT_QTFR:\r | |
4783 | { /* check redundant double repeat. */\r | |
4784 | /* verbose warn (?:.?)? etc... but not warn (.?)? etc... */\r | |
4785 | QtfrNode* qnt = NQTFR(target);\r | |
4786 | int nestq_num = popular_quantifier_num(qn);\r | |
4787 | int targetq_num = popular_quantifier_num(qnt);\r | |
4788 | \r | |
4789 | #ifdef USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR\r | |
4790 | if (!IS_QUANTIFIER_BY_NUMBER(qn) && !IS_QUANTIFIER_BY_NUMBER(qnt) &&\r | |
4791 | IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT)) {\r | |
4792 | UChar buf[WARN_BUFSIZE];\r | |
4793 | \r | |
4794 | switch(ReduceTypeTable[targetq_num][nestq_num]) {\r | |
4795 | case RQ_ASIS:\r | |
4796 | break;\r | |
4797 | \r | |
4798 | case RQ_DEL:\r | |
4799 | if (onig_verb_warn != onig_null_warn) {\r | |
4800 | onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,\r | |
4801 | env->pattern, env->pattern_end,\r | |
4802 | (UChar* )"redundant nested repeat operator");\r | |
4803 | (*onig_verb_warn)((char* )buf);\r | |
4804 | }\r | |
4805 | goto warn_exit;\r | |
4806 | break;\r | |
4807 | \r | |
4808 | default:\r | |
4809 | if (onig_verb_warn != onig_null_warn) {\r | |
4810 | onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,\r | |
4811 | env->pattern, env->pattern_end,\r | |
4812 | (UChar* )"nested repeat operator %s and %s was replaced with '%s'",\r | |
4813 | PopularQStr[targetq_num], PopularQStr[nestq_num],\r | |
4814 | ReduceQStr[ReduceTypeTable[targetq_num][nestq_num]]);\r | |
4815 | (*onig_verb_warn)((char* )buf);\r | |
4816 | }\r | |
4817 | goto warn_exit;\r | |
4818 | break;\r | |
4819 | }\r | |
4820 | }\r | |
4821 | \r | |
4822 | warn_exit:\r | |
4823 | #endif\r | |
4824 | if (targetq_num >= 0) {\r | |
4825 | if (nestq_num >= 0) {\r | |
4826 | onig_reduce_nested_quantifier(qnode, target);\r | |
4827 | goto q_exit;\r | |
4828 | }\r | |
4829 | else if (targetq_num == 1 || targetq_num == 2) { /* * or + */\r | |
4830 | /* (?:a*){n,m}, (?:a+){n,m} => (?:a*){n,n}, (?:a+){n,n} */\r | |
4831 | if (! IS_REPEAT_INFINITE(qn->upper) && qn->upper > 1 && qn->greedy) {\r | |
4832 | qn->upper = (qn->lower == 0 ? 1 : qn->lower);\r | |
4833 | }\r | |
4834 | }\r | |
4835 | }\r | |
4836 | }\r | |
4837 | break;\r | |
4838 | \r | |
4839 | default:\r | |
4840 | break;\r | |
4841 | }\r | |
4842 | \r | |
4843 | qn->target = target;\r | |
4844 | q_exit:\r | |
4845 | return 0;\r | |
4846 | }\r | |
4847 | \r | |
4848 | \r | |
4849 | #ifdef USE_SHARED_CCLASS_TABLE\r | |
4850 | \r | |
4851 | #define THRESHOLD_RANGE_NUM_FOR_SHARE_CCLASS 8\r | |
4852 | \r | |
4853 | /* for ctype node hash table */\r | |
4854 | \r | |
4855 | typedef struct {\r | |
4856 | OnigEncoding enc;\r | |
4857 | int not;\r | |
4858 | int type;\r | |
4859 | } type_cclass_key;\r | |
4860 | \r | |
4861 | static int type_cclass_cmp(type_cclass_key* x, type_cclass_key* y)\r | |
4862 | {\r | |
4863 | if (x->type != y->type) return 1;\r | |
4864 | if (x->enc != y->enc) return 1;\r | |
4865 | if (x->not != y->not) return 1;\r | |
4866 | return 0;\r | |
4867 | }\r | |
4868 | \r | |
4869 | static int type_cclass_hash(type_cclass_key* key)\r | |
4870 | {\r | |
4871 | int i, val;\r | |
4872 | UChar *p;\r | |
