1 | /**********************************************************************
|
---|
2 | regparse.c - Onigmo (Oniguruma-mod) (regular expression library)
|
---|
3 | **********************************************************************/
|
---|
4 | /*-
|
---|
5 | * Copyright (c) 2002-2008 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
|
---|
6 | * Copyright (c) 2011-2016 K.Takata <kentkt AT csc DOT jp>
|
---|
7 | * All rights reserved.
|
---|
8 | *
|
---|
9 | * Redistribution and use in source and binary forms, with or without
|
---|
10 | * modification, are permitted provided that the following conditions
|
---|
11 | * are met:
|
---|
12 | * 1. Redistributions of source code must retain the above copyright
|
---|
13 | * notice, this list of conditions and the following disclaimer.
|
---|
14 | * 2. Redistributions in binary form must reproduce the above copyright
|
---|
15 | * notice, this list of conditions and the following disclaimer in the
|
---|
16 | * documentation and/or other materials provided with the distribution.
|
---|
17 | *
|
---|
18 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
---|
19 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
---|
20 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
---|
21 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
---|
22 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
---|
23 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
---|
24 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
---|
25 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
---|
26 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
---|
27 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
---|
28 | * SUCH DAMAGE.
|
---|
29 | */
|
---|
30 |
|
---|
31 | #include "regparse.h"
|
---|
32 | #include <stdarg.h>
|
---|
33 |
|
---|
34 | #define WARN_BUFSIZE 256
|
---|
35 |
|
---|
36 | #define CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
|
---|
37 |
|
---|
38 |
|
---|
39 | const OnigSyntaxType OnigSyntaxRuby = {
|
---|
40 | (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY |
|
---|
41 | ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 |
|
---|
42 | ONIG_SYN_OP_ESC_X_BRACE_HEX8 | ONIG_SYN_OP_ESC_CONTROL_CHARS |
|
---|
43 | ONIG_SYN_OP_ESC_C_CONTROL )
|
---|
44 | & ~ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END )
|
---|
45 | , ( ONIG_SYN_OP2_QMARK_GROUP_EFFECT |
|
---|
46 | ONIG_SYN_OP2_OPTION_RUBY |
|
---|
47 | ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP | ONIG_SYN_OP2_ESC_K_NAMED_BACKREF |
|
---|
48 | ONIG_SYN_OP2_ESC_G_SUBEXP_CALL |
|
---|
49 | ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY |
|
---|
50 | ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT |
|
---|
51 | ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT |
|
---|
52 | ONIG_SYN_OP2_CCLASS_SET_OP | ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL |
|
---|
53 | ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META | ONIG_SYN_OP2_ESC_V_VTAB |
|
---|
54 | ONIG_SYN_OP2_ESC_H_XDIGIT |
|
---|
55 | #ifndef RUBY
|
---|
56 | ONIG_SYN_OP2_ESC_U_HEX4 |
|
---|
57 | #endif
|
---|
58 | ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER |
|
---|
59 | ONIG_SYN_OP2_QMARK_LPAREN_CONDITION |
|
---|
60 | ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK |
|
---|
61 | ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP |
|
---|
62 | ONIG_SYN_OP2_QMARK_TILDE_ABSENT )
|
---|
63 | , ( SYN_GNU_REGEX_BV |
|
---|
64 | ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV |
|
---|
65 | ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND |
|
---|
66 | ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP |
|
---|
67 | ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME |
|
---|
68 | ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY |
|
---|
69 | ONIG_SYN_WARN_CC_OP_NOT_ESCAPED |
|
---|
70 | ONIG_SYN_WARN_CC_DUP |
|
---|
71 | ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT )
|
---|
72 | , ( ONIG_OPTION_ASCII_RANGE | ONIG_OPTION_POSIX_BRACKET_ALL_RANGE |
|
---|
73 | ONIG_OPTION_WORD_BOUND_ALL_RANGE )
|
---|
74 | ,
|
---|
75 | {
|
---|
76 | (OnigCodePoint )'\\' /* esc */
|
---|
77 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar '.' */
|
---|
78 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anytime '*' */
|
---|
79 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* zero or one time '?' */
|
---|
80 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* one or more time '+' */
|
---|
81 | , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR /* anychar anytime */
|
---|
82 | }
|
---|
83 | };
|
---|
84 |
|
---|
85 | const OnigSyntaxType* OnigDefaultSyntax = ONIG_SYNTAX_RUBY;
|
---|
86 |
|
---|
87 | extern void onig_null_warn(const char* s ARG_UNUSED) { }
|
---|
88 |
|
---|
89 | #ifdef DEFAULT_WARN_FUNCTION
|
---|
90 | static OnigWarnFunc onig_warn = (OnigWarnFunc )DEFAULT_WARN_FUNCTION;
|
---|
91 | #else
|
---|
92 | static OnigWarnFunc onig_warn = onig_null_warn;
|
---|
93 | #endif
|
---|
94 |
|
---|
95 | #ifdef DEFAULT_VERB_WARN_FUNCTION
|
---|
96 | static OnigWarnFunc onig_verb_warn = (OnigWarnFunc )DEFAULT_VERB_WARN_FUNCTION;
|
---|
97 | #else
|
---|
98 | static OnigWarnFunc onig_verb_warn = onig_null_warn;
|
---|
99 | #endif
|
---|
100 |
|
---|
101 | extern void onig_set_warn_func(OnigWarnFunc f)
|
---|
102 | {
|
---|
103 | onig_warn = f;
|
---|
104 | }
|
---|
105 |
|
---|
106 | extern void onig_set_verb_warn_func(OnigWarnFunc f)
|
---|
107 | {
|
---|
108 | onig_verb_warn = f;
|
---|
109 | }
|
---|
110 |
|
---|
111 | static void CC_DUP_WARN(ScanEnv *env);
|
---|
112 |
|
---|
113 |
|
---|
114 | static unsigned int ParseDepthLimit = DEFAULT_PARSE_DEPTH_LIMIT;
|
---|
115 |
|
---|
116 | extern unsigned int
|
---|
117 | onig_get_parse_depth_limit(void)
|
---|
118 | {
|
---|
119 | return ParseDepthLimit;
|
---|
120 | }
|
---|
121 |
|
---|
122 | extern int
|
---|
123 | onig_set_parse_depth_limit(unsigned int depth)
|
---|
124 | {
|
---|
125 | if (depth == 0)
|
---|
126 | ParseDepthLimit = DEFAULT_PARSE_DEPTH_LIMIT;
|
---|
127 | else
|
---|
128 | ParseDepthLimit = depth;
|
---|
129 | return 0;
|
---|
130 | }
|
---|
131 |
|
---|
132 |
|
---|
133 | static void
|
---|
134 | bbuf_free(BBuf* bbuf)
|
---|
135 | {
|
---|
136 | if (IS_NOT_NULL(bbuf)) {
|
---|
137 | if (IS_NOT_NULL(bbuf->p)) xfree(bbuf->p);
|
---|
138 | xfree(bbuf);
|
---|
139 | }
|
---|
140 | }
|
---|
141 |
|
---|
142 | static int
|
---|
143 | bbuf_clone(BBuf** rto, BBuf* from)
|
---|
144 | {
|
---|
145 | int r;
|
---|
146 | BBuf *to;
|
---|
147 |
|
---|
148 | *rto = to = (BBuf* )xmalloc(sizeof(BBuf));
|
---|
149 | CHECK_NULL_RETURN_MEMERR(to);
|
---|
150 | r = BBUF_INIT(to, from->alloc);
|
---|
151 | if (r != 0) return r;
|
---|
152 | to->used = from->used;
|
---|
153 | xmemcpy(to->p, from->p, from->used);
|
---|
154 | return 0;
|
---|
155 | }
|
---|
156 |
|
---|
157 | #define BACKREF_REL_TO_ABS(rel_no, env) \
|
---|
158 | ((env)->num_mem + 1 + (rel_no))
|
---|
159 |
|
---|
160 | #define ONOFF(v,f,negative) (negative) ? ((v) &= ~(f)) : ((v) |= (f))
|
---|
161 |
|
---|
162 | #define MBCODE_START_POS(enc) \
|
---|
163 | (OnigCodePoint )(ONIGENC_MBC_MINLEN(enc) > 1 ? 0 : 0x80)
|
---|
164 |
|
---|
165 | #define SET_ALL_MULTI_BYTE_RANGE(enc, pbuf) \
|
---|
166 | add_code_range_to_buf(pbuf, env, MBCODE_START_POS(enc), ONIG_LAST_CODE_POINT)
|
---|
167 |
|
---|
168 | #define ADD_ALL_MULTI_BYTE_RANGE(enc, mbuf) do {\
|
---|
169 | if (! ONIGENC_IS_SINGLEBYTE(enc)) {\
|
---|
170 | r = SET_ALL_MULTI_BYTE_RANGE(enc, &(mbuf));\
|
---|
171 | if (r) return r;\
|
---|
172 | }\
|
---|
173 | } while (0)
|
---|
174 |
|
---|
175 |
|
---|
176 | #define BITSET_SET_BIT_CHKDUP(bs, pos) do { \
|
---|
177 | if (BITSET_AT(bs, pos)) CC_DUP_WARN(env); \
|
---|
178 | BS_ROOM(bs, pos) |= BS_BIT(pos); \
|
---|
179 | } while (0)
|
---|
180 |
|
---|
181 | #define BITSET_IS_EMPTY(bs,empty) do {\
|
---|
182 | int i;\
|
---|
183 | empty = 1;\
|
---|
184 | for (i = 0; i < BITSET_SIZE; i++) {\
|
---|
185 | if ((bs)[i] != 0) {\
|
---|
186 | empty = 0; break;\
|
---|
187 | }\
|
---|
188 | }\
|
---|
189 | } while (0)
|
---|
190 |
|
---|
191 | static void
|
---|
192 | bitset_set_range(ScanEnv *env, BitSetRef bs, int from, int to)
|
---|
193 | {
|
---|
194 | int i;
|
---|
195 | for (i = from; i <= to && i < SINGLE_BYTE_SIZE; i++) {
|
---|
196 | BITSET_SET_BIT_CHKDUP(bs, i);
|
---|
197 | }
|
---|
198 | }
|
---|
199 |
|
---|
200 | #if 0
|
---|
201 | static void
|
---|
202 | bitset_set_all(BitSetRef bs)
|
---|
203 | {
|
---|
204 | int i;
|
---|
205 | for (i = 0; i < BITSET_SIZE; i++) { bs[i] = ~((Bits )0); }
|
---|
206 | }
|
---|
207 | #endif
|
---|
208 |
|
---|
209 | static void
|
---|
210 | bitset_invert(BitSetRef bs)
|
---|
211 | {
|
---|
212 | int i;
|
---|
213 | for (i = 0; i < BITSET_SIZE; i++) { bs[i] = ~(bs[i]); }
|
---|
214 | }
|
---|
215 |
|
---|
216 | static void
|
---|
217 | bitset_invert_to(BitSetRef from, BitSetRef to)
|
---|
218 | {
|
---|
219 | int i;
|
---|
220 | for (i = 0; i < BITSET_SIZE; i++) { to[i] = ~(from[i]); }
|
---|
221 | }
|
---|
222 |
|
---|
223 | static void
|
---|
224 | bitset_and(BitSetRef dest, BitSetRef bs)
|
---|
225 | {
|
---|
226 | int i;
|
---|
227 | for (i = 0; i < BITSET_SIZE; i++) { dest[i] &= bs[i]; }
|
---|
228 | }
|
---|
229 |
|
---|
230 | static void
|
---|
231 | bitset_or(BitSetRef dest, BitSetRef bs)
|
---|
232 | {
|
---|
233 | int i;
|
---|
234 | for (i = 0; i < BITSET_SIZE; i++) { dest[i] |= bs[i]; }
|
---|
235 | }
|
---|
236 |
|
---|
237 | static void
|
---|
238 | bitset_copy(BitSetRef dest, BitSetRef bs)
|
---|
239 | {
|
---|
240 | int i;
|
---|
241 | for (i = 0; i < BITSET_SIZE; i++) { dest[i] = bs[i]; }
|
---|
242 | }
|
---|
243 |
|
---|
244 | #if defined(USE_NAMED_GROUP) && !defined(USE_ST_LIBRARY)
|
---|
245 | extern int
|
---|
246 | onig_strncmp(const UChar* s1, const UChar* s2, int n)
|
---|
247 | {
|
---|
248 | int x;
|
---|
249 |
|
---|
250 | while (n-- > 0) {
|
---|
251 | x = *s2++ - *s1++;
|
---|
252 | if (x) return x;
|
---|
253 | }
|
---|
254 | return 0;
|
---|
255 | }
|
---|
256 | #endif
|
---|
257 |
|
---|
258 | extern void
|
---|
259 | onig_strcpy(UChar* dest, const UChar* src, const UChar* end)
|
---|
260 | {
|
---|
261 | ptrdiff_t len = end - src;
|
---|
262 | if (len > 0) {
|
---|
263 | xmemcpy(dest, src, len);
|
---|
264 | dest[len] = (UChar )0;
|
---|
265 | }
|
---|
266 | }
|
---|
267 |
|
---|
268 | #ifdef USE_NAMED_GROUP
|
---|
269 | static UChar*
|
---|
270 | strdup_with_null(OnigEncoding enc, UChar* s, UChar* end)
|
---|
271 | {
|
---|
272 | ptrdiff_t slen;
|
---|
273 | int term_len, i;
|
---|
274 | UChar *r;
|
---|
275 |
|
---|
276 | slen = end - s;
|
---|
277 | term_len = ONIGENC_MBC_MINLEN(enc);
|
---|
278 |
|
---|
279 | r = (UChar* )xmalloc(slen + term_len);
|
---|
280 | CHECK_NULL_RETURN(r);
|
---|
281 | xmemcpy(r, s, slen);
|
---|
282 |
|
---|
283 | for (i = 0; i < term_len; i++)
|
---|
284 | r[slen + i] = (UChar )0;
|
---|
285 |
|
---|
286 | return r;
|
---|
287 | }
|
---|
288 | #endif
|
---|
289 |
|
---|
290 | /* scan pattern methods */
|
---|
291 | #define PEND_VALUE 0
|
---|
292 |
|
---|
293 | #ifdef __GNUC__
|
---|
294 | /* get rid of Wunused-but-set-variable and Wuninitialized */
|
---|
295 | # define PFETCH_READY UChar* pfetch_prev = NULL; (void)pfetch_prev
|
---|
296 | #else
|
---|
297 | # define PFETCH_READY UChar* pfetch_prev
|
---|
298 | #endif
|
---|
299 | #define PEND (p < end ? 0 : 1)
|
---|
300 | #define PUNFETCH p = pfetch_prev
|
---|
301 | #define PINC do { \
|
---|
302 | pfetch_prev = p; \
|
---|
303 | p += enclen(enc, p, end); \
|
---|
304 | } while (0)
|
---|
305 | #define PFETCH(c) do { \
|
---|
306 | c = ((enc->max_enc_len == 1) ? *p : ONIGENC_MBC_TO_CODE(enc, p, end)); \
|
---|
307 | pfetch_prev = p; \
|
---|
308 | p += enclen(enc, p, end); \
|
---|
309 | } while (0)
|
---|
310 |
|
---|
311 | #define PINC_S do { \
|
---|
312 | p += enclen(enc, p, end); \
|
---|
313 | } while (0)
|
---|
314 | #define PFETCH_S(c) do { \
|
---|
315 | c = ((enc->max_enc_len == 1) ? *p : ONIGENC_MBC_TO_CODE(enc, p, end)); \
|
---|
316 | p += enclen(enc, p, end); \
|
---|
317 | } while (0)
|
---|
318 |
|
---|
319 | #define PPEEK (p < end ? ONIGENC_MBC_TO_CODE(enc, p, end) : PEND_VALUE)
|
---|
320 | #define PPEEK_IS(c) (PPEEK == (OnigCodePoint )c)
|
---|
321 |
|
---|
322 | static UChar*
|
---|
323 | strcat_capa(UChar* dest, UChar* dest_end, const UChar* src, const UChar* src_end,
|
---|
324 | size_t capa)
|
---|
325 | {
|
---|
326 | UChar* r;
|
---|
327 |
|
---|
328 | if (dest)
|
---|
329 | r = (UChar* )xrealloc(dest, capa + 1);
|
---|
330 | else
|
---|
331 | r = (UChar* )xmalloc(capa + 1);
|
---|
332 |
|
---|
333 | CHECK_NULL_RETURN(r);
|
---|
334 | onig_strcpy(r + (dest_end - dest), src, src_end);
|
---|
335 | return r;
|
---|
336 | }
|
---|
337 |
|
---|
338 | /* dest on static area */
|
---|
339 | static UChar*
|
---|
340 | strcat_capa_from_static(UChar* dest, UChar* dest_end,
|
---|
341 | const UChar* src, const UChar* src_end, size_t capa)
|
---|
342 | {
|
---|
343 | UChar* r;
|
---|
344 |
|
---|
345 | r = (UChar* )xmalloc(capa + 1);
|
---|
346 | CHECK_NULL_RETURN(r);
|
---|
347 | onig_strcpy(r, dest, dest_end);
|
---|
348 | onig_strcpy(r + (dest_end - dest), src, src_end);
|
---|
349 | return r;
|
---|
350 | }
|
---|
351 |
|
---|
352 |
|
---|
353 | #ifdef USE_ST_LIBRARY
|
---|
354 |
|
---|
355 | # ifdef RUBY
|
---|
356 | # include "ruby/st.h"
|
---|
357 | # else
|
---|
358 | # include "st.h"
|
---|
359 | # endif
|
---|
360 |
|
---|
361 | typedef struct {
|
---|
362 | const UChar* s;
|
---|
363 | const UChar* end;
|
---|
364 | } st_str_end_key;
|
---|
365 |
|
---|
366 | static int
|
---|
367 | str_end_cmp(st_data_t xp, st_data_t yp)
|
---|
368 | {
|
---|
369 | const st_str_end_key *x, *y;
|
---|
370 | const UChar *p, *q;
|
---|
371 | int c;
|
---|
372 |
|
---|
373 | x = (const st_str_end_key *)xp;
|
---|
374 | y = (const st_str_end_key *)yp;
|
---|
375 | if ((x->end - x->s) != (y->end - y->s))
|
---|
376 | return 1;
|
---|
377 |
|
---|
378 | p = x->s;
|
---|
379 | q = y->s;
|
---|
380 | while (p < x->end) {
|
---|
381 | c = (int )*p - (int )*q;
|
---|
382 | if (c != 0) return c;
|
---|
383 |
|
---|
384 | p++; q++;
|
---|
385 | }
|
---|
386 |
|
---|
387 | return 0;
|
---|
388 | }
|
---|
389 |
|
---|
390 | static st_index_t
|
---|
391 | str_end_hash(st_data_t xp)
|
---|
392 | {
|
---|
393 | const st_str_end_key *x = (const st_str_end_key *)xp;
|
---|
394 | const UChar *p;
|
---|
395 | st_index_t val = 0;
|
---|
396 |
|
---|
397 | p = x->s;
|
---|
398 | while (p < x->end) {
|
---|
399 | val = val * 997 + (int )*p++;
|
---|
400 | }
|
---|
401 |
|
---|
402 | return val + (val >> 5);
|
---|
403 | }
|
---|
404 |
|
---|
405 | extern hash_table_type*
|
---|
406 | onig_st_init_strend_table_with_size(st_index_t size)
|
---|
407 | {
|
---|
408 | static const struct st_hash_type hashType = {
|
---|
409 | str_end_cmp,
|
---|
410 | str_end_hash,
|
---|
411 | };
|
---|
412 |
|
---|
413 | return (hash_table_type* )
|
---|
414 | onig_st_init_table_with_size(&hashType, size);
|
---|
415 | }
|
---|
416 |
|
---|
417 | extern int
|
---|
418 | onig_st_lookup_strend(hash_table_type* table, const UChar* str_key,
|
---|
419 | const UChar* end_key, hash_data_type *value)
|
---|
420 | {
|
---|
421 | st_str_end_key key;
|
---|
422 |
|
---|
423 | key.s = (UChar* )str_key;
|
---|
424 | key.end = (UChar* )end_key;
|
---|
425 |
|
---|
426 | return onig_st_lookup(table, (st_data_t )(&key), value);
|
---|
427 | }
|
---|
428 |
|
---|
429 | extern int
|
---|
430 | onig_st_insert_strend(hash_table_type* table, const UChar* str_key,
|
---|
431 | const UChar* end_key, hash_data_type value)
|
---|
432 | {
|
---|
433 | st_str_end_key* key;
|
---|
434 | int result;
|
---|
435 |
|
---|
436 | key = (st_str_end_key* )xmalloc(sizeof(st_str_end_key));
|
---|
437 | key->s = (UChar* )str_key;
|
---|
438 | key->end = (UChar* )end_key;
|
---|
439 | result = onig_st_insert(table, (st_data_t )key, value);
|
---|
440 | if (result) {
|
---|
441 | xfree(key);
|
---|
442 | }
|
---|
443 | return result;
|
---|
444 | }
|
---|
445 |
|
---|
446 | #endif /* USE_ST_LIBRARY */
|
---|
447 |
|
---|
448 |
|
---|
449 | #ifdef USE_NAMED_GROUP
|
---|
450 |
|
---|
451 | # define INIT_NAME_BACKREFS_ALLOC_NUM 8
|
---|
452 |
|
---|
453 | typedef struct {
|
---|
454 | UChar* name;
|
---|
455 | size_t name_len; /* byte length */
|
---|
456 | int back_num; /* number of backrefs */
|
---|
457 | int back_alloc;
|
---|
458 | int back_ref1;
|
---|
459 | int* back_refs;
|
---|
460 | } NameEntry;
|
---|
461 |
|
---|
462 | # ifdef USE_ST_LIBRARY
|
---|
463 |
|
---|
464 | typedef st_table NameTable;
|
---|
465 | typedef st_data_t HashDataType; /* 1.6 st.h doesn't define st_data_t type */
|
---|
466 |
|
---|
467 | # ifdef ONIG_DEBUG
|
---|
468 | static int
|
---|
469 | i_print_name_entry(UChar* key, NameEntry* e, void* arg)
|
---|
470 | {
|
---|
471 | int i;
|
---|
472 | FILE* fp = (FILE* )arg;
|
---|
473 |
|
---|
474 | fprintf(fp, "%s: ", e->name);
|
---|
475 | if (e->back_num == 0)
|
---|
476 | fputs("-", fp);
|
---|
477 | else if (e->back_num == 1)
|
---|
478 | fprintf(fp, "%d", e->back_ref1);
|
---|
479 | else {
|
---|
480 | for (i = 0; i < e->back_num; i++) {
|
---|
481 | if (i > 0) fprintf(fp, ", ");
|
---|
482 | fprintf(fp, "%d", e->back_refs[i]);
|
---|
483 | }
|
---|
484 | }
|
---|
485 | fputs("\n", fp);
|
---|
486 | return ST_CONTINUE;
|
---|
487 | }
|
---|
488 |
|
---|
489 | extern int
|
---|
490 | onig_print_names(FILE* fp, regex_t* reg)
|
---|
491 | {
|
---|
492 | NameTable* t = (NameTable* )reg->name_table;
|
---|
493 |
|
---|
494 | if (IS_NOT_NULL(t)) {
|
---|
495 | fprintf(fp, "name table\n");
|
---|
496 | onig_st_foreach(t, i_print_name_entry, (HashDataType )fp);
|
---|
497 | fputs("\n", fp);
|
---|
498 | }
|
---|
499 | return 0;
|
---|
500 | }
|
---|
501 | # endif /* ONIG_DEBUG */
|
---|
502 |
|
---|
503 | static int
|
---|
504 | i_free_name_entry(UChar* key, NameEntry* e, void* arg ARG_UNUSED)
|
---|
505 | {
|
---|
506 | xfree(e->name);
|
---|
507 | if (IS_NOT_NULL(e->back_refs)) xfree(e->back_refs);
|
---|
508 | xfree(key);
|
---|
509 | xfree(e);
|
---|
510 | return ST_DELETE;
|
---|
511 | }
|
---|
512 |
|
---|
513 | static int
|
---|
514 | names_clear(regex_t* reg)
|
---|
515 | {
|
---|
516 | NameTable* t = (NameTable* )reg->name_table;
|
---|
517 |
|
---|
518 | if (IS_NOT_NULL(t)) {
|
---|
519 | onig_st_foreach(t, i_free_name_entry, 0);
|
---|
520 | }
|
---|
521 | return 0;
|
---|
522 | }
|
---|
523 |
|
---|
524 | extern int
|
---|
525 | onig_names_free(regex_t* reg)
|
---|
526 | {
|
---|
527 | int r;
|
---|
528 | NameTable* t;
|
---|
529 |
|
---|
530 | r = names_clear(reg);
|
---|
531 | if (r) return r;
|
---|
532 |
|
---|
533 | t = (NameTable* )reg->name_table;
|
---|
534 | if (IS_NOT_NULL(t)) onig_st_free_table(t);
|
---|
535 | reg->name_table = (void* )NULL;
|
---|
536 | return 0;
|
---|
537 | }
|
---|
538 |
|
---|
539 | static NameEntry*
|
---|
540 | name_find(regex_t* reg, const UChar* name, const UChar* name_end)
|
---|
541 | {
|
---|
542 | NameEntry* e;
|
---|
543 | NameTable* t = (NameTable* )reg->name_table;
|
---|
544 |
|
---|
545 | e = (NameEntry* )NULL;
|
---|
546 | if (IS_NOT_NULL(t)) {
|
---|
547 | onig_st_lookup_strend(t, name, name_end, (HashDataType* )((void* )(&e)));
|
---|
548 | }
|
---|
549 | return e;
|
---|
550 | }
|
---|
551 |
|
---|
552 | typedef struct {
|
---|
553 | int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*);
|
---|
554 | regex_t* reg;
|
---|
555 | void* arg;
|
---|
556 | int ret;
|
---|
557 | OnigEncoding enc;
|
---|
558 | } INamesArg;
|
---|
559 |
|
---|
560 | static int
|
---|
561 | i_names(UChar* key ARG_UNUSED, NameEntry* e, INamesArg* arg)
|
---|
562 | {
|
---|
563 | int r = (*(arg->func))(e->name,
|
---|
564 | e->name + e->name_len,
|
---|
565 | e->back_num,
|
---|
566 | (e->back_num > 1 ? e->back_refs : &(e->back_ref1)),
|
---|
567 | arg->reg, arg->arg);
|
---|
568 | if (r != 0) {
|
---|
569 | arg->ret = r;
|
---|
570 | return ST_STOP;
|
---|
571 | }
|
---|
572 | return ST_CONTINUE;
|
---|
573 | }
|
---|
574 |
|
---|
575 | extern int
|
---|
576 | onig_foreach_name(regex_t* reg,
|
---|
577 | int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)
|
---|
578 | {
|
---|
579 | INamesArg narg;
|
---|
580 | NameTable* t = (NameTable* )reg->name_table;
|
---|
581 |
|
---|
582 | narg.ret = 0;
|
---|
583 | if (IS_NOT_NULL(t)) {
|
---|
584 | narg.func = func;
|
---|
585 | narg.reg = reg;
|
---|
586 | narg.arg = arg;
|
---|
587 | narg.enc = reg->enc; /* should be pattern encoding. */
|
---|
588 | onig_st_foreach(t, i_names, (HashDataType )&narg);
|
---|
589 | }
|
---|
590 | return narg.ret;
|
---|
591 | }
|
---|
592 |
|
---|
593 | static int
|
---|
594 | i_renumber_name(UChar* key ARG_UNUSED, NameEntry* e, GroupNumRemap* map)
|
---|
595 | {
|
---|
596 | int i;
|
---|
597 |
|
---|
598 | if (e->back_num > 1) {
|
---|
599 | for (i = 0; i < e->back_num; i++) {
|
---|
600 | e->back_refs[i] = map[e->back_refs[i]].new_val;
|
---|
601 | }
|
---|
602 | }
|
---|
603 | else if (e->back_num == 1) {
|
---|
604 | e->back_ref1 = map[e->back_ref1].new_val;
|
---|
605 | }
|
---|
606 |
|
---|
607 | return ST_CONTINUE;
|
---|
608 | }
|
---|
609 |
|
---|
610 | extern int
|
---|
611 | onig_renumber_name_table(regex_t* reg, GroupNumRemap* map)
|
---|
612 | {
|
---|
613 | NameTable* t = (NameTable* )reg->name_table;
|
---|
614 |
|
---|
615 | if (IS_NOT_NULL(t)) {
|
---|
616 | onig_st_foreach(t, i_renumber_name, (HashDataType )map);
|
---|
617 | }
|
---|
618 | return 0;
|
---|
619 | }
|
---|
620 |
|
---|
621 |
|
---|
622 | extern int
|
---|
623 | onig_number_of_names(const regex_t* reg)
|
---|
624 | {
|
---|
625 | NameTable* t = (NameTable* )reg->name_table;
|
---|
626 |
|
---|
627 | if (IS_NOT_NULL(t))
|
---|
628 | return (int )t->num_entries;
|
---|
629 | else
|
---|
630 | return 0;
|
---|
631 | }
|
---|
632 |
|
---|
633 | # else /* USE_ST_LIBRARY */
|
---|
634 |
|
---|
635 | # define INIT_NAMES_ALLOC_NUM 8
|
---|
636 |
|
---|
637 | typedef struct {
|
---|
638 | NameEntry* e;
|
---|
639 | int num;
|
---|
640 | int alloc;
|
---|
641 | } NameTable;
|
---|
642 |
|
---|
643 | # ifdef ONIG_DEBUG
|
---|
644 | extern int
|
---|
645 | onig_print_names(FILE* fp, regex_t* reg)
|
---|
646 | {
|
---|
647 | int i, j;
|
---|
648 | NameEntry* e;
|
---|
649 | NameTable* t = (NameTable* )reg->name_table;
|
---|
650 |
|
---|
651 | if (IS_NOT_NULL(t) && t->num > 0) {
|
---|
652 | fprintf(fp, "name table\n");
|
---|
653 | for (i = 0; i < t->num; i++) {
|
---|
654 | e = &(t->e[i]);
|
---|
655 | fprintf(fp, "%s: ", e->name);
|
---|
656 | if (e->back_num == 0) {
|
---|
657 | fputs("-", fp);
|
---|
658 | }
|
---|
659 | else if (e->back_num == 1) {
|
---|
660 | fprintf(fp, "%d", e->back_ref1);
|
---|
661 | }
|
---|
662 | else {
|
---|
663 | for (j = 0; j < e->back_num; j++) {
|
---|
664 | if (j > 0) fprintf(fp, ", ");
|
---|
665 | fprintf(fp, "%d", e->back_refs[j]);
|
---|
666 | }
|
---|
667 | }
|
---|
668 | fputs("\n", fp);
|
---|
669 | }
|
---|
670 | fputs("\n", fp);
|
---|
671 | }
|
---|
672 | return 0;
|
---|
673 | }
|
---|
674 | # endif
|
---|
675 |
|
---|
676 | static int
|
---|
677 | names_clear(regex_t* reg)
|
---|
678 | {
|
---|
679 | int i;
|
---|
680 | NameEntry* e;
|
---|
681 | NameTable* t = (NameTable* )reg->name_table;
|
---|
682 |
|
---|
683 | if (IS_NOT_NULL(t)) {
|
---|
684 | for (i = 0; i < t->num; i++) {
|
---|
685 | e = &(t->e[i]);