4873 | \r | |
4874 | val = 0;\r | |
4875 | \r | |
4876 | p = (UChar* )&(key->enc);\r | |
4877 | for (i = 0; i < (int )sizeof(key->enc); i++) {\r | |
4878 | val = val * 997 + (int )*p++;\r | |
4879 | }\r | |
4880 | \r | |
4881 | p = (UChar* )(&key->type);\r | |
4882 | for (i = 0; i < (int )sizeof(key->type); i++) {\r | |
4883 | val = val * 997 + (int )*p++;\r | |
4884 | }\r | |
4885 | \r | |
4886 | val += key->not;\r | |
4887 | return val + (val >> 5);\r | |
4888 | }\r | |
4889 | \r | |
4890 | static struct st_hash_type type_type_cclass_hash = {\r | |
4891 | type_cclass_cmp,\r | |
4892 | type_cclass_hash,\r | |
4893 | };\r | |
4894 | \r | |
4895 | static st_table* OnigTypeCClassTable;\r | |
4896 | \r | |
4897 | \r | |
4898 | static int\r | |
4899 | i_free_shared_class(type_cclass_key* key, Node* node, void* arg ARG_UNUSED)\r | |
4900 | {\r | |
4901 | if (IS_NOT_NULL(node)) {\r | |
4902 | CClassNode* cc = NCCLASS(node);\r | |
4903 | if (IS_NOT_NULL(cc->mbuf)) xfree(cc->mbuf);\r | |
4904 | xfree(node);\r | |
4905 | }\r | |
4906 | \r | |
4907 | if (IS_NOT_NULL(key)) xfree(key);\r | |
4908 | return ST_DELETE;\r | |
4909 | }\r | |
4910 | \r | |
4911 | extern int\r | |
4912 | onig_free_shared_cclass_table(void)\r | |
4913 | {\r | |
4914 | if (IS_NOT_NULL(OnigTypeCClassTable)) {\r | |
4915 | onig_st_foreach(OnigTypeCClassTable, i_free_shared_class, 0);\r | |
4916 | onig_st_free_table(OnigTypeCClassTable);\r | |
4917 | OnigTypeCClassTable = NULL;\r | |
4918 | }\r | |
4919 | \r | |
4920 | return 0;\r | |
4921 | }\r | |
4922 | \r | |
4923 | #endif /* USE_SHARED_CCLASS_TABLE */\r | |
4924 | \r | |
4925 | \r | |
4926 | #ifndef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS\r | |
4927 | static int\r | |
4928 | clear_not_flag_cclass(CClassNode* cc, OnigEncoding enc)\r | |
4929 | {\r | |
4930 | BBuf *tbuf;\r | |
4931 | int r;\r | |
4932 | \r | |
4933 | if (IS_NCCLASS_NOT(cc)) {\r | |
4934 | bitset_invert(cc->bs);\r | |
4935 | \r | |
4936 | if (! ONIGENC_IS_SINGLEBYTE(enc)) {\r | |
4937 | r = not_code_range_buf(enc, cc->mbuf, &tbuf);\r | |
4938 | if (r != 0) return r;\r | |
4939 | \r | |
4940 | bbuf_free(cc->mbuf);\r | |
4941 | cc->mbuf = tbuf;\r | |
4942 | }\r | |
4943 | \r | |
4944 | NCCLASS_CLEAR_NOT(cc);\r | |
4945 | }\r | |
4946 | \r | |
4947 | return 0;\r | |
4948 | }\r | |
4949 | #endif /* CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS */\r | |
4950 | \r | |
4951 | typedef struct {\r | |
4952 | ScanEnv* env;\r | |
4953 | CClassNode* cc;\r | |
4954 | Node* alt_root;\r | |
4955 | Node** ptail;\r | |
4956 | } IApplyCaseFoldArg;\r | |
4957 | \r | |
4958 | static int\r | |
4959 | i_apply_case_fold(OnigCodePoint from, OnigCodePoint to[],\r | |
4960 | int to_len, void* arg)\r | |
4961 | {\r | |
4962 | IApplyCaseFoldArg* iarg;\r | |
4963 | ScanEnv* env;\r | |
4964 | CClassNode* cc;\r | |
4965 | BitSetRef bs;\r | |
4966 | \r | |
4967 | iarg = (IApplyCaseFoldArg* )arg;\r | |
4968 | env = iarg->env;\r | |
4969 | cc = iarg->cc;\r | |
4970 | bs = cc->bs;\r | |
4971 | \r | |
4972 | if (to_len == 1) {\r | |
4973 | int is_in = onig_is_code_in_cc(env->enc, from, cc);\r | |
4974 | #ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS\r | |