|
---|
686 | if (IS_NOT_NULL(e->name)) {
|
---|
687 | xfree(e->name);
|
---|
688 | e->name = NULL;
|
---|
689 | e->name_len = 0;
|
---|
690 | e->back_num = 0;
|
---|
691 | e->back_alloc = 0;
|
---|
692 | if (IS_NOT_NULL(e->back_refs)) xfree(e->back_refs);
|
---|
693 | e->back_refs = (int* )NULL;
|
---|
694 | }
|
---|
695 | }
|
---|
696 | if (IS_NOT_NULL(t->e)) {
|
---|
697 | xfree(t->e);
|
---|
698 | t->e = NULL;
|
---|
699 | }
|
---|
700 | t->num = 0;
|
---|
701 | }
|
---|
702 | return 0;
|
---|
703 | }
|
---|
704 |
|
---|
705 | extern int
|
---|
706 | onig_names_free(regex_t* reg)
|
---|
707 | {
|
---|
708 | int r;
|
---|
709 | NameTable* t;
|
---|
710 |
|
---|
711 | r = names_clear(reg);
|
---|
712 | if (r) return r;
|
---|
713 |
|
---|
714 | t = (NameTable* )reg->name_table;
|
---|
715 | if (IS_NOT_NULL(t)) xfree(t);
|
---|
716 | reg->name_table = NULL;
|
---|
717 | return 0;
|
---|
718 | }
|
---|
719 |
|
---|
720 | static NameEntry*
|
---|
721 | name_find(regex_t* reg, const UChar* name, const UChar* name_end)
|
---|
722 | {
|
---|
723 | int i, len;
|
---|
724 | NameEntry* e;
|
---|
725 | NameTable* t = (NameTable* )reg->name_table;
|
---|
726 |
|
---|
727 | if (IS_NOT_NULL(t)) {
|
---|
728 | len = name_end - name;
|
---|
729 | for (i = 0; i < t->num; i++) {
|
---|
730 | e = &(t->e[i]);
|
---|
731 | if (len == e->name_len && onig_strncmp(name, e->name, len) == 0)
|
---|
732 | return e;
|
---|
733 | }
|
---|
734 | }
|
---|
735 | return (NameEntry* )NULL;
|
---|
736 | }
|
---|
737 |
|
---|
738 | extern int
|
---|
739 | onig_foreach_name(regex_t* reg,
|
---|
740 | int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)
|
---|
741 | {
|
---|
742 | int i, r;
|
---|
743 | NameEntry* e;
|
---|
744 | NameTable* t = (NameTable* )reg->name_table;
|
---|
745 |
|
---|
746 | if (IS_NOT_NULL(t)) {
|
---|
747 | for (i = 0; i < t->num; i++) {
|
---|
748 | e = &(t->e[i]);
|
---|
749 | r = (*func)(e->name, e->name + e->name_len, e->back_num,
|
---|
750 | (e->back_num > 1 ? e->back_refs : &(e->back_ref1)),
|
---|
751 | reg, arg);
|
---|
752 | if (r != 0) return r;
|
---|
753 | }
|
---|
754 | }
|
---|
755 | return 0;
|
---|
756 | }
|
---|
757 |
|
---|
758 | extern int
|
---|
759 | onig_number_of_names(const regex_t* reg)
|
---|
760 | {
|
---|
761 | NameTable* t = (NameTable* )reg->name_table;
|
---|
762 |
|
---|
763 | if (IS_NOT_NULL(t))
|
---|
764 | return t->num;
|
---|
765 | else
|
---|
766 | return 0;
|
---|
767 | }
|
---|
768 |
|
---|
769 | # endif /* else USE_ST_LIBRARY */
|
---|
770 |
|
---|
771 | static int
|
---|
772 | name_add(regex_t* reg, UChar* name, UChar* name_end, int backref, ScanEnv* env)
|
---|
773 | {
|
---|
774 | int alloc;
|
---|
775 | NameEntry* e;
|
---|
776 | NameTable* t = (NameTable* )reg->name_table;
|
---|
777 |
|
---|
778 | if (name_end - name <= 0)
|
---|
779 | return ONIGERR_EMPTY_GROUP_NAME;
|
---|
780 |
|
---|
781 | e = name_find(reg, name, name_end);
|
---|
782 | if (IS_NULL(e)) {
|
---|
783 | # ifdef USE_ST_LIBRARY
|
---|
784 | if (IS_NULL(t)) {
|
---|
785 | t = onig_st_init_strend_table_with_size(5);
|
---|
786 | reg->name_table = (void* )t;
|
---|
787 | }
|
---|
788 | e = (NameEntry* )xmalloc(sizeof(NameEntry));
|
---|
789 | CHECK_NULL_RETURN_MEMERR(e);
|
---|
790 |
|
---|
791 | e->name = strdup_with_null(reg->enc, name, name_end);
|
---|
792 | if (IS_NULL(e->name)) {
|
---|
793 | xfree(e);
|
---|
794 | return ONIGERR_MEMORY;
|
---|
795 | }
|
---|
796 | onig_st_insert_strend(t, e->name, (e->name + (name_end - name)),
|
---|
797 | (HashDataType )e);
|
---|
798 |
|
---|
799 | e->name_len = name_end - name;
|
---|
800 | e->back_num = 0;
|
---|
801 | e->back_alloc = 0;
|
---|
802 | e->back_refs = (int* )NULL;
|
---|
803 |
|
---|
804 | # else
|
---|
805 |
|
---|
806 | if (IS_NULL(t)) {
|
---|
807 | alloc = INIT_NAMES_ALLOC_NUM;
|
---|
808 | t = (NameTable* )xmalloc(sizeof(NameTable));
|
---|
809 | CHECK_NULL_RETURN_MEMERR(t);
|
---|
810 | t->e = NULL;
|
---|
811 | t->alloc = 0;
|
---|
812 | t->num = 0;
|
---|
813 |
|
---|
814 | t->e = (NameEntry* )xmalloc(sizeof(NameEntry) * alloc);
|
---|
815 | if (IS_NULL(t->e)) {
|
---|
816 | xfree(t);
|
---|
817 | return ONIGERR_MEMORY;
|
---|
818 | }
|
---|
819 | t->alloc = alloc;
|
---|
820 | reg->name_table = t;
|
---|
821 | goto clear;
|
---|
822 | }
|
---|
823 | else if (t->num == t->alloc) {
|
---|
824 | int i;
|
---|
825 | NameEntry* p;
|
---|
826 |
|
---|
827 | alloc = t->alloc * 2;
|
---|
828 | p = (NameEntry* )xrealloc(t->e, sizeof(NameEntry) * alloc);
|
---|
829 | CHECK_NULL_RETURN_MEMERR(p);
|
---|
830 | t->e = p;
|
---|
831 | t->alloc = alloc;
|
---|
832 |
|
---|
833 | clear:
|
---|
834 | for (i = t->num; i < t->alloc; i++) {
|
---|
835 | t->e[i].name = NULL;
|
---|
836 | t->e[i].name_len = 0;
|
---|
837 | t->e[i].back_num = 0;
|
---|
838 | t->e[i].back_alloc = 0;
|
---|
839 | t->e[i].back_refs = (int* )NULL;
|
---|
840 | }
|
---|
841 | }
|
---|
842 | e = &(t->e[t->num]);
|
---|
843 | t->num++;
|
---|
844 | e->name = strdup_with_null(reg->enc, name, name_end);
|
---|
845 | if (IS_NULL(e->name)) return ONIGERR_MEMORY;
|
---|
846 | e->name_len = name_end - name;
|
---|
847 | # endif
|
---|
848 | }
|
---|
849 |
|
---|
850 | if (e->back_num >= 1 &&
|
---|
851 | ! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME)) {
|
---|
852 | onig_scan_env_set_error_string(env, ONIGERR_MULTIPLEX_DEFINED_NAME,
|
---|
853 | name, name_end);
|
---|
854 | return ONIGERR_MULTIPLEX_DEFINED_NAME;
|
---|
855 | }
|
---|
856 |
|
---|
857 | e->back_num++;
|
---|
858 | if (e->back_num == 1) {
|
---|
859 | e->back_ref1 = backref;
|
---|
860 | }
|
---|
861 | else {
|
---|
862 | if (e->back_num == 2) {
|
---|
863 | alloc = INIT_NAME_BACKREFS_ALLOC_NUM;
|
---|
864 | e->back_refs = (int* )xmalloc(sizeof(int) * alloc);
|
---|
865 | CHECK_NULL_RETURN_MEMERR(e->back_refs);
|
---|
866 | e->back_alloc = alloc;
|
---|
867 | e->back_refs[0] = e->back_ref1;
|
---|
868 | e->back_refs[1] = backref;
|
---|
869 | }
|
---|
870 | else {
|
---|
871 | if (e->back_num > e->back_alloc) {
|
---|
872 | int* p;
|
---|
873 | alloc = e->back_alloc * 2;
|
---|
874 | p = (int* )xrealloc(e->back_refs, sizeof(int) * alloc);
|
---|
875 | CHECK_NULL_RETURN_MEMERR(p);
|
---|
876 | e->back_refs = p;
|
---|
877 | e->back_alloc = alloc;
|
---|
878 | }
|
---|
879 | e->back_refs[e->back_num - 1] = backref;
|
---|
880 | }
|
---|
881 | }
|
---|
882 |
|
---|
883 | return 0;
|
---|
884 | }
|
---|
885 |
|
---|
886 | extern int
|
---|
887 | onig_name_to_group_numbers(regex_t* reg, const UChar* name,
|
---|
888 | const UChar* name_end, int** nums)
|
---|
889 | {
|
---|
890 | NameEntry* e = name_find(reg, name, name_end);
|
---|
891 |
|
---|
892 | if (IS_NULL(e)) return ONIGERR_UNDEFINED_NAME_REFERENCE;
|
---|
893 |
|
---|
894 | switch (e->back_num) {
|
---|
895 | case 0:
|
---|
896 | *nums = 0;
|
---|
897 | break;
|
---|
898 | case 1:
|
---|
899 | *nums = &(e->back_ref1);
|
---|
900 | break;
|
---|
901 | default:
|
---|
902 | *nums = e->back_refs;
|
---|
903 | break;
|
---|
904 | }
|
---|
905 | return e->back_num;
|
---|
906 | }
|
---|
907 |
|
---|
908 | extern int
|
---|
909 | onig_name_to_backref_number(regex_t* reg, const UChar* name,
|
---|
910 | const UChar* name_end, const OnigRegion *region)
|
---|
911 | {
|
---|
912 | int i, n, *nums;
|
---|
913 |
|
---|
914 | n = onig_name_to_group_numbers(reg, name, name_end, &nums);
|
---|
915 | if (n < 0)
|
---|
916 | return n;
|
---|
917 | else if (n == 0)
|
---|
918 | return ONIGERR_PARSER_BUG;
|
---|
919 | else if (n == 1)
|
---|
920 | return nums[0];
|
---|
921 | else {
|
---|
922 | if (IS_NOT_NULL(region)) {
|
---|
923 | for (i = n - 1; i >= 0; i--) {
|
---|
924 | if (region->beg[nums[i]] != ONIG_REGION_NOTPOS)
|
---|
925 | return nums[i];
|
---|
926 | }
|
---|
927 | }
|
---|
928 | return nums[n - 1];
|
---|
929 | }
|
---|
930 | }
|
---|
931 |
|
---|
932 | #else /* USE_NAMED_GROUP */
|
---|
933 |
|
---|
934 | extern int
|
---|
935 | onig_name_to_group_numbers(regex_t* reg, const UChar* name,
|
---|
936 | const UChar* name_end, int** nums)
|
---|
937 | {
|
---|
938 | return ONIG_NO_SUPPORT_CONFIG;
|
---|
939 | }
|
---|
940 |
|
---|
941 | extern int
|
---|
942 | onig_name_to_backref_number(regex_t* reg, const UChar* name,
|
---|
943 | const UChar* name_end, const OnigRegion* region)
|
---|
944 | {
|
---|
945 | return ONIG_NO_SUPPORT_CONFIG;
|
---|
946 | }
|
---|
947 |
|
---|
948 | extern int
|
---|
949 | onig_foreach_name(regex_t* reg,
|
---|
950 | int (*func)(const UChar*, const UChar*,int,int*,regex_t*,void*), void* arg)
|
---|
951 | {
|
---|
952 | return ONIG_NO_SUPPORT_CONFIG;
|
---|
953 | }
|
---|
954 |
|
---|
955 | extern int
|
---|
956 | onig_number_of_names(const regex_t* reg)
|
---|
957 | {
|
---|
958 | return 0;
|
---|
959 | }
|
---|
960 | #endif /* else USE_NAMED_GROUP */
|
---|
961 |
|
---|
962 | extern int
|
---|
963 | onig_noname_group_capture_is_active(const regex_t* reg)
|
---|
964 | {
|
---|
965 | if (ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_DONT_CAPTURE_GROUP))
|
---|
966 | return 0;
|
---|
967 |
|
---|
968 | #ifdef USE_NAMED_GROUP
|
---|
969 | if (onig_number_of_names(reg) > 0 &&
|
---|
970 | IS_SYNTAX_BV(reg->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) &&
|
---|
971 | !ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_CAPTURE_GROUP)) {
|
---|
972 | return 0;
|
---|
973 | }
|
---|
974 | #endif
|
---|
975 |
|
---|
976 | return 1;
|
---|
977 | }
|
---|
978 |
|
---|
979 |
|
---|
980 | #define INIT_SCANENV_MEMNODES_ALLOC_SIZE 16
|
---|
981 |
|
---|
982 | static void
|
---|
983 | scan_env_clear(ScanEnv* env)
|
---|
984 | {
|
---|
985 | int i;
|
---|
986 |
|
---|
987 | BIT_STATUS_CLEAR(env->capture_history);
|
---|
988 | BIT_STATUS_CLEAR(env->bt_mem_start);
|
---|
989 | BIT_STATUS_CLEAR(env->bt_mem_end);
|
---|
990 | BIT_STATUS_CLEAR(env->backrefed_mem);
|
---|
991 | env->error = (UChar* )NULL;
|
---|
992 | env->error_end = (UChar* )NULL;
|
---|
993 | env->num_call = 0;
|
---|
994 | env->num_mem = 0;
|
---|
995 | #ifdef USE_NAMED_GROUP
|
---|
996 | env->num_named = 0;
|
---|
997 | #endif
|
---|
998 | env->mem_alloc = 0;
|
---|
999 | env->mem_nodes_dynamic = (Node** )NULL;
|
---|
1000 |
|
---|
1001 | for (i = 0; i < SCANENV_MEMNODES_SIZE; i++)
|
---|
1002 | env->mem_nodes_static[i] = NULL_NODE;
|
---|
1003 |
|
---|
1004 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
1005 | env->num_comb_exp_check = 0;
|
---|
1006 | env->comb_exp_max_regnum = 0;
|
---|
1007 | env->curr_max_regnum = 0;
|
---|
1008 | env->has_recursion = 0;
|
---|
1009 | #endif
|
---|
1010 | env->parse_depth = 0;
|
---|
1011 | env->warnings_flag = 0;
|
---|
1012 | }
|
---|
1013 |
|
---|
1014 | static int
|
---|
1015 | scan_env_add_mem_entry(ScanEnv* env)
|
---|
1016 | {
|
---|
1017 | int i, need, alloc;
|
---|
1018 | Node** p;
|
---|
1019 |
|
---|
1020 | need = env->num_mem + 1;
|
---|
1021 | if (need > ONIG_MAX_CAPTURE_GROUP_NUM)
|
---|
1022 | return ONIGERR_TOO_MANY_CAPTURE_GROUPS;
|
---|
1023 | if (need >= SCANENV_MEMNODES_SIZE) {
|
---|
1024 | if (env->mem_alloc <= need) {
|
---|
1025 | if (IS_NULL(env->mem_nodes_dynamic)) {
|
---|
1026 | alloc = INIT_SCANENV_MEMNODES_ALLOC_SIZE;
|
---|
1027 | p = (Node** )xmalloc(sizeof(Node*) * alloc);
|
---|
1028 | CHECK_NULL_RETURN_MEMERR(p);
|
---|
1029 | xmemcpy(p, env->mem_nodes_static,
|
---|
1030 | sizeof(Node*) * SCANENV_MEMNODES_SIZE);
|
---|
1031 | }
|
---|
1032 | else {
|
---|
1033 | alloc = env->mem_alloc * 2;
|
---|
1034 | p = (Node** )xrealloc(env->mem_nodes_dynamic, sizeof(Node*) * alloc);
|
---|
1035 | CHECK_NULL_RETURN_MEMERR(p);
|
---|
1036 | }
|
---|
1037 |
|
---|
1038 | for (i = env->num_mem + 1; i < alloc; i++)
|
---|
1039 | p[i] = NULL_NODE;
|
---|
1040 |
|
---|
1041 | env->mem_nodes_dynamic = p;
|
---|
1042 | env->mem_alloc = alloc;
|
---|
1043 | }
|
---|
1044 | }
|
---|
1045 |
|
---|
1046 | env->num_mem++;
|
---|
1047 | return env->num_mem;
|
---|
1048 | }
|
---|
1049 |
|
---|
1050 | static int
|
---|
1051 | scan_env_set_mem_node(ScanEnv* env, int num, Node* node)
|
---|
1052 | {
|
---|
1053 | if (env->num_mem >= num)
|
---|
1054 | SCANENV_MEM_NODES(env)[num] = node;
|
---|
1055 | else
|
---|
1056 | return ONIGERR_PARSER_BUG;
|
---|
1057 | return 0;
|
---|
1058 | }
|
---|
1059 |
|
---|
1060 |
|
---|
1061 | extern void
|
---|
1062 | onig_node_free(Node* node)
|
---|
1063 | {
|
---|
1064 | start:
|
---|
1065 | if (IS_NULL(node)) return ;
|
---|
1066 |
|
---|
1067 | switch (NTYPE(node)) {
|
---|
1068 | case NT_STR:
|
---|
1069 | if (NSTR(node)->capa != 0 &&
|
---|
1070 | IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) {
|
---|
1071 | xfree(NSTR(node)->s);
|
---|
1072 | }
|
---|
1073 | break;
|
---|
1074 |
|
---|
1075 | case NT_LIST:
|
---|
1076 | case NT_ALT:
|
---|
1077 | onig_node_free(NCAR(node));
|
---|
1078 | {
|
---|
1079 | Node* next_node = NCDR(node);
|
---|
1080 |
|
---|
1081 | xfree(node);
|
---|
1082 | node = next_node;
|
---|
1083 | goto start;
|
---|
1084 | }
|
---|
1085 | break;
|
---|
1086 |
|
---|
1087 | case NT_CCLASS:
|
---|
1088 | {
|
---|
1089 | CClassNode* cc = NCCLASS(node);
|
---|
1090 |
|
---|
1091 | if (cc->mbuf)
|
---|
1092 | bbuf_free(cc->mbuf);
|
---|
1093 | }
|
---|
1094 | break;
|
---|
1095 |
|
---|
1096 | case NT_QTFR:
|
---|
1097 | if (NQTFR(node)->target)
|
---|
1098 | onig_node_free(NQTFR(node)->target);
|
---|
1099 | break;
|
---|
1100 |
|
---|
1101 | case NT_ENCLOSE:
|
---|
1102 | if (NENCLOSE(node)->target)
|
---|
1103 | onig_node_free(NENCLOSE(node)->target);
|
---|
1104 | break;
|
---|
1105 |
|
---|
1106 | case NT_BREF:
|
---|
1107 | if (IS_NOT_NULL(NBREF(node)->back_dynamic))
|
---|
1108 | xfree(NBREF(node)->back_dynamic);
|
---|
1109 | break;
|
---|
1110 |
|
---|
1111 | case NT_ANCHOR:
|
---|
1112 | if (NANCHOR(node)->target)
|
---|
1113 | onig_node_free(NANCHOR(node)->target);
|
---|
1114 | break;
|
---|
1115 | }
|
---|
1116 |
|
---|
1117 | xfree(node);
|
---|
1118 | }
|
---|
1119 |
|
---|
1120 | static Node*
|
---|
1121 | node_new(void)
|
---|
1122 | {
|
---|
1123 | Node* node;
|
---|
1124 |
|
---|
1125 | node = (Node* )xmalloc(sizeof(Node));
|
---|
1126 | /* xmemset(node, 0, sizeof(Node)); */
|
---|
1127 | return node;
|
---|
1128 | }
|
---|
1129 |
|
---|
1130 | static void
|
---|
1131 | initialize_cclass(CClassNode* cc)
|
---|
1132 | {
|
---|
1133 | BITSET_CLEAR(cc->bs);
|
---|
1134 | /* cc->base.flags = 0; */
|
---|
1135 | cc->flags = 0;
|
---|
1136 | cc->mbuf = NULL;
|
---|
1137 | }
|
---|
1138 |
|
---|
1139 | static Node*
|
---|
1140 | node_new_cclass(void)
|
---|
1141 | {
|
---|
1142 | Node* node = node_new();
|
---|
1143 | CHECK_NULL_RETURN(node);
|
---|
1144 |
|
---|
1145 | SET_NTYPE(node, NT_CCLASS);
|
---|
1146 | initialize_cclass(NCCLASS(node));
|
---|
1147 | return node;
|
---|
1148 | }
|
---|
1149 |
|
---|
1150 | static Node*
|
---|
1151 | node_new_ctype(int type, int not, int ascii_range)
|
---|
1152 | {
|
---|
1153 | Node* node = node_new();
|
---|
1154 | CHECK_NULL_RETURN(node);
|
---|
1155 |
|
---|
1156 | SET_NTYPE(node, NT_CTYPE);
|
---|
1157 | NCTYPE(node)->ctype = type;
|
---|
1158 | NCTYPE(node)->not = not;
|
---|
1159 | NCTYPE(node)->ascii_range = ascii_range;
|
---|
1160 | return node;
|
---|
1161 | }
|
---|
1162 |
|
---|
1163 | static Node*
|
---|
1164 | node_new_anychar(void)
|
---|
1165 | {
|
---|
1166 | Node* node = node_new();
|
---|
1167 | CHECK_NULL_RETURN(node);
|
---|
1168 |
|
---|
1169 | SET_NTYPE(node, NT_CANY);
|
---|
1170 | return node;
|
---|
1171 | }
|
---|
1172 |
|
---|
1173 | static Node*
|
---|
1174 | node_new_list(Node* left, Node* right)
|
---|
1175 | {
|
---|
1176 | Node* node = node_new();
|
---|
1177 | CHECK_NULL_RETURN(node);
|
---|
1178 |
|
---|
1179 | SET_NTYPE(node, NT_LIST);
|
---|
1180 | NCAR(node) = left;
|
---|
1181 | NCDR(node) = right;
|
---|
1182 | return node;
|
---|
1183 | }
|
---|
1184 |
|
---|
1185 | extern Node*
|
---|
1186 | onig_node_new_list(Node* left, Node* right)
|
---|
1187 | {
|
---|
1188 | return node_new_list(left, right);
|
---|
1189 | }
|
---|
1190 |
|
---|
1191 | extern Node*
|
---|
1192 | onig_node_list_add(Node* list, Node* x)
|
---|
1193 | {
|
---|
1194 | Node *n;
|
---|
1195 |
|
---|
1196 | n = onig_node_new_list(x, NULL);
|
---|
1197 | if (IS_NULL(n)) return NULL_NODE;
|
---|
1198 |
|
---|
1199 | if (IS_NOT_NULL(list)) {
|
---|
1200 | while (IS_NOT_NULL(NCDR(list)))
|
---|
1201 | list = NCDR(list);
|
---|
1202 |
|
---|
1203 | NCDR(list) = n;
|
---|
1204 | }
|
---|
1205 |
|
---|
1206 | return n;
|
---|
1207 | }
|
---|
1208 |
|
---|
1209 | extern Node*
|
---|
1210 | onig_node_new_alt(Node* left, Node* right)
|
---|
1211 | {
|
---|
1212 | Node* node = node_new();
|
---|
1213 | CHECK_NULL_RETURN(node);
|
---|
1214 |
|
---|
1215 | SET_NTYPE(node, NT_ALT);
|
---|
1216 | NCAR(node) = left;
|
---|
1217 | NCDR(node) = right;
|
---|
1218 | return node;
|
---|
1219 | }
|
---|
1220 |
|
---|
1221 | extern Node*
|
---|
1222 | onig_node_new_anchor(int type)
|
---|
1223 | {
|
---|
1224 | Node* node = node_new();
|
---|
1225 | CHECK_NULL_RETURN(node);
|
---|
1226 |
|
---|
1227 | SET_NTYPE(node, NT_ANCHOR);
|
---|
1228 | NANCHOR(node)->type = type;
|
---|
1229 | NANCHOR(node)->target = NULL;
|
---|
1230 | NANCHOR(node)->char_len = -1;
|
---|
1231 | NANCHOR(node)->ascii_range = 0;
|
---|
1232 | return node;
|
---|
1233 | }
|
---|
1234 |
|
---|
1235 | static Node*
|
---|
1236 | node_new_backref(int back_num, int* backrefs, int by_name,
|
---|
1237 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
1238 | int exist_level, int nest_level,
|
---|
1239 | #endif
|
---|
1240 | ScanEnv* env)
|
---|
1241 | {
|
---|
1242 | int i;
|
---|
1243 | Node* node = node_new();
|
---|
1244 |
|
---|
1245 | CHECK_NULL_RETURN(node);
|
---|
1246 |
|
---|
1247 | SET_NTYPE(node, NT_BREF);
|
---|
1248 | NBREF(node)->state = 0;
|
---|
1249 | NBREF(node)->back_num = back_num;
|
---|
1250 | NBREF(node)->back_dynamic = (int* )NULL;
|
---|
1251 | if (by_name != 0)
|
---|
1252 | NBREF(node)->state |= NST_NAME_REF;
|
---|
1253 |
|
---|
1254 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
1255 | if (exist_level != 0) {
|
---|
1256 | NBREF(node)->state |= NST_NEST_LEVEL;
|
---|
1257 | NBREF(node)->nest_level = nest_level;
|
---|
1258 | }
|
---|
1259 | #endif
|
---|
1260 |
|
---|
1261 | for (i = 0; i < back_num; i++) {
|
---|
1262 | if (backrefs[i] <= env->num_mem &&
|
---|
1263 | IS_NULL(SCANENV_MEM_NODES(env)[backrefs[i]])) {
|
---|
1264 | NBREF(node)->state |= NST_RECURSION; /* /...(\1).../ */
|
---|
1265 | break;
|
---|
1266 | }
|
---|
1267 | }
|
---|
1268 |
|
---|
1269 | if (back_num <= NODE_BACKREFS_SIZE) {
|
---|
1270 | for (i = 0; i < back_num; i++)
|
---|
1271 | NBREF(node)->back_static[i] = backrefs[i];
|
---|
1272 | }
|
---|
1273 | else {
|
---|
1274 | int* p = (int* )xmalloc(sizeof(int) * back_num);
|
---|
1275 | if (IS_NULL(p)) {
|
---|
1276 | onig_node_free(node);
|
---|
1277 | return NULL;
|
---|
1278 | }
|
---|
1279 | NBREF(node)->back_dynamic = p;
|
---|
1280 | for (i = 0; i < back_num; i++)
|
---|
1281 | p[i] = backrefs[i];
|
---|
1282 | }
|
---|
1283 | return node;
|
---|
1284 | }
|
---|
1285 |
|
---|
1286 | #ifdef USE_SUBEXP_CALL
|
---|
1287 | static Node*
|
---|
1288 | node_new_call(UChar* name, UChar* name_end, int gnum)
|
---|
1289 | {
|
---|
1290 | Node* node = node_new();
|
---|
1291 | CHECK_NULL_RETURN(node);
|
---|
1292 |
|
---|
1293 | SET_NTYPE(node, NT_CALL);
|
---|
1294 | NCALL(node)->state = 0;
|
---|
1295 | NCALL(node)->target = NULL_NODE;
|
---|
1296 | NCALL(node)->name = name;
|
---|
1297 | NCALL(node)->name_end = name_end;
|
---|
1298 | NCALL(node)->group_num = gnum; /* call by number if gnum != 0 */
|
---|
1299 | return node;
|
---|
1300 | }
|
---|
1301 | #endif
|
---|
1302 |
|
---|
1303 | static Node*
|
---|
1304 | node_new_quantifier(int lower, int upper, int by_number)
|
---|
1305 | {
|
---|
1306 | Node* node = node_new();
|
---|
1307 | CHECK_NULL_RETURN(node);
|
---|
1308 |
|
---|
1309 | SET_NTYPE(node, NT_QTFR);
|
---|
1310 | NQTFR(node)->state = 0;
|
---|
1311 | NQTFR(node)->target = NULL;
|
---|
1312 | NQTFR(node)->lower = lower;
|
---|
1313 | NQTFR(node)->upper = upper;
|
---|
1314 | NQTFR(node)->greedy = 1;
|
---|
1315 | NQTFR(node)->target_empty_info = NQ_TARGET_ISNOT_EMPTY;
|
---|
1316 | NQTFR(node)->head_exact = NULL_NODE;
|
---|
1317 | NQTFR(node)->next_head_exact = NULL_NODE;
|
---|
1318 | NQTFR(node)->is_refered = 0;
|
---|
1319 | if (by_number != 0)
|
---|
1320 | NQTFR(node)->state |= NST_BY_NUMBER;
|
---|
1321 |
|
---|
1322 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
1323 | NQTFR(node)->comb_exp_check_num = 0;
|
---|
1324 | #endif
|
---|
1325 |
|
---|
1326 | return node;
|
---|
1327 | }
|
---|
1328 |
|
---|
1329 | static Node*
|
---|
1330 | node_new_enclose(int type)
|
---|
1331 | {
|
---|
1332 | Node* node = node_new();
|
---|
1333 | CHECK_NULL_RETURN(node);
|
---|
1334 |
|
---|
1335 | SET_NTYPE(node, NT_ENCLOSE);
|
---|
1336 | NENCLOSE(node)->type = type;
|
---|
1337 | NENCLOSE(node)->state = 0;
|
---|
1338 | NENCLOSE(node)->regnum = 0;
|
---|
1339 | NENCLOSE(node)->option = 0;
|
---|
1340 | NENCLOSE(node)->target = NULL;
|
---|
1341 | NENCLOSE(node)->call_addr = -1;
|
---|
1342 | NENCLOSE(node)->opt_count = 0;
|
---|
1343 | return node;
|
---|
1344 | }
|
---|
1345 |
|
---|
1346 | extern Node*
|
---|
1347 | onig_node_new_enclose(int type)
|
---|
1348 | {
|
---|
1349 | return node_new_enclose(type);
|
---|
1350 | }
|
---|
1351 |
|
---|
1352 | static Node*
|
---|
1353 | node_new_enclose_memory(OnigOptionType option, int is_named)
|
---|
1354 | {
|
---|
1355 | Node* node = node_new_enclose(ENCLOSE_MEMORY);
|
---|
1356 | CHECK_NULL_RETURN(node);
|
---|
1357 | if (is_named != 0)
|
---|
1358 | SET_ENCLOSE_STATUS(node, NST_NAMED_GROUP);
|
---|
1359 |
|
---|
1360 | #ifdef USE_SUBEXP_CALL
|
---|
1361 | NENCLOSE(node)->option = option;
|
---|
1362 | #endif
|
---|
1363 | return node;
|
---|
1364 | }
|
---|
1365 |
|
---|
1366 | static Node*
|
---|
1367 | node_new_option(OnigOptionType option)
|
---|
1368 | {
|
---|
1369 | Node* node = node_new_enclose(ENCLOSE_OPTION);
|
---|
1370 | CHECK_NULL_RETURN(node);
|
---|
1371 | NENCLOSE(node)->option = option;
|
---|
1372 | return node;
|
---|
1373 | }
|
---|
1374 |
|
---|
1375 | extern int
|
---|
1376 | onig_node_str_cat(Node* node, const UChar* s, const UChar* end)
|
---|
1377 | {
|
---|
1378 | ptrdiff_t addlen = end - s;
|
---|
1379 |
|
---|
1380 | if (addlen > 0) {
|
---|
1381 | ptrdiff_t len = NSTR(node)->end - NSTR(node)->s;
|
---|
1382 |
|
---|
1383 | if (NSTR(node)->capa > 0 || (len + addlen > NODE_STR_BUF_SIZE - 1)) {
|
---|
1384 | UChar* p;
|
---|
1385 | ptrdiff_t capa = len + addlen + NODE_STR_MARGIN;
|
---|
1386 |
|
---|
1387 | if (capa <= NSTR(node)->capa) {
|
---|
1388 | onig_strcpy(NSTR(node)->s + len, s, end);
|
---|
1389 | }
|
---|
1390 | else {
|
---|
1391 | if (NSTR(node)->s == NSTR(node)->buf)
|
---|
1392 | p = strcat_capa_from_static(NSTR(node)->s, NSTR(node)->end,
|
---|
1393 | s, end, capa);
|
---|
1394 | else
|
---|
1395 | p = strcat_capa(NSTR(node)->s, NSTR(node)->end, s, end, capa);
|
---|
1396 |
|
---|
1397 | CHECK_NULL_RETURN_MEMERR(p);
|
---|
1398 | NSTR(node)->s = p;
|
---|
1399 | NSTR(node)->capa = (int )capa;
|
---|
1400 | }
|
---|
1401 | }
|
---|
1402 | else {
|
---|
1403 | onig_strcpy(NSTR(node)->s + len, s, end);
|
---|
1404 | }
|
---|
1405 | NSTR(node)->end = NSTR(node)->s + len + addlen;
|
---|
1406 | }
|
---|
1407 |
|
---|
1408 | return 0;
|
---|
1409 | }
|
---|
1410 |
|
---|
1411 | extern int
|
---|
1412 | onig_node_str_set(Node* node, const UChar* s, const UChar* end)
|
---|
1413 | {
|
---|
1414 | onig_node_str_clear(node);
|
---|
1415 | return onig_node_str_cat(node, s, end);
|
---|
1416 | }
|
---|
1417 |
|
---|
1418 | static int
|
---|
1419 | node_str_cat_char(Node* node, UChar c)
|
---|
1420 | {
|
---|
1421 | UChar s[1];
|
---|
1422 |
|
---|
1423 | s[0] = c;
|
---|
1424 | return onig_node_str_cat(node, s, s + 1);
|
---|
1425 | }
|
---|
1426 |
|
---|
1427 | static int
|
---|
1428 | node_str_cat_codepoint(Node* node, OnigEncoding enc, OnigCodePoint c)
|
---|
1429 | {
|
---|
1430 | UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
|
---|
1431 | int num = ONIGENC_CODE_TO_MBC(enc, c, buf);
|
---|
1432 | if (num < 0) return num;
|
---|
1433 | return onig_node_str_cat(node, buf, buf + num);
|
---|
1434 | }
|
---|
1435 |
|
---|
1436 | #if 0
|
---|
1437 | extern void
|
---|
1438 | onig_node_conv_to_str_node(Node* node, int flag)
|
---|
1439 | {
|
---|
1440 | SET_NTYPE(node, NT_STR);
|
---|
1441 | NSTR(node)->flag = flag;
|
---|
1442 | NSTR(node)->capa = 0;
|
---|
1443 | NSTR(node)->s = NSTR(node)->buf;
|
---|
1444 | NSTR(node)->end = NSTR(node)->buf;
|
---|
1445 | }
|
---|
1446 | #endif
|
---|
1447 |
|
---|
1448 | extern void
|
---|
1449 | onig_node_str_clear(Node* node)
|
---|
1450 | {
|
---|
1451 | if (NSTR(node)->capa != 0 &&
|
---|
1452 | IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) {
|
---|
1453 | xfree(NSTR(node)->s);
|
---|
1454 | }
|
---|
1455 |
|
---|
1456 | NSTR(node)->capa = 0;
|
---|
1457 | NSTR(node)->flag = 0;
|
---|
1458 | NSTR(node)->s = NSTR(node)->buf;
|
---|
1459 | NSTR(node)->end = NSTR(node)->buf;
|
---|
1460 | }
|
---|
1461 |
|
---|
1462 | static Node*
|
---|
1463 | node_new_str(const UChar* s, const UChar* end)
|
---|
1464 | {
|
---|
1465 | Node* node = node_new();
|
---|
1466 | CHECK_NULL_RETURN(node);
|
---|
1467 |
|
---|
1468 | SET_NTYPE(node, NT_STR);
|
---|
1469 | NSTR(node)->capa = 0;
|
---|
1470 | NSTR(node)->flag = 0;
|
---|
1471 | NSTR(node)->s = NSTR(node)->buf;
|
---|
1472 | NSTR(node)->end = NSTR(node)->buf;
|
---|
1473 | if (onig_node_str_cat(node, s, end)) {
|
---|
1474 | onig_node_free(node);
|
---|
1475 | return NULL;
|
---|
1476 | }
|
---|
1477 | return node;
|
---|
1478 | }
|
---|
1479 |
|
---|
1480 | extern Node*
|
---|
1481 | onig_node_new_str(const UChar* s, const UChar* end)
|
---|
1482 | {
|
---|
1483 | return node_new_str(s, end);
|
---|
1484 | }
|
---|
1485 |
|
---|
1486 | static Node*
|
---|
1487 | node_new_str_raw(UChar* s, UChar* end)
|
---|
1488 | {
|
---|
1489 | Node* node = node_new_str(s, end);
|
---|
1490 | if (IS_NOT_NULL(node))
|
---|
1491 | NSTRING_SET_RAW(node);
|
---|
1492 | return node;
|
---|
1493 | }
|
---|
1494 |
|
---|
1495 | static Node*
|
---|
1496 | node_new_empty(void)
|
---|
1497 | {
|
---|
1498 | return node_new_str(NULL, NULL);
|
---|
1499 | }
|
---|
1500 |
|
---|
1501 | static Node*
|
---|
1502 | node_new_str_raw_char(UChar c)
|
---|
1503 | {
|
---|
1504 | UChar p[1];
|
---|
1505 |
|
---|
1506 | p[0] = c;
|
---|
1507 | return node_new_str_raw(p, p + 1);
|
---|
1508 | }
|
---|
1509 |
|
---|
1510 | static Node*
|
---|
1511 | str_node_split_last_char(StrNode* sn, OnigEncoding enc)
|
---|
1512 | {
|
---|
1513 | const UChar *p;
|
---|
1514 | Node* n = NULL_NODE;
|
---|
1515 |
|
---|
1516 | if (sn->end > sn->s) {
|
---|
1517 | p = onigenc_get_prev_char_head(enc, sn->s, sn->end, sn->end);
|
---|
1518 | if (p && p > sn->s) { /* can be split. */
|
---|
1519 | n = node_new_str(p, sn->end);
|
---|
1520 | if (IS_NOT_NULL(n) && (sn->flag & NSTR_RAW) != 0)
|
---|
1521 | NSTRING_SET_RAW(n);
|
---|
1522 | sn->end = (UChar* )p;
|
---|
1523 | }
|
---|
1524 | }
|
---|
1525 | return n;
|
---|
1526 | }
|
---|
1527 |
|
---|
1528 | static int
|
---|
1529 | str_node_can_be_split(StrNode* sn, OnigEncoding enc)
|
---|
1530 | {
|
---|
1531 | if (sn->end > sn->s) {
|
---|
1532 | return ((enclen(enc, sn->s, sn->end) < sn->end - sn->s) ? 1 : 0);
|
---|
1533 | }
|
---|
1534 | return 0;
|
---|
1535 | }
|
---|
1536 |
|
---|
1537 | #ifdef USE_PAD_TO_SHORT_BYTE_CHAR
|
---|
1538 | static int
|
---|
1539 | node_str_head_pad(StrNode* sn, int num, UChar val)
|
---|
1540 | {
|
---|
1541 | UChar buf[NODE_STR_BUF_SIZE];
|
---|
1542 | int i, len;
|
---|
1543 |
|
---|
1544 | len = sn->end - sn->s;
|
---|
1545 | onig_strcpy(buf, sn->s, sn->end);
|
---|
1546 | onig_strcpy(&(sn->s[num]), buf, buf + len);
|
---|
1547 | sn->end += num;
|
---|
1548 |
|
---|
1549 | for (i = 0; i < num; i++) {
|
---|
1550 | sn->s[i] = val;
|
---|
1551 | }
|
---|
1552 | }
|
---|
1553 | #endif
|
---|
1554 |
|
---|
1555 | extern int
|
---|
1556 | onig_scan_unsigned_number(UChar** src, const UChar* end, OnigEncoding enc)
|
---|
1557 | {
|
---|
1558 | unsigned int num, val;
|
---|
1559 | OnigCodePoint c;
|
---|
1560 | UChar* p = *src;
|
---|
1561 | PFETCH_READY;
|
---|
1562 |
|
---|
1563 | num = 0;
|
---|
1564 | while (!PEND) {
|
---|
1565 | PFETCH(c);
|
---|
1566 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
|
---|
1567 | val = (unsigned int )DIGITVAL(c);
|
---|
1568 | if ((INT_MAX_LIMIT - val) / 10UL < num)
|
---|
1569 | return -1; /* overflow */
|
---|
1570 |
|
---|
1571 | num = num * 10 + val;
|
---|
1572 | }
|
---|
1573 | else {
|
---|
1574 | PUNFETCH;
|
---|
1575 | break;
|
---|
1576 | }
|
---|
1577 | }
|
---|
1578 | *src = p;
|
---|
1579 | return num;
|
---|
1580 | }
|
---|
1581 |
|
---|
1582 | static int
|
---|
1583 | scan_unsigned_hexadecimal_number(UChar** src, UChar* end, int minlen,
|
---|
1584 | int maxlen, OnigEncoding enc)
|
---|
1585 | {
|
---|
1586 | OnigCodePoint c;
|
---|
1587 | unsigned int num, val;
|
---|
1588 | int restlen;
|
---|
1589 | UChar* p = *src;
|
---|
1590 | PFETCH_READY;
|
---|
1591 |
|
---|
1592 | restlen = maxlen - minlen;
|
---|
1593 | num = 0;
|
---|
1594 | while (!PEND && maxlen-- != 0) {
|
---|
1595 | PFETCH(c);
|
---|
1596 | if (ONIGENC_IS_CODE_XDIGIT(enc, c)) {
|
---|
1597 | val = (unsigned int )XDIGITVAL(enc,c);
|
---|
1598 | if ((INT_MAX_LIMIT - val) / 16UL < num)
|
---|
1599 | return -1; /* overflow */
|
---|
1600 |
|
---|
1601 | num = (num << 4) + XDIGITVAL(enc,c);
|
---|
1602 | }
|
---|
1603 | else {
|
---|
1604 | PUNFETCH;
|
---|
1605 | maxlen++;
|
---|
1606 | break;
|
---|
1607 | }
|
---|
1608 | }
|
---|
1609 | if (maxlen > restlen)
|
---|
1610 | return -2; /* not enough digits */
|
---|
1611 | *src = p;
|
---|
1612 | return num;
|
---|
1613 | }
|
---|
1614 |
|
---|
1615 | static int
|
---|
1616 | scan_unsigned_octal_number(UChar** src, UChar* end, int maxlen,
|
---|
1617 | OnigEncoding enc)
|
---|
1618 | {
|
---|
1619 | OnigCodePoint c;
|
---|
1620 | unsigned int num, val;
|
---|
1621 | UChar* p = *src;
|
---|
1622 | PFETCH_READY;
|
---|
1623 |
|
---|
1624 | num = 0;
|
---|
1625 | while (!PEND && maxlen-- != 0) {
|
---|
1626 | PFETCH(c);
|
---|
1627 | if (ONIGENC_IS_CODE_DIGIT(enc, c) && c < '8') {
|
---|
1628 | val = ODIGITVAL(c);
|
---|
1629 | if ((INT_MAX_LIMIT - val) / 8UL < num)
|
---|
1630 | return -1; /* overflow */
|
---|
1631 |
|
---|
1632 | num = (num << 3) + val;
|
---|
1633 | }
|
---|
1634 | else {
|
---|
1635 | PUNFETCH;
|
---|
1636 | break;
|
---|
1637 | }
|
---|
1638 | }
|
---|
1639 | *src = p;
|
---|
1640 | return num;
|
---|
1641 | }
|
---|
1642 |
|
---|
1643 |
|
---|
1644 | #define BBUF_WRITE_CODE_POINT(bbuf,pos,code) \
|
---|
1645 | BBUF_WRITE(bbuf, pos, &(code), SIZE_CODE_POINT)
|
---|
1646 |
|
---|
1647 | /* data format:
|
---|
1648 | [n][from-1][to-1][from-2][to-2] ... [from-n][to-n]
|
---|
1649 | (all data size is OnigCodePoint)
|
---|
1650 | */
|
---|
1651 | static int
|
---|
1652 | new_code_range(BBuf** pbuf)
|
---|
1653 | {
|
---|
1654 | #define INIT_MULTI_BYTE_RANGE_SIZE (SIZE_CODE_POINT * 5)
|
---|
1655 | int r;
|
---|
1656 | OnigCodePoint n;
|
---|
1657 | BBuf* bbuf;
|
---|
1658 |
|
---|
1659 | bbuf = *pbuf = (BBuf* )xmalloc(sizeof(BBuf));
|
---|
1660 | CHECK_NULL_RETURN_MEMERR(*pbuf);
|
---|
1661 | r = BBUF_INIT(*pbuf, INIT_MULTI_BYTE_RANGE_SIZE);
|
---|
1662 | if (r) return r;
|
---|
1663 |
|
---|
1664 | n = 0;
|
---|
1665 | BBUF_WRITE_CODE_POINT(bbuf, 0, n);
|
---|
1666 | return 0;
|
---|
1667 | }
|
---|
1668 |
|
---|
1669 | static int
|
---|
1670 | add_code_range_to_buf0(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePoint to,
|
---|
1671 | int checkdup)
|
---|
1672 | {
|
---|
1673 | int r, inc_n, pos;
|
---|
1674 | OnigCodePoint low, high, bound, x;
|
---|
1675 | OnigCodePoint n, *data;
|
---|
1676 | BBuf* bbuf;
|
---|
1677 |
|
---|
1678 | if (from > to) {
|
---|
1679 | n = from; from = to; to = n;
|
---|
1680 | }
|
---|
1681 |
|
---|
1682 | if (IS_NULL(*pbuf)) {
|
---|
1683 | r = new_code_range(pbuf);
|
---|
1684 | if (r) return r;
|
---|
1685 | bbuf = *pbuf;
|
---|
1686 | n = 0;
|
---|
1687 | }
|
---|
1688 | else {
|
---|
1689 | bbuf = *pbuf;
|
---|
1690 | GET_CODE_POINT(n, bbuf->p);
|
---|
1691 | }
|
---|
1692 | data = (OnigCodePoint* )(bbuf->p);
|
---|
1693 | data++;
|
---|
1694 |
|
---|
1695 | bound = (from == 0) ? 0 : n;
|
---|
1696 | for (low = 0; low < bound; ) {
|
---|
1697 | x = (low + bound) >> 1;
|
---|
1698 | if (from - 1 > data[x*2 + 1])
|
---|
1699 | low = x + 1;
|
---|
1700 | else
|
---|
1701 | bound = x;
|
---|
1702 | }
|
---|
1703 |
|
---|
1704 | high = (to == ONIG_LAST_CODE_POINT) ? n : low;
|
---|
1705 | for (bound = n; high < bound; ) {
|
---|
1706 | x = (high + bound) >> 1;
|
---|
1707 | if (to + 1 >= data[x*2])
|
---|
1708 | high = x + 1;
|
---|
1709 | else
|
---|
1710 | bound = x;
|
---|
1711 | }
|
---|
1712 | /* data[(low-1)*2+1] << from <= data[low*2]
|
---|
1713 | * data[(high-1)*2+1] <= to << data[high*2]
|
---|
1714 | */
|
---|
1715 |
|
---|
1716 | inc_n = low + 1 - high;
|
---|
1717 | if (n + inc_n > ONIG_MAX_MULTI_BYTE_RANGES_NUM)
|
---|
1718 | return ONIGERR_TOO_MANY_MULTI_BYTE_RANGES;
|
---|
1719 |
|
---|
1720 | if (inc_n != 1) {
|
---|
1721 | if (checkdup && from <= data[low*2+1]
|
---|
1722 | && (data[low*2] <= from || data[low*2+1] <= to))
|
---|
1723 | CC_DUP_WARN(env);
|
---|
1724 | if (from > data[low*2])
|
---|
1725 | from = data[low*2];
|
---|
1726 | if (to < data[(high - 1)*2 + 1])
|
---|
1727 | to = data[(high - 1)*2 + 1];
|
---|
1728 | }
|
---|
1729 |
|
---|
1730 | if (inc_n != 0) {
|
---|
1731 | int from_pos = SIZE_CODE_POINT * (1 + high * 2);
|
---|
1732 | int to_pos = SIZE_CODE_POINT * (1 + (low + 1) * 2);
|
---|
1733 |
|
---|
1734 | if (inc_n > 0) {
|
---|
1735 | if (high < n) {
|
---|
1736 | int size = (n - high) * 2 * SIZE_CODE_POINT;
|
---|
1737 | BBUF_MOVE_RIGHT(bbuf, from_pos, to_pos, size);
|
---|
1738 | }
|
---|
1739 | }
|
---|
1740 | else {
|
---|
1741 | BBUF_MOVE_LEFT_REDUCE(bbuf, from_pos, to_pos);
|
---|
1742 | }
|
---|
1743 | }
|
---|
1744 |
|
---|
1745 | pos = SIZE_CODE_POINT * (1 + low * 2);
|
---|
1746 | BBUF_ENSURE_SIZE(bbuf, pos + SIZE_CODE_POINT * 2);
|
---|
1747 | BBUF_WRITE_CODE_POINT(bbuf, pos, from);
|
---|
1748 | BBUF_WRITE_CODE_POINT(bbuf, pos + SIZE_CODE_POINT, to);
|
---|
1749 | n += inc_n;
|
---|
1750 | BBUF_WRITE_CODE_POINT(bbuf, 0, n);
|
---|
1751 |
|
---|
1752 | return 0;
|
---|
1753 | }
|
---|
1754 |
|
---|
1755 | static int
|
---|
1756 | add_code_range_to_buf(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePoint to)
|
---|
1757 | {
|
---|
1758 | return add_code_range_to_buf0(pbuf, env, from, to, 1);
|
---|
1759 | }
|
---|
1760 |
|
---|
1761 | static int
|
---|
1762 | add_code_range0(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePoint to, int checkdup)
|
---|
1763 | {
|
---|
1764 | if (from > to) {
|
---|
1765 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
|
---|
1766 | return 0;
|
---|
1767 | else
|
---|
1768 | return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
|
---|
1769 | }
|
---|
1770 |
|
---|
1771 | return add_code_range_to_buf0(pbuf, env, from, to, checkdup);
|
---|
1772 | }
|
---|
1773 |
|
---|
1774 | static int
|
---|
1775 | add_code_range(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePoint to)
|
---|
1776 | {
|
---|
1777 | return add_code_range0(pbuf, env, from, to, 1);
|
---|
1778 | }
|
---|
1779 |
|
---|
1780 | static int
|
---|
1781 | not_code_range_buf(OnigEncoding enc, BBuf* bbuf, BBuf** pbuf, ScanEnv* env)
|
---|
1782 | {
|
---|
1783 | int r, i, n;
|
---|
1784 | OnigCodePoint pre, from, *data, to = 0;
|
---|
1785 |
|
---|
1786 | *pbuf = (BBuf* )NULL;
|
---|
1787 | if (IS_NULL(bbuf)) {
|
---|
1788 | set_all:
|
---|
1789 | return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
|
---|
1790 | }
|
---|
1791 |
|
---|
1792 | data = (OnigCodePoint* )(bbuf->p);
|
---|
1793 | GET_CODE_POINT(n, data);
|
---|
1794 | data++;
|
---|
1795 | if (n <= 0) goto set_all;
|
---|
1796 |
|
---|
1797 | r = 0;
|
---|
1798 | pre = MBCODE_START_POS(enc);
|
---|
1799 | for (i = 0; i < n; i++) {
|
---|
1800 | from = data[i*2];
|
---|
1801 | to = data[i*2+1];
|
---|
1802 | if (pre <= from - 1) {
|
---|
1803 | r = add_code_range_to_buf(pbuf, env, pre, from - 1);
|
---|
1804 | if (r != 0) return r;
|
---|
1805 | }
|
---|
1806 | if (to == ONIG_LAST_CODE_POINT) break;
|
---|
1807 | pre = to + 1;
|
---|
1808 | }
|
---|
1809 | if (to < ONIG_LAST_CODE_POINT) {
|
---|
1810 | r = add_code_range_to_buf(pbuf, env, to + 1, ONIG_LAST_CODE_POINT);
|
---|
1811 | }
|
---|
1812 | return r;
|
---|
1813 | }
|
---|
1814 |
|
---|
1815 | #define SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2) do {\
|
---|
1816 | BBuf *tbuf; \
|
---|
1817 | int tnot; \
|
---|
1818 | tnot = not1; not1 = not2; not2 = tnot; \
|
---|
1819 | tbuf = bbuf1; bbuf1 = bbuf2; bbuf2 = tbuf; \
|
---|
1820 | } while (0)
|
---|
1821 |
|
---|
1822 | static int
|
---|
1823 | or_code_range_buf(OnigEncoding enc, BBuf* bbuf1, int not1,
|
---|
1824 | BBuf* bbuf2, int not2, BBuf** pbuf, ScanEnv* env)
|
---|
1825 | {
|
---|
1826 | int r;
|
---|
1827 | OnigCodePoint i, n1, *data1;
|
---|
1828 | OnigCodePoint from, to;
|
---|
1829 |
|
---|
1830 | *pbuf = (BBuf* )NULL;
|
---|
1831 | if (IS_NULL(bbuf1) && IS_NULL(bbuf2)) {
|
---|
1832 | if (not1 != 0 || not2 != 0)
|
---|
1833 | return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
|
---|
1834 | return 0;
|
---|
1835 | }
|
---|
1836 |
|
---|
1837 | r = 0;
|
---|
1838 | if (IS_NULL(bbuf2))
|
---|
1839 | SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
|
---|
1840 |
|
---|
1841 | if (IS_NULL(bbuf1)) {
|
---|
1842 | if (not1 != 0) {
|
---|
1843 | return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
|
---|
1844 | }
|
---|
1845 | else {
|
---|
1846 | if (not2 == 0) {
|
---|
1847 | return bbuf_clone(pbuf, bbuf2);
|
---|
1848 | }
|
---|
1849 | else {
|
---|
1850 | return not_code_range_buf(enc, bbuf2, pbuf, env);
|
---|
1851 | }
|
---|
1852 | }
|
---|
1853 | }
|
---|
1854 |
|
---|
1855 | if (not1 != 0)
|
---|
1856 | SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
|
---|
1857 |
|
---|
1858 | data1 = (OnigCodePoint* )(bbuf1->p);
|
---|
1859 | GET_CODE_POINT(n1, data1);
|
---|
1860 | data1++;
|
---|
1861 |
|
---|
1862 | if (not2 == 0 && not1 == 0) { /* 1 OR 2 */
|
---|
1863 | r = bbuf_clone(pbuf, bbuf2);
|
---|
1864 | }
|
---|
1865 | else if (not1 == 0) { /* 1 OR (not 2) */
|
---|
1866 | r = not_code_range_buf(enc, bbuf2, pbuf, env);
|
---|
1867 | }
|
---|
1868 | if (r != 0) return r;
|
---|
1869 |
|
---|
1870 | for (i = 0; i < n1; i++) {
|
---|
1871 | from = data1[i*2];
|
---|
1872 | to = data1[i*2+1];
|
---|
1873 | r = add_code_range_to_buf(pbuf, env, from, to);
|
---|
1874 | if (r != 0) return r;
|
---|
1875 | }
|
---|
1876 | return 0;
|
---|
1877 | }
|
---|
1878 |
|
---|
1879 | static int
|
---|
1880 | and_code_range1(BBuf** pbuf, ScanEnv* env, OnigCodePoint from1, OnigCodePoint to1,
|
---|
1881 | OnigCodePoint* data, int n)
|
---|
1882 | {
|
---|
1883 | int i, r;
|
---|
1884 | OnigCodePoint from2, to2;
|
---|
1885 |
|
---|
1886 | for (i = 0; i < n; i++) {
|
---|
1887 | from2 = data[i*2];
|
---|
1888 | to2 = data[i*2+1];
|
---|
1889 | if (from2 < from1) {
|
---|
1890 | if (to2 < from1) continue;
|
---|
1891 | else {
|
---|
1892 | from1 = to2 + 1;
|
---|
1893 | }
|
---|
1894 | }
|
---|
1895 | else if (from2 <= to1) {
|
---|
1896 | if (to2 < to1) {
|
---|
1897 | if (from1 <= from2 - 1) {
|
---|
1898 | r = add_code_range_to_buf(pbuf, env, from1, from2-1);
|
---|
1899 | if (r != 0) return r;
|
---|
1900 | }
|
---|
1901 | from1 = to2 + 1;
|
---|
1902 | }
|
---|
1903 | else {
|
---|
1904 | to1 = from2 - 1;
|
---|
1905 | }
|
---|
1906 | }
|
---|
1907 | else {
|
---|
1908 | from1 = from2;
|
---|
1909 | }
|
---|
1910 | if (from1 > to1) break;
|
---|
1911 | }
|
---|
1912 | if (from1 <= to1) {
|
---|
1913 | r = add_code_range_to_buf(pbuf, env, from1, to1);
|
---|
1914 | if (r != 0) return r;
|
---|
1915 | }
|
---|
1916 | return 0;
|
---|
1917 | }
|
---|
1918 |
|
---|
1919 | static int
|
---|
1920 | and_code_range_buf(BBuf* bbuf1, int not1, BBuf* bbuf2, int not2, BBuf** pbuf, ScanEnv* env)
|
---|
1921 | {
|
---|
1922 | int r;
|
---|
1923 | OnigCodePoint i, j, n1, n2, *data1, *data2;
|
---|
1924 | OnigCodePoint from, to, from1, to1, from2, to2;
|
---|
1925 |
|
---|
1926 | *pbuf = (BBuf* )NULL;
|
---|
1927 | if (IS_NULL(bbuf1)) {
|
---|
1928 | if (not1 != 0 && IS_NOT_NULL(bbuf2)) /* not1 != 0 -> not2 == 0 */
|
---|
1929 | return bbuf_clone(pbuf, bbuf2);
|
---|
1930 | return 0;
|
---|
1931 | }
|
---|
1932 | else if (IS_NULL(bbuf2)) {
|
---|
1933 | if (not2 != 0)
|
---|
1934 | return bbuf_clone(pbuf, bbuf1);
|
---|
1935 | return 0;
|
---|
1936 | }
|
---|
1937 |
|
---|
1938 | if (not1 != 0)
|
---|
1939 | SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
|
---|
1940 |
|
---|
1941 | data1 = (OnigCodePoint* )(bbuf1->p);
|
---|
1942 | data2 = (OnigCodePoint* )(bbuf2->p);
|
---|
1943 | GET_CODE_POINT(n1, data1);
|
---|
1944 | GET_CODE_POINT(n2, data2);
|
---|
1945 | data1++;
|
---|
1946 | data2++;
|
---|
1947 |
|
---|
1948 | if (not2 == 0 && not1 == 0) { /* 1 AND 2 */
|
---|
1949 | for (i = 0; i < n1; i++) {
|
---|
1950 | from1 = data1[i*2];
|
---|
1951 | to1 = data1[i*2+1];
|
---|
1952 | for (j = 0; j < n2; j++) {
|
---|
1953 | from2 = data2[j*2];
|
---|
1954 | to2 = data2[j*2+1];
|
---|
1955 | if (from2 > to1) break;
|
---|
1956 | if (to2 < from1) continue;
|
---|
1957 | from = MAX(from1, from2);
|
---|
1958 | to = MIN(to1, to2);
|
---|
1959 | r = add_code_range_to_buf(pbuf, env, from, to);
|
---|
1960 | if (r != 0) return r;
|
---|
1961 | }
|
---|
1962 | }
|
---|
1963 | }
|
---|
1964 | else if (not1 == 0) { /* 1 AND (not 2) */
|
---|
1965 | for (i = 0; i < n1; i++) {
|
---|
1966 | from1 = data1[i*2];
|
---|
1967 | to1 = data1[i*2+1];
|
---|
1968 | r = and_code_range1(pbuf, env, from1, to1, data2, n2);
|
---|
1969 | if (r != 0) return r;
|
---|
1970 | }
|
---|
1971 | }
|
---|
1972 |
|
---|
1973 | return 0;
|
---|
1974 | }
|
---|
1975 |
|
---|
1976 | static int
|
---|
1977 | and_cclass(CClassNode* dest, CClassNode* cc, ScanEnv* env)
|
---|
1978 | {
|
---|
1979 | OnigEncoding enc = env->enc;
|
---|
1980 | int r, not1, not2;
|
---|
1981 | BBuf *buf1, *buf2, *pbuf = 0;
|
---|
1982 | BitSetRef bsr1, bsr2;
|
---|
1983 | BitSet bs1, bs2;
|
---|
1984 |
|
---|
1985 | not1 = IS_NCCLASS_NOT(dest);
|
---|
1986 | bsr1 = dest->bs;
|
---|
1987 | buf1 = dest->mbuf;
|
---|
1988 | not2 = IS_NCCLASS_NOT(cc);
|
---|
1989 | bsr2 = cc->bs;
|
---|
1990 | buf2 = cc->mbuf;
|
---|
1991 |
|
---|
1992 | if (not1 != 0) {
|
---|
1993 | bitset_invert_to(bsr1, bs1);
|
---|
1994 | bsr1 = bs1;
|
---|
1995 | }
|
---|
1996 | if (not2 != 0) {
|
---|
1997 | bitset_invert_to(bsr2, bs2);
|
---|
1998 | bsr2 = bs2;
|
---|
1999 | }
|
---|
2000 | bitset_and(bsr1, bsr2);
|
---|
2001 | if (bsr1 != dest->bs) {
|
---|
2002 | bitset_copy(dest->bs, bsr1);
|
---|
2003 | bsr1 = dest->bs;
|
---|
2004 | }
|
---|
2005 | if (not1 != 0) {
|
---|
2006 | bitset_invert(dest->bs);
|
---|
2007 | }
|
---|
2008 |
|
---|
2009 | if (! ONIGENC_IS_SINGLEBYTE(enc)) {
|
---|
2010 | if (not1 != 0 && not2 != 0) {
|
---|
2011 | r = or_code_range_buf(enc, buf1, 0, buf2, 0, &pbuf, env);
|
---|
2012 | }
|
---|
2013 | else {
|
---|
2014 | r = and_code_range_buf(buf1, not1, buf2, not2, &pbuf, env);
|
---|
2015 | if (r == 0 && not1 != 0) {
|
---|
2016 | BBuf *tbuf = 0;
|
---|
2017 | r = not_code_range_buf(enc, pbuf, &tbuf, env);
|
---|
2018 | bbuf_free(pbuf);
|
---|
2019 | pbuf = tbuf;
|
---|
2020 | }
|
---|
2021 | }
|
---|
2022 | if (r != 0) {
|
---|
2023 | bbuf_free(pbuf);
|
---|
2024 | return r;
|
---|
2025 | }
|
---|
2026 |
|
---|
2027 | dest->mbuf = pbuf;
|
---|
2028 | bbuf_free(buf1);
|
---|
2029 | return r;
|
---|
2030 | }
|
---|
2031 | return 0;
|
---|
2032 | }
|
---|
2033 |
|
---|
2034 | static int
|
---|
2035 | or_cclass(CClassNode* dest, CClassNode* cc, ScanEnv* env)
|
---|
2036 | {
|
---|
2037 | OnigEncoding enc = env->enc;
|
---|
2038 | int r, not1, not2;
|
---|
2039 | BBuf *buf1, *buf2, *pbuf = 0;
|
---|
2040 | BitSetRef bsr1, bsr2;
|
---|
2041 | BitSet bs1, bs2;
|
---|
2042 |
|
---|
2043 | not1 = IS_NCCLASS_NOT(dest);
|
---|
2044 | bsr1 = dest->bs;
|
---|
2045 | buf1 = dest->mbuf;
|
---|
2046 | not2 = IS_NCCLASS_NOT(cc);
|
---|
2047 | bsr2 = cc->bs;
|
---|
2048 | buf2 = cc->mbuf;
|
---|
2049 |
|
---|
2050 | if (not1 != 0) {
|
---|
2051 | bitset_invert_to(bsr1, bs1);
|
---|
2052 | bsr1 = bs1;
|
---|
2053 | }
|
---|
2054 | if (not2 != 0) {
|
---|
2055 | bitset_invert_to(bsr2, bs2);
|
---|
2056 | bsr2 = bs2;
|
---|
2057 | }
|
---|
2058 | bitset_or(bsr1, bsr2);
|
---|
2059 | if (bsr1 != dest->bs) {
|
---|
2060 | bitset_copy(dest->bs, bsr1);
|
---|
2061 | bsr1 = dest->bs;
|
---|
2062 | }
|
---|
2063 | if (not1 != 0) {
|
---|
2064 | bitset_invert(dest->bs);
|
---|
2065 | }
|
---|
2066 |
|
---|
2067 | if (! ONIGENC_IS_SINGLEBYTE(enc)) {
|
---|
2068 | if (not1 != 0 && not2 != 0) {
|
---|
2069 | r = and_code_range_buf(buf1, 0, buf2, 0, &pbuf, env);
|
---|
2070 | }
|
---|
2071 | else {
|
---|
2072 | r = or_code_range_buf(enc, buf1, not1, buf2, not2, &pbuf, env);
|
---|
2073 | if (r == 0 && not1 != 0) {
|
---|
2074 | BBuf *tbuf = 0;
|
---|
2075 | r = not_code_range_buf(enc, pbuf, &tbuf, env);
|
---|
2076 | bbuf_free(pbuf);
|
---|
2077 | pbuf = tbuf;
|
---|
2078 | }
|
---|
2079 | }
|
---|
2080 | if (r != 0) {
|
---|
2081 | bbuf_free(pbuf);
|
---|
2082 | return r;
|
---|
2083 | }
|
---|
2084 |
|
---|
2085 | dest->mbuf = pbuf;
|
---|
2086 | bbuf_free(buf1);
|
---|
2087 | return r;
|
---|
2088 | }
|
---|
2089 | else
|
---|
2090 | return 0;
|
---|
2091 | }
|
---|
2092 |
|
---|
2093 | static void UNKNOWN_ESC_WARN(ScanEnv *env, int c);
|
---|
2094 |
|
---|
2095 | static OnigCodePoint
|
---|
2096 | conv_backslash_value(OnigCodePoint c, ScanEnv* env)
|
---|
2097 | {
|
---|
2098 | if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_CONTROL_CHARS)) {
|
---|
2099 | switch (c) {
|
---|
2100 | case 'n': return '\n';
|
---|
2101 | case 't': return '\t';
|
---|
2102 | case 'r': return '\r';
|
---|
2103 | case 'f': return '\f';
|
---|
2104 | case 'a': return '\007';
|
---|
2105 | case 'b': return '\010';
|
---|
2106 | case 'e': return '\033';
|
---|
2107 | case 'v':
|
---|
2108 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_V_VTAB))
|
---|
2109 | return '\v';
|
---|
2110 | break;
|
---|
2111 |
|
---|
2112 | default:
|
---|
2113 | if (('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z'))
|
---|
2114 | UNKNOWN_ESC_WARN(env, c);
|
---|
2115 | break;
|
---|
2116 | }
|
---|
2117 | }
|
---|
2118 | return c;
|
---|
2119 | }
|
---|
2120 |
|
---|
2121 | #ifdef USE_NO_INVALID_QUANTIFIER
|
---|