4975 | if ((is_in != 0 && !IS_NCCLASS_NOT(cc)) ||\r | |
4976 | (is_in == 0 && IS_NCCLASS_NOT(cc))) {\r | |
4977 | if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {\r | |
4978 | add_code_range(&(cc->mbuf), env, *to, *to);\r | |
4979 | }\r | |
4980 | else {\r | |
4981 | BITSET_SET_BIT(bs, *to);\r | |
4982 | }\r | |
4983 | }\r | |
4984 | #else\r | |
4985 | if (is_in != 0) {\r | |
4986 | if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {\r | |
4987 | if (IS_NCCLASS_NOT(cc)) clear_not_flag_cclass(cc, env->enc);\r | |
4988 | add_code_range(&(cc->mbuf), env, *to, *to);\r | |
4989 | }\r | |
4990 | else {\r | |
4991 | if (IS_NCCLASS_NOT(cc)) {\r | |
4992 | BITSET_CLEAR_BIT(bs, *to);\r | |
4993 | }\r | |
4994 | else\r | |
4995 | BITSET_SET_BIT(bs, *to);\r | |
4996 | }\r | |
4997 | }\r | |
4998 | #endif /* CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS */\r | |
4999 | }\r | |
5000 | else {\r | |
5001 | int r, i, len;\r | |
5002 | UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];\r | |
5003 | Node *snode = NULL_NODE;\r | |
5004 | \r | |
5005 | if (onig_is_code_in_cc(env->enc, from, cc)\r | |
5006 | #ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS\r | |
5007 | && !IS_NCCLASS_NOT(cc)\r | |
5008 | #endif\r | |
5009 | ) {\r | |
5010 | for (i = 0; i < to_len; i++) {\r | |
5011 | len = ONIGENC_CODE_TO_MBC(env->enc, to[i], buf);\r | |
5012 | if (i == 0) {\r | |
5013 | snode = onig_node_new_str(buf, buf + len);\r | |
5014 | CHECK_NULL_RETURN_MEMERR(snode);\r | |
5015 | \r | |
5016 | /* char-class expanded multi-char only\r | |
5017 | compare with string folded at match time. */\r | |
5018 | NSTRING_SET_AMBIG(snode);\r | |
5019 | }\r | |
5020 | else {\r | |
5021 | r = onig_node_str_cat(snode, buf, buf + len);\r | |
5022 | if (r < 0) {\r | |
5023 | onig_node_free(snode);\r | |
5024 | return r;\r | |
5025 | }\r | |
5026 | }\r | |
5027 | }\r | |
5028 | \r | |
5029 | *(iarg->ptail) = onig_node_new_alt(snode, NULL_NODE);\r | |
5030 | CHECK_NULL_RETURN_MEMERR(*(iarg->ptail));\r | |
5031 | iarg->ptail = &(NCDR((*(iarg->ptail))));\r | |
5032 | }\r | |
5033 | }\r | |
5034 | \r | |
5035 | return 0;\r | |
5036 | }\r | |
5037 | \r | |
5038 | static int\r | |
5039 | parse_exp(Node** np, OnigToken* tok, int term,\r | |
5040 | UChar** src, UChar* end, ScanEnv* env)\r | |
5041 | {\r | |
5042 | int r, len, group = 0;\r | |
5043 | Node* qn;\r | |
5044 | Node** targetp;\r | |
5045 | \r | |
5046 | *np = NULL;\r | |
5047 | if (tok->type == (enum TokenSyms )term)\r | |
5048 | goto end_of_token;\r | |
5049 | \r | |
5050 | switch (tok->type) {\r | |
5051 | case TK_ALT:\r | |
5052 | case TK_EOT:\r | |
5053 | end_of_token:\r | |
5054 | *np = node_new_empty();\r | |
5055 | return tok->type;\r | |
5056 | break;\r | |
5057 | \r | |
5058 | case TK_SUBEXP_OPEN:\r | |
5059 | r = parse_enclose(np, tok, TK_SUBEXP_CLOSE, src, end, env);\r | |
5060 | if (r < 0) return r;\r | |
5061 | if (r == 1) group = 1;\r | |
5062 | else if (r == 2) { /* option only */\r | |
5063 | Node* target;\r | |
5064 | OnigOptionType prev = env->option;\r | |
5065 | \r | |
5066 | env->option = NENCLOSE(*np)->option;\r | |
5067 | r = fetch_token(tok, src, end, env);\r | |