2122 | # define is_invalid_quantifier_target(node) 0
|
---|
2123 | #else
|
---|
2124 | static int
|
---|
2125 | is_invalid_quantifier_target(Node* node)
|
---|
2126 | {
|
---|
2127 | switch (NTYPE(node)) {
|
---|
2128 | case NT_ANCHOR:
|
---|
2129 | return 1;
|
---|
2130 | break;
|
---|
2131 |
|
---|
2132 | case NT_ENCLOSE:
|
---|
2133 | /* allow enclosed elements */
|
---|
2134 | /* return is_invalid_quantifier_target(NENCLOSE(node)->target); */
|
---|
2135 | break;
|
---|
2136 |
|
---|
2137 | case NT_LIST:
|
---|
2138 | do {
|
---|
2139 | if (! is_invalid_quantifier_target(NCAR(node))) return 0;
|
---|
2140 | } while (IS_NOT_NULL(node = NCDR(node)));
|
---|
2141 | return 0;
|
---|
2142 | break;
|
---|
2143 |
|
---|
2144 | case NT_ALT:
|
---|
2145 | do {
|
---|
2146 | if (is_invalid_quantifier_target(NCAR(node))) return 1;
|
---|
2147 | } while (IS_NOT_NULL(node = NCDR(node)));
|
---|
2148 | break;
|
---|
2149 |
|
---|
2150 | default:
|
---|
2151 | break;
|
---|
2152 | }
|
---|
2153 | return 0;
|
---|
2154 | }
|
---|
2155 | #endif
|
---|
2156 |
|
---|
2157 | /* ?:0, *:1, +:2, ??:3, *?:4, +?:5 */
|
---|
2158 | static int
|
---|
2159 | popular_quantifier_num(QtfrNode* q)
|
---|
2160 | {
|
---|
2161 | if (q->greedy) {
|
---|
2162 | if (q->lower == 0) {
|
---|
2163 | if (q->upper == 1) return 0;
|
---|
2164 | else if (IS_REPEAT_INFINITE(q->upper)) return 1;
|
---|
2165 | }
|
---|
2166 | else if (q->lower == 1) {
|
---|
2167 | if (IS_REPEAT_INFINITE(q->upper)) return 2;
|
---|
2168 | }
|
---|
2169 | }
|
---|
2170 | else {
|
---|
2171 | if (q->lower == 0) {
|
---|
2172 | if (q->upper == 1) return 3;
|
---|
2173 | else if (IS_REPEAT_INFINITE(q->upper)) return 4;
|
---|
2174 | }
|
---|
2175 | else if (q->lower == 1) {
|
---|
2176 | if (IS_REPEAT_INFINITE(q->upper)) return 5;
|
---|
2177 | }
|
---|
2178 | }
|
---|
2179 | return -1;
|
---|
2180 | }
|
---|
2181 |
|
---|
2182 |
|
---|
2183 | enum ReduceType {
|
---|
2184 | RQ_ASIS = 0, /* as is */
|
---|
2185 | RQ_DEL = 1, /* delete parent */
|
---|
2186 | RQ_A, /* to '*' */
|
---|
2187 | RQ_AQ, /* to '*?' */
|
---|
2188 | RQ_QQ, /* to '??' */
|
---|
2189 | RQ_P_QQ, /* to '+)??' */
|
---|
2190 | RQ_PQ_Q /* to '+?)?' */
|
---|
2191 | };
|
---|
2192 |
|
---|
2193 | static enum ReduceType const ReduceTypeTable[6][6] = {
|
---|
2194 | /* '?', '*', '+', '??', '*?', '+?' p / c */
|
---|
2195 | {RQ_DEL, RQ_A, RQ_A, RQ_QQ, RQ_AQ, RQ_ASIS}, /* '?' */
|
---|
2196 | {RQ_DEL, RQ_DEL, RQ_DEL, RQ_P_QQ, RQ_P_QQ, RQ_DEL}, /* '*' */
|
---|
2197 | {RQ_A, RQ_A, RQ_DEL, RQ_ASIS, RQ_P_QQ, RQ_DEL}, /* '+' */
|
---|
2198 | {RQ_DEL, RQ_AQ, RQ_AQ, RQ_DEL, RQ_AQ, RQ_AQ}, /* '??' */
|
---|
2199 | {RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL}, /* '*?' */
|
---|
2200 | {RQ_ASIS, RQ_PQ_Q, RQ_DEL, RQ_AQ, RQ_AQ, RQ_DEL} /* '+?' */
|
---|
2201 | };
|
---|
2202 |
|
---|
2203 | extern void
|
---|
2204 | onig_reduce_nested_quantifier(Node* pnode, Node* cnode)
|
---|
2205 | {
|
---|
2206 | int pnum, cnum;
|
---|
2207 | QtfrNode *p, *c;
|
---|
2208 |
|
---|
2209 | p = NQTFR(pnode);
|
---|
2210 | c = NQTFR(cnode);
|
---|
2211 | pnum = popular_quantifier_num(p);
|
---|
2212 | cnum = popular_quantifier_num(c);
|
---|
2213 | if (pnum < 0 || cnum < 0) return ;
|
---|
2214 |
|
---|
2215 | switch (ReduceTypeTable[cnum][pnum]) {
|
---|
2216 | case RQ_DEL:
|
---|
2217 | *pnode = *cnode;
|
---|
2218 | break;
|
---|
2219 | case RQ_A:
|
---|
2220 | p->target = c->target;
|
---|
2221 | p->lower = 0; p->upper = REPEAT_INFINITE; p->greedy = 1;
|
---|
2222 | break;
|
---|
2223 | case RQ_AQ:
|
---|
2224 | p->target = c->target;
|
---|
2225 | p->lower = 0; p->upper = REPEAT_INFINITE; p->greedy = 0;
|
---|
2226 | break;
|
---|
2227 | case RQ_QQ:
|
---|
2228 | p->target = c->target;
|
---|
2229 | p->lower = 0; p->upper = 1; p->greedy = 0;
|
---|
2230 | break;
|
---|
2231 | case RQ_P_QQ:
|
---|
2232 | p->target = cnode;
|
---|
2233 | p->lower = 0; p->upper = 1; p->greedy = 0;
|
---|
2234 | c->lower = 1; c->upper = REPEAT_INFINITE; c->greedy = 1;
|
---|
2235 | return ;
|
---|
2236 | break;
|
---|
2237 | case RQ_PQ_Q:
|
---|
2238 | p->target = cnode;
|
---|
2239 | p->lower = 0; p->upper = 1; p->greedy = 1;
|
---|
2240 | c->lower = 1; c->upper = REPEAT_INFINITE; c->greedy = 0;
|
---|
2241 | return ;
|
---|
2242 | break;
|
---|
2243 | case RQ_ASIS:
|
---|
2244 | p->target = cnode;
|
---|
2245 | return ;
|
---|
2246 | break;
|
---|
2247 | }
|
---|
2248 |
|
---|
2249 | c->target = NULL_NODE;
|
---|
2250 | onig_node_free(cnode);
|
---|
2251 | }
|
---|
2252 |
|
---|
2253 |
|
---|
2254 | enum TokenSyms {
|
---|
2255 | TK_EOT = 0, /* end of token */
|
---|
2256 | TK_RAW_BYTE = 1,
|
---|
2257 | TK_CHAR,
|
---|
2258 | TK_STRING,
|
---|
2259 | TK_CODE_POINT,
|
---|
2260 | TK_ANYCHAR,
|
---|
2261 | TK_CHAR_TYPE,
|
---|
2262 | TK_BACKREF,
|
---|
2263 | TK_CALL,
|
---|
2264 | TK_ANCHOR,
|
---|
2265 | TK_OP_REPEAT,
|
---|
2266 | TK_INTERVAL,
|
---|
2267 | TK_ANYCHAR_ANYTIME, /* SQL '%' == .* */
|
---|
2268 | TK_ALT,
|
---|
2269 | TK_SUBEXP_OPEN,
|
---|
2270 | TK_SUBEXP_CLOSE,
|
---|
2271 | TK_CC_OPEN,
|
---|
2272 | TK_QUOTE_OPEN,
|
---|
2273 | TK_CHAR_PROPERTY, /* \p{...}, \P{...} */
|
---|
2274 | TK_LINEBREAK,
|
---|
2275 | TK_EXTENDED_GRAPHEME_CLUSTER,
|
---|
2276 | TK_KEEP,
|
---|
2277 | /* in cc */
|
---|
2278 | TK_CC_CLOSE,
|
---|
2279 | TK_CC_RANGE,
|
---|
2280 | TK_POSIX_BRACKET_OPEN,
|
---|
2281 | TK_CC_AND, /* && */
|
---|
2282 | TK_CC_CC_OPEN /* [ */
|
---|
2283 | };
|
---|
2284 |
|
---|
2285 | typedef struct {
|
---|
2286 | enum TokenSyms type;
|
---|
2287 | int escaped;
|
---|
2288 | int base; /* is number: 8, 16 (used in [....]) */
|
---|
2289 | UChar* backp;
|
---|
2290 | union {
|
---|
2291 | UChar* s;
|
---|
2292 | int c;
|
---|
2293 | OnigCodePoint code;
|
---|
2294 | struct {
|
---|
2295 | int subtype;
|
---|
2296 | int ascii_range;
|
---|
2297 | } anchor;
|
---|
2298 | struct {
|
---|
2299 | int lower;
|
---|
2300 | int upper;
|
---|
2301 | int greedy;
|
---|
2302 | int possessive;
|
---|
2303 | } repeat;
|
---|
2304 | struct {
|
---|
2305 | int num;
|
---|
2306 | int ref1;
|
---|
2307 | int* refs;
|
---|
2308 | int by_name;
|
---|
2309 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
2310 | int exist_level;
|
---|
2311 | int level; /* \k<name+n> */
|
---|
2312 | #endif
|
---|
2313 | } backref;
|
---|
2314 | struct {
|
---|
2315 | UChar* name;
|
---|
2316 | UChar* name_end;
|
---|
2317 | int gnum;
|
---|
2318 | int rel;
|
---|
2319 | } call;
|
---|
2320 | struct {
|
---|
2321 | int ctype;
|
---|
2322 | int not;
|
---|
2323 | } prop;
|
---|
2324 | } u;
|
---|
2325 | } OnigToken;
|
---|
2326 |
|
---|
2327 |
|
---|
2328 | static int
|
---|
2329 | fetch_range_quantifier(UChar** src, UChar* end, OnigToken* tok, ScanEnv* env)
|
---|
2330 | {
|
---|
2331 | int low, up, syn_allow, non_low = 0;
|
---|
2332 | int r = 0;
|
---|
2333 | OnigCodePoint c;
|
---|
2334 | OnigEncoding enc = env->enc;
|
---|
2335 | UChar* p = *src;
|
---|
2336 | PFETCH_READY;
|
---|
2337 |
|
---|
2338 | syn_allow = IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_INVALID_INTERVAL);
|
---|
2339 |
|
---|
2340 | if (PEND) {
|
---|
2341 | if (syn_allow)
|
---|
2342 | return 1; /* "....{" : OK! */
|
---|
2343 | else
|
---|
2344 | return ONIGERR_END_PATTERN_AT_LEFT_BRACE; /* "....{" syntax error */
|
---|
2345 | }
|
---|
2346 |
|
---|
2347 | if (! syn_allow) {
|
---|
2348 | c = PPEEK;
|
---|
2349 | if (c == ')' || c == '(' || c == '|') {
|
---|
2350 | return ONIGERR_END_PATTERN_AT_LEFT_BRACE;
|
---|
2351 | }
|
---|
2352 | }
|
---|
2353 |
|
---|
2354 | low = onig_scan_unsigned_number(&p, end, env->enc);
|
---|
2355 | if (low < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
|
---|
2356 | if (low > ONIG_MAX_REPEAT_NUM)
|
---|
2357 | return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
|
---|
2358 |
|
---|
2359 | if (p == *src) { /* can't read low */
|
---|
2360 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV)) {
|
---|
2361 | /* allow {,n} as {0,n} */
|
---|
2362 | low = 0;
|
---|
2363 | non_low = 1;
|
---|
2364 | }
|
---|
2365 | else
|
---|
2366 | goto invalid;
|
---|
2367 | }
|
---|
2368 |
|
---|
2369 | if (PEND) goto invalid;
|
---|
2370 | PFETCH(c);
|
---|
2371 | if (c == ',') {
|
---|
2372 | UChar* prev = p;
|
---|
2373 | up = onig_scan_unsigned_number(&p, end, env->enc);
|
---|
2374 | if (up < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
|
---|
2375 | if (up > ONIG_MAX_REPEAT_NUM)
|
---|
2376 | return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
|
---|
2377 |
|
---|
2378 | if (p == prev) {
|
---|
2379 | if (non_low != 0)
|
---|
2380 | goto invalid;
|
---|
2381 | up = REPEAT_INFINITE; /* {n,} : {n,infinite} */
|
---|
2382 | }
|
---|
2383 | }
|
---|
2384 | else {
|
---|
2385 | if (non_low != 0)
|
---|
2386 | goto invalid;
|
---|
2387 |
|
---|
2388 | PUNFETCH;
|
---|
2389 | up = low; /* {n} : exact n times */
|
---|
2390 | r = 2; /* fixed */
|
---|
2391 | }
|
---|
2392 |
|
---|
2393 | if (PEND) goto invalid;
|
---|
2394 | PFETCH(c);
|
---|
2395 | if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) {
|
---|
2396 | if (c != MC_ESC(env->syntax)) goto invalid;
|
---|
2397 | if (PEND) goto invalid;
|
---|
2398 | PFETCH(c);
|
---|
2399 | }
|
---|
2400 | if (c != '}') goto invalid;
|
---|
2401 |
|
---|
2402 | if (!IS_REPEAT_INFINITE(up) && low > up) {
|
---|
2403 | return ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE;
|
---|
2404 | }
|
---|
2405 |
|
---|
2406 | tok->type = TK_INTERVAL;
|
---|
2407 | tok->u.repeat.lower = low;
|
---|
2408 | tok->u.repeat.upper = up;
|
---|
2409 | *src = p;
|
---|
2410 | return r; /* 0: normal {n,m}, 2: fixed {n} */
|
---|
2411 |
|
---|
2412 | invalid:
|
---|
2413 | if (syn_allow)
|
---|
2414 | return 1; /* OK */
|
---|
2415 | else
|
---|
2416 | return ONIGERR_INVALID_REPEAT_RANGE_PATTERN;
|
---|
2417 | }
|
---|
2418 |
|
---|
2419 | /* \M-, \C-, \c, or \... */
|
---|
2420 | static int
|
---|
2421 | fetch_escaped_value(UChar** src, UChar* end, ScanEnv* env, OnigCodePoint* val)
|
---|
2422 | {
|
---|
2423 | int v;
|
---|
2424 | OnigCodePoint c;
|
---|
2425 | OnigEncoding enc = env->enc;
|
---|
2426 | UChar* p = *src;
|
---|
2427 |
|
---|
2428 | if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;
|
---|
2429 |
|
---|
2430 | PFETCH_S(c);
|
---|
2431 | switch (c) {
|
---|
2432 | case 'M':
|
---|
2433 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META)) {
|
---|
2434 | if (PEND) return ONIGERR_END_PATTERN_AT_META;
|
---|
2435 | PFETCH_S(c);
|
---|
2436 | if (c != '-') return ONIGERR_META_CODE_SYNTAX;
|
---|
2437 | if (PEND) return ONIGERR_END_PATTERN_AT_META;
|
---|
2438 | PFETCH_S(c);
|
---|
2439 | if (c == MC_ESC(env->syntax)) {
|
---|
2440 | v = fetch_escaped_value(&p, end, env, &c);
|
---|
2441 | if (v < 0) return v;
|
---|
2442 | }
|
---|
2443 | c = ((c & 0xff) | 0x80);
|
---|
2444 | }
|
---|
2445 | else
|
---|
2446 | goto backslash;
|
---|
2447 | break;
|
---|
2448 |
|
---|
2449 | case 'C':
|
---|
2450 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL)) {
|
---|
2451 | if (PEND) return ONIGERR_END_PATTERN_AT_CONTROL;
|
---|
2452 | PFETCH_S(c);
|
---|
2453 | if (c != '-') return ONIGERR_CONTROL_CODE_SYNTAX;
|
---|
2454 | goto control;
|
---|
2455 | }
|
---|
2456 | else
|
---|
2457 | goto backslash;
|
---|
2458 |
|
---|
2459 | case 'c':
|
---|
2460 | if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_C_CONTROL)) {
|
---|
2461 | control:
|
---|
2462 | if (PEND) return ONIGERR_END_PATTERN_AT_CONTROL;
|
---|
2463 | PFETCH_S(c);
|
---|
2464 | if (c == '?') {
|
---|
2465 | c = 0177;
|
---|
2466 | }
|
---|
2467 | else {
|
---|
2468 | if (c == MC_ESC(env->syntax)) {
|
---|
2469 | v = fetch_escaped_value(&p, end, env, &c);
|
---|
2470 | if (v < 0) return v;
|
---|
2471 | }
|
---|
2472 | c &= 0x9f;
|
---|
2473 | }
|
---|
2474 | break;
|
---|
2475 | }
|
---|
2476 | /* fall through */
|
---|
2477 |
|
---|
2478 | default:
|
---|
2479 | {
|
---|
2480 | backslash:
|
---|
2481 | c = conv_backslash_value(c, env);
|
---|
2482 | }
|
---|
2483 | break;
|
---|
2484 | }
|
---|
2485 |
|
---|
2486 | *src = p;
|
---|
2487 | *val = c;
|
---|
2488 | return 0;
|
---|
2489 | }
|
---|
2490 |
|
---|
2491 | static int fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env);
|
---|
2492 |
|
---|
2493 | static OnigCodePoint
|
---|
2494 | get_name_end_code_point(OnigCodePoint start)
|
---|
2495 | {
|
---|
2496 | switch (start) {
|
---|
2497 | case '<': return (OnigCodePoint )'>'; break;
|
---|
2498 | case '\'': return (OnigCodePoint )'\''; break;
|
---|
2499 | case '(': return (OnigCodePoint )')'; break;
|
---|
2500 | case '{': return (OnigCodePoint )'}'; break;
|
---|
2501 | default:
|
---|
2502 | break;
|
---|
2503 | }
|
---|
2504 |
|
---|
2505 | return (OnigCodePoint )0;
|
---|
2506 | }
|
---|
2507 |
|
---|
2508 | #ifdef USE_NAMED_GROUP
|
---|
2509 | # ifdef RUBY
|
---|
2510 | # define ONIGENC_IS_CODE_NAME(enc, c) TRUE
|
---|
2511 | # else
|
---|
2512 | # define ONIGENC_IS_CODE_NAME(enc, c) ONIGENC_IS_CODE_WORD(enc, c)
|
---|
2513 | # endif
|
---|
2514 |
|
---|
2515 | # ifdef USE_BACKREF_WITH_LEVEL
|
---|
2516 | /*
|
---|
2517 | \k<name+n>, \k<name-n>
|
---|
2518 | \k<num+n>, \k<num-n>
|
---|
2519 | \k<-num+n>, \k<-num-n>
|
---|
2520 | */
|
---|
2521 | static int
|
---|
2522 | fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end,
|
---|
2523 | UChar** rname_end, ScanEnv* env,
|
---|
2524 | int* rback_num, int* rlevel)
|
---|
2525 | {
|
---|
2526 | int r, sign, is_num, exist_level;
|
---|
2527 | OnigCodePoint end_code;
|
---|
2528 | OnigCodePoint c = 0;
|
---|
2529 | OnigEncoding enc = env->enc;
|
---|
2530 | UChar *name_end;
|
---|
2531 | UChar *pnum_head;
|
---|
2532 | UChar *p = *src;
|
---|
2533 | PFETCH_READY;
|
---|
2534 |
|
---|
2535 | *rback_num = 0;
|
---|
2536 | is_num = exist_level = 0;
|
---|
2537 | sign = 1;
|
---|
2538 | pnum_head = *src;
|
---|
2539 |
|
---|
2540 | end_code = get_name_end_code_point(start_code);
|
---|
2541 |
|
---|
2542 | name_end = end;
|
---|
2543 | r = 0;
|
---|
2544 | if (PEND) {
|
---|
2545 | return ONIGERR_EMPTY_GROUP_NAME;
|
---|
2546 | }
|
---|
2547 | else {
|
---|
2548 | PFETCH(c);
|
---|
2549 | if (c == end_code)
|
---|
2550 | return ONIGERR_EMPTY_GROUP_NAME;
|
---|
2551 |
|
---|
2552 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
|
---|
2553 | is_num = 1;
|
---|
2554 | }
|
---|
2555 | else if (c == '-') {
|
---|
2556 | is_num = 2;
|
---|
2557 | sign = -1;
|
---|
2558 | pnum_head = p;
|
---|
2559 | }
|
---|
2560 | else if (!ONIGENC_IS_CODE_NAME(enc, c)) {
|
---|
2561 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
|
---|
2562 | }
|
---|
2563 | }
|
---|
2564 |
|
---|
2565 | while (!PEND) {
|
---|
2566 | name_end = p;
|
---|
2567 | PFETCH(c);
|
---|
2568 | if (c == end_code || c == ')' || c == '+' || c == '-') {
|
---|
2569 | if (is_num == 2) r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2570 | break;
|
---|
2571 | }
|
---|
2572 |
|
---|
2573 | if (is_num != 0) {
|
---|
2574 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
|
---|
2575 | is_num = 1;
|
---|
2576 | }
|
---|
2577 | else {
|
---|
2578 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2579 | is_num = 0;
|
---|
2580 | }
|
---|
2581 | }
|
---|
2582 | else if (!ONIGENC_IS_CODE_NAME(enc, c)) {
|
---|
2583 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
|
---|
2584 | }
|
---|
2585 | }
|
---|
2586 |
|
---|
2587 | if (r == 0 && c != end_code) {
|
---|
2588 | if (c == '+' || c == '-') {
|
---|
2589 | int level;
|
---|
2590 | int flag = (c == '-' ? -1 : 1);
|
---|
2591 |
|
---|
2592 | if (PEND) {
|
---|
2593 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
|
---|
2594 | goto end;
|
---|
2595 | }
|
---|
2596 | PFETCH(c);
|
---|
2597 | if (! ONIGENC_IS_CODE_DIGIT(enc, c)) goto err;
|
---|
2598 | PUNFETCH;
|
---|
2599 | level = onig_scan_unsigned_number(&p, end, enc);
|
---|
2600 | if (level < 0) return ONIGERR_TOO_BIG_NUMBER;
|
---|
2601 | *rlevel = (level * flag);
|
---|
2602 | exist_level = 1;
|
---|
2603 |
|
---|
2604 | if (!PEND) {
|
---|
2605 | PFETCH(c);
|
---|
2606 | if (c == end_code)
|
---|
2607 | goto end;
|
---|
2608 | }
|
---|
2609 | }
|
---|
2610 |
|
---|
2611 | err:
|
---|
2612 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2613 | name_end = end;
|
---|
2614 | }
|
---|
2615 |
|
---|
2616 | end:
|
---|
2617 | if (r == 0) {
|
---|
2618 | if (is_num != 0) {
|
---|
2619 | *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
|
---|
2620 | if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;
|
---|
2621 | else if (*rback_num == 0) goto err;
|
---|
2622 |
|
---|
2623 | *rback_num *= sign;
|
---|
2624 | }
|
---|
2625 |
|
---|
2626 | *rname_end = name_end;
|
---|
2627 | *src = p;
|
---|
2628 | return (exist_level ? 1 : 0);
|
---|
2629 | }
|
---|
2630 | else {
|
---|
2631 | onig_scan_env_set_error_string(env, r, *src, name_end);
|
---|
2632 | return r;
|
---|
2633 | }
|
---|
2634 | }
|
---|
2635 | # endif /* USE_BACKREF_WITH_LEVEL */
|
---|
2636 |
|
---|
2637 | /*
|
---|
2638 | ref: 0 -> define name (don't allow number name)
|
---|
2639 | 1 -> reference name (allow number name)
|
---|
2640 | */
|
---|
2641 | static int
|
---|
2642 | fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,
|
---|
2643 | UChar** rname_end, ScanEnv* env, int* rback_num, int ref)
|
---|
2644 | {
|
---|
2645 | int r, is_num, sign;
|
---|
2646 | OnigCodePoint end_code;
|
---|
2647 | OnigCodePoint c = 0;
|
---|
2648 | OnigEncoding enc = env->enc;
|
---|
2649 | UChar *name_end;
|
---|
2650 | UChar *pnum_head;
|
---|
2651 | UChar *p = *src;
|
---|
2652 |
|
---|
2653 | *rback_num = 0;
|
---|
2654 |
|
---|
2655 | end_code = get_name_end_code_point(start_code);
|
---|
2656 |
|
---|
2657 | name_end = end;
|
---|
2658 | pnum_head = *src;
|
---|
2659 | r = 0;
|
---|
2660 | is_num = 0;
|
---|
2661 | sign = 1;
|
---|
2662 | if (PEND) {
|
---|
2663 | return ONIGERR_EMPTY_GROUP_NAME;
|
---|
2664 | }
|
---|
2665 | else {
|
---|
2666 | PFETCH_S(c);
|
---|
2667 | if (c == end_code)
|
---|
2668 | return ONIGERR_EMPTY_GROUP_NAME;
|
---|
2669 |
|
---|
2670 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
|
---|
2671 | if (ref == 1)
|
---|
2672 | is_num = 1;
|
---|
2673 | else {
|
---|
2674 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2675 | is_num = 0;
|
---|
2676 | }
|
---|
2677 | }
|
---|
2678 | else if (c == '-') {
|
---|
2679 | if (ref == 1) {
|
---|
2680 | is_num = 2;
|
---|
2681 | sign = -1;
|
---|
2682 | pnum_head = p;
|
---|
2683 | }
|
---|
2684 | else {
|
---|
2685 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2686 | is_num = 0;
|
---|
2687 | }
|
---|
2688 | }
|
---|
2689 | else if (!ONIGENC_IS_CODE_NAME(enc, c)) {
|
---|
2690 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
|
---|
2691 | }
|
---|
2692 | }
|
---|
2693 |
|
---|
2694 | if (r == 0) {
|
---|
2695 | while (!PEND) {
|
---|
2696 | name_end = p;
|
---|
2697 | PFETCH_S(c);
|
---|
2698 | if (c == end_code || c == ')') {
|
---|
2699 | if (is_num == 2) {
|
---|
2700 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2701 | goto teardown;
|
---|
2702 | }
|
---|
2703 | break;
|
---|
2704 | }
|
---|
2705 |
|
---|
2706 | if (is_num != 0) {
|
---|
2707 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
|
---|
2708 | is_num = 1;
|
---|
2709 | }
|
---|
2710 | else {
|
---|
2711 | if (!ONIGENC_IS_CODE_WORD(enc, c))
|
---|
2712 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
|
---|
2713 | else
|
---|
2714 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2715 | goto teardown;
|
---|
2716 | }
|
---|
2717 | }
|
---|
2718 | else {
|
---|
2719 | if (!ONIGENC_IS_CODE_NAME(enc, c)) {
|
---|
2720 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
|
---|
2721 | goto teardown;
|
---|
2722 | }
|
---|
2723 | }
|
---|
2724 | }
|
---|
2725 |
|
---|
2726 | if (c != end_code) {
|
---|
2727 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2728 | name_end = end;
|
---|
2729 | goto err;
|
---|
2730 | }
|
---|
2731 |
|
---|
2732 | if (is_num != 0) {
|
---|
2733 | *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
|
---|
2734 | if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;
|
---|
2735 | else if (*rback_num == 0) {
|
---|
2736 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2737 | goto err;
|
---|
2738 | }
|
---|
2739 |
|
---|
2740 | *rback_num *= sign;
|
---|
2741 | }
|
---|
2742 |
|
---|
2743 | *rname_end = name_end;
|
---|
2744 | *src = p;
|
---|
2745 | return 0;
|
---|
2746 | }
|
---|
2747 | else {
|
---|
2748 | teardown:
|
---|
2749 | while (!PEND) {
|
---|
2750 | name_end = p;
|
---|
2751 | PFETCH_S(c);
|
---|
2752 | if (c == end_code || c == ')')
|
---|
2753 | break;
|
---|
2754 | }
|
---|
2755 | if (PEND)
|
---|
2756 | name_end = end;
|
---|
2757 |
|
---|
2758 | err:
|
---|
2759 | onig_scan_env_set_error_string(env, r, *src, name_end);
|
---|
2760 | return r;
|
---|
2761 | }
|
---|
2762 | }
|
---|
2763 | #else
|
---|
2764 | static int
|
---|
2765 | fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,
|
---|
2766 | UChar** rname_end, ScanEnv* env, int* rback_num, int ref)
|
---|
2767 | {
|
---|
2768 | int r, is_num, sign;
|
---|
2769 | OnigCodePoint end_code;
|
---|
2770 | OnigCodePoint c = 0;
|
---|
2771 | UChar *name_end;
|
---|
2772 | OnigEncoding enc = env->enc;
|
---|
2773 | UChar *pnum_head;
|
---|
2774 | UChar *p = *src;
|
---|
2775 | PFETCH_READY;
|
---|
2776 |
|
---|
2777 | *rback_num = 0;
|
---|
2778 |
|
---|
2779 | end_code = get_name_end_code_point(start_code);
|
---|
2780 |
|
---|
2781 | *rname_end = name_end = end;
|
---|
2782 | r = 0;
|
---|
2783 | pnum_head = *src;
|
---|
2784 | is_num = 0;
|
---|
2785 | sign = 1;
|
---|
2786 |
|
---|
2787 | if (PEND) {
|
---|
2788 | return ONIGERR_EMPTY_GROUP_NAME;
|
---|
2789 | }
|
---|
2790 | else {
|
---|
2791 | PFETCH(c);
|
---|
2792 | if (c == end_code)
|
---|
2793 | return ONIGERR_EMPTY_GROUP_NAME;
|
---|
2794 |
|
---|
2795 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
|
---|
2796 | is_num = 1;
|
---|
2797 | }
|
---|
2798 | else if (c == '-') {
|
---|
2799 | is_num = 2;
|
---|
2800 | sign = -1;
|
---|
2801 | pnum_head = p;
|
---|
2802 | }
|
---|
2803 | else {
|
---|
2804 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
|
---|
2805 | }
|
---|
2806 | }
|
---|
2807 |
|
---|
2808 | while (!PEND) {
|
---|
2809 | name_end = p;
|
---|
2810 |
|
---|
2811 | PFETCH(c);
|
---|
2812 | if (c == end_code || c == ')') break;
|
---|
2813 | if (! ONIGENC_IS_CODE_DIGIT(enc, c))
|
---|
2814 | r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
|
---|
2815 | }
|
---|
2816 | if (r == 0 && c != end_code) {
|
---|
2817 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2818 | name_end = end;
|
---|
2819 | }
|
---|
2820 |
|
---|
2821 | if (r == 0) {
|
---|
2822 | *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
|
---|
2823 | if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;
|
---|
2824 | else if (*rback_num == 0) {
|
---|
2825 | r = ONIGERR_INVALID_GROUP_NAME;
|
---|
2826 | goto err;
|
---|
2827 | }
|
---|
2828 | *rback_num *= sign;
|
---|
2829 |
|
---|
2830 | *rname_end = name_end;
|
---|
2831 | *src = p;
|
---|
2832 | return 0;
|
---|
2833 | }
|
---|
2834 | else {
|
---|
2835 | err:
|
---|
2836 | onig_scan_env_set_error_string(env, r, *src, name_end);
|
---|
2837 | return r;
|
---|
2838 | }
|
---|
2839 | }
|
---|
2840 | #endif /* USE_NAMED_GROUP */
|
---|
2841 |
|
---|
2842 |
|
---|
2843 | static void
|
---|
2844 | onig_syntax_warn(ScanEnv *env, const char *fmt, ...)
|
---|
2845 | {
|
---|
2846 | va_list args;
|
---|
2847 | UChar buf[WARN_BUFSIZE];
|
---|
2848 | va_start(args, fmt);
|
---|
2849 | onig_vsnprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,
|
---|
2850 | env->pattern, env->pattern_end,
|
---|
2851 | (const UChar *)fmt, args);
|
---|
2852 | va_end(args);
|
---|
2853 | #ifdef RUBY
|
---|
2854 | if (env->sourcefile == NULL)
|
---|
2855 | rb_warn("%s", (char *)buf);
|
---|
2856 | else
|
---|
2857 | rb_compile_warn(env->sourcefile, env->sourceline, "%s", (char *)buf);
|
---|
2858 | #else
|
---|
2859 | (*onig_warn)((char* )buf);
|
---|
2860 | #endif
|
---|
2861 | }
|
---|
2862 |
|
---|
2863 | static void
|
---|
2864 | CC_ESC_WARN(ScanEnv *env, UChar *c)
|
---|
2865 | {
|
---|
2866 | if (onig_warn == onig_null_warn) return ;
|
---|
2867 |
|
---|
2868 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED) &&
|
---|
2869 | IS_SYNTAX_BV(env->syntax, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC)) {
|
---|
2870 | onig_syntax_warn(env, "character class has '%s' without escape", c);
|
---|
2871 | }
|
---|
2872 | }
|
---|
2873 |
|
---|
2874 | static void
|
---|
2875 | CLOSE_BRACKET_WITHOUT_ESC_WARN(ScanEnv* env, UChar* c)
|
---|
2876 | {
|
---|
2877 | if (onig_warn == onig_null_warn) return ;
|
---|
2878 |
|
---|
2879 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED)) {
|
---|
2880 | onig_syntax_warn(env, "regular expression has '%s' without escape", c);
|
---|
2881 | }
|
---|
2882 | }
|
---|
2883 |
|
---|
2884 | #ifndef RTEST
|
---|
2885 | # define RTEST(v) 1
|
---|
2886 | #endif
|
---|
2887 |
|
---|
2888 | static void
|
---|
2889 | CC_DUP_WARN(ScanEnv *env)
|
---|
2890 | {
|
---|
2891 | if (onig_warn == onig_null_warn || !RTEST(ruby_verbose)) return ;
|
---|
2892 |
|
---|
2893 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_DUP) &&
|
---|
2894 | !(env->warnings_flag & ONIG_SYN_WARN_CC_DUP)) {
|
---|
2895 | env->warnings_flag |= ONIG_SYN_WARN_CC_DUP;
|
---|
2896 | onig_syntax_warn(env, "character class has duplicated range");
|
---|
2897 | }
|
---|
2898 | }
|
---|
2899 |
|
---|
2900 | static void
|
---|
2901 | UNKNOWN_ESC_WARN(ScanEnv *env, int c)
|
---|
2902 | {
|
---|
2903 | if (onig_warn == onig_null_warn || !RTEST(ruby_verbose)) return ;
|
---|
2904 | onig_syntax_warn(env, "Unknown escape \\%c is ignored", c);
|
---|
2905 | }
|
---|
2906 |
|
---|
2907 | static UChar*
|
---|
2908 | find_str_position(OnigCodePoint s[], int n, UChar* from, UChar* to,
|
---|
2909 | UChar **next, OnigEncoding enc)
|
---|
2910 | {
|
---|
2911 | int i;
|
---|
2912 | OnigCodePoint x;
|
---|
2913 | UChar *q;
|
---|
2914 | UChar *p = from;
|
---|
2915 |
|
---|
2916 | while (p < to) {
|
---|
2917 | x = ONIGENC_MBC_TO_CODE(enc, p, to);
|
---|
2918 | q = p + enclen(enc, p, to);
|
---|
2919 | if (x == s[0]) {
|
---|
2920 | for (i = 1; i < n && q < to; i++) {
|
---|
2921 | x = ONIGENC_MBC_TO_CODE(enc, q, to);
|
---|
2922 | if (x != s[i]) break;
|
---|
2923 | q += enclen(enc, q, to);
|
---|
2924 | }
|
---|
2925 | if (i >= n) {
|
---|
2926 | if (IS_NOT_NULL(next))
|
---|
2927 | *next = q;
|
---|
2928 | return p;
|
---|
2929 | }
|
---|
2930 | }
|
---|
2931 | p = q;
|
---|
2932 | }
|
---|
2933 | return NULL_UCHARP;
|
---|
2934 | }
|
---|
2935 |
|
---|
2936 | static int
|
---|
2937 | str_exist_check_with_esc(OnigCodePoint s[], int n, UChar* from, UChar* to,
|
---|
2938 | OnigCodePoint bad, OnigEncoding enc, const OnigSyntaxType* syn)
|
---|
2939 | {
|
---|
2940 | int i, in_esc;
|
---|
2941 | OnigCodePoint x;
|
---|
2942 | UChar *q;
|
---|
2943 | UChar *p = from;
|
---|
2944 |
|
---|
2945 | in_esc = 0;
|
---|
2946 | while (p < to) {
|
---|
2947 | if (in_esc) {
|
---|
2948 | in_esc = 0;
|
---|
2949 | p += enclen(enc, p, to);
|
---|
2950 | }
|
---|
2951 | else {
|
---|
2952 | x = ONIGENC_MBC_TO_CODE(enc, p, to);
|
---|
2953 | q = p + enclen(enc, p, to);
|
---|
2954 | if (x == s[0]) {
|
---|
2955 | for (i = 1; i < n && q < to; i++) {
|
---|
2956 | x = ONIGENC_MBC_TO_CODE(enc, q, to);
|
---|
2957 | if (x != s[i]) break;
|
---|
2958 | q += enclen(enc, q, to);
|
---|
2959 | }
|
---|
2960 | if (i >= n) return 1;
|
---|
2961 | p += enclen(enc, p, to);
|
---|
2962 | }
|
---|
2963 | else {
|
---|
2964 | x = ONIGENC_MBC_TO_CODE(enc, p, to);
|
---|
2965 | if (x == bad) return 0;
|
---|
2966 | else if (x == MC_ESC(syn)) in_esc = 1;
|
---|
2967 | p = q;
|
---|
2968 | }
|
---|
2969 | }
|
---|
2970 | }
|
---|
2971 | return 0;
|
---|
2972 | }
|
---|
2973 |
|
---|
2974 | static int
|
---|
2975 | fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
|
---|
2976 | {
|
---|
2977 | int num;
|
---|
2978 | OnigCodePoint c, c2;
|
---|
2979 | const OnigSyntaxType* syn = env->syntax;
|
---|
2980 | OnigEncoding enc = env->enc;
|
---|
2981 | UChar* prev;
|
---|
2982 | UChar* p = *src;
|
---|
2983 | PFETCH_READY;
|
---|
2984 |
|
---|
2985 | if (PEND) {
|
---|
2986 | tok->type = TK_EOT;
|
---|
2987 | return tok->type;
|
---|
2988 | }
|
---|
2989 |
|
---|
2990 | PFETCH(c);
|
---|
2991 | tok->type = TK_CHAR;
|
---|
2992 | tok->base = 0;
|
---|
2993 | tok->u.c = c;
|
---|
2994 | tok->escaped = 0;
|
---|
2995 |
|
---|
2996 | if (c == ']') {
|
---|
2997 | tok->type = TK_CC_CLOSE;
|
---|
2998 | }
|
---|
2999 | else if (c == '-') {
|
---|
3000 | tok->type = TK_CC_RANGE;
|
---|
3001 | }
|
---|
3002 | else if (c == MC_ESC(syn)) {
|
---|
3003 | if (! IS_SYNTAX_BV(syn, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC))
|
---|
3004 | goto end;
|
---|
3005 |
|
---|
3006 | if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;
|
---|
3007 |
|
---|
3008 | PFETCH(c);
|
---|
3009 | tok->escaped = 1;
|
---|
3010 | tok->u.c = c;
|
---|
3011 | switch (c) {
|
---|
3012 | case 'w':
|
---|
3013 | tok->type = TK_CHAR_TYPE;
|
---|
3014 | tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
|
---|
3015 | tok->u.prop.not = 0;
|
---|
3016 | break;
|
---|
3017 | case 'W':
|
---|
3018 | tok->type = TK_CHAR_TYPE;
|
---|
3019 | tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
|
---|
3020 | tok->u.prop.not = 1;
|
---|
3021 | break;
|
---|
3022 | case 'd':
|
---|
3023 | tok->type = TK_CHAR_TYPE;
|
---|
3024 | tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
|
---|
3025 | tok->u.prop.not = 0;
|
---|
3026 | break;
|
---|
3027 | case 'D':
|
---|
3028 | tok->type = TK_CHAR_TYPE;
|
---|
3029 | tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
|
---|
3030 | tok->u.prop.not = 1;
|
---|
3031 | break;
|
---|
3032 | case 's':
|
---|
3033 | tok->type = TK_CHAR_TYPE;
|
---|
3034 | tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
|
---|
3035 | tok->u.prop.not = 0;
|
---|
3036 | break;
|
---|
3037 | case 'S':
|
---|
3038 | tok->type = TK_CHAR_TYPE;
|
---|
3039 | tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
|
---|
3040 | tok->u.prop.not = 1;
|
---|
3041 | break;
|
---|
3042 | case 'h':
|
---|
3043 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;
|
---|
3044 | tok->type = TK_CHAR_TYPE;
|
---|
3045 | tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
|
---|
3046 | tok->u.prop.not = 0;
|
---|
3047 | break;
|
---|
3048 | case 'H':
|
---|
3049 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;
|
---|
3050 | tok->type = TK_CHAR_TYPE;
|
---|
3051 | tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
|
---|
3052 | tok->u.prop.not = 1;
|
---|
3053 | break;
|
---|
3054 |
|
---|
3055 | case 'p':
|
---|
3056 | case 'P':
|
---|
3057 | if (PEND) break;
|
---|
3058 |
|
---|
3059 | c2 = PPEEK;
|
---|
3060 | if (c2 == '{' &&
|
---|
3061 | IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) {
|
---|
3062 | PINC;
|
---|
3063 | tok->type = TK_CHAR_PROPERTY;
|
---|
3064 | tok->u.prop.not = (c == 'P' ? 1 : 0);
|
---|
3065 |
|
---|
3066 | if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) {
|
---|
3067 | PFETCH(c2);
|
---|
3068 | if (c2 == '^') {
|
---|
3069 | tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0);
|
---|
3070 | }
|
---|
3071 | else
|
---|
3072 | PUNFETCH;
|
---|
3073 | }
|
---|
3074 | }
|
---|
3075 | else {
|
---|
3076 | onig_syntax_warn(env, "invalid Unicode Property \\%c", c);
|
---|
3077 | }
|
---|
3078 | break;
|
---|
3079 |
|
---|
3080 | case 'x':
|
---|
3081 | if (PEND) break;
|
---|
3082 |
|
---|
3083 | prev = p;
|
---|
3084 | if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {
|
---|
3085 | PINC;
|
---|
3086 | num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc);
|
---|
3087 | if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
|
---|
3088 | if (!PEND) {
|
---|
3089 | c2 = PPEEK;
|
---|
3090 | if (ONIGENC_IS_CODE_XDIGIT(enc, c2))
|
---|
3091 | return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
|
---|
3092 | }
|
---|
3093 |
|
---|
3094 | if (p > prev + enclen(enc, prev, end) && !PEND && (PPEEK_IS('}'))) {
|
---|
3095 | PINC;
|
---|
3096 | tok->type = TK_CODE_POINT;
|
---|
3097 | tok->base = 16;
|
---|
3098 | tok->u.code = (OnigCodePoint )num;
|
---|
3099 | }
|
---|
3100 | else {
|
---|
3101 | /* can't read nothing or invalid format */
|
---|
3102 | p = prev;
|
---|
3103 | }
|
---|
3104 | }
|
---|
3105 | else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {
|
---|
3106 | num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc);
|
---|
3107 | if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
|
---|
3108 | if (p == prev) { /* can't read nothing. */
|
---|
3109 | num = 0; /* but, it's not error */
|
---|
3110 | }
|
---|
3111 | tok->type = TK_RAW_BYTE;
|
---|
3112 | tok->base = 16;
|
---|
3113 | tok->u.c = num;
|
---|
3114 | }
|
---|
3115 | break;
|
---|
3116 |
|
---|
3117 | case 'u':
|
---|
3118 | if (PEND) break;
|
---|
3119 |
|
---|
3120 | prev = p;
|
---|
3121 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {
|
---|
3122 | num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc);
|
---|
3123 | if (num < -1) return ONIGERR_TOO_SHORT_DIGITS;
|
---|
3124 | else if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
|
---|
3125 | if (p == prev) { /* can't read nothing. */
|
---|
3126 | num = 0; /* but, it's not error */
|
---|
3127 | }
|
---|
3128 | tok->type = TK_CODE_POINT;
|
---|
3129 | tok->base = 16;
|
---|
3130 | tok->u.code = (OnigCodePoint )num;
|
---|
3131 | }
|
---|
3132 | break;
|
---|
3133 |
|
---|
3134 | case 'o':
|
---|
3135 | if (PEND) break;
|
---|
3136 |
|
---|
3137 | prev = p;
|
---|
3138 | if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_O_BRACE_OCTAL)) {
|
---|
3139 | PINC;
|
---|
3140 | num = scan_unsigned_octal_number(&p, end, 11, enc);
|
---|
3141 | if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
|
---|
3142 | if (!PEND) {
|
---|
3143 | c2 = PPEEK;
|
---|
3144 | if (ONIGENC_IS_CODE_DIGIT(enc, c2) && c2 < '8')
|
---|
3145 | return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
|
---|
3146 | }
|
---|
3147 |
|
---|
3148 | if (p > prev + enclen(enc, prev, end) && !PEND && (PPEEK_IS('}'))) {
|
---|
3149 | PINC;
|
---|
3150 | tok->type = TK_CODE_POINT;
|
---|
3151 | tok->base = 8;
|
---|
3152 | tok->u.code = (OnigCodePoint )num;
|
---|
3153 | }
|
---|
3154 | else {
|
---|
3155 | /* can't read nothing or invalid format */
|
---|
3156 | p = prev;
|
---|
3157 | }
|
---|
3158 | }
|
---|
3159 | break;
|
---|
3160 |
|
---|
3161 | case '0':
|
---|
3162 | case '1': case '2': case '3': case '4': case '5': case '6': case '7':
|
---|
3163 | if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {
|
---|
3164 | PUNFETCH;
|
---|
3165 | prev = p;
|
---|
3166 | num = scan_unsigned_octal_number(&p, end, 3, enc);
|
---|
3167 | if (num < 0 || 0xff < num) return ONIGERR_TOO_BIG_NUMBER;
|
---|
3168 | if (p == prev) { /* can't read nothing. */
|
---|
3169 | num = 0; /* but, it's not error */
|
---|
3170 | }
|
---|
3171 | tok->type = TK_RAW_BYTE;
|
---|
3172 | tok->base = 8;
|
---|
3173 | tok->u.c = num;
|
---|
3174 | }
|
---|
3175 | break;
|
---|
3176 |
|
---|
3177 | default:
|
---|
3178 | PUNFETCH;
|
---|
3179 | num = fetch_escaped_value(&p, end, env, &c2);
|
---|
3180 | if (num < 0) return num;
|
---|
3181 | if ((OnigCodePoint )tok->u.c != c2) {
|
---|
3182 | tok->u.code = (OnigCodePoint )c2;
|
---|
3183 | tok->type = TK_CODE_POINT;
|
---|
3184 | }
|
---|
3185 | break;
|
---|
3186 | }
|
---|
3187 | }
|
---|
3188 | else if (c == '[') {
|
---|
3189 | if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_POSIX_BRACKET) && (PPEEK_IS(':'))) {
|
---|
3190 | OnigCodePoint send[] = { (OnigCodePoint )':', (OnigCodePoint )']' };
|
---|
3191 | tok->backp = p; /* point at '[' is read */
|
---|
3192 | PINC;
|
---|
3193 | if (str_exist_check_with_esc(send, 2, p, end,
|
---|
3194 | (OnigCodePoint )']', enc, syn)) {
|
---|
3195 | tok->type = TK_POSIX_BRACKET_OPEN;
|
---|
3196 | }
|
---|
3197 | else {
|
---|
3198 | PUNFETCH;
|
---|
3199 | goto cc_in_cc;
|
---|
3200 | }
|
---|
3201 | }
|
---|
3202 | else {
|
---|
3203 | cc_in_cc:
|
---|
3204 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP)) {
|
---|
3205 | tok->type = TK_CC_CC_OPEN;
|
---|
3206 | }
|
---|
3207 | else {
|
---|
3208 | CC_ESC_WARN(env, (UChar* )"[");
|
---|
3209 | }
|
---|
3210 | }
|
---|
3211 | }
|
---|
3212 | else if (c == '&') {
|
---|
3213 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP) &&
|
---|
3214 | !PEND && (PPEEK_IS('&'))) {
|
---|
3215 | PINC;
|
---|
3216 | tok->type = TK_CC_AND;
|
---|
3217 | }
|
---|
3218 | }
|
---|
3219 |
|
---|
3220 | end:
|
---|
3221 | *src = p;
|
---|
3222 | return tok->type;
|
---|
3223 | }
|
---|
3224 |
|
---|
3225 | #ifdef USE_NAMED_GROUP
|
---|
3226 | static int
|
---|
3227 | fetch_named_backref_token(OnigCodePoint c, OnigToken* tok, UChar** src,
|
---|
3228 | UChar* end, ScanEnv* env)
|
---|
3229 | {
|
---|
3230 | int r, num;
|
---|
3231 | const OnigSyntaxType* syn = env->syntax;
|
---|
3232 | UChar* prev;
|
---|
3233 | UChar* p = *src;
|
---|
3234 | UChar* name_end;
|
---|
3235 | int* backs;
|
---|
3236 | int back_num;
|
---|
3237 |
|
---|
3238 | prev = p;
|
---|
3239 |
|
---|
3240 | # ifdef USE_BACKREF_WITH_LEVEL
|
---|
3241 | name_end = NULL_UCHARP; /* no need. escape gcc warning. */
|
---|
3242 | r = fetch_name_with_level(c, &p, end, &name_end,
|
---|
3243 | env, &back_num, &tok->u.backref.level);
|
---|
3244 | if (r == 1) tok->u.backref.exist_level = 1;
|
---|
3245 | else tok->u.backref.exist_level = 0;
|
---|
3246 | # else
|
---|
3247 | r = fetch_name(&p, end, &name_end, env, &back_num, 1);
|
---|
3248 | # endif
|
---|
3249 | if (r < 0) return r;
|
---|
3250 |
|
---|
3251 | if (back_num != 0) {
|
---|
3252 | if (back_num < 0) {
|
---|
3253 | back_num = BACKREF_REL_TO_ABS(back_num, env);
|
---|
3254 | if (back_num <= 0)
|
---|
3255 | return ONIGERR_INVALID_BACKREF;
|
---|
3256 | }
|
---|
3257 |
|
---|
3258 | if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
|
---|
3259 | if (back_num > env->num_mem ||
|
---|
3260 | IS_NULL(SCANENV_MEM_NODES(env)[back_num]))
|
---|
3261 | return ONIGERR_INVALID_BACKREF;
|
---|
3262 | }
|
---|
3263 | tok->type = TK_BACKREF;
|
---|
3264 | tok->u.backref.by_name = 0;
|
---|
3265 | tok->u.backref.num = 1;
|
---|
3266 | tok->u.backref.ref1 = back_num;
|
---|
3267 | }
|
---|
3268 | else {
|
---|
3269 | num = onig_name_to_group_numbers(env->reg, prev, name_end, &backs);
|
---|
3270 | if (num <= 0) {
|
---|
3271 | onig_scan_env_set_error_string(env,
|
---|
3272 | ONIGERR_UNDEFINED_NAME_REFERENCE, prev, name_end);
|
---|
3273 | return ONIGERR_UNDEFINED_NAME_REFERENCE;
|
---|
3274 | }
|
---|
3275 | if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
|
---|
3276 | int i;
|
---|
3277 | for (i = 0; i < num; i++) {
|
---|
3278 | if (backs[i] > env->num_mem ||
|
---|
3279 | IS_NULL(SCANENV_MEM_NODES(env)[backs[i]]))
|
---|
3280 | return ONIGERR_INVALID_BACKREF;
|
---|
3281 | }
|
---|
3282 | }
|
---|
3283 |
|
---|
3284 | tok->type = TK_BACKREF;
|
---|
3285 | tok->u.backref.by_name = 1;
|
---|
3286 | if (num == 1 || IS_SYNTAX_BV(syn, ONIG_SYN_USE_LEFT_MOST_NAMED_GROUP)) {
|
---|
3287 | tok->u.backref.num = 1;
|
---|
3288 | tok->u.backref.ref1 = backs[0];
|
---|
3289 | }
|
---|
3290 | else {
|
---|
3291 | tok->u.backref.num = num;
|
---|
3292 | tok->u.backref.refs = backs;
|
---|
3293 | }
|
---|
3294 | }
|
---|
3295 | *src = p;
|
---|
3296 | return 0;
|
---|
3297 | }
|
---|
3298 | #endif
|
---|
3299 |
|
---|
3300 | static int
|
---|
3301 | fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
|
---|
3302 | {
|
---|
3303 | int r, num;
|
---|
3304 | OnigCodePoint c;
|
---|
3305 | OnigEncoding enc = env->enc;
|
---|
3306 | const OnigSyntaxType* syn = env->syntax;
|
---|
3307 | UChar* prev;
|
---|
3308 | UChar* p = *src;
|
---|
3309 | PFETCH_READY;
|
---|
3310 |
|
---|
3311 | start:
|
---|
3312 | if (PEND) {
|
---|
3313 | tok->type = TK_EOT;
|
---|
3314 | return tok->type;
|
---|
3315 | }
|
---|
3316 |
|
---|
3317 | tok->type = TK_STRING;
|
---|
3318 | tok->base = 0;
|
---|
3319 | tok->backp = p;
|
---|
3320 |
|
---|
3321 | PFETCH(c);
|
---|
3322 | if (IS_MC_ESC_CODE(c, syn)) {
|
---|
3323 | if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;
|
---|
3324 |
|
---|
3325 | tok->backp = p;
|
---|
3326 | PFETCH(c);
|
---|
3327 |
|
---|
3328 | tok->u.c = c;
|
---|
3329 | tok->escaped = 1;
|
---|
3330 | switch (c) {
|
---|
3331 | case '*':
|
---|
3332 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF)) break;
|
---|
3333 | tok->type = TK_OP_REPEAT;
|
---|
3334 | tok->u.repeat.lower = 0;
|
---|
3335 | tok->u.repeat.upper = REPEAT_INFINITE;
|
---|
3336 | goto greedy_check;
|
---|
3337 | break;
|
---|
3338 |
|
---|
3339 | case '+':
|
---|
3340 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_PLUS_ONE_INF)) break;
|
---|
3341 | tok->type = TK_OP_REPEAT;
|
---|
3342 | tok->u.repeat.lower = 1;
|
---|
3343 | tok->u.repeat.upper = REPEAT_INFINITE;
|
---|
3344 | goto greedy_check;
|
---|
3345 | break;
|
---|
3346 |
|
---|
3347 | case '?':
|
---|
3348 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_QMARK_ZERO_ONE)) break;
|
---|
3349 | tok->type = TK_OP_REPEAT;
|
---|
3350 | tok->u.repeat.lower = 0;
|
---|
3351 | tok->u.repeat.upper = 1;
|
---|
3352 | greedy_check:
|
---|
3353 | if (!PEND && PPEEK_IS('?') &&
|
---|
3354 | IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_NON_GREEDY)) {
|
---|
3355 | PFETCH(c);
|
---|
3356 | tok->u.repeat.greedy = 0;
|
---|
3357 | tok->u.repeat.possessive = 0;
|
---|
3358 | }
|
---|
3359 | else {
|
---|
3360 | possessive_check:
|
---|
3361 | if (!PEND && PPEEK_IS('+') &&
|
---|
3362 | ((IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT) &&
|
---|
3363 | tok->type != TK_INTERVAL) ||
|
---|
3364 | (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL) &&
|
---|
3365 | tok->type == TK_INTERVAL))) {
|
---|
3366 | PFETCH(c);
|
---|
3367 | tok->u.repeat.greedy = 1;
|
---|
3368 | tok->u.repeat.possessive = 1;
|
---|
3369 | }
|
---|
3370 | else {
|
---|
3371 | tok->u.repeat.greedy = 1;
|
---|
3372 | tok->u.repeat.possessive = 0;
|
---|
3373 | }
|
---|
3374 | }
|
---|
3375 | break;
|
---|
3376 |
|
---|
3377 | case '{':
|
---|
3378 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) break;
|
---|
3379 | r = fetch_range_quantifier(&p, end, tok, env);
|
---|
3380 | if (r < 0) return r; /* error */
|
---|
3381 | if (r == 0) goto greedy_check;
|
---|
3382 | else if (r == 2) { /* {n} */
|
---|
3383 | if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY))
|
---|
3384 | goto possessive_check;
|
---|
3385 |
|
---|
3386 | goto greedy_check;
|
---|
3387 | }
|
---|
3388 | /* r == 1 : normal char */
|
---|
3389 | break;
|
---|
3390 |
|
---|
3391 | case '|':
|
---|
3392 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_VBAR_ALT)) break;
|
---|
3393 | tok->type = TK_ALT;
|
---|
3394 | break;
|
---|
3395 |
|
---|
3396 | case '(':
|
---|
3397 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LPAREN_SUBEXP)) break;
|
---|
3398 | tok->type = TK_SUBEXP_OPEN;
|
---|
3399 | break;
|
---|
3400 |
|
---|
3401 | case ')':
|
---|
3402 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LPAREN_SUBEXP)) break;
|
---|
3403 | tok->type = TK_SUBEXP_CLOSE;
|
---|
3404 | break;
|
---|
3405 |
|
---|
3406 | case 'w':
|
---|
3407 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD)) break;
|
---|
3408 | tok->type = TK_CHAR_TYPE;
|
---|
3409 | tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
|
---|
3410 | tok->u.prop.not = 0;
|
---|
3411 | break;
|
---|
3412 |
|
---|
3413 | case 'W':
|
---|
3414 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD)) break;
|
---|
3415 | tok->type = TK_CHAR_TYPE;
|
---|
3416 | tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
|
---|
3417 | tok->u.prop.not = 1;
|
---|
3418 | break;
|
---|
3419 |
|
---|
3420 | case 'b':
|
---|
3421 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_B_WORD_BOUND)) break;
|
---|
3422 | tok->type = TK_ANCHOR;
|
---|
3423 | tok->u.anchor.subtype = ANCHOR_WORD_BOUND;
|
---|
3424 | tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option)
|
---|
3425 | && ! IS_WORD_BOUND_ALL_RANGE(env->option);
|
---|
3426 | break;
|
---|
3427 |
|
---|
3428 | case 'B':
|
---|
3429 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_B_WORD_BOUND)) break;
|
---|
3430 | tok->type = TK_ANCHOR;
|
---|
3431 | tok->u.anchor.subtype = ANCHOR_NOT_WORD_BOUND;
|
---|
3432 | tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option)
|
---|
3433 | && ! IS_WORD_BOUND_ALL_RANGE(env->option);
|
---|
3434 | break;
|
---|
3435 |
|
---|
3436 | #ifdef USE_WORD_BEGIN_END
|
---|
3437 | case '<':
|
---|
3438 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END)) break;
|
---|
3439 | tok->type = TK_ANCHOR;
|
---|
3440 | tok->u.anchor.subtype = ANCHOR_WORD_BEGIN;
|
---|
3441 | tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option);
|
---|
3442 | break;
|
---|
3443 |
|
---|
3444 | case '>':
|
---|
3445 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END)) break;
|
---|
3446 | tok->type = TK_ANCHOR;
|
---|
3447 | tok->u.anchor.subtype = ANCHOR_WORD_END;
|
---|
3448 | tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option);
|
---|
3449 | break;
|
---|
3450 | #endif
|
---|
3451 |
|
---|
3452 | case 's':
|
---|
3453 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE)) break;
|
---|
3454 | tok->type = TK_CHAR_TYPE;
|
---|
3455 | tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
|
---|
3456 | tok->u.prop.not = 0;
|
---|
3457 | break;
|
---|
3458 |
|
---|
3459 | case 'S':
|
---|
3460 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE)) break;
|
---|
3461 | tok->type = TK_CHAR_TYPE;
|
---|
3462 | tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
|
---|
3463 | tok->u.prop.not = 1;
|
---|
3464 | break;
|
---|
3465 |
|
---|
3466 | case 'd':
|
---|
3467 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT)) break;
|
---|
3468 | tok->type = TK_CHAR_TYPE;
|
---|
3469 | tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
|
---|
3470 | tok->u.prop.not = 0;
|
---|
3471 | break;
|
---|
3472 |
|
---|
3473 | case 'D':
|
---|
3474 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT)) break;
|
---|
3475 | tok->type = TK_CHAR_TYPE;
|
---|
3476 | tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
|
---|
3477 | tok->u.prop.not = 1;
|
---|
3478 | break;
|
---|
3479 |
|
---|
3480 | case 'h':
|
---|
3481 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;
|
---|
3482 | tok->type = TK_CHAR_TYPE;
|
---|
3483 | tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
|
---|
3484 | tok->u.prop.not = 0;
|
---|
3485 | break;
|
---|
3486 |
|
---|
3487 | case 'H':
|
---|
3488 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT)) break;
|
---|
3489 | tok->type = TK_CHAR_TYPE;
|
---|
3490 | tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
|
---|
3491 | tok->u.prop.not = 1;
|
---|
3492 | break;
|
---|
3493 |
|
---|
3494 | case 'A':
|
---|
3495 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;
|
---|
3496 | begin_buf:
|
---|
3497 | tok->type = TK_ANCHOR;
|
---|
3498 | tok->u.anchor.subtype = ANCHOR_BEGIN_BUF;
|
---|
3499 | break;
|
---|
3500 |
|
---|
3501 | case 'Z':
|
---|
3502 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;
|
---|
3503 | tok->type = TK_ANCHOR;
|
---|
3504 | tok->u.anchor.subtype = ANCHOR_SEMI_END_BUF;
|
---|
3505 | break;
|
---|
3506 |
|
---|
3507 | case 'z':
|
---|
3508 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR)) break;
|
---|
3509 | end_buf:
|
---|
3510 | tok->type = TK_ANCHOR;
|
---|
3511 | tok->u.anchor.subtype = ANCHOR_END_BUF;
|
---|
3512 | break;
|
---|
3513 |
|
---|
3514 | case 'G':
|
---|
3515 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR)) break;
|
---|
3516 | tok->type = TK_ANCHOR;
|
---|
3517 | tok->u.anchor.subtype = ANCHOR_BEGIN_POSITION;
|
---|
3518 | break;
|
---|
3519 |
|
---|
3520 | case '`':
|
---|
3521 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR)) break;
|
---|
3522 | goto begin_buf;
|
---|
3523 | break;
|
---|
3524 |
|
---|
3525 | case '\'':
|
---|
3526 | if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR)) break;
|
---|
3527 | goto end_buf;
|
---|
3528 | break;
|
---|
3529 |
|
---|
3530 | case 'x':
|
---|
3531 | if (PEND) break;
|
---|
3532 |
|
---|
3533 | prev = p;
|
---|
3534 | if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {
|
---|
3535 | PINC;
|
---|
3536 | num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc);
|
---|
3537 | if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
|
---|
3538 | if (!PEND) {
|
---|
3539 | if (ONIGENC_IS_CODE_XDIGIT(enc, PPEEK))
|
---|
3540 | return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
|
---|
3541 | }
|
---|
3542 |
|
---|
3543 | if ((p > prev + enclen(enc, prev, end)) && !PEND && PPEEK_IS('}')) {
|
---|
3544 | PINC;
|
---|
3545 | tok->type = TK_CODE_POINT;
|
---|
3546 | tok->u.code = (OnigCodePoint )num;
|
---|
3547 | }
|
---|
3548 | else {
|
---|
3549 | /* can't read nothing or invalid format */
|
---|
3550 | p = prev;
|
---|
3551 | }
|
---|
3552 | }
|
---|
3553 | else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {
|
---|
3554 | num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc);
|
---|
3555 | if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
|
---|
3556 | if (p == prev) { /* can't read nothing. */
|
---|
3557 | num = 0; /* but, it's not error */
|
---|
3558 | }
|
---|
3559 | tok->type = TK_RAW_BYTE;
|
---|
3560 | tok->base = 16;
|
---|
3561 | tok->u.c = num;
|
---|
3562 | }
|
---|
3563 | break;
|
---|
3564 |
|
---|
3565 | case 'u':
|
---|
3566 | if (PEND) break;
|
---|
3567 |
|
---|
3568 | prev = p;
|
---|
3569 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {
|
---|
3570 | num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc);
|
---|
3571 | if (num < -1) return ONIGERR_TOO_SHORT_DIGITS;
|
---|
3572 | else if (num < 0) return ONIGERR_TOO_BIG_NUMBER;
|
---|
3573 | if (p == prev) { /* can't read nothing. */
|
---|
3574 | num = 0; /* but, it's not error */
|
---|
3575 | }
|
---|
3576 | tok->type = TK_CODE_POINT;
|
---|
3577 | tok->base = 16;
|
---|
3578 | tok->u.code = (OnigCodePoint )num;
|
---|
3579 | }
|
---|
3580 | break;
|
---|
3581 |
|
---|
3582 | case 'o':
|
---|
3583 | if (PEND) break;
|
---|
3584 |
|
---|
3585 | prev = p;
|
---|
3586 | if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_O_BRACE_OCTAL)) {
|
---|
3587 | PINC;
|
---|
3588 | num = scan_unsigned_octal_number(&p, end, 11, enc);
|
---|
3589 | if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
|
---|
3590 | if (!PEND) {
|
---|
3591 | OnigCodePoint c = PPEEK;
|
---|
3592 | if (ONIGENC_IS_CODE_DIGIT(enc, c) && c < '8')
|
---|
3593 | return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
|
---|
3594 | }
|
---|
3595 |
|
---|
3596 | if ((p > prev + enclen(enc, prev, end)) && !PEND && PPEEK_IS('}')) {
|
---|
3597 | PINC;
|
---|
3598 | tok->type = TK_CODE_POINT;
|
---|
3599 | tok->u.code = (OnigCodePoint )num;
|
---|
3600 | }
|
---|
3601 | else {
|
---|
3602 | /* can't read nothing or invalid format */
|
---|
3603 | p = prev;
|
---|
3604 | }
|
---|
3605 | }
|
---|
3606 | break;
|
---|
3607 |
|
---|
3608 | case '1': case '2': case '3': case '4':
|
---|
3609 | case '5': case '6': case '7': case '8': case '9':
|
---|
3610 | PUNFETCH;
|
---|
3611 | prev = p;
|
---|
3612 | num = onig_scan_unsigned_number(&p, end, enc);
|
---|
3613 | if (num < 0 || num > ONIG_MAX_BACKREF_NUM) {
|
---|
3614 | goto skip_backref;
|
---|
3615 | }
|
---|
3616 |
|
---|
3617 | if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_DECIMAL_BACKREF) &&
|
---|
3618 | (num <= env->num_mem || num <= 9)) { /* This spec. from GNU regex */
|
---|
3619 | if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
|
---|
3620 | if (num > env->num_mem || IS_NULL(SCANENV_MEM_NODES(env)[num]))
|
---|
3621 | return ONIGERR_INVALID_BACKREF;
|
---|
3622 | }
|
---|
3623 |
|
---|
3624 | tok->type = TK_BACKREF;
|
---|
3625 | tok->u.backref.num = 1;
|
---|
3626 | tok->u.backref.ref1 = num;
|
---|
3627 | tok->u.backref.by_name = 0;
|
---|
3628 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
3629 | tok->u.backref.exist_level = 0;
|
---|
3630 | #endif
|
---|
3631 | break;
|
---|
3632 | }
|
---|
3633 |
|
---|
3634 | skip_backref:
|
---|
3635 | if (c == '8' || c == '9') {
|
---|
3636 | /* normal char */
|
---|
3637 | p = prev; PINC;
|
---|
3638 | break;
|
---|
3639 | }
|
---|
3640 |
|
---|
3641 | p = prev;
|
---|
3642 | /* fall through */
|
---|
3643 | case '0':
|
---|
3644 | if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {
|
---|
3645 | prev = p;
|
---|
3646 | num = scan_unsigned_octal_number(&p, end, (c == '0' ? 2:3), enc);
|
---|
3647 | if (num < 0 || 0xff < num) return ONIGERR_TOO_BIG_NUMBER;
|
---|
3648 | if (p == prev) { /* can't read nothing. */
|
---|
3649 | num = 0; /* but, it's not error */
|
---|
3650 | }
|
---|
3651 | tok->type = TK_RAW_BYTE;
|
---|
3652 | tok->base = 8;
|
---|
3653 | tok->u.c = num;
|
---|
3654 | }
|
---|
3655 | else if (c != '0') {
|
---|
3656 | PINC;
|
---|
3657 | }
|
---|
3658 | break;
|
---|
3659 |
|
---|
3660 | #ifdef USE_NAMED_GROUP
|
---|
3661 | case 'k':
|
---|
3662 | if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_K_NAMED_BACKREF)) {
|
---|
3663 | PFETCH(c);
|
---|
3664 | if (c == '<' || c == '\'') {
|
---|
3665 | r = fetch_named_backref_token(c, tok, &p, end, env);
|
---|
3666 | if (r < 0) return r;
|
---|
3667 | }
|
---|
3668 | else {
|
---|
3669 | PUNFETCH;
|
---|
3670 | onig_syntax_warn(env, "invalid back reference");
|
---|
3671 | }
|
---|
3672 | }
|
---|
3673 | break;
|
---|
3674 | #endif
|
---|
3675 |
|
---|
3676 | #if defined(USE_SUBEXP_CALL) || defined(USE_NAMED_GROUP)
|
---|
3677 | case 'g':
|
---|
3678 | # ifdef USE_NAMED_GROUP
|
---|
3679 | if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_G_BRACE_BACKREF)) {
|
---|
3680 | PFETCH(c);
|
---|
3681 | if (c == '{') {