5068 | if (r < 0) return r;\r | |
5069 | r = parse_subexp(&target, tok, term, src, end, env);\r | |
5070 | env->option = prev;\r | |
5071 | if (r < 0) return r;\r | |
5072 | NENCLOSE(*np)->target = target; \r | |
5073 | return tok->type;\r | |
5074 | }\r | |
5075 | break;\r | |
5076 | \r | |
5077 | case TK_SUBEXP_CLOSE:\r | |
5078 | if (! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP))\r | |
5079 | return ONIGERR_UNMATCHED_CLOSE_PARENTHESIS;\r | |
5080 | \r | |
5081 | if (tok->escaped) goto tk_raw_byte;\r | |
5082 | else goto tk_byte;\r | |
5083 | break;\r | |
5084 | \r | |
5085 | case TK_STRING:\r | |
5086 | tk_byte:\r | |
5087 | {\r | |
5088 | *np = node_new_str(tok->backp, *src);\r | |
5089 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5090 | \r | |
5091 | while (1) {\r | |
5092 | r = fetch_token(tok, src, end, env);\r | |
5093 | if (r < 0) return r;\r | |
5094 | if (r != TK_STRING) break;\r | |
5095 | \r | |
5096 | r = onig_node_str_cat(*np, tok->backp, *src);\r | |
5097 | if (r < 0) return r;\r | |
5098 | }\r | |
5099 | \r | |
5100 | string_end:\r | |
5101 | targetp = np;\r | |
5102 | goto repeat;\r | |
5103 | }\r | |
5104 | break;\r | |
5105 | \r | |
5106 | case TK_RAW_BYTE:\r | |
5107 | tk_raw_byte:\r | |
5108 | {\r | |
5109 | *np = node_new_str_raw_char((UChar )tok->u.c);\r | |
5110 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5111 | len = 1;\r | |
5112 | while (1) {\r | |
5113 | if (len >= ONIGENC_MBC_MINLEN(env->enc)) {\r | |
5114 | if (len == enclen(env->enc, NSTR(*np)->s)) {\r | |
5115 | r = fetch_token(tok, src, end, env);\r | |
5116 | NSTRING_CLEAR_RAW(*np);\r | |
5117 | goto string_end;\r | |
5118 | }\r | |
5119 | }\r | |
5120 | \r | |
5121 | r = fetch_token(tok, src, end, env);\r | |
5122 | if (r < 0) return r;\r | |
5123 | if (r != TK_RAW_BYTE) {\r | |
5124 | /* Don't use this, it is wrong for little endian encodings. */\r | |
5125 | #ifdef USE_PAD_TO_SHORT_BYTE_CHAR\r | |
5126 | int rem;\r | |
5127 | if (len < ONIGENC_MBC_MINLEN(env->enc)) {\r | |
5128 | rem = ONIGENC_MBC_MINLEN(env->enc) - len;\r | |
5129 | (void )node_str_head_pad(NSTR(*np), rem, (UChar )0);\r | |
5130 | if (len + rem == enclen(env->enc, NSTR(*np)->s)) {\r | |
5131 | NSTRING_CLEAR_RAW(*np);\r | |
5132 | goto string_end;\r | |
5133 | }\r | |
5134 | }\r | |
5135 | #endif\r | |
5136 | return ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;\r | |
5137 | }\r | |
5138 | \r | |
5139 | r = node_str_cat_char(*np, (UChar )tok->u.c);\r | |
5140 | if (r < 0) return r;\r | |
5141 | \r | |
5142 | len++;\r | |
5143 | }\r | |
5144 | }\r | |
5145 | break;\r | |
5146 | \r | |
5147 | case TK_CODE_POINT:\r | |
5148 | {\r | |
5149 | UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];\r | |
5150 | int num = ONIGENC_CODE_TO_MBC(env->enc, tok->u.code, buf);\r | |
5151 | if (num < 0) return num;\r | |
5152 | #ifdef NUMBERED_CHAR_IS_NOT_CASE_AMBIG\r | |
5153 | *np = node_new_str_raw(buf, buf + num);\r | |
5154 | #else\r | |
5155 | *np = node_new_str(buf, buf + num);\r | |
5156 | #endif\r | |
5157 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5158 | }\r | |
5159 | break;\r | |
5160 | \r | |
5161 | case TK_QUOTE_OPEN:\r | |
5162 | {\r | |
5163 | OnigCodePoint end_op[2];\r | |
5164 | UChar *qstart, *qend, *nextp;\r | |
5165 | \r | |
5166 | end_op[0] = (OnigCodePoint )MC_ESC(env->syntax);\r | |
5167 | end_op[1] = (OnigCodePoint )'E';\r | |
5168 | qstart = *src;\r | |
5169 | qend = find_str_position(end_op, 2, qstart, end, &nextp, env->enc);\r | |
5170 | if (IS_NULL(qend)) {\r | |
5171 | nextp = qend = end;\r | |
5172 | }\r | |
5173 | *np = node_new_str(qstart, qend);\r | |
5174 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5175 | *src = nextp;\r | |
5176 | }\r | |
5177 | break;\r | |
5178 | \r | |
5179 | case TK_CHAR_TYPE:\r | |
5180 | {\r | |
5181 | switch (tok->u.prop.ctype) {\r | |
5182 | case ONIGENC_CTYPE_WORD:\r | |
5183 | *np = node_new_ctype(tok->u.prop.ctype, tok->u.prop.not);\r | |
5184 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5185 | break;\r | |
5186 | \r | |
5187 | case ONIGENC_CTYPE_SPACE:\r | |
5188 | case ONIGENC_CTYPE_DIGIT:\r | |
5189 | case ONIGENC_CTYPE_XDIGIT:\r | |
5190 | {\r | |
5191 | CClassNode* cc;\r | |
5192 | \r | |
5193 | #ifdef USE_SHARED_CCLASS_TABLE\r | |
5194 | const OnigCodePoint *mbr;\r | |
5195 | OnigCodePoint sb_out;\r | |
5196 | \r | |
5197 | r = ONIGENC_GET_CTYPE_CODE_RANGE(env->enc, tok->u.prop.ctype,\r | |
5198 | &sb_out, &mbr);\r | |
5199 | if (r == 0 &&\r | |
5200 | ONIGENC_CODE_RANGE_NUM(mbr)\r | |
5201 | >= THRESHOLD_RANGE_NUM_FOR_SHARE_CCLASS) {\r | |
5202 | type_cclass_key key;\r | |
5203 | type_cclass_key* new_key;\r | |
5204 | \r | |
5205 | key.enc = env->enc;\r | |
5206 | key.not = tok->u.prop.not;\r | |
5207 | key.type = tok->u.prop.ctype;\r | |
5208 | \r | |
5209 | THREAD_ATOMIC_START;\r | |
5210 | \r | |
5211 | if (IS_NULL(OnigTypeCClassTable)) {\r | |
5212 | OnigTypeCClassTable\r | |
5213 | = onig_st_init_table_with_size(&type_type_cclass_hash, 10);\r | |
5214 | if (IS_NULL(OnigTypeCClassTable)) {\r | |
5215 | THREAD_ATOMIC_END;\r | |
5216 | return ONIGERR_MEMORY;\r | |
5217 | }\r | |
5218 | }\r | |
5219 | else {\r | |
5220 | if (onig_st_lookup(OnigTypeCClassTable, (st_data_t )&key,\r | |
5221 | (st_data_t* )np)) {\r | |
5222 | THREAD_ATOMIC_END;\r | |
5223 | break;\r | |
5224 | }\r | |
5225 | }\r | |
5226 | \r | |
5227 | *np = node_new_cclass_by_codepoint_range(tok->u.prop.not,\r | |
5228 | sb_out, mbr);\r | |
5229 | if (IS_NULL(*np)) {\r | |
5230 | THREAD_ATOMIC_END;\r | |
5231 | return ONIGERR_MEMORY;\r | |
5232 | }\r | |
5233 | \r | |
5234 | cc = NCCLASS(*np);\r | |
5235 | NCCLASS_SET_SHARE(cc);\r | |
5236 | new_key = (type_cclass_key* )xmalloc(sizeof(type_cclass_key));\r | |
5237 | xmemcpy(new_key, &key, sizeof(type_cclass_key));\r | |
5238 | onig_st_add_direct(OnigTypeCClassTable, (st_data_t )new_key,\r | |
5239 | (st_data_t )*np);\r | |
5240 | \r | |
5241 | THREAD_ATOMIC_END;\r | |
5242 | }\r | |
5243 | else {\r | |
5244 | #endif\r | |
5245 | *np = node_new_cclass();\r | |
5246 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5247 | cc = NCCLASS(*np);\r | |
5248 | add_ctype_to_cc(cc, tok->u.prop.ctype, 0, env);\r | |
5249 | if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);\r | |
5250 | #ifdef USE_SHARED_CCLASS_TABLE\r | |
5251 | }\r | |
5252 | #endif\r | |
5253 | }\r | |