|
---|
3682 | r = fetch_named_backref_token(c, tok, &p, end, env);
|
---|
3683 | if (r < 0) return r;
|
---|
3684 | }
|
---|
3685 | else
|
---|
3686 | PUNFETCH;
|
---|
3687 | }
|
---|
3688 | # endif
|
---|
3689 | # ifdef USE_SUBEXP_CALL
|
---|
3690 | if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_G_SUBEXP_CALL)) {
|
---|
3691 | PFETCH(c);
|
---|
3692 | if (c == '<' || c == '\'') {
|
---|
3693 | int gnum = -1, rel = 0;
|
---|
3694 | UChar* name_end;
|
---|
3695 | OnigCodePoint cnext;
|
---|
3696 |
|
---|
3697 | cnext = PPEEK;
|
---|
3698 | if (cnext == '0') {
|
---|
3699 | PINC;
|
---|
3700 | if (PPEEK_IS(get_name_end_code_point(c))) { /* \g<0>, \g'0' */
|
---|
3701 | PINC;
|
---|
3702 | name_end = p;
|
---|
3703 | gnum = 0;
|
---|
3704 | }
|
---|
3705 | }
|
---|
3706 | else if (cnext == '+') {
|
---|
3707 | PINC;
|
---|
3708 | rel = 1;
|
---|
3709 | }
|
---|
3710 | prev = p;
|
---|
3711 | if (gnum < 0) {
|
---|
3712 | r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &gnum, 1);
|
---|
3713 | if (r < 0) return r;
|
---|
3714 | }
|
---|
3715 |
|
---|
3716 | tok->type = TK_CALL;
|
---|
3717 | tok->u.call.name = prev;
|
---|
3718 | tok->u.call.name_end = name_end;
|
---|
3719 | tok->u.call.gnum = gnum;
|
---|
3720 | tok->u.call.rel = rel;
|
---|
3721 | }
|
---|
3722 | else {
|
---|
3723 | onig_syntax_warn(env, "invalid subexp call");
|
---|
3724 | PUNFETCH;
|
---|
3725 | }
|
---|
3726 | }
|
---|
3727 | # endif
|
---|
3728 | break;
|
---|
3729 | #endif
|
---|
3730 |
|
---|
3731 | case 'Q':
|
---|
3732 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE)) {
|
---|
3733 | tok->type = TK_QUOTE_OPEN;
|
---|
3734 | }
|
---|
3735 | break;
|
---|
3736 |
|
---|
3737 | case 'p':
|
---|
3738 | case 'P':
|
---|
3739 | if (PPEEK_IS('{') &&
|
---|
3740 | IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) {
|
---|
3741 | PINC;
|
---|
3742 | tok->type = TK_CHAR_PROPERTY;
|
---|
3743 | tok->u.prop.not = (c == 'P' ? 1 : 0);
|
---|
3744 |
|
---|
3745 | if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) {
|
---|
3746 | PFETCH(c);
|
---|
3747 | if (c == '^') {
|
---|
3748 | tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0);
|
---|
3749 | }
|
---|
3750 | else
|
---|
3751 | PUNFETCH;
|
---|
3752 | }
|
---|
3753 | }
|
---|
3754 | else {
|
---|
3755 | onig_syntax_warn(env, "invalid Unicode Property \\%c", c);
|
---|
3756 | }
|
---|
3757 | break;
|
---|
3758 |
|
---|
3759 | case 'R':
|
---|
3760 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK)) {
|
---|
3761 | tok->type = TK_LINEBREAK;
|
---|
3762 | }
|
---|
3763 | break;
|
---|
3764 |
|
---|
3765 | case 'X':
|
---|
3766 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER)) {
|
---|
3767 | tok->type = TK_EXTENDED_GRAPHEME_CLUSTER;
|
---|
3768 | }
|
---|
3769 | break;
|
---|
3770 |
|
---|
3771 | case 'K':
|
---|
3772 | if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP)) {
|
---|
3773 | tok->type = TK_KEEP;
|
---|
3774 | }
|
---|
3775 | break;
|
---|
3776 |
|
---|
3777 | default:
|
---|
3778 | {
|
---|
3779 | OnigCodePoint c2;
|
---|
3780 |
|
---|
3781 | PUNFETCH;
|
---|
3782 | num = fetch_escaped_value(&p, end, env, &c2);
|
---|
3783 | if (num < 0) return num;
|
---|
3784 | /* set_raw: */
|
---|
3785 | if ((OnigCodePoint )tok->u.c != c2) {
|
---|
3786 | tok->type = TK_CODE_POINT;
|
---|
3787 | tok->u.code = (OnigCodePoint )c2;
|
---|
3788 | }
|
---|
3789 | else { /* string */
|
---|
3790 | p = tok->backp + enclen(enc, tok->backp, end);
|
---|
3791 | }
|
---|
3792 | }
|
---|
3793 | break;
|
---|
3794 | }
|
---|
3795 | }
|
---|
3796 | else {
|
---|
3797 | tok->u.c = c;
|
---|
3798 | tok->escaped = 0;
|
---|
3799 |
|
---|
3800 | #ifdef USE_VARIABLE_META_CHARS
|
---|
3801 | if ((c != ONIG_INEFFECTIVE_META_CHAR) &&
|
---|
3802 | IS_SYNTAX_OP(syn, ONIG_SYN_OP_VARIABLE_META_CHARACTERS)) {
|
---|
3803 | if (c == MC_ANYCHAR(syn))
|
---|
3804 | goto any_char;
|
---|
3805 | else if (c == MC_ANYTIME(syn))
|
---|
3806 | goto anytime;
|
---|
3807 | else if (c == MC_ZERO_OR_ONE_TIME(syn))
|
---|
3808 | goto zero_or_one_time;
|
---|
3809 | else if (c == MC_ONE_OR_MORE_TIME(syn))
|
---|
3810 | goto one_or_more_time;
|
---|
3811 | else if (c == MC_ANYCHAR_ANYTIME(syn)) {
|
---|
3812 | tok->type = TK_ANYCHAR_ANYTIME;
|
---|
3813 | goto out;
|
---|
3814 | }
|
---|
3815 | }
|
---|
3816 | #endif
|
---|
3817 |
|
---|
3818 | switch (c) {
|
---|
3819 | case '.':
|
---|
3820 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_DOT_ANYCHAR)) break;
|
---|
3821 | #ifdef USE_VARIABLE_META_CHARS
|
---|
3822 | any_char:
|
---|
3823 | #endif
|
---|
3824 | tok->type = TK_ANYCHAR;
|
---|
3825 | break;
|
---|
3826 |
|
---|
3827 | case '*':
|
---|
3828 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ASTERISK_ZERO_INF)) break;
|
---|
3829 | #ifdef USE_VARIABLE_META_CHARS
|
---|
3830 | anytime:
|
---|
3831 | #endif
|
---|
3832 | tok->type = TK_OP_REPEAT;
|
---|
3833 | tok->u.repeat.lower = 0;
|
---|
3834 | tok->u.repeat.upper = REPEAT_INFINITE;
|
---|
3835 | goto greedy_check;
|
---|
3836 | break;
|
---|
3837 |
|
---|
3838 | case '+':
|
---|
3839 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_PLUS_ONE_INF)) break;
|
---|
3840 | #ifdef USE_VARIABLE_META_CHARS
|
---|
3841 | one_or_more_time:
|
---|
3842 | #endif
|
---|
3843 | tok->type = TK_OP_REPEAT;
|
---|
3844 | tok->u.repeat.lower = 1;
|
---|
3845 | tok->u.repeat.upper = REPEAT_INFINITE;
|
---|
3846 | goto greedy_check;
|
---|
3847 | break;
|
---|
3848 |
|
---|
3849 | case '?':
|
---|
3850 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_ZERO_ONE)) break;
|
---|
3851 | #ifdef USE_VARIABLE_META_CHARS
|
---|
3852 | zero_or_one_time:
|
---|
3853 | #endif
|
---|
3854 | tok->type = TK_OP_REPEAT;
|
---|
3855 | tok->u.repeat.lower = 0;
|
---|
3856 | tok->u.repeat.upper = 1;
|
---|
3857 | goto greedy_check;
|
---|
3858 | break;
|
---|
3859 |
|
---|
3860 | case '{':
|
---|
3861 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACE_INTERVAL)) break;
|
---|
3862 | r = fetch_range_quantifier(&p, end, tok, env);
|
---|
3863 | if (r < 0) return r; /* error */
|
---|
3864 | if (r == 0) goto greedy_check;
|
---|
3865 | else if (r == 2) { /* {n} */
|
---|
3866 | if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY))
|
---|
3867 | goto possessive_check;
|
---|
3868 |
|
---|
3869 | goto greedy_check;
|
---|
3870 | }
|
---|
3871 | /* r == 1 : normal char */
|
---|
3872 | break;
|
---|
3873 |
|
---|
3874 | case '|':
|
---|
3875 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_VBAR_ALT)) break;
|
---|
3876 | tok->type = TK_ALT;
|
---|
3877 | break;
|
---|
3878 |
|
---|
3879 | case '(':
|
---|
3880 | if (PPEEK_IS('?') &&
|
---|
3881 | IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) {
|
---|
3882 | PINC;
|
---|
3883 | if (PPEEK_IS('#')) {
|
---|
3884 | PFETCH(c);
|
---|
3885 | while (1) {
|
---|
3886 | if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;
|
---|
3887 | PFETCH(c);
|
---|
3888 | if (c == MC_ESC(syn)) {
|
---|
3889 | if (!PEND) PFETCH(c);
|
---|
3890 | }
|
---|
3891 | else {
|
---|
3892 | if (c == ')') break;
|
---|
3893 | }
|
---|
3894 | }
|
---|
3895 | goto start;
|
---|
3896 | }
|
---|
3897 | #ifdef USE_PERL_SUBEXP_CALL
|
---|
3898 | /* (?&name), (?n), (?R), (?0), (?+n), (?-n) */
|
---|
3899 | c = PPEEK;
|
---|
3900 | if ((c == '&' || c == 'R' || ONIGENC_IS_CODE_DIGIT(enc, c)) &&
|
---|
3901 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_SUBEXP_CALL)) {
|
---|
3902 | /* (?&name), (?n), (?R), (?0) */
|
---|
3903 | int gnum;
|
---|
3904 | UChar *name;
|
---|
3905 | UChar *name_end;
|
---|
3906 |
|
---|
3907 | if (c == 'R' || c == '0') {
|
---|
3908 | PINC; /* skip 'R' / '0' */
|
---|
3909 | if (!PPEEK_IS(')')) return ONIGERR_INVALID_GROUP_NAME;
|
---|
3910 | PINC; /* skip ')' */
|
---|
3911 | name_end = name = p;
|
---|
3912 | gnum = 0;
|
---|
3913 | }
|
---|
3914 | else {
|
---|
3915 | int numref = 1;
|
---|
3916 | if (c == '&') { /* (?&name) */
|
---|
3917 | PINC;
|
---|
3918 | numref = 0; /* don't allow number name */
|
---|
3919 | }
|
---|
3920 | name = p;
|
---|
3921 | r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, numref);
|
---|
3922 | if (r < 0) return r;
|
---|
3923 | }
|
---|
3924 |
|
---|
3925 | tok->type = TK_CALL;
|
---|
3926 | tok->u.call.name = name;
|
---|
3927 | tok->u.call.name_end = name_end;
|
---|
3928 | tok->u.call.gnum = gnum;
|
---|
3929 | tok->u.call.rel = 0;
|
---|
3930 | break;
|
---|
3931 | }
|
---|
3932 | else if ((c == '-' || c == '+') &&
|
---|
3933 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_SUBEXP_CALL)) {
|
---|
3934 | /* (?+n), (?-n) */
|
---|
3935 | int gnum;
|
---|
3936 | UChar *name;
|
---|
3937 | UChar *name_end;
|
---|
3938 | OnigCodePoint cnext;
|
---|
3939 | PFETCH_READY;
|
---|
3940 |
|
---|
3941 | PINC; /* skip '-' / '+' */
|
---|
3942 | cnext = PPEEK;
|
---|
3943 | if (ONIGENC_IS_CODE_DIGIT(enc, cnext)) {
|
---|
3944 | if (c == '-') PUNFETCH;
|
---|
3945 | name = p;
|
---|
3946 | r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, 1);
|
---|
3947 | if (r < 0) return r;
|
---|
3948 |
|
---|
3949 | tok->type = TK_CALL;
|
---|
3950 | tok->u.call.name = name;
|
---|
3951 | tok->u.call.name_end = name_end;
|
---|
3952 | tok->u.call.gnum = gnum;
|
---|
3953 | tok->u.call.rel = 1;
|
---|
3954 | break;
|
---|
3955 | }
|
---|
3956 | }
|
---|
3957 | #endif /* USE_PERL_SUBEXP_CALL */
|
---|
3958 | #ifdef USE_CAPITAL_P_NAMED_GROUP
|
---|
3959 | if (PPEEK_IS('P') &&
|
---|
3960 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP)) {
|
---|
3961 | int gnum;
|
---|
3962 | UChar *name;
|
---|
3963 | UChar *name_end;
|
---|
3964 | PFETCH_READY;
|
---|
3965 |
|
---|
3966 | PINC; /* skip 'P' */
|
---|
3967 | if (PEND) return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
3968 | PFETCH(c);
|
---|
3969 | if (c == '=') { /* (?P=name): backref */
|
---|
3970 | r = fetch_named_backref_token((OnigCodePoint )'(', tok, &p, end, env);
|
---|
3971 | if (r < 0) return r;
|
---|
3972 | break;
|
---|
3973 | }
|
---|
3974 | else if (c == '>') { /* (?P>name): subexp call */
|
---|
3975 | name = p;
|
---|
3976 | r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, 0);
|
---|
3977 | if (r < 0) return r;
|
---|
3978 |
|
---|
3979 | tok->type = TK_CALL;
|
---|
3980 | tok->u.call.name = name;
|
---|
3981 | tok->u.call.name_end = name_end;
|
---|
3982 | tok->u.call.gnum = gnum;
|
---|
3983 | tok->u.call.rel = 0;
|
---|
3984 | break;
|
---|
3985 | }
|
---|
3986 | }
|
---|
3987 | #endif /* USE_CAPITAL_P_NAMED_GROUP */
|
---|
3988 | PUNFETCH;
|
---|
3989 | }
|
---|
3990 |
|
---|
3991 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP)) break;
|
---|
3992 | tok->type = TK_SUBEXP_OPEN;
|
---|
3993 | break;
|
---|
3994 |
|
---|
3995 | case ')':
|
---|
3996 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP)) break;
|
---|
3997 | tok->type = TK_SUBEXP_CLOSE;
|
---|
3998 | break;
|
---|
3999 |
|
---|
4000 | case '^':
|
---|
4001 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR)) break;
|
---|
4002 | tok->type = TK_ANCHOR;
|
---|
4003 | tok->u.anchor.subtype = (IS_SINGLELINE(env->option)
|
---|
4004 | ? ANCHOR_BEGIN_BUF : ANCHOR_BEGIN_LINE);
|
---|
4005 | break;
|
---|
4006 |
|
---|
4007 | case '$':
|
---|
4008 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR)) break;
|
---|
4009 | tok->type = TK_ANCHOR;
|
---|
4010 | tok->u.anchor.subtype = (IS_SINGLELINE(env->option)
|
---|
4011 | ? ANCHOR_SEMI_END_BUF : ANCHOR_END_LINE);
|
---|
4012 | break;
|
---|
4013 |
|
---|
4014 | case '[':
|
---|
4015 | if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACKET_CC)) break;
|
---|
4016 | tok->type = TK_CC_OPEN;
|
---|
4017 | break;
|
---|
4018 |
|
---|
4019 | case ']':
|
---|
4020 | if (*src > env->pattern) /* /].../ is allowed. */
|
---|
4021 | CLOSE_BRACKET_WITHOUT_ESC_WARN(env, (UChar* )"]");
|
---|
4022 | break;
|
---|
4023 |
|
---|
4024 | case '#':
|
---|
4025 | if (IS_EXTEND(env->option)) {
|
---|
4026 | while (!PEND) {
|
---|
4027 | PFETCH(c);
|
---|
4028 | if (ONIGENC_IS_CODE_NEWLINE(enc, c))
|
---|
4029 | break;
|
---|
4030 | }
|
---|
4031 | goto start;
|
---|
4032 | break;
|
---|
4033 | }
|
---|
4034 | break;
|
---|
4035 |
|
---|
4036 | case ' ': case '\t': case '\n': case '\r': case '\f':
|
---|
4037 | if (IS_EXTEND(env->option))
|
---|
4038 | goto start;
|
---|
4039 | break;
|
---|
4040 |
|
---|
4041 | default:
|
---|
4042 | /* string */
|
---|
4043 | break;
|
---|
4044 | }
|
---|
4045 | }
|
---|
4046 |
|
---|
4047 | #ifdef USE_VARIABLE_META_CHARS
|
---|
4048 | out:
|
---|
4049 | #endif
|
---|
4050 | *src = p;
|
---|
4051 | return tok->type;
|
---|
4052 | }
|
---|
4053 |
|
---|
4054 | static int
|
---|
4055 | add_ctype_to_cc_by_range(CClassNode* cc, int ctype ARG_UNUSED, int not,
|
---|
4056 | ScanEnv* env,
|
---|
4057 | OnigCodePoint sb_out, const OnigCodePoint mbr[])
|
---|
4058 | {
|
---|
4059 | int i, r;
|
---|
4060 | OnigCodePoint j;
|
---|
4061 |
|
---|
4062 | int n = ONIGENC_CODE_RANGE_NUM(mbr);
|
---|
4063 |
|
---|
4064 | if (not == 0) {
|
---|
4065 | for (i = 0; i < n; i++) {
|
---|
4066 | for (j = ONIGENC_CODE_RANGE_FROM(mbr, i);
|
---|
4067 | j <= ONIGENC_CODE_RANGE_TO(mbr, i); j++) {
|
---|
4068 | if (j >= sb_out) {
|
---|
4069 | if (j > ONIGENC_CODE_RANGE_FROM(mbr, i)) {
|
---|
4070 | r = add_code_range_to_buf(&(cc->mbuf), env, j,
|
---|
4071 | ONIGENC_CODE_RANGE_TO(mbr, i));
|
---|
4072 | if (r != 0) return r;
|
---|
4073 | i++;
|
---|
4074 | }
|
---|
4075 |
|
---|
4076 | goto sb_end;
|
---|
4077 | }
|
---|
4078 | BITSET_SET_BIT_CHKDUP(cc->bs, j);
|
---|
4079 | }
|
---|
4080 | }
|
---|
4081 |
|
---|
4082 | sb_end:
|
---|
4083 | for ( ; i < n; i++) {
|
---|
4084 | r = add_code_range_to_buf(&(cc->mbuf), env,
|
---|
4085 | ONIGENC_CODE_RANGE_FROM(mbr, i),
|
---|
4086 | ONIGENC_CODE_RANGE_TO(mbr, i));
|
---|
4087 | if (r != 0) return r;
|
---|
4088 | }
|
---|
4089 | }
|
---|
4090 | else {
|
---|
4091 | OnigCodePoint prev = 0;
|
---|
4092 |
|
---|
4093 | for (i = 0; i < n; i++) {
|
---|
4094 | for (j = prev;
|
---|
4095 | j < ONIGENC_CODE_RANGE_FROM(mbr, i); j++) {
|
---|
4096 | if (j >= sb_out) {
|
---|
4097 | goto sb_end2;
|
---|
4098 | }
|
---|
4099 | BITSET_SET_BIT_CHKDUP(cc->bs, j);
|
---|
4100 | }
|
---|
4101 | prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1;
|
---|
4102 | }
|
---|
4103 | for (j = prev; j < sb_out; j++) {
|
---|
4104 | BITSET_SET_BIT_CHKDUP(cc->bs, j);
|
---|
4105 | }
|
---|
4106 |
|
---|
4107 | sb_end2:
|
---|
4108 | prev = sb_out;
|
---|
4109 |
|
---|
4110 | for (i = 0; i < n; i++) {
|
---|
4111 | if (prev < ONIGENC_CODE_RANGE_FROM(mbr, i)) {
|
---|
4112 | r = add_code_range_to_buf(&(cc->mbuf), env, prev,
|
---|
4113 | ONIGENC_CODE_RANGE_FROM(mbr, i) - 1);
|
---|
4114 | if (r != 0) return r;
|
---|
4115 | }
|
---|
4116 | prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1;
|
---|
4117 | }
|
---|
4118 | if (prev < 0x7fffffff) {
|
---|
4119 | r = add_code_range_to_buf(&(cc->mbuf), env, prev, 0x7fffffff);
|
---|
4120 | if (r != 0) return r;
|
---|
4121 | }
|
---|
4122 | }
|
---|
4123 |
|
---|
4124 | return 0;
|
---|
4125 | }
|
---|
4126 |
|
---|
4127 | static int
|
---|
4128 | add_ctype_to_cc(CClassNode* cc, int ctype, int not, int ascii_range, ScanEnv* env)
|
---|
4129 | {
|
---|
4130 | int maxcode;
|
---|
4131 | int c, r;
|
---|
4132 | const OnigCodePoint *ranges;
|
---|
4133 | OnigCodePoint sb_out;
|
---|
4134 | OnigEncoding enc = env->enc;
|
---|
4135 |
|
---|
4136 | r = ONIGENC_GET_CTYPE_CODE_RANGE(enc, ctype, &sb_out, &ranges);
|
---|
4137 | if (r == 0) {
|
---|
4138 | if (ascii_range) {
|
---|
4139 | CClassNode ccwork;
|
---|
4140 | initialize_cclass(&ccwork);
|
---|
4141 | r = add_ctype_to_cc_by_range(&ccwork, ctype, not, env, sb_out,
|
---|
4142 | ranges);
|
---|
4143 | if (r == 0) {
|
---|
4144 | if (not) {
|
---|
4145 | r = add_code_range_to_buf0(&(ccwork.mbuf), env, 0x80, ONIG_LAST_CODE_POINT, FALSE);
|
---|
4146 | }
|
---|
4147 | else {
|
---|
4148 | CClassNode ccascii;
|
---|
4149 | initialize_cclass(&ccascii);
|
---|
4150 | if (ONIGENC_MBC_MINLEN(env->enc) > 1) {
|
---|
4151 | r = add_code_range(&(ccascii.mbuf), env, 0x00, 0x7F);
|
---|
4152 | }
|
---|
4153 | else {
|
---|
4154 | bitset_set_range(env, ccascii.bs, 0x00, 0x7F);
|
---|
4155 | r = 0;
|
---|
4156 | }
|
---|
4157 | if (r == 0) {
|
---|
4158 | r = and_cclass(&ccwork, &ccascii, env);
|
---|
4159 | }
|
---|
4160 | if (IS_NOT_NULL(ccascii.mbuf)) bbuf_free(ccascii.mbuf);
|
---|
4161 | }
|
---|
4162 | if (r == 0) {
|
---|
4163 | r = or_cclass(cc, &ccwork, env);
|
---|
4164 | }
|
---|
4165 | if (IS_NOT_NULL(ccwork.mbuf)) bbuf_free(ccwork.mbuf);
|
---|
4166 | }
|
---|
4167 | }
|
---|
4168 | else {
|
---|
4169 | r = add_ctype_to_cc_by_range(cc, ctype, not, env, sb_out, ranges);
|
---|
4170 | }
|
---|
4171 | return r;
|
---|
4172 | }
|
---|
4173 | else if (r != ONIG_NO_SUPPORT_CONFIG) {
|
---|
4174 | return r;
|
---|
4175 | }
|
---|
4176 |
|
---|
4177 | maxcode = ascii_range ? 0x80 : SINGLE_BYTE_SIZE;
|
---|
4178 | r = 0;
|
---|
4179 | switch (ctype) {
|
---|
4180 | case ONIGENC_CTYPE_ALPHA:
|
---|
4181 | case ONIGENC_CTYPE_BLANK:
|
---|
4182 | case ONIGENC_CTYPE_CNTRL:
|
---|
4183 | case ONIGENC_CTYPE_DIGIT:
|
---|
4184 | case ONIGENC_CTYPE_LOWER:
|
---|
4185 | case ONIGENC_CTYPE_PUNCT:
|
---|
4186 | case ONIGENC_CTYPE_SPACE:
|
---|
4187 | case ONIGENC_CTYPE_UPPER:
|
---|
4188 | case ONIGENC_CTYPE_XDIGIT:
|
---|
4189 | case ONIGENC_CTYPE_ASCII:
|
---|
4190 | case ONIGENC_CTYPE_ALNUM:
|
---|
4191 | if (not != 0) {
|
---|
4192 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
|
---|
4193 | if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
|
---|
4194 | BITSET_SET_BIT_CHKDUP(cc->bs, c);
|
---|
4195 | }
|
---|
4196 | ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
|
---|
4197 | }
|
---|
4198 | else {
|
---|
4199 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
|
---|
4200 | if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
|
---|
4201 | BITSET_SET_BIT_CHKDUP(cc->bs, c);
|
---|
4202 | }
|
---|
4203 | }
|
---|
4204 | break;
|
---|
4205 |
|
---|
4206 | case ONIGENC_CTYPE_GRAPH:
|
---|
4207 | case ONIGENC_CTYPE_PRINT:
|
---|
4208 | if (not != 0) {
|
---|
4209 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
|
---|
4210 | if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype)
|
---|
4211 | || c >= maxcode)
|
---|
4212 | BITSET_SET_BIT_CHKDUP(cc->bs, c);
|
---|
4213 | }
|
---|
4214 | if (ascii_range)
|
---|
4215 | ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
|
---|
4216 | }
|
---|
4217 | else {
|
---|
4218 | for (c = 0; c < maxcode; c++) {
|
---|
4219 | if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
|
---|
4220 | BITSET_SET_BIT_CHKDUP(cc->bs, c);
|
---|
4221 | }
|
---|
4222 | if (! ascii_range)
|
---|
4223 | ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
|
---|
4224 | }
|
---|
4225 | break;
|
---|
4226 |
|
---|
4227 | case ONIGENC_CTYPE_WORD:
|
---|
4228 | if (not == 0) {
|
---|
4229 | for (c = 0; c < maxcode; c++) {
|
---|
4230 | if (ONIGENC_IS_CODE_WORD(enc, c)) BITSET_SET_BIT_CHKDUP(cc->bs, c);
|
---|
4231 | }
|
---|
4232 | if (! ascii_range)
|
---|
4233 | ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
|
---|
4234 | }
|
---|
4235 | else {
|
---|
4236 | for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
|
---|
4237 | if ((ONIGENC_CODE_TO_MBCLEN(enc, c) > 0) /* check invalid code point */
|
---|
4238 | && (! ONIGENC_IS_CODE_WORD(enc, c) || c >= maxcode))
|
---|
4239 | BITSET_SET_BIT_CHKDUP(cc->bs, c);
|
---|
4240 | }
|
---|
4241 | if (ascii_range)
|
---|
4242 | ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
|
---|
4243 | }
|
---|
4244 | break;
|
---|
4245 |
|
---|
4246 | default:
|
---|
4247 | return ONIGERR_PARSER_BUG;
|
---|
4248 | break;
|
---|
4249 | }
|
---|
4250 |
|
---|
4251 | return r;
|
---|
4252 | }
|
---|
4253 |
|
---|
4254 | static int
|
---|
4255 | parse_posix_bracket(CClassNode* cc, CClassNode* asc_cc,
|
---|
4256 | UChar** src, UChar* end, ScanEnv* env)
|
---|
4257 | {
|
---|
4258 | #define POSIX_BRACKET_CHECK_LIMIT_LENGTH 20
|
---|
4259 | #define POSIX_BRACKET_NAME_MIN_LEN 4
|
---|
4260 |
|
---|
4261 | static const PosixBracketEntryType PBS[] = {
|
---|
4262 | POSIX_BRACKET_ENTRY_INIT("alnum", ONIGENC_CTYPE_ALNUM),
|
---|
4263 | POSIX_BRACKET_ENTRY_INIT("alpha", ONIGENC_CTYPE_ALPHA),
|
---|
4264 | POSIX_BRACKET_ENTRY_INIT("blank", ONIGENC_CTYPE_BLANK),
|
---|
4265 | POSIX_BRACKET_ENTRY_INIT("cntrl", ONIGENC_CTYPE_CNTRL),
|
---|
4266 | POSIX_BRACKET_ENTRY_INIT("digit", ONIGENC_CTYPE_DIGIT),
|
---|
4267 | POSIX_BRACKET_ENTRY_INIT("graph", ONIGENC_CTYPE_GRAPH),
|
---|
4268 | POSIX_BRACKET_ENTRY_INIT("lower", ONIGENC_CTYPE_LOWER),
|
---|
4269 | POSIX_BRACKET_ENTRY_INIT("print", ONIGENC_CTYPE_PRINT),
|
---|
4270 | POSIX_BRACKET_ENTRY_INIT("punct", ONIGENC_CTYPE_PUNCT),
|
---|
4271 | POSIX_BRACKET_ENTRY_INIT("space", ONIGENC_CTYPE_SPACE),
|
---|
4272 | POSIX_BRACKET_ENTRY_INIT("upper", ONIGENC_CTYPE_UPPER),
|
---|
4273 | POSIX_BRACKET_ENTRY_INIT("xdigit", ONIGENC_CTYPE_XDIGIT),
|
---|
4274 | POSIX_BRACKET_ENTRY_INIT("ascii", ONIGENC_CTYPE_ASCII),
|
---|
4275 | POSIX_BRACKET_ENTRY_INIT("word", ONIGENC_CTYPE_WORD),
|
---|
4276 | };
|
---|
4277 |
|
---|
4278 | const PosixBracketEntryType *pb;
|
---|
4279 | int not, i, r;
|
---|
4280 | int ascii_range;
|
---|
4281 | OnigCodePoint c;
|
---|
4282 | OnigEncoding enc = env->enc;
|
---|
4283 | UChar *p = *src;
|
---|
4284 |
|
---|
4285 | if (PPEEK_IS('^')) {
|
---|
4286 | PINC_S;
|
---|
4287 | not = 1;
|
---|
4288 | }
|
---|
4289 | else
|
---|
4290 | not = 0;
|
---|
4291 |
|
---|
4292 | if (onigenc_strlen(enc, p, end) < POSIX_BRACKET_NAME_MIN_LEN + 3)
|
---|
4293 | goto not_posix_bracket;
|
---|
4294 |
|
---|
4295 | ascii_range = IS_ASCII_RANGE(env->option) &&
|
---|
4296 | ! IS_POSIX_BRACKET_ALL_RANGE(env->option);
|
---|
4297 | for (pb = PBS; pb < PBS + numberof(PBS); pb++) {
|
---|
4298 | if (onigenc_with_ascii_strncmp(enc, p, end, pb->name, pb->len) == 0) {
|
---|
4299 | p = (UChar* )onigenc_step(enc, p, end, pb->len);
|
---|
4300 | if (onigenc_with_ascii_strncmp(enc, p, end, (UChar* )":]", 2) != 0)
|
---|
4301 | return ONIGERR_INVALID_POSIX_BRACKET_TYPE;
|
---|
4302 |
|
---|
4303 | r = add_ctype_to_cc(cc, pb->ctype, not, ascii_range, env);
|
---|
4304 | if (r != 0) return r;
|
---|
4305 |
|
---|
4306 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4307 | if (pb->ctype != ONIGENC_CTYPE_WORD &&
|
---|
4308 | pb->ctype != ONIGENC_CTYPE_ASCII &&
|
---|
4309 | !ascii_range)
|
---|
4310 | r = add_ctype_to_cc(asc_cc, pb->ctype, not, ascii_range, env);
|
---|
4311 | if (r != 0) return r;
|
---|
4312 | }
|
---|
4313 |
|
---|
4314 | PINC_S; PINC_S;
|
---|
4315 | *src = p;
|
---|
4316 | return 0;
|
---|
4317 | }
|
---|
4318 | }
|
---|
4319 |
|
---|
4320 | not_posix_bracket:
|
---|
4321 | c = 0;
|
---|
4322 | i = 0;
|
---|
4323 | while (!PEND && ((c = PPEEK) != ':') && c != ']') {
|
---|
4324 | PINC_S;
|
---|
4325 | if (++i > POSIX_BRACKET_CHECK_LIMIT_LENGTH) break;
|
---|
4326 | }
|
---|
4327 | if (c == ':' && ! PEND) {
|
---|
4328 | PINC_S;
|
---|
4329 | if (! PEND) {
|
---|
4330 | PFETCH_S(c);
|
---|
4331 | if (c == ']')
|
---|
4332 | return ONIGERR_INVALID_POSIX_BRACKET_TYPE;
|
---|
4333 | }
|
---|
4334 | }
|
---|
4335 |
|
---|
4336 | return 1; /* 1: is not POSIX bracket, but no error. */
|
---|
4337 | }
|
---|
4338 |
|
---|
4339 | static int
|
---|
4340 | fetch_char_property_to_ctype(UChar** src, UChar* end, ScanEnv* env)
|
---|
4341 | {
|
---|
4342 | int r;
|
---|
4343 | OnigCodePoint c;
|
---|
4344 | OnigEncoding enc = env->enc;
|
---|
4345 | UChar *prev, *start, *p = *src;
|
---|
4346 |
|
---|
4347 | r = 0;
|
---|
4348 | start = prev = p;
|
---|
4349 |
|
---|
4350 | while (!PEND) {
|
---|
4351 | prev = p;
|
---|
4352 | PFETCH_S(c);
|
---|
4353 | if (c == '}') {
|
---|
4354 | r = ONIGENC_PROPERTY_NAME_TO_CTYPE(enc, start, prev);
|
---|
4355 | if (r < 0) break;
|
---|
4356 |
|
---|
4357 | *src = p;
|
---|
4358 | return r;
|
---|
4359 | }
|
---|
4360 | else if (c == '(' || c == ')' || c == '{' || c == '|') {
|
---|
4361 | r = ONIGERR_INVALID_CHAR_PROPERTY_NAME;
|
---|
4362 | break;
|
---|
4363 | }
|
---|
4364 | }
|
---|
4365 |
|
---|
4366 | onig_scan_env_set_error_string(env, r, *src, prev);
|
---|
4367 | return r;
|
---|
4368 | }
|
---|
4369 |
|
---|
4370 | static int cclass_case_fold(Node** np, CClassNode* cc, CClassNode* asc_cc, ScanEnv* env);
|
---|
4371 |
|
---|
4372 | static int
|
---|
4373 | parse_char_property(Node** np, OnigToken* tok, UChar** src, UChar* end,
|
---|
4374 | ScanEnv* env)
|
---|
4375 | {
|
---|
4376 | int r, ctype;
|
---|
4377 | CClassNode* cc;
|
---|
4378 |
|
---|
4379 | ctype = fetch_char_property_to_ctype(src, end, env);
|
---|
4380 | if (ctype < 0) return ctype;
|
---|
4381 |
|
---|
4382 | *np = node_new_cclass();
|
---|
4383 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
4384 | cc = NCCLASS(*np);
|
---|
4385 | r = add_ctype_to_cc(cc, ctype, 0, 0, env);
|
---|
4386 | if (r != 0) return r;
|
---|
4387 | if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
|
---|
4388 |
|
---|
4389 | if (IS_IGNORECASE(env->option)) {
|
---|
4390 | if (ctype != ONIGENC_CTYPE_ASCII)
|
---|
4391 | r = cclass_case_fold(np, cc, cc, env);
|
---|
4392 | }
|
---|
4393 | return r;
|
---|
4394 | }
|
---|
4395 |
|
---|
4396 |
|
---|
4397 | enum CCSTATE {
|
---|
4398 | CCS_VALUE,
|
---|
4399 | CCS_RANGE,
|
---|
4400 | CCS_COMPLETE,
|
---|
4401 | CCS_START
|
---|
4402 | };
|
---|
4403 |
|
---|
4404 | enum CCVALTYPE {
|
---|
4405 | CCV_SB,
|
---|
4406 | CCV_CODE_POINT,
|
---|
4407 | CCV_CLASS
|
---|
4408 | };
|
---|
4409 |
|
---|
4410 | static int
|
---|
4411 | next_state_class(CClassNode* cc, CClassNode* asc_cc,
|
---|
4412 | OnigCodePoint* vs, enum CCVALTYPE* type,
|
---|
4413 | enum CCSTATE* state, ScanEnv* env)
|
---|
4414 | {
|
---|
4415 | int r;
|
---|
4416 |
|
---|
4417 | if (*state == CCS_RANGE)
|
---|
4418 | return ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE;
|
---|
4419 |
|
---|
4420 | if (*state == CCS_VALUE && *type != CCV_CLASS) {
|
---|
4421 | if (*type == CCV_SB) {
|
---|
4422 | BITSET_SET_BIT_CHKDUP(cc->bs, (int )(*vs));
|
---|
4423 | if (IS_NOT_NULL(asc_cc))
|
---|
4424 | BITSET_SET_BIT(asc_cc->bs, (int )(*vs));
|
---|
4425 | }
|
---|
4426 | else if (*type == CCV_CODE_POINT) {
|
---|
4427 | r = add_code_range(&(cc->mbuf), env, *vs, *vs);
|
---|
4428 | if (r < 0) return r;
|
---|
4429 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4430 | r = add_code_range0(&(asc_cc->mbuf), env, *vs, *vs, 0);
|
---|
4431 | if (r < 0) return r;
|
---|
4432 | }
|
---|
4433 | }
|
---|
4434 | }
|
---|
4435 |
|
---|
4436 | *state = CCS_VALUE;
|
---|
4437 | *type = CCV_CLASS;
|
---|
4438 | return 0;
|
---|
4439 | }
|
---|
4440 |
|
---|
4441 | static int
|
---|
4442 | next_state_val(CClassNode* cc, CClassNode* asc_cc,
|
---|
4443 | OnigCodePoint *from, OnigCodePoint to,
|
---|
4444 | int* from_israw, int to_israw,
|
---|
4445 | enum CCVALTYPE intype, enum CCVALTYPE* type,
|
---|
4446 | enum CCSTATE* state, ScanEnv* env)
|
---|
4447 | {
|
---|
4448 | int r;
|
---|
4449 |
|
---|
4450 | switch (*state) {
|
---|
4451 | case CCS_VALUE:
|
---|
4452 | if (*type == CCV_SB) {
|
---|
4453 | BITSET_SET_BIT_CHKDUP(cc->bs, (int )(*from));
|
---|
4454 | if (IS_NOT_NULL(asc_cc))
|
---|
4455 | BITSET_SET_BIT(asc_cc->bs, (int )(*from));
|
---|
4456 | }
|
---|
4457 | else if (*type == CCV_CODE_POINT) {
|
---|
4458 | r = add_code_range(&(cc->mbuf), env, *from, *from);
|
---|
4459 | if (r < 0) return r;
|
---|
4460 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4461 | r = add_code_range0(&(asc_cc->mbuf), env, *from, *from, 0);
|
---|
4462 | if (r < 0) return r;
|
---|
4463 | }
|
---|
4464 | }
|
---|
4465 | break;
|
---|
4466 |
|
---|
4467 | case CCS_RANGE:
|
---|
4468 | if (intype == *type) {
|
---|
4469 | if (intype == CCV_SB) {
|
---|
4470 | if (*from > 0xff || to > 0xff)
|
---|
4471 | return ONIGERR_INVALID_CODE_POINT_VALUE;
|
---|
4472 |
|
---|
4473 | if (*from > to) {
|
---|
4474 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
|
---|
4475 | goto ccs_range_end;
|
---|
4476 | else
|
---|
4477 | return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
|
---|
4478 | }
|
---|
4479 | bitset_set_range(env, cc->bs, (int )*from, (int )to);
|
---|
4480 | if (IS_NOT_NULL(asc_cc))
|
---|
4481 | bitset_set_range(env, asc_cc->bs, (int )*from, (int )to);
|
---|
4482 | }
|
---|
4483 | else {
|
---|
4484 | r = add_code_range(&(cc->mbuf), env, *from, to);
|
---|
4485 | if (r < 0) return r;
|
---|
4486 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4487 | r = add_code_range0(&(asc_cc->mbuf), env, *from, to, 0);
|
---|
4488 | if (r < 0) return r;
|
---|
4489 | }
|
---|
4490 | }
|
---|
4491 | }
|
---|
4492 | else {
|
---|
4493 | if (*from > to) {
|
---|
4494 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
|
---|
4495 | goto ccs_range_end;
|
---|
4496 | else
|
---|
4497 | return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
|
---|
4498 | }
|
---|
4499 | bitset_set_range(env, cc->bs, (int )*from, (int )(to < 0xff ? to : 0xff));
|
---|
4500 | r = add_code_range(&(cc->mbuf), env, (OnigCodePoint )*from, to);
|
---|
4501 | if (r < 0) return r;
|
---|
4502 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4503 | bitset_set_range(env, asc_cc->bs, (int )*from, (int )(to < 0xff ? to : 0xff));
|
---|
4504 | r = add_code_range0(&(asc_cc->mbuf), env, (OnigCodePoint )*from, to, 0);
|
---|
4505 | if (r < 0) return r;
|
---|
4506 | }
|
---|
4507 | }
|
---|
4508 | ccs_range_end:
|
---|
4509 | *state = CCS_COMPLETE;
|
---|
4510 | break;
|
---|
4511 |
|
---|
4512 | case CCS_COMPLETE:
|
---|
4513 | case CCS_START:
|
---|
4514 | *state = CCS_VALUE;
|
---|
4515 | break;
|
---|
4516 |
|
---|
4517 | default:
|
---|
4518 | break;
|
---|
4519 | }
|
---|
4520 |
|
---|
4521 | *from_israw = to_israw;
|
---|
4522 | *from = to;
|
---|
4523 | *type = intype;
|
---|
4524 | return 0;
|
---|
4525 | }
|
---|
4526 |
|
---|
4527 | static int
|
---|
4528 | code_exist_check(OnigCodePoint c, UChar* from, UChar* end, int ignore_escaped,
|
---|
4529 | ScanEnv* env)
|
---|
4530 | {
|
---|
4531 | int in_esc;
|
---|
4532 | OnigCodePoint code;
|
---|
4533 | OnigEncoding enc = env->enc;
|
---|
4534 | UChar* p = from;
|
---|
4535 |
|
---|
4536 | in_esc = 0;
|
---|
4537 | while (! PEND) {
|
---|
4538 | if (ignore_escaped && in_esc) {
|
---|
4539 | in_esc = 0;
|
---|
4540 | }
|
---|
4541 | else {
|
---|
4542 | PFETCH_S(code);
|
---|
4543 | if (code == c) return 1;
|
---|
4544 | if (code == MC_ESC(env->syntax)) in_esc = 1;
|
---|
4545 | }