5254 | break;\r | |
5255 | \r | |
5256 | default:\r | |
5257 | return ONIGERR_PARSER_BUG;\r | |
5258 | break;\r | |
5259 | }\r | |
5260 | }\r | |
5261 | break;\r | |
5262 | \r | |
5263 | case TK_CHAR_PROPERTY:\r | |
5264 | r = parse_char_property(np, tok, src, end, env);\r | |
5265 | if (r != 0) return r;\r | |
5266 | break;\r | |
5267 | \r | |
5268 | case TK_CC_OPEN:\r | |
5269 | {\r | |
5270 | CClassNode* cc;\r | |
5271 | \r | |
5272 | r = parse_char_class(np, tok, src, end, env);\r | |
5273 | if (r != 0) return r;\r | |
5274 | \r | |
5275 | cc = NCCLASS(*np);\r | |
5276 | if (IS_IGNORECASE(env->option)) {\r | |
5277 | IApplyCaseFoldArg iarg;\r | |
5278 | \r | |
5279 | iarg.env = env;\r | |
5280 | iarg.cc = cc;\r | |
5281 | iarg.alt_root = NULL_NODE;\r | |
5282 | iarg.ptail = &(iarg.alt_root);\r | |
5283 | \r | |
5284 | r = ONIGENC_APPLY_ALL_CASE_FOLD(env->enc, env->case_fold_flag,\r | |
5285 | i_apply_case_fold, &iarg);\r | |
5286 | if (r != 0) {\r | |
5287 | onig_node_free(iarg.alt_root);\r | |
5288 | return r;\r | |
5289 | }\r | |
5290 | if (IS_NOT_NULL(iarg.alt_root)) {\r | |
5291 | Node* work = onig_node_new_alt(*np, iarg.alt_root);\r | |
5292 | if (IS_NULL(work)) {\r | |
5293 | onig_node_free(iarg.alt_root);\r | |
5294 | return ONIGERR_MEMORY;\r | |
5295 | }\r | |
5296 | *np = work;\r | |
5297 | }\r | |
5298 | }\r | |
5299 | }\r | |
5300 | break;\r | |
5301 | \r | |
5302 | case TK_ANYCHAR:\r | |
5303 | *np = node_new_anychar();\r | |
5304 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5305 | break;\r | |
5306 | \r | |
5307 | case TK_ANYCHAR_ANYTIME:\r | |
5308 | *np = node_new_anychar();\r | |
5309 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5310 | qn = node_new_quantifier(0, REPEAT_INFINITE, 0);\r | |
5311 | CHECK_NULL_RETURN_MEMERR(qn);\r | |
5312 | NQTFR(qn)->target = *np;\r | |
5313 | *np = qn;\r | |
5314 | break;\r | |
5315 | \r | |
5316 | case TK_BACKREF:\r | |
5317 | len = tok->u.backref.num;\r | |
5318 | *np = node_new_backref(len,\r | |
5319 | (len > 1 ? tok->u.backref.refs : &(tok->u.backref.ref1)),\r | |
5320 | tok->u.backref.by_name,\r | |
5321 | #ifdef USE_BACKREF_WITH_LEVEL\r | |
5322 | tok->u.backref.exist_level,\r | |
5323 | tok->u.backref.level,\r | |
5324 | #endif\r | |
5325 | env);\r | |
5326 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5327 | break;\r | |
5328 | \r | |
5329 | #ifdef USE_SUBEXP_CALL\r | |
5330 | case TK_CALL:\r | |
5331 | {\r | |
5332 | int gnum = tok->u.call.gnum;\r | |
5333 | \r | |
5334 | if (gnum < 0) {\r | |
5335 | gnum = BACKREF_REL_TO_ABS(gnum, env);\r | |
5336 | if (gnum <= 0)\r | |
5337 | return ONIGERR_INVALID_BACKREF;\r | |
5338 | }\r | |
5339 | *np = node_new_call(tok->u.call.name, tok->u.call.name_end, gnum);\r | |
5340 | CHECK_NULL_RETURN_MEMERR(*np);\r | |
5341 | env->num_call++;\r | |
5342 | }\r | |
5343 | break;\r | |
5344 | #endif\r | |
5345 | \r | |
5346 | case TK_ANCHOR:\r | |
5347 | *np = onig_node_new_anchor(tok->u.anchor);\r | |
5348 | break;\r | |
5349 | \r | |
5350 | case TK_OP_REPEAT:\r | |
5351 | case TK_INTERVAL:\r | |
5352 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS)) {\r | |
5353 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS))\r | |
5354 | return ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED;\r | |
5355 | else\r | |
5356 | *np = node_new_empty();\r | |
5357 | }\r | |
5358 | else {\r | |
5359 | goto tk_byte;\r | |
5360 | }\r | |
5361 | break;\r | |
5362 | \r | |
5363 | default:\r | |
5364 | return ONIGERR_PARSER_BUG;\r | |
5365 | break;\r | |
5366 | }\r | |
5367 | \r | |
5368 | {\r | |
5369 | targetp = np;\r | |
5370 | \r | |
5371 | re_entry:\r | |
5372 | r = fetch_token(tok, src, end, env);\r | |
5373 | if (r < 0) return r;\r | |
5374 | \r | |
5375 | repeat:\r | |
5376 | if (r == TK_OP_REPEAT || r == TK_INTERVAL) {\r | |
5377 | if (is_invalid_quantifier_target(*targetp))\r | |
5378 | return ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID;\r | |
5379 | \r | |
5380 | qn = node_new_quantifier(tok->u.repeat.lower, tok->u.repeat.upper,\r | |
5381 | (r == TK_INTERVAL ? 1 : 0));\r | |
5382 | CHECK_NULL_RETURN_MEMERR(qn);\r | |
5383 | NQTFR(qn)->greedy = tok->u.repeat.greedy;\r | |
5384 | r = set_quantifier(qn, *targetp, group, env);\r | |
5385 | if (r < 0) {\r | |
5386 | onig_node_free(qn);\r | |
5387 | return r;\r | |
5388 | }\r | |
5389 | \r | |
5390 | if (tok->u.repeat.possessive != 0) {\r | |
5391 | Node* en;\r | |
5392 | en = node_new_enclose(ENCLOSE_STOP_BACKTRACK);\r | |
5393 | if (IS_NULL(en)) {\r | |
5394 | onig_node_free(qn);\r | |
5395 | return ONIGERR_MEMORY;\r | |
5396 | }\r | |
5397 | NENCLOSE(en)->target = qn;\r | |
5398 | qn = en;\r | |
5399 | }\r | |
5400 | \r | |
5401 | if (r == 0) {\r | |
5402 | *targetp = qn;\r | |
5403 | }\r | |
5404 | else if (r == 1) {\r | |
5405 | onig_node_free(qn);\r | |
5406 | }\r | |
5407 | else if (r == 2) { /* split case: /abc+/ */\r | |
5408 | Node *tmp;\r | |
5409 | \r | |
5410 | *targetp = node_new_list(*targetp, NULL);\r | |
5411 | if (IS_NULL(*targetp)) {\r | |
5412 | onig_node_free(qn);\r | |
5413 | return ONIGERR_MEMORY;\r | |
5414 | }\r | |
5415 | tmp = NCDR(*targetp) = node_new_list(qn, NULL);\r | |
5416 | if (IS_NULL(tmp)) {\r | |
5417 | onig_node_free(qn);\r | |
5418 | return ONIGERR_MEMORY;\r | |
5419 | }\r | |
5420 | targetp = &(NCAR(tmp));\r | |
5421 | }\r | |
5422 | goto re_entry;\r | |
5423 | }\r | |
5424 | }\r | |
5425 | \r | |
5426 | return r;\r | |
5427 | }\r | |
5428 | \r | |
5429 | static int\r | |
5430 | parse_branch(Node** top, OnigToken* tok, int term,\r | |
5431 | UChar** src, UChar* end, ScanEnv* env)\r | |
5432 | {\r | |
5433 | int r;\r | |
5434 | Node *node, **headp;\r | |
5435 | \r | |
5436 | *top = NULL;\r | |
5437 | r = parse_exp(&node, tok, term, src, end, env);\r | |
5438 | if (r < 0) return r;\r | |
5439 | \r | |
5440 | if (r == TK_EOT || r == term || r == TK_ALT) {\r | |
5441 | *top = node;\r | |
5442 | }\r | |
5443 | else {\r | |
5444 | *top = node_new_list(node, NULL);\r | |
5445 | headp = &(NCDR(*top));\r | |
5446 | while (r != TK_EOT && r != term && r != TK_ALT) {\r | |
5447 | r = parse_exp(&node, tok, term, src, end, env);\r | |
5448 | if (r < 0) return r;\r | |
5449 | \r | |
5450 | if (NTYPE(node) == NT_LIST) {\r | |
5451 | *headp = node;\r | |
5452 | while (IS_NOT_NULL(NCDR(node))) node = NCDR(node);\r | |
5453 | headp = &(NCDR(node));\r | |