|
---|
4546 | }
|
---|
4547 | return 0;
|
---|
4548 | }
|
---|
4549 |
|
---|
4550 | static int
|
---|
4551 | parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* end,
|
---|
4552 | ScanEnv* env)
|
---|
4553 | {
|
---|
4554 | int r, neg, len, fetched, and_start;
|
---|
4555 | OnigCodePoint v, vs;
|
---|
4556 | UChar *p;
|
---|
4557 | Node* node;
|
---|
4558 | Node* asc_node;
|
---|
4559 | CClassNode *cc, *prev_cc;
|
---|
4560 | CClassNode *asc_cc, *asc_prev_cc;
|
---|
4561 | CClassNode work_cc, asc_work_cc;
|
---|
4562 |
|
---|
4563 | enum CCSTATE state;
|
---|
4564 | enum CCVALTYPE val_type, in_type;
|
---|
4565 | int val_israw, in_israw;
|
---|
4566 |
|
---|
4567 | *np = *asc_np = NULL_NODE;
|
---|
4568 | env->parse_depth++;
|
---|
4569 | if (env->parse_depth > ParseDepthLimit)
|
---|
4570 | return ONIGERR_PARSE_DEPTH_LIMIT_OVER;
|
---|
4571 | prev_cc = asc_prev_cc = (CClassNode* )NULL;
|
---|
4572 | r = fetch_token_in_cc(tok, src, end, env);
|
---|
4573 | if (r == TK_CHAR && tok->u.c == '^' && tok->escaped == 0) {
|
---|
4574 | neg = 1;
|
---|
4575 | r = fetch_token_in_cc(tok, src, end, env);
|
---|
4576 | }
|
---|
4577 | else {
|
---|
4578 | neg = 0;
|
---|
4579 | }
|
---|
4580 |
|
---|
4581 | if (r < 0) return r;
|
---|
4582 | if (r == TK_CC_CLOSE) {
|
---|
4583 | if (! code_exist_check((OnigCodePoint )']',
|
---|
4584 | *src, env->pattern_end, 1, env))
|
---|
4585 | return ONIGERR_EMPTY_CHAR_CLASS;
|
---|
4586 |
|
---|
4587 | CC_ESC_WARN(env, (UChar* )"]");
|
---|
4588 | r = tok->type = TK_CHAR; /* allow []...] */
|
---|
4589 | }
|
---|
4590 |
|
---|
4591 | *np = node = node_new_cclass();
|
---|
4592 | CHECK_NULL_RETURN_MEMERR(node);
|
---|
4593 | cc = NCCLASS(node);
|
---|
4594 |
|
---|
4595 | if (IS_IGNORECASE(env->option)) {
|
---|
4596 | *asc_np = asc_node = node_new_cclass();
|
---|
4597 | CHECK_NULL_RETURN_MEMERR(asc_node);
|
---|
4598 | asc_cc = NCCLASS(asc_node);
|
---|
4599 | }
|
---|
4600 | else {
|
---|
4601 | asc_node = NULL_NODE;
|
---|
4602 | asc_cc = NULL;
|
---|
4603 | }
|
---|
4604 |
|
---|
4605 | and_start = 0;
|
---|
4606 | state = CCS_START;
|
---|
4607 | p = *src;
|
---|
4608 | while (r != TK_CC_CLOSE) {
|
---|
4609 | fetched = 0;
|
---|
4610 | switch (r) {
|
---|
4611 | case TK_CHAR:
|
---|
4612 | if ((tok->u.code >= SINGLE_BYTE_SIZE) ||
|
---|
4613 | (len = ONIGENC_CODE_TO_MBCLEN(env->enc, tok->u.c)) > 1) {
|
---|
4614 | in_type = CCV_CODE_POINT;
|
---|
4615 | }
|
---|
4616 | else if (len < 0) {
|
---|
4617 | r = len;
|
---|
4618 | goto err;
|
---|
4619 | }
|
---|
4620 | else {
|
---|
4621 | sb_char:
|
---|
4622 | in_type = CCV_SB;
|
---|
4623 | }
|
---|
4624 | v = (OnigCodePoint )tok->u.c;
|
---|
4625 | in_israw = 0;
|
---|
4626 | goto val_entry2;
|
---|
4627 | break;
|
---|
4628 |
|
---|
4629 | case TK_RAW_BYTE:
|
---|
4630 | /* tok->base != 0 : octal or hexadec. */
|
---|
4631 | if (! ONIGENC_IS_SINGLEBYTE(env->enc) && tok->base != 0) {
|
---|
4632 | UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
|
---|
4633 | UChar* bufe = buf + ONIGENC_CODE_TO_MBC_MAXLEN;
|
---|
4634 | UChar* psave = p;
|
---|
4635 | int i, base = tok->base;
|
---|
4636 |
|
---|
4637 | buf[0] = (UChar )tok->u.c;
|
---|
4638 | for (i = 1; i < ONIGENC_MBC_MAXLEN(env->enc); i++) {
|
---|
4639 | r = fetch_token_in_cc(tok, &p, end, env);
|
---|
4640 | if (r < 0) goto err;
|
---|
4641 | if (r != TK_RAW_BYTE || tok->base != base) {
|
---|
4642 | fetched = 1;
|
---|
4643 | break;
|
---|
4644 | }
|
---|
4645 | buf[i] = (UChar )tok->u.c;
|
---|
4646 | }
|
---|
4647 |
|
---|
4648 | if (i < ONIGENC_MBC_MINLEN(env->enc)) {
|
---|
4649 | r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
|
---|
4650 | goto err;
|
---|
4651 | }
|
---|
4652 |
|
---|
4653 | len = enclen(env->enc, buf, buf + i);
|
---|
4654 | if (i < len) {
|
---|
4655 | r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
|
---|
4656 | goto err;
|
---|
4657 | }
|
---|
4658 | else if (i > len) { /* fetch back */
|
---|
4659 | p = psave;
|
---|
4660 | for (i = 1; i < len; i++) {
|
---|
4661 | (void)fetch_token_in_cc(tok, &p, end, env);
|
---|
4662 | /* no need to check the retun value (already checked above) */
|
---|
4663 | }
|
---|
4664 | fetched = 0;
|
---|
4665 | }
|
---|
4666 |
|
---|
4667 | if (i == 1) {
|
---|
4668 | v = (OnigCodePoint )buf[0];
|
---|
4669 | goto raw_single;
|
---|
4670 | }
|
---|
4671 | else {
|
---|
4672 | v = ONIGENC_MBC_TO_CODE(env->enc, buf, bufe);
|
---|
4673 | in_type = CCV_CODE_POINT;
|
---|
4674 | }
|
---|
4675 | }
|
---|
4676 | else {
|
---|
4677 | v = (OnigCodePoint )tok->u.c;
|
---|
4678 | raw_single:
|
---|
4679 | in_type = CCV_SB;
|
---|
4680 | }
|
---|
4681 | in_israw = 1;
|
---|
4682 | goto val_entry2;
|
---|
4683 | break;
|
---|
4684 |
|
---|
4685 | case TK_CODE_POINT:
|
---|
4686 | v = tok->u.code;
|
---|
4687 | in_israw = 1;
|
---|
4688 | val_entry:
|
---|
4689 | len = ONIGENC_CODE_TO_MBCLEN(env->enc, v);
|
---|
4690 | if (len < 0) {
|
---|
4691 | r = len;
|
---|
4692 | goto err;
|
---|
4693 | }
|
---|
4694 | in_type = (len == 1 ? CCV_SB : CCV_CODE_POINT);
|
---|
4695 | val_entry2:
|
---|
4696 | r = next_state_val(cc, asc_cc, &vs, v, &val_israw, in_israw, in_type, &val_type,
|
---|
4697 | &state, env);
|
---|
4698 | if (r != 0) goto err;
|
---|
4699 | break;
|
---|
4700 |
|
---|
4701 | case TK_POSIX_BRACKET_OPEN:
|
---|
4702 | r = parse_posix_bracket(cc, asc_cc, &p, end, env);
|
---|
4703 | if (r < 0) goto err;
|
---|
4704 | if (r == 1) { /* is not POSIX bracket */
|
---|
4705 | CC_ESC_WARN(env, (UChar* )"[");
|
---|
4706 | p = tok->backp;
|
---|
4707 | v = (OnigCodePoint )tok->u.c;
|
---|
4708 | in_israw = 0;
|
---|
4709 | goto val_entry;
|
---|
4710 | }
|
---|
4711 | goto next_class;
|
---|
4712 | break;
|
---|
4713 |
|
---|
4714 | case TK_CHAR_TYPE:
|
---|
4715 | r = add_ctype_to_cc(cc, tok->u.prop.ctype, tok->u.prop.not,
|
---|
4716 | IS_ASCII_RANGE(env->option), env);
|
---|
4717 | if (r != 0) return r;
|
---|
4718 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4719 | if (tok->u.prop.ctype != ONIGENC_CTYPE_WORD)
|
---|
4720 | r = add_ctype_to_cc(asc_cc, tok->u.prop.ctype, tok->u.prop.not,
|
---|
4721 | IS_ASCII_RANGE(env->option), env);
|
---|
4722 | if (r != 0) return r;
|
---|
4723 | }
|
---|
4724 |
|
---|
4725 | next_class:
|
---|
4726 | r = next_state_class(cc, asc_cc, &vs, &val_type, &state, env);
|
---|
4727 | if (r != 0) goto err;
|
---|
4728 | break;
|
---|
4729 |
|
---|
4730 | case TK_CHAR_PROPERTY:
|
---|
4731 | {
|
---|
4732 | int ctype;
|
---|
4733 |
|
---|
4734 | ctype = fetch_char_property_to_ctype(&p, end, env);
|
---|
4735 | if (ctype < 0) return ctype;
|
---|
4736 | r = add_ctype_to_cc(cc, ctype, tok->u.prop.not, 0, env);
|
---|
4737 | if (r != 0) return r;
|
---|
4738 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4739 | if (ctype != ONIGENC_CTYPE_ASCII)
|
---|
4740 | r = add_ctype_to_cc(asc_cc, ctype, tok->u.prop.not, 0, env);
|
---|
4741 | if (r != 0) return r;
|
---|
4742 | }
|
---|
4743 | goto next_class;
|
---|
4744 | }
|
---|
4745 | break;
|
---|
4746 |
|
---|
4747 | case TK_CC_RANGE:
|
---|
4748 | if (state == CCS_VALUE) {
|
---|
4749 | r = fetch_token_in_cc(tok, &p, end, env);
|
---|
4750 | if (r < 0) goto err;
|
---|
4751 | fetched = 1;
|
---|
4752 | if (r == TK_CC_CLOSE) { /* allow [x-] */
|
---|
4753 | range_end_val:
|
---|
4754 | v = (OnigCodePoint )'-';
|
---|
4755 | in_israw = 0;
|
---|
4756 | goto val_entry;
|
---|
4757 | }
|
---|
4758 | else if (r == TK_CC_AND) {
|
---|
4759 | CC_ESC_WARN(env, (UChar* )"-");
|
---|
4760 | goto range_end_val;
|
---|
4761 | }
|
---|
4762 |
|
---|
4763 | if (val_type == CCV_CLASS) {
|
---|
4764 | r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS;
|
---|
4765 | goto err;
|
---|
4766 | }
|
---|
4767 |
|
---|
4768 | state = CCS_RANGE;
|
---|
4769 | }
|
---|
4770 | else if (state == CCS_START) {
|
---|
4771 | /* [-xa] is allowed */
|
---|
4772 | v = (OnigCodePoint )tok->u.c;
|
---|
4773 | in_israw = 0;
|
---|
4774 |
|
---|
4775 | r = fetch_token_in_cc(tok, &p, end, env);
|
---|
4776 | if (r < 0) goto err;
|
---|
4777 | fetched = 1;
|
---|
4778 | /* [--x] or [a&&-x] is warned. */
|
---|
4779 | if (r == TK_CC_RANGE || and_start != 0)
|
---|
4780 | CC_ESC_WARN(env, (UChar* )"-");
|
---|
4781 |
|
---|
4782 | goto val_entry;
|
---|
4783 | }
|
---|
4784 | else if (state == CCS_RANGE) {
|
---|
4785 | CC_ESC_WARN(env, (UChar* )"-");
|
---|
4786 | goto sb_char; /* [!--x] is allowed */
|
---|
4787 | }
|
---|
4788 | else { /* CCS_COMPLETE */
|
---|
4789 | r = fetch_token_in_cc(tok, &p, end, env);
|
---|
4790 | if (r < 0) goto err;
|
---|
4791 | fetched = 1;
|
---|
4792 | if (r == TK_CC_CLOSE) goto range_end_val; /* allow [a-b-] */
|
---|
4793 | else if (r == TK_CC_AND) {
|
---|
4794 | CC_ESC_WARN(env, (UChar* )"-");
|
---|
4795 | goto range_end_val;
|
---|
4796 | }
|
---|
4797 |
|
---|
4798 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC)) {
|
---|
4799 | CC_ESC_WARN(env, (UChar* )"-");
|
---|
4800 | goto range_end_val; /* [0-9-a] is allowed as [0-9\-a] */
|
---|
4801 | }
|
---|
4802 | r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS;
|
---|
4803 | goto err;
|
---|
4804 | }
|
---|
4805 | break;
|
---|
4806 |
|
---|
4807 | case TK_CC_CC_OPEN: /* [ */
|
---|
4808 | {
|
---|
4809 | Node *anode, *aasc_node;
|
---|
4810 | CClassNode* acc;
|
---|
4811 |
|
---|
4812 | r = parse_char_class(&anode, &aasc_node, tok, &p, end, env);
|
---|
4813 | if (r == 0) {
|
---|
4814 | acc = NCCLASS(anode);
|
---|
4815 | r = or_cclass(cc, acc, env);
|
---|
4816 | }
|
---|
4817 | if (r == 0 && IS_NOT_NULL(aasc_node)) {
|
---|
4818 | acc = NCCLASS(aasc_node);
|
---|
4819 | r = or_cclass(asc_cc, acc, env);
|
---|
4820 | }
|
---|
4821 | onig_node_free(anode);
|
---|
4822 | onig_node_free(aasc_node);
|
---|
4823 | if (r != 0) goto err;
|
---|
4824 | }
|
---|
4825 | break;
|
---|
4826 |
|
---|
4827 | case TK_CC_AND: /* && */
|
---|
4828 | {
|
---|
4829 | if (state == CCS_VALUE) {
|
---|
4830 | r = next_state_val(cc, asc_cc, &vs, 0, &val_israw, 0, val_type,
|
---|
4831 | &val_type, &state, env);
|
---|
4832 | if (r != 0) goto err;
|
---|
4833 | }
|
---|
4834 | /* initialize local variables */
|
---|
4835 | and_start = 1;
|
---|
4836 | state = CCS_START;
|
---|
4837 |
|
---|
4838 | if (IS_NOT_NULL(prev_cc)) {
|
---|
4839 | r = and_cclass(prev_cc, cc, env);
|
---|
4840 | if (r != 0) goto err;
|
---|
4841 | bbuf_free(cc->mbuf);
|
---|
4842 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4843 | r = and_cclass(asc_prev_cc, asc_cc, env);
|
---|
4844 | if (r != 0) goto err;
|
---|
4845 | bbuf_free(asc_cc->mbuf);
|
---|
4846 | }
|
---|
4847 | }
|
---|
4848 | else {
|
---|
4849 | prev_cc = cc;
|
---|
4850 | cc = &work_cc;
|
---|
4851 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4852 | asc_prev_cc = asc_cc;
|
---|
4853 | asc_cc = &asc_work_cc;
|
---|
4854 | }
|
---|
4855 | }
|
---|
4856 | initialize_cclass(cc);
|
---|
4857 | if (IS_NOT_NULL(asc_cc))
|
---|
4858 | initialize_cclass(asc_cc);
|
---|
4859 | }
|
---|
4860 | break;
|
---|
4861 |
|
---|
4862 | case TK_EOT:
|
---|
4863 | r = ONIGERR_PREMATURE_END_OF_CHAR_CLASS;
|
---|
4864 | goto err;
|
---|
4865 | break;
|
---|
4866 | default:
|
---|
4867 | r = ONIGERR_PARSER_BUG;
|
---|
4868 | goto err;
|
---|
4869 | break;
|
---|
4870 | }
|
---|
4871 |
|
---|
4872 | if (fetched)
|
---|
4873 | r = tok->type;
|
---|
4874 | else {
|
---|
4875 | r = fetch_token_in_cc(tok, &p, end, env);
|
---|
4876 | if (r < 0) goto err;
|
---|
4877 | }
|
---|
4878 | }
|
---|
4879 |
|
---|
4880 | if (state == CCS_VALUE) {
|
---|
4881 | r = next_state_val(cc, asc_cc, &vs, 0, &val_israw, 0, val_type,
|
---|
4882 | &val_type, &state, env);
|
---|
4883 | if (r != 0) goto err;
|
---|
4884 | }
|
---|
4885 |
|
---|
4886 | if (IS_NOT_NULL(prev_cc)) {
|
---|
4887 | r = and_cclass(prev_cc, cc, env);
|
---|
4888 | if (r != 0) goto err;
|
---|
4889 | bbuf_free(cc->mbuf);
|
---|
4890 | cc = prev_cc;
|
---|
4891 | if (IS_NOT_NULL(asc_cc)) {
|
---|
4892 | r = and_cclass(asc_prev_cc, asc_cc, env);
|
---|
4893 | if (r != 0) goto err;
|
---|
4894 | bbuf_free(asc_cc->mbuf);
|
---|
4895 | asc_cc = asc_prev_cc;
|
---|
4896 | }
|
---|
4897 | }
|
---|
4898 |
|
---|
4899 | if (neg != 0) {
|
---|
4900 | NCCLASS_SET_NOT(cc);
|
---|
4901 | if (IS_NOT_NULL(asc_cc))
|
---|
4902 | NCCLASS_SET_NOT(asc_cc);
|
---|
4903 | }
|
---|
4904 | else {
|
---|
4905 | NCCLASS_CLEAR_NOT(cc);
|
---|
4906 | if (IS_NOT_NULL(asc_cc))
|
---|
4907 | NCCLASS_CLEAR_NOT(asc_cc);
|
---|
4908 | }
|
---|
4909 | if (IS_NCCLASS_NOT(cc) &&
|
---|
4910 | IS_SYNTAX_BV(env->syntax, ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC)) {
|
---|
4911 | int is_empty;
|
---|
4912 |
|
---|
4913 | is_empty = (IS_NULL(cc->mbuf) ? 1 : 0);
|
---|
4914 | if (is_empty != 0)
|
---|
4915 | BITSET_IS_EMPTY(cc->bs, is_empty);
|
---|
4916 |
|
---|
4917 | if (is_empty == 0) {
|
---|
4918 | #define NEWLINE_CODE 0x0a
|
---|
4919 |
|
---|
4920 | if (ONIGENC_IS_CODE_NEWLINE(env->enc, NEWLINE_CODE)) {
|
---|
4921 | if (ONIGENC_CODE_TO_MBCLEN(env->enc, NEWLINE_CODE) == 1)
|
---|
4922 | BITSET_SET_BIT_CHKDUP(cc->bs, NEWLINE_CODE);
|
---|
4923 | else {
|
---|
4924 | r = add_code_range(&(cc->mbuf), env, NEWLINE_CODE, NEWLINE_CODE);
|
---|
4925 | if (r < 0) goto err;
|
---|
4926 | }
|
---|
4927 | }
|
---|
4928 | }
|
---|
4929 | }
|
---|
4930 | *src = p;
|
---|
4931 | env->parse_depth--;
|
---|
4932 | return 0;
|
---|
4933 |
|
---|
4934 | err:
|
---|
4935 | if (cc != NCCLASS(*np))
|
---|
4936 | bbuf_free(cc->mbuf);
|
---|
4937 | if (IS_NOT_NULL(asc_cc) && (asc_cc != NCCLASS(*asc_np)))
|
---|
4938 | bbuf_free(asc_cc->mbuf);
|
---|
4939 | return r;
|
---|
4940 | }
|
---|
4941 |
|
---|
4942 | static int parse_subexp(Node** top, OnigToken* tok, int term,
|
---|
4943 | UChar** src, UChar* end, ScanEnv* env);
|
---|
4944 |
|
---|
4945 | static int
|
---|
4946 | parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end,
|
---|
4947 | ScanEnv* env)
|
---|
4948 | {
|
---|
4949 | int r = 0, num;
|
---|
4950 | Node *target, *work1 = NULL, *work2 = NULL;
|
---|
4951 | OnigOptionType option;
|
---|
4952 | OnigCodePoint c;
|
---|
4953 | OnigEncoding enc = env->enc;
|
---|
4954 |
|
---|
4955 | #ifdef USE_NAMED_GROUP
|
---|
4956 | int list_capture;
|
---|
4957 | #endif
|
---|
4958 |
|
---|
4959 | UChar* p = *src;
|
---|
4960 | PFETCH_READY;
|
---|
4961 |
|
---|
4962 | *np = NULL;
|
---|
4963 | if (PEND) return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
|
---|
4964 |
|
---|
4965 | option = env->option;
|
---|
4966 | if (PPEEK_IS('?') &&
|
---|
4967 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) {
|
---|
4968 | PINC;
|
---|
4969 | if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;
|
---|
4970 |
|
---|
4971 | PFETCH(c);
|
---|
4972 | switch (c) {
|
---|
4973 | case ':': /* (?:...) grouping only */
|
---|
4974 | group:
|
---|
4975 | r = fetch_token(tok, &p, end, env);
|
---|
4976 | if (r < 0) return r;
|
---|
4977 | r = parse_subexp(np, tok, term, &p, end, env);
|
---|
4978 | if (r < 0) return r;
|
---|
4979 | *src = p;
|
---|
4980 | return 1; /* group */
|
---|
4981 | break;
|
---|
4982 |
|
---|
4983 | case '=':
|
---|
4984 | *np = onig_node_new_anchor(ANCHOR_PREC_READ);
|
---|
4985 | break;
|
---|
4986 | case '!': /* preceding read */
|
---|
4987 | *np = onig_node_new_anchor(ANCHOR_PREC_READ_NOT);
|
---|
4988 | break;
|
---|
4989 | case '>': /* (?>...) stop backtrack */
|
---|
4990 | *np = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
|
---|
4991 | break;
|
---|
4992 | case '~': /* (?~...) absent operator */
|
---|
4993 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_TILDE_ABSENT)) {
|
---|
4994 | *np = node_new_enclose(ENCLOSE_ABSENT);
|
---|
4995 | }
|
---|
4996 | else {
|
---|
4997 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
4998 | }
|
---|
4999 | break;
|
---|
5000 |
|
---|
5001 | #ifdef USE_NAMED_GROUP
|
---|
5002 | case '\'':
|
---|
5003 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
|
---|
5004 | goto named_group1;
|
---|
5005 | }
|
---|
5006 | else
|
---|
5007 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5008 | break;
|
---|
5009 |
|
---|
5010 | # ifdef USE_CAPITAL_P_NAMED_GROUP
|
---|
5011 | case 'P': /* (?P<name>...) */
|
---|
5012 | if (!PEND &&
|
---|
5013 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP)) {
|
---|
5014 | PFETCH(c);
|
---|
5015 | if (c == '<') goto named_group1;
|
---|
5016 | }
|
---|
5017 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5018 | break;
|
---|
5019 | # endif
|
---|
5020 | #endif
|
---|
5021 |
|
---|
5022 | case '<': /* look behind (?<=...), (?<!...) */
|
---|
5023 | if (PEND) return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
|
---|
5024 | PFETCH(c);
|
---|
5025 | if (c == '=')
|
---|
5026 | *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND);
|
---|
5027 | else if (c == '!')
|
---|
5028 | *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND_NOT);
|
---|
5029 | #ifdef USE_NAMED_GROUP
|
---|
5030 | else { /* (?<name>...) */
|
---|
5031 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
|
---|
5032 | UChar *name;
|
---|
5033 | UChar *name_end;
|
---|
5034 |
|
---|
5035 | PUNFETCH;
|
---|
5036 | c = '<';
|
---|
5037 |
|
---|
5038 | named_group1:
|
---|
5039 | list_capture = 0;
|
---|
5040 |
|
---|
5041 | # ifdef USE_CAPTURE_HISTORY
|
---|
5042 | named_group2:
|
---|
5043 | # endif
|
---|
5044 | name = p;
|
---|
5045 | r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &num, 0);
|
---|
5046 | if (r < 0) return r;
|
---|
5047 |
|
---|
5048 | num = scan_env_add_mem_entry(env);
|
---|
5049 | if (num < 0) return num;
|
---|
5050 | if (list_capture != 0 && num >= (int )BIT_STATUS_BITS_NUM)
|
---|
5051 | return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;
|
---|
5052 |
|
---|
5053 | r = name_add(env->reg, name, name_end, num, env);
|
---|
5054 | if (r != 0) return r;
|
---|
5055 | *np = node_new_enclose_memory(env->option, 1);
|
---|
5056 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
5057 | NENCLOSE(*np)->regnum = num;
|
---|
5058 | if (list_capture != 0)
|
---|
5059 | BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num);
|
---|
5060 | env->num_named++;
|
---|
5061 | }
|
---|
5062 | else {
|
---|
5063 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5064 | }
|
---|
5065 | }
|
---|
5066 | #else
|
---|
5067 | else {
|
---|
5068 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5069 | }
|
---|
5070 | #endif
|
---|
5071 | break;
|
---|
5072 |
|
---|
5073 | #ifdef USE_CAPTURE_HISTORY
|
---|
5074 | case '@':
|
---|
5075 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY)) {
|
---|
5076 | # ifdef USE_NAMED_GROUP
|
---|
5077 | if (!PEND &&
|
---|
5078 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
|
---|
5079 | PFETCH(c);
|
---|
5080 | if (c == '<' || c == '\'') {
|
---|
5081 | list_capture = 1;
|
---|
5082 | goto named_group2; /* (?@<name>...) */
|
---|
5083 | }
|
---|
5084 | PUNFETCH;
|
---|
5085 | }
|
---|
5086 | # endif
|
---|
5087 | *np = node_new_enclose_memory(env->option, 0);
|
---|
5088 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
5089 | num = scan_env_add_mem_entry(env);
|
---|
5090 | if (num < 0) return num;
|
---|
5091 | if (num >= (int )BIT_STATUS_BITS_NUM)
|
---|
5092 | return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;
|
---|
5093 |
|
---|
5094 | NENCLOSE(*np)->regnum = num;
|
---|
5095 | BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num);
|
---|
5096 | }
|
---|
5097 | else {
|
---|
5098 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5099 | }
|
---|
5100 | break;
|
---|
5101 | #endif /* USE_CAPTURE_HISTORY */
|
---|
5102 |
|
---|
5103 | case '(': /* conditional expression: (?(cond)yes), (?(cond)yes|no) */
|
---|
5104 | if (!PEND &&
|
---|
5105 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LPAREN_CONDITION)) {
|
---|
5106 | UChar *name = NULL;
|
---|
5107 | UChar *name_end;
|
---|
5108 | PFETCH(c);
|
---|
5109 | if (ONIGENC_IS_CODE_DIGIT(enc, c)) { /* (n) */
|
---|
5110 | PUNFETCH;
|
---|
5111 | r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &num, 1);
|
---|
5112 | if (r < 0) return r;
|
---|
5113 | #if 0
|
---|
5114 | /* Relative number is not currently supported. (same as Perl) */
|
---|
5115 | if (num < 0) {
|
---|
5116 | num = BACKREF_REL_TO_ABS(num, env);
|
---|
5117 | if (num <= 0)
|
---|
5118 | return ONIGERR_INVALID_BACKREF;
|
---|
5119 | }
|
---|
5120 | #endif
|
---|
5121 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_STRICT_CHECK_BACKREF)) {
|
---|
5122 | if (num > env->num_mem ||
|
---|
5123 | IS_NULL(SCANENV_MEM_NODES(env)[num]))
|
---|
5124 | return ONIGERR_INVALID_BACKREF;
|
---|
5125 | }
|
---|
5126 | }
|
---|
5127 | #ifdef USE_NAMED_GROUP
|
---|
5128 | else if (c == '<' || c == '\'') { /* (<name>), ('name') */
|
---|
5129 | name = p;
|
---|
5130 | r = fetch_named_backref_token(c, tok, &p, end, env);
|
---|
5131 | if (r < 0) return r;
|
---|
5132 | if (!PPEEK_IS(')')) return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5133 | PINC;
|
---|
5134 |
|
---|
5135 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_USE_LEFT_MOST_NAMED_GROUP)) {
|
---|
5136 | num = tok->u.backref.ref1;
|
---|
5137 | }
|
---|
5138 | else {
|
---|
5139 | /* FIXME:
|
---|
5140 | * Use left most named group for now. This is the same as Perl.
|
---|
5141 | * However this should use the same strategy as normal back-
|
---|
5142 | * references on Ruby syntax; search right to left. */
|
---|
5143 | int len = tok->u.backref.num;
|
---|
5144 | num = len > 1 ? tok->u.backref.refs[0] : tok->u.backref.ref1;
|
---|
5145 | }
|
---|
5146 | }
|
---|
5147 | #endif
|
---|
5148 | else
|
---|
5149 | return ONIGERR_INVALID_CONDITION_PATTERN;
|
---|
5150 | *np = node_new_enclose(ENCLOSE_CONDITION);
|
---|
5151 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
5152 | NENCLOSE(*np)->regnum = num;
|
---|
5153 | if (IS_NOT_NULL(name)) NENCLOSE(*np)->state |= NST_NAME_REF;
|
---|
5154 | }
|
---|
5155 | else
|
---|
5156 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5157 | break;
|
---|
5158 |
|
---|
5159 | #if 0
|
---|
5160 | case '|': /* branch reset: (?|...) */
|
---|
5161 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_VBAR_BRANCH_RESET)) {
|
---|
5162 | /* TODO */
|
---|
5163 | }
|
---|
5164 | else
|
---|
5165 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5166 | break;
|
---|
5167 | #endif
|
---|
5168 |
|
---|
5169 | case '^': /* loads default options */
|
---|
5170 | if (!PEND && IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
|
---|
5171 | /* d-imsx */
|
---|
5172 | ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1);
|
---|
5173 | ONOFF(option, ONIG_OPTION_IGNORECASE, 1);
|
---|
5174 | ONOFF(option, ONIG_OPTION_SINGLELINE, 0);
|
---|
5175 | ONOFF(option, ONIG_OPTION_MULTILINE, 1);
|
---|
5176 | ONOFF(option, ONIG_OPTION_EXTEND, 1);
|
---|
5177 | PFETCH(c);
|
---|
5178 | }
|
---|
5179 | #if 0
|
---|
5180 | else if (!PEND && IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) {
|
---|
5181 | /* d-imx */
|
---|
5182 | ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0);
|
---|
5183 | ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 0);
|
---|
5184 | ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 0);
|
---|
5185 | ONOFF(option, ONIG_OPTION_IGNORECASE, 1);
|
---|
5186 | ONOFF(option, ONIG_OPTION_MULTILINE, 1);
|
---|
5187 | ONOFF(option, ONIG_OPTION_EXTEND, 1);
|
---|
5188 | PFETCH(c);
|
---|
5189 | }
|
---|
5190 | #endif
|
---|
5191 | else {
|
---|
5192 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5193 | }
|
---|
5194 | /* fall through */
|
---|
5195 | #ifdef USE_POSIXLINE_OPTION
|
---|
5196 | case 'p':
|
---|
5197 | #endif
|
---|
5198 | case '-': case 'i': case 'm': case 's': case 'x':
|
---|
5199 | case 'a': case 'd': case 'l': case 'u':
|
---|
5200 | {
|
---|
5201 | int neg = 0;
|
---|
5202 |
|
---|
5203 | while (1) {
|
---|
5204 | switch (c) {
|
---|
5205 | case ':':
|
---|
5206 | case ')':
|
---|
5207 | break;
|
---|
5208 |
|
---|
5209 | case '-': neg = 1; break;
|
---|
5210 | case 'x': ONOFF(option, ONIG_OPTION_EXTEND, neg); break;
|
---|
5211 | case 'i': ONOFF(option, ONIG_OPTION_IGNORECASE, neg); break;
|
---|
5212 | case 's':
|
---|
5213 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
|
---|
5214 | ONOFF(option, ONIG_OPTION_MULTILINE, neg);
|
---|
5215 | }
|
---|
5216 | else
|
---|
5217 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5218 | break;
|
---|
5219 |
|
---|
5220 | case 'm':
|
---|
5221 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
|
---|
5222 | ONOFF(option, ONIG_OPTION_SINGLELINE, (neg == 0 ? 1 : 0));
|
---|
5223 | }
|
---|
5224 | else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) {
|
---|
5225 | ONOFF(option, ONIG_OPTION_MULTILINE, neg);
|
---|
5226 | }
|
---|
5227 | else
|
---|
5228 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5229 | break;
|
---|
5230 | #ifdef USE_POSIXLINE_OPTION
|
---|
5231 | case 'p':
|
---|
5232 | ONOFF(option, ONIG_OPTION_MULTILINE|ONIG_OPTION_SINGLELINE, neg);
|
---|
5233 | break;
|
---|
5234 | #endif
|
---|
5235 |
|
---|
5236 | case 'a': /* limits \d, \s, \w and POSIX brackets to ASCII range */
|
---|
5237 | if ((IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) ||
|
---|
5238 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) &&
|
---|
5239 | (neg == 0)) {
|
---|
5240 | ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0);
|
---|
5241 | ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 1);
|
---|
5242 | ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 1);
|
---|
5243 | }
|
---|
5244 | else
|
---|
5245 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5246 | break;
|
---|
5247 |
|
---|
5248 | case 'u':
|
---|
5249 | if ((IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) ||
|
---|
5250 | IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) &&
|
---|
5251 | (neg == 0)) {
|
---|
5252 | ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1);
|
---|
5253 | ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 1);
|
---|
5254 | ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 1);
|
---|
5255 | }
|
---|
5256 | else
|
---|
5257 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5258 | break;
|
---|
5259 |
|
---|
5260 | case 'd':
|
---|
5261 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) &&
|
---|
5262 | (neg == 0)) {
|
---|
5263 | ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1);
|
---|
5264 | }
|
---|
5265 | else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY) &&
|
---|
5266 | (neg == 0)) {
|
---|
5267 | ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0);
|
---|
5268 | ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 0);
|
---|
5269 | ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 0);
|
---|
5270 | }
|
---|
5271 | else
|
---|
5272 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5273 | break;
|
---|
5274 |
|
---|
5275 | case 'l':
|
---|
5276 | if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) && (neg == 0)) {
|
---|
5277 | ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1);
|
---|
5278 | }
|
---|
5279 | else
|
---|
5280 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5281 | break;
|
---|
5282 |
|
---|
5283 | default:
|
---|
5284 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5285 | }
|
---|
5286 |
|
---|
5287 | if (c == ')') {
|
---|
5288 | *np = node_new_option(option);
|
---|
5289 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
5290 | *src = p;
|
---|
5291 | return 2; /* option only */
|
---|
5292 | }
|
---|
5293 | else if (c == ':') {
|
---|
5294 | OnigOptionType prev = env->option;
|
---|
5295 |
|
---|
5296 | env->option = option;
|
---|
5297 | r = fetch_token(tok, &p, end, env);
|
---|
5298 | if (r < 0) {
|
---|
5299 | env->option = prev;
|
---|
5300 | return r;
|
---|
5301 | }
|
---|
5302 | r = parse_subexp(&target, tok, term, &p, end, env);
|
---|
5303 | env->option = prev;
|
---|
5304 | if (r < 0) return r;
|
---|
5305 | *np = node_new_option(option);
|
---|
5306 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
5307 | NENCLOSE(*np)->target = target;
|
---|
5308 | *src = p;
|
---|
5309 | return 0;
|
---|
5310 | }
|
---|
5311 |
|
---|
5312 | if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;
|
---|
5313 | PFETCH(c);
|
---|
5314 | }
|
---|
5315 | }
|
---|
5316 | break;
|
---|
5317 |
|
---|
5318 | default:
|
---|
5319 | return ONIGERR_UNDEFINED_GROUP_OPTION;
|
---|
5320 | }
|
---|
5321 | }
|
---|
5322 | else {
|
---|
5323 | if (ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_DONT_CAPTURE_GROUP))
|
---|
5324 | goto group;
|
---|
5325 |
|
---|
5326 | *np = node_new_enclose_memory(env->option, 0);
|
---|
5327 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
5328 | num = scan_env_add_mem_entry(env);
|
---|
5329 | if (num < 0) return num;
|
---|
5330 | NENCLOSE(*np)->regnum = num;
|
---|
5331 | }
|
---|
5332 |
|
---|
5333 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
5334 | r = fetch_token(tok, &p, end, env);
|
---|
5335 | if (r < 0) return r;
|
---|
5336 | r = parse_subexp(&target, tok, term, &p, end, env);
|
---|
5337 | if (r < 0) {
|
---|
5338 | onig_node_free(target);
|
---|
5339 | return r;
|
---|
5340 | }
|
---|
5341 |
|
---|
5342 | if (NTYPE(*np) == NT_ANCHOR)
|
---|
5343 | NANCHOR(*np)->target = target;
|
---|
5344 | else {
|
---|
5345 | NENCLOSE(*np)->target = target;
|
---|
5346 | if (NENCLOSE(*np)->type == ENCLOSE_MEMORY) {
|
---|
5347 | /* Don't move this to previous of parse_subexp() */
|
---|
5348 | r = scan_env_set_mem_node(env, NENCLOSE(*np)->regnum, *np);
|
---|
5349 | if (r != 0) return r;
|
---|
5350 | }
|
---|
5351 | else if (NENCLOSE(*np)->type == ENCLOSE_CONDITION) {
|
---|
5352 | if (NTYPE(target) != NT_ALT) {
|
---|
5353 | /* convert (?(cond)yes) to (?(cond)yes|empty) */
|
---|
5354 | work1 = node_new_empty();
|
---|
5355 | if (IS_NULL(work1)) goto err;
|
---|
5356 | work2 = onig_node_new_alt(work1, NULL_NODE);
|
---|
5357 | if (IS_NULL(work2)) goto err;
|
---|
5358 | work1 = onig_node_new_alt(target, work2);
|
---|
5359 | if (IS_NULL(work1)) goto err;
|
---|
5360 | NENCLOSE(*np)->target = work1;
|
---|
5361 | }
|
---|
5362 | }
|
---|
5363 | }
|
---|
5364 |
|
---|
5365 | *src = p;
|
---|
5366 | return 0;
|
---|
5367 |
|
---|
5368 | err:
|
---|
5369 | onig_node_free(work1);
|
---|
5370 | onig_node_free(work2);
|
---|
5371 | onig_node_free(*np);
|
---|
5372 | *np = NULL;
|
---|
5373 | return ONIGERR_MEMORY;
|
---|
5374 | }
|
---|
5375 |
|
---|
5376 | static const char* const PopularQStr[] = {
|
---|
5377 | "?", "*", "+", "??", "*?", "+?"