5454 | }\r | |
5455 | else {\r | |
5456 | *headp = node_new_list(node, NULL);\r | |
5457 | headp = &(NCDR(*headp));\r | |
5458 | }\r | |
5459 | }\r | |
5460 | }\r | |
5461 | \r | |
5462 | return r;\r | |
5463 | }\r | |
5464 | \r | |
5465 | /* term_tok: TK_EOT or TK_SUBEXP_CLOSE */\r | |
5466 | static int\r | |
5467 | parse_subexp(Node** top, OnigToken* tok, int term,\r | |
5468 | UChar** src, UChar* end, ScanEnv* env)\r | |
5469 | {\r | |
5470 | int r;\r | |
5471 | Node *node, **headp;\r | |
5472 | \r | |
5473 | *top = NULL;\r | |
5474 | r = parse_branch(&node, tok, term, src, end, env);\r | |
5475 | if (r < 0) {\r | |
5476 | onig_node_free(node);\r | |
5477 | return r;\r | |
5478 | }\r | |
5479 | \r | |
5480 | if (r == term) {\r | |
5481 | *top = node;\r | |
5482 | }\r | |
5483 | else if (r == TK_ALT) {\r | |
5484 | *top = onig_node_new_alt(node, NULL);\r | |
5485 | headp = &(NCDR(*top));\r | |
5486 | while (r == TK_ALT) {\r | |
5487 | r = fetch_token(tok, src, end, env);\r | |
5488 | if (r < 0) return r;\r | |
5489 | r = parse_branch(&node, tok, term, src, end, env);\r | |
5490 | if (r < 0) return r;\r | |
5491 | \r | |
5492 | *headp = onig_node_new_alt(node, NULL);\r | |
5493 | headp = &(NCDR(*headp));\r | |
5494 | }\r | |
5495 | \r | |
5496 | if (tok->type != (enum TokenSyms )term)\r | |
5497 | goto err;\r | |
5498 | }\r | |
5499 | else {\r | |
5500 | err:\r | |
5501 | if (term == TK_SUBEXP_CLOSE)\r | |
5502 | return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;\r | |
5503 | else\r | |
5504 | return ONIGERR_PARSER_BUG;\r | |
5505 | }\r | |
5506 | \r | |
5507 | return r;\r | |
5508 | }\r | |
5509 | \r | |
5510 | static int\r | |
5511 | parse_regexp(Node** top, UChar** src, UChar* end, ScanEnv* env)\r | |
5512 | {\r | |
5513 | int r;\r | |
5514 | OnigToken tok;\r | |
5515 | \r | |
5516 | r = fetch_token(&tok, src, end, env);\r | |
5517 | if (r < 0) return r;\r | |
5518 | r = parse_subexp(top, &tok, TK_EOT, src, end, env);\r | |
5519 | if (r < 0) return r;\r | |
5520 | return 0;\r | |
5521 | }\r | |
5522 | \r | |
5523 | extern int\r | |
5524 | onig_parse_make_tree(Node** root, const UChar* pattern, const UChar* end,\r | |
5525 | regex_t* reg, ScanEnv* env)\r | |
5526 | {\r | |
5527 | int r;\r | |
5528 | UChar* p;\r | |
5529 | \r | |
5530 | #ifdef USE_NAMED_GROUP\r | |
5531 | names_clear(reg);\r | |
5532 | #endif\r | |
5533 | \r | |
5534 | scan_env_clear(env);\r | |
5535 | env->option = reg->options;\r | |
5536 | env->case_fold_flag = reg->case_fold_flag;\r | |
5537 | env->enc = reg->enc;\r | |
5538 | env->syntax = reg->syntax;\r | |
5539 | env->pattern = (UChar* )pattern;\r | |
5540 | env->pattern_end = (UChar* )end;\r | |
5541 | env->reg = reg;\r | |
5542 | \r | |
5543 | *root = NULL;\r | |
5544 | p = (UChar* )pattern;\r | |
5545 | r = parse_regexp(root, &p, (UChar* )end, env);\r | |
5546 | reg->num_mem = env->num_mem;\r | |
5547 | return r;\r | |
5548 | }\r | |
5549 | \r | |
5550 | extern void\r | |
5551 | onig_scan_env_set_error_string(ScanEnv* env, int ecode ARG_UNUSED,\r | |
5552 | UChar* arg, UChar* arg_end)\r | |
5553 | {\r | |
5554 | env->error = arg;\r | |
5555 | env->error_end = arg_end;\r | |
5556 | }\r |