|
---|
5378 | };
|
---|
5379 |
|
---|
5380 | static const char* const ReduceQStr[] = {
|
---|
5381 | "", "", "*", "*?", "??", "+ and ??", "+? and ?"
|
---|
5382 | };
|
---|
5383 |
|
---|
5384 | static int
|
---|
5385 | set_quantifier(Node* qnode, Node* target, int group, ScanEnv* env)
|
---|
5386 | {
|
---|
5387 | QtfrNode* qn;
|
---|
5388 |
|
---|
5389 | qn = NQTFR(qnode);
|
---|
5390 | if (qn->lower == 1 && qn->upper == 1) {
|
---|
5391 | return 1;
|
---|
5392 | }
|
---|
5393 |
|
---|
5394 | switch (NTYPE(target)) {
|
---|
5395 | case NT_STR:
|
---|
5396 | if (! group) {
|
---|
5397 | StrNode* sn = NSTR(target);
|
---|
5398 | if (str_node_can_be_split(sn, env->enc)) {
|
---|
5399 | Node* n = str_node_split_last_char(sn, env->enc);
|
---|
5400 | if (IS_NOT_NULL(n)) {
|
---|
5401 | qn->target = n;
|
---|
5402 | return 2;
|
---|
5403 | }
|
---|
5404 | }
|
---|
5405 | }
|
---|
5406 | break;
|
---|
5407 |
|
---|
5408 | case NT_QTFR:
|
---|
5409 | { /* check redundant double repeat. */
|
---|
5410 | /* verbose warn (?:.?)? etc... but not warn (.?)? etc... */
|
---|
5411 | QtfrNode* qnt = NQTFR(target);
|
---|
5412 | int nestq_num = popular_quantifier_num(qn);
|
---|
5413 | int targetq_num = popular_quantifier_num(qnt);
|
---|
5414 |
|
---|
5415 | #ifdef USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
|
---|
5416 | if (nestq_num >= 0 && targetq_num >= 0 &&
|
---|
5417 | IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT)) {
|
---|
5418 | switch (ReduceTypeTable[targetq_num][nestq_num]) {
|
---|
5419 | case RQ_ASIS:
|
---|
5420 | break;
|
---|
5421 |
|
---|
5422 | case RQ_DEL:
|
---|
5423 | if (onig_warn != onig_null_warn) {
|
---|
5424 | onig_syntax_warn(env, "regular expression has redundant nested repeat operator '%s'",
|
---|
5425 | PopularQStr[targetq_num]);
|
---|
5426 | }
|
---|
5427 | goto warn_exit;
|
---|
5428 | break;
|
---|
5429 |
|
---|
5430 | default:
|
---|
5431 | if (onig_warn != onig_null_warn) {
|
---|
5432 | onig_syntax_warn(env, "nested repeat operator '%s' and '%s' was replaced with '%s' in regular expression",
|
---|
5433 | PopularQStr[targetq_num], PopularQStr[nestq_num],
|
---|
5434 | ReduceQStr[ReduceTypeTable[targetq_num][nestq_num]]);
|
---|
5435 | }
|
---|
5436 | goto warn_exit;
|
---|
5437 | break;
|
---|
5438 | }
|
---|
5439 | }
|
---|
5440 |
|
---|
5441 | warn_exit:
|
---|
5442 | #endif
|
---|
5443 | if (targetq_num >= 0) {
|
---|
5444 | if (nestq_num >= 0) {
|
---|
5445 | onig_reduce_nested_quantifier(qnode, target);
|
---|
5446 | goto q_exit;
|
---|
5447 | }
|
---|
5448 | else if (targetq_num == 1 || targetq_num == 2) { /* * or + */
|
---|
5449 | /* (?:a*){n,m}, (?:a+){n,m} => (?:a*){n,n}, (?:a+){n,n} */
|
---|
5450 | if (! IS_REPEAT_INFINITE(qn->upper) && qn->upper > 1 && qn->greedy) {
|
---|
5451 | qn->upper = (qn->lower == 0 ? 1 : qn->lower);
|
---|
5452 | }
|
---|
5453 | }
|
---|
5454 | }
|
---|
5455 | }
|
---|
5456 | break;
|
---|
5457 |
|
---|
5458 | default:
|
---|
5459 | break;
|
---|
5460 | }
|
---|
5461 |
|
---|
5462 | qn->target = target;
|
---|
5463 | q_exit:
|
---|
5464 | return 0;
|
---|
5465 | }
|
---|
5466 |
|
---|
5467 |
|
---|
5468 | #ifndef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
|
---|
5469 | static int
|
---|
5470 | clear_not_flag_cclass(CClassNode* cc, OnigEncoding enc)
|
---|
5471 | {
|
---|
5472 | BBuf *tbuf;
|
---|
5473 | int r;
|
---|
5474 |
|
---|
5475 | if (IS_NCCLASS_NOT(cc)) {
|
---|
5476 | bitset_invert(cc->bs);
|
---|
5477 |
|
---|
5478 | if (! ONIGENC_IS_SINGLEBYTE(enc)) {
|
---|
5479 | r = not_code_range_buf(enc, cc->mbuf, &tbuf);
|
---|
5480 | if (r != 0) return r;
|
---|
5481 |
|
---|
5482 | bbuf_free(cc->mbuf);
|
---|
5483 | cc->mbuf = tbuf;
|
---|
5484 | }
|
---|
5485 |
|
---|
5486 | NCCLASS_CLEAR_NOT(cc);
|
---|
5487 | }
|
---|
5488 |
|
---|
5489 | return 0;
|
---|
5490 | }
|
---|
5491 | #endif /* CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS */
|
---|
5492 |
|
---|
5493 | typedef struct {
|
---|
5494 | ScanEnv* env;
|
---|
5495 | CClassNode* cc;
|
---|
5496 | CClassNode* asc_cc;
|
---|
5497 | Node* alt_root;
|
---|
5498 | Node** ptail;
|
---|
5499 | } IApplyCaseFoldArg;
|
---|
5500 |
|
---|
5501 | static int
|
---|
5502 | i_apply_case_fold(OnigCodePoint from, OnigCodePoint to[],
|
---|
5503 | int to_len, void* arg)
|
---|
5504 | {
|
---|
5505 | IApplyCaseFoldArg* iarg;
|
---|
5506 | ScanEnv* env;
|
---|
5507 | CClassNode* cc;
|
---|
5508 | CClassNode* asc_cc;
|
---|
5509 | BitSetRef bs;
|
---|
5510 | int add_flag, r;
|
---|
5511 |
|
---|
5512 | iarg = (IApplyCaseFoldArg* )arg;
|
---|
5513 | env = iarg->env;
|
---|
5514 | cc = iarg->cc;
|
---|
5515 | asc_cc = iarg->asc_cc;
|
---|
5516 | bs = cc->bs;
|
---|
5517 |
|
---|
5518 | if (IS_NULL(asc_cc)) {
|
---|
5519 | add_flag = 0;
|
---|
5520 | }
|
---|
5521 | else if (ONIGENC_IS_ASCII_CODE(from) == ONIGENC_IS_ASCII_CODE(*to)) {
|
---|
5522 | add_flag = 1;
|
---|
5523 | }
|
---|
5524 | else {
|
---|
5525 | add_flag = onig_is_code_in_cc(env->enc, from, asc_cc);
|
---|
5526 | if (IS_NCCLASS_NOT(asc_cc))
|
---|
5527 | add_flag = !add_flag;
|
---|
5528 | }
|
---|
5529 |
|
---|
5530 | if (to_len == 1) {
|
---|
5531 | int is_in = onig_is_code_in_cc(env->enc, from, cc);
|
---|
5532 | #ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
|
---|
5533 | if ((is_in != 0 && !IS_NCCLASS_NOT(cc)) ||
|
---|
5534 | (is_in == 0 && IS_NCCLASS_NOT(cc))) {
|
---|
5535 | if (add_flag) {
|
---|
5536 | if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {
|
---|
5537 | r = add_code_range0(&(cc->mbuf), env, *to, *to, 0);
|
---|
5538 | if (r < 0) return r;
|
---|
5539 | }
|
---|
5540 | else {
|
---|
5541 | BITSET_SET_BIT(bs, *to);
|
---|
5542 | }
|
---|
5543 | }
|
---|
5544 | }
|
---|
5545 | #else
|
---|
5546 | if (is_in != 0) {
|
---|
5547 | if (add_flag) {
|
---|
5548 | if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {
|
---|
5549 | if (IS_NCCLASS_NOT(cc)) clear_not_flag_cclass(cc, env->enc);
|
---|
5550 | r = add_code_range0(&(cc->mbuf), env, *to, *to, 0);
|
---|
5551 | if (r < 0) return r;
|
---|
5552 | }
|
---|
5553 | else {
|
---|
5554 | if (IS_NCCLASS_NOT(cc)) {
|
---|
5555 | BITSET_CLEAR_BIT(bs, *to);
|
---|
5556 | }
|
---|
5557 | else {
|
---|
5558 | BITSET_SET_BIT(bs, *to);
|
---|
5559 | }
|
---|
5560 | }
|
---|
5561 | }
|
---|
5562 | }
|
---|
5563 | #endif /* CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS */
|
---|
5564 | }
|
---|
5565 | else {
|
---|
5566 | int r, i, len;
|
---|
5567 | UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
|
---|
5568 | Node *snode = NULL_NODE;
|
---|
5569 |
|
---|
5570 | if (onig_is_code_in_cc(env->enc, from, cc)
|
---|
5571 | #ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
|
---|
5572 | && !IS_NCCLASS_NOT(cc)
|
---|
5573 | #endif
|
---|
5574 | ) {
|
---|
5575 | for (i = 0; i < to_len; i++) {
|
---|
5576 | len = ONIGENC_CODE_TO_MBC(env->enc, to[i], buf);
|
---|
5577 | if (i == 0) {
|
---|
5578 | snode = onig_node_new_str(buf, buf + len);
|
---|
5579 | CHECK_NULL_RETURN_MEMERR(snode);
|
---|
5580 |
|
---|
5581 | /* char-class expanded multi-char only
|
---|
5582 | compare with string folded at match time. */
|
---|
5583 | NSTRING_SET_AMBIG(snode);
|
---|
5584 | }
|
---|
5585 | else {
|
---|
5586 | r = onig_node_str_cat(snode, buf, buf + len);
|
---|
5587 | if (r < 0) {
|
---|
5588 | onig_node_free(snode);
|
---|
5589 | return r;
|
---|
5590 | }
|
---|
5591 | }
|
---|
5592 | }
|
---|
5593 |
|
---|
5594 | *(iarg->ptail) = onig_node_new_alt(snode, NULL_NODE);
|
---|
5595 | CHECK_NULL_RETURN_MEMERR(*(iarg->ptail));
|
---|
5596 | iarg->ptail = &(NCDR((*(iarg->ptail))));
|
---|
5597 | }
|
---|
5598 | }
|
---|
5599 |
|
---|
5600 | return 0;
|
---|
5601 | }
|
---|
5602 |
|
---|
5603 | static int
|
---|
5604 | cclass_case_fold(Node** np, CClassNode* cc, CClassNode* asc_cc, ScanEnv* env)
|
---|
5605 | {
|
---|
5606 | int r;
|
---|
5607 | IApplyCaseFoldArg iarg;
|
---|
5608 |
|
---|
5609 | iarg.env = env;
|
---|
5610 | iarg.cc = cc;
|
---|
5611 | iarg.asc_cc = asc_cc;
|
---|
5612 | iarg.alt_root = NULL_NODE;
|
---|
5613 | iarg.ptail = &(iarg.alt_root);
|
---|
5614 |
|
---|
5615 | r = ONIGENC_APPLY_ALL_CASE_FOLD(env->enc, env->case_fold_flag,
|
---|
5616 | i_apply_case_fold, &iarg);
|
---|
5617 | if (r != 0) {
|
---|
5618 | onig_node_free(iarg.alt_root);
|
---|
5619 | return r;
|
---|
5620 | }
|
---|
5621 | if (IS_NOT_NULL(iarg.alt_root)) {
|
---|
5622 | Node* work = onig_node_new_alt(*np, iarg.alt_root);
|
---|
5623 | if (IS_NULL(work)) {
|
---|
5624 | onig_node_free(iarg.alt_root);
|
---|
5625 | return ONIGERR_MEMORY;
|
---|
5626 | }
|
---|
5627 | *np = work;
|
---|
5628 | }
|
---|
5629 | return r;
|
---|
5630 | }
|
---|
5631 |
|
---|
5632 | static int
|
---|
5633 | node_linebreak(Node** np, ScanEnv* env)
|
---|
5634 | {
|
---|
5635 | /* same as (?>\x0D\x0A|[\x0A-\x0D\x{85}\x{2028}\x{2029}]) */
|
---|
5636 | Node* left = NULL;
|
---|
5637 | Node* right = NULL;
|
---|
5638 | Node* target1 = NULL;
|
---|
5639 | Node* target2 = NULL;
|
---|
5640 | CClassNode* cc;
|
---|
5641 | int num1, num2, r;
|
---|
5642 | UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN * 2];
|
---|
5643 |
|
---|
5644 | /* \x0D\x0A */
|
---|
5645 | num1 = ONIGENC_CODE_TO_MBC(env->enc, 0x0D, buf);
|
---|
5646 | if (num1 < 0) return num1;
|
---|
5647 | num2 = ONIGENC_CODE_TO_MBC(env->enc, 0x0A, buf + num1);
|
---|
5648 | if (num2 < 0) return num2;
|
---|
5649 | left = node_new_str_raw(buf, buf + num1 + num2);
|
---|
5650 | if (IS_NULL(left)) goto err;
|
---|
5651 |
|
---|
5652 | /* [\x0A-\x0D] or [\x0A-\x0D\x{85}\x{2028}\x{2029}] */
|
---|
5653 | right = node_new_cclass();
|
---|
5654 | if (IS_NULL(right)) goto err;
|
---|
5655 | cc = NCCLASS(right);
|
---|
5656 | if (ONIGENC_MBC_MINLEN(env->enc) > 1) {
|
---|
5657 | r = add_code_range(&(cc->mbuf), env, 0x0A, 0x0D);
|
---|
5658 | if (r != 0) goto err;
|
---|
5659 | }
|
---|
5660 | else {
|
---|
5661 | bitset_set_range(env, cc->bs, 0x0A, 0x0D);
|
---|
5662 | }
|
---|
5663 |
|
---|
5664 | /* TODO: move this block to enc/unicode.c */
|
---|
5665 | if (ONIGENC_IS_UNICODE(env->enc)) {
|
---|
5666 | /* UTF-8, UTF-16BE/LE, UTF-32BE/LE */
|
---|
5667 | r = add_code_range(&(cc->mbuf), env, 0x85, 0x85);
|
---|
5668 | if (r != 0) goto err;
|
---|
5669 | r = add_code_range(&(cc->mbuf), env, 0x2028, 0x2029);
|
---|
5670 | if (r != 0) goto err;
|
---|
5671 | }
|
---|
5672 |
|
---|
5673 | /* ...|... */
|
---|
5674 | target1 = onig_node_new_alt(right, NULL_NODE);
|
---|
5675 | if (IS_NULL(target1)) goto err;
|
---|
5676 | right = NULL;
|
---|
5677 | target2 = onig_node_new_alt(left, target1);
|
---|
5678 | if (IS_NULL(target2)) goto err;
|
---|
5679 | left = NULL;
|
---|
5680 | target1 = NULL;
|
---|
5681 |
|
---|
5682 | /* (?>...) */
|
---|
5683 | *np = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
|
---|
5684 | if (IS_NULL(*np)) goto err;
|
---|
5685 | NENCLOSE(*np)->target = target2;
|
---|
5686 | return ONIG_NORMAL;
|
---|
5687 |
|
---|
5688 | err:
|
---|
5689 | onig_node_free(left);
|
---|
5690 | onig_node_free(right);
|
---|
5691 | onig_node_free(target1);
|
---|
5692 | onig_node_free(target2);
|
---|
5693 | return ONIGERR_MEMORY;
|
---|
5694 | }
|
---|
5695 |
|
---|
5696 | static int
|
---|
5697 | propname2ctype(ScanEnv* env, const char* propname)
|
---|
5698 | {
|
---|
5699 | UChar* name = (UChar* )propname;
|
---|
5700 | int ctype = env->enc->property_name_to_ctype(ONIG_ENCODING_ASCII,
|
---|
5701 | name, name + strlen(propname));
|
---|
5702 | return ctype;
|
---|
5703 | }
|
---|
5704 |
|
---|
5705 | static int
|
---|
5706 | node_extended_grapheme_cluster(Node** np, ScanEnv* env)
|
---|
5707 | {
|
---|
5708 | Node* tmp = NULL;
|
---|
5709 | Node* np1 = NULL;
|
---|
5710 | Node* list = NULL;
|
---|
5711 | Node* list2 = NULL;
|
---|
5712 | Node* alt = NULL;
|
---|
5713 | Node* alt2 = NULL;
|
---|
5714 | BBuf *pbuf1 = NULL;
|
---|
5715 | int r = 0;
|
---|
5716 | int num1;
|
---|
5717 | UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN * 2];
|
---|
5718 | OnigOptionType option;
|
---|
5719 |
|
---|
5720 | #ifdef USE_UNICODE_PROPERTIES
|
---|
5721 | if (ONIGENC_IS_UNICODE(env->enc)) {
|
---|
5722 | /* UTF-8, UTF-16BE/LE, UTF-32BE/LE */
|
---|
5723 | CClassNode* cc;
|
---|
5724 | OnigCodePoint sb_out = (ONIGENC_MBC_MINLEN(env->enc) > 1) ? 0x00 : 0x80;
|
---|
5725 | int extend = propname2ctype(env, "Grapheme_Cluster_Break=Extend");
|
---|
5726 |
|
---|
5727 | /* Prepend*
|
---|
5728 | * ( RI-sequence | Hangul-Syllable | !Control )
|
---|
5729 | * ( Grapheme_Extend | SpacingMark )* */
|
---|
5730 |
|
---|
5731 | /* ( Grapheme_Extend | SpacingMark )* */
|
---|
5732 | np1 = node_new_cclass();
|
---|
5733 | if (IS_NULL(np1)) goto err;
|
---|
5734 | cc = NCCLASS(np1);
|
---|
5735 | r = add_ctype_to_cc(cc, extend, 0, 0, env);
|
---|
5736 | if (r != 0) goto err;
|
---|
5737 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=SpacingMark"), 0, 0, env);
|
---|
5738 | if (r != 0) goto err;
|
---|
5739 | r = add_code_range(&(cc->mbuf), env, 0x200D, 0x200D);
|
---|
5740 | if (r != 0) goto err;
|
---|
5741 |
|
---|
5742 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
5743 | if (IS_NULL(tmp)) goto err;
|
---|
5744 | NQTFR(tmp)->target = np1;
|
---|
5745 | np1 = tmp;
|
---|
5746 |
|
---|
5747 | tmp = node_new_list(np1, NULL_NODE);
|
---|
5748 | if (IS_NULL(tmp)) goto err;
|
---|
5749 | list = tmp;
|
---|
5750 | np1 = NULL;
|
---|
5751 |
|
---|
5752 | /* ( RI-sequence | Hangul-Syllable | !Control ) */
|
---|
5753 | /* !Control */
|
---|
5754 | np1 = node_new_cclass();
|
---|
5755 | if (IS_NULL(np1)) goto err;
|
---|
5756 | cc = NCCLASS(np1);
|
---|
5757 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=Control"), 1, 0, env);
|
---|
5758 | if (r != 0) goto err;
|
---|
5759 | if (ONIGENC_MBC_MINLEN(env->enc) > 1) {
|
---|
5760 | BBuf *pbuf2 = NULL;
|
---|
5761 | r = add_code_range(&pbuf1, env, 0x0a, 0x0a);
|
---|
5762 | if (r != 0) goto err;
|
---|
5763 | r = add_code_range(&pbuf1, env, 0x0d, 0x0d);
|
---|
5764 | if (r != 0) goto err;
|
---|
5765 | r = and_code_range_buf(cc->mbuf, 0, pbuf1, 1, &pbuf2, env);
|
---|
5766 | if (r != 0) {
|
---|
5767 | bbuf_free(pbuf2);
|
---|
5768 | goto err;
|
---|
5769 | }
|
---|
5770 | bbuf_free(pbuf1);
|
---|
5771 | pbuf1 = NULL;
|
---|
5772 | bbuf_free(cc->mbuf);
|
---|
5773 | cc->mbuf = pbuf2;
|
---|
5774 | }
|
---|
5775 | else {
|
---|
5776 | BITSET_CLEAR_BIT(cc->bs, 0x0a);
|
---|
5777 | BITSET_CLEAR_BIT(cc->bs, 0x0d);
|
---|
5778 | }
|
---|
5779 |
|
---|
5780 | tmp = onig_node_new_alt(np1, NULL_NODE);
|
---|
5781 | if (IS_NULL(tmp)) goto err;
|
---|
5782 | alt = tmp;
|
---|
5783 | np1 = NULL;
|
---|
5784 |
|
---|
5785 | /* Hangul-Syllable
|
---|
5786 | * := L* V+ T*
|
---|
5787 | * | L* LV V* T*
|
---|
5788 | * | L* LVT T*
|
---|
5789 | * | L+
|
---|
5790 | * | T+ */
|
---|
5791 |
|
---|
5792 | /* T+ */
|
---|
5793 | np1 = node_new_cclass();
|
---|
5794 | if (IS_NULL(np1)) goto err;
|
---|
5795 | cc = NCCLASS(np1);
|
---|
5796 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=T"), 0, 0, env);
|
---|
5797 | if (r != 0) goto err;
|
---|
5798 |
|
---|
5799 | tmp = node_new_quantifier(1, REPEAT_INFINITE, 0);
|
---|
5800 | if (IS_NULL(tmp)) goto err;
|
---|
5801 | NQTFR(tmp)->target = np1;
|
---|
5802 | np1 = tmp;
|
---|
5803 |
|
---|
5804 | tmp = onig_node_new_alt(np1, alt);
|
---|
5805 | if (IS_NULL(tmp)) goto err;
|
---|
5806 | alt = tmp;
|
---|
5807 | np1 = NULL;
|
---|
5808 |
|
---|
5809 | /* L+ */
|
---|
5810 | np1 = node_new_cclass();
|
---|
5811 | if (IS_NULL(np1)) goto err;
|
---|
5812 | cc = NCCLASS(np1);
|
---|
5813 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=L"), 0, 0, env);
|
---|
5814 | if (r != 0) goto err;
|
---|
5815 |
|
---|
5816 | tmp = node_new_quantifier(1, REPEAT_INFINITE, 0);
|
---|
5817 | if (IS_NULL(tmp)) goto err;
|
---|
5818 | NQTFR(tmp)->target = np1;
|
---|
5819 | np1 = tmp;
|
---|
5820 |
|
---|
5821 | tmp = onig_node_new_alt(np1, alt);
|
---|
5822 | if (IS_NULL(tmp)) goto err;
|
---|
5823 | alt = tmp;
|
---|
5824 | np1 = NULL;
|
---|
5825 |
|
---|
5826 | /* L* LVT T* */
|
---|
5827 | np1 = node_new_cclass();
|
---|
5828 | if (IS_NULL(np1)) goto err;
|
---|
5829 | cc = NCCLASS(np1);
|
---|
5830 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=T"), 0, 0, env);
|
---|
5831 | if (r != 0) goto err;
|
---|
5832 |
|
---|
5833 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
5834 | if (IS_NULL(tmp)) goto err;
|
---|
5835 | NQTFR(tmp)->target = np1;
|
---|
5836 | np1 = tmp;
|
---|
5837 |
|
---|
5838 | tmp = node_new_list(np1, NULL_NODE);
|
---|
5839 | if (IS_NULL(tmp)) goto err;
|
---|
5840 | list2 = tmp;
|
---|
5841 | np1 = NULL;
|
---|
5842 |
|
---|
5843 | np1 = node_new_cclass();
|
---|
5844 | if (IS_NULL(np1)) goto err;
|
---|
5845 | cc = NCCLASS(np1);
|
---|
5846 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=LVT"), 0, 0, env);
|
---|
5847 | if (r != 0) goto err;
|
---|
5848 |
|
---|
5849 | tmp = node_new_list(np1, list2);
|
---|
5850 | if (IS_NULL(tmp)) goto err;
|
---|
5851 | list2 = tmp;
|
---|
5852 | np1 = NULL;
|
---|
5853 |
|
---|
5854 | np1 = node_new_cclass();
|
---|
5855 | if (IS_NULL(np1)) goto err;
|
---|
5856 | cc = NCCLASS(np1);
|
---|
5857 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=L"), 0, 0, env);
|
---|
5858 | if (r != 0) goto err;
|
---|
5859 |
|
---|
5860 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
5861 | if (IS_NULL(tmp)) goto err;
|
---|
5862 | NQTFR(tmp)->target = np1;
|
---|
5863 | np1 = tmp;
|
---|
5864 |
|
---|
5865 | tmp = node_new_list(np1, list2);
|
---|
5866 | if (IS_NULL(tmp)) goto err;
|
---|
5867 | list2 = tmp;
|
---|
5868 | np1 = NULL;
|
---|
5869 |
|
---|
5870 | tmp = onig_node_new_alt(list2, alt);
|
---|
5871 | if (IS_NULL(tmp)) goto err;
|
---|
5872 | alt = tmp;
|
---|
5873 | list2 = NULL;
|
---|
5874 |
|
---|
5875 | /* L* LV V* T* */
|
---|
5876 | np1 = node_new_cclass();
|
---|
5877 | if (IS_NULL(np1)) goto err;
|
---|
5878 | cc = NCCLASS(np1);
|
---|
5879 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=T"), 0, 0, env);
|
---|
5880 | if (r != 0) goto err;
|
---|
5881 |
|
---|
5882 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
5883 | if (IS_NULL(tmp)) goto err;
|
---|
5884 | NQTFR(tmp)->target = np1;
|
---|
5885 | np1 = tmp;
|
---|
5886 |
|
---|
5887 | tmp = node_new_list(np1, NULL_NODE);
|
---|
5888 | if (IS_NULL(tmp)) goto err;
|
---|
5889 | list2 = tmp;
|
---|
5890 | np1 = NULL;
|
---|
5891 |
|
---|
5892 | np1 = node_new_cclass();
|
---|
5893 | if (IS_NULL(np1)) goto err;
|
---|
5894 | cc = NCCLASS(np1);
|
---|
5895 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=V"), 0, 0, env);
|
---|
5896 | if (r != 0) goto err;
|
---|
5897 |
|
---|
5898 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
5899 | if (IS_NULL(tmp)) goto err;
|
---|
5900 | NQTFR(tmp)->target = np1;
|
---|
5901 | np1 = tmp;
|
---|
5902 |
|
---|
5903 | tmp = node_new_list(np1, list2);
|
---|
5904 | if (IS_NULL(tmp)) goto err;
|
---|
5905 | list2 = tmp;
|
---|
5906 | np1 = NULL;
|
---|
5907 |
|
---|
5908 | np1 = node_new_cclass();
|
---|
5909 | if (IS_NULL(np1)) goto err;
|
---|
5910 | cc = NCCLASS(np1);
|
---|
5911 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=LV"), 0, 0, env);
|
---|
5912 | if (r != 0) goto err;
|
---|
5913 |
|
---|
5914 | tmp = node_new_list(np1, list2);
|
---|
5915 | if (IS_NULL(tmp)) goto err;
|
---|
5916 | list2 = tmp;
|
---|
5917 | np1 = NULL;
|
---|
5918 |
|
---|
5919 | np1 = node_new_cclass();
|
---|
5920 | if (IS_NULL(np1)) goto err;
|
---|
5921 | cc = NCCLASS(np1);
|
---|
5922 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=L"), 0, 0, env);
|
---|
5923 | if (r != 0) goto err;
|
---|
5924 |
|
---|
5925 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
5926 | if (IS_NULL(tmp)) goto err;
|
---|
5927 | NQTFR(tmp)->target = np1;
|
---|
5928 | np1 = tmp;
|
---|
5929 |
|
---|
5930 | tmp = node_new_list(np1, list2);
|
---|
5931 | if (IS_NULL(tmp)) goto err;
|
---|
5932 | list2 = tmp;
|
---|
5933 | np1 = NULL;
|
---|
5934 |
|
---|
5935 | tmp = onig_node_new_alt(list2, alt);
|
---|
5936 | if (IS_NULL(tmp)) goto err;
|
---|
5937 | alt = tmp;
|
---|
5938 | list2 = NULL;
|
---|
5939 |
|
---|
5940 | /* L* V+ T* */
|
---|
5941 | np1 = node_new_cclass();
|
---|
5942 | if (IS_NULL(np1)) goto err;
|
---|
5943 | cc = NCCLASS(np1);
|
---|
5944 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=T"), 0, 0, env);
|
---|
5945 | if (r != 0) goto err;
|
---|
5946 |
|
---|
5947 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
5948 | if (IS_NULL(tmp)) goto err;
|
---|
5949 | NQTFR(tmp)->target = np1;
|
---|
5950 | np1 = tmp;
|
---|
5951 |
|
---|
5952 | tmp = node_new_list(np1, NULL_NODE);
|
---|
5953 | if (IS_NULL(tmp)) goto err;
|
---|
5954 | list2 = tmp;
|
---|
5955 | np1 = NULL;
|
---|
5956 |
|
---|
5957 | np1 = node_new_cclass();
|
---|
5958 | if (IS_NULL(np1)) goto err;
|
---|
5959 | cc = NCCLASS(np1);
|
---|
5960 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=V"), 0, 0, env);
|
---|
5961 | if (r != 0) goto err;
|
---|
5962 |
|
---|
5963 | tmp = node_new_quantifier(1, REPEAT_INFINITE, 0);
|
---|
5964 | if (IS_NULL(tmp)) goto err;
|
---|
5965 | NQTFR(tmp)->target = np1;
|
---|
5966 | np1 = tmp;
|
---|
5967 |
|
---|
5968 | tmp = node_new_list(np1, list2);
|
---|
5969 | if (IS_NULL(tmp)) goto err;
|
---|
5970 | list2 = tmp;
|
---|
5971 | np1 = NULL;
|
---|
5972 |
|
---|
5973 | np1 = node_new_cclass();
|
---|
5974 | if (IS_NULL(np1)) goto err;
|
---|
5975 | cc = NCCLASS(np1);
|
---|
5976 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=L"), 0, 0, env);
|
---|
5977 | if (r != 0) goto err;
|
---|
5978 |
|
---|
5979 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
5980 | if (IS_NULL(tmp)) goto err;
|
---|
5981 | NQTFR(tmp)->target = np1;
|
---|
5982 | np1 = tmp;
|
---|
5983 |
|
---|
5984 | tmp = node_new_list(np1, list2);
|
---|
5985 | if (IS_NULL(tmp)) goto err;
|
---|
5986 | list2 = tmp;
|
---|
5987 | np1 = NULL;
|
---|
5988 |
|
---|
5989 | tmp = onig_node_new_alt(list2, alt);
|
---|
5990 | if (IS_NULL(tmp)) goto err;
|
---|
5991 | alt = tmp;
|
---|
5992 | list2 = NULL;
|
---|
5993 |
|
---|
5994 | /* Emoji sequence := (E_Base | EBG) Extend* E_Modifier?
|
---|
5995 | * (ZWJ (Glue_After_Zwj | EBG Extend* E_Modifier?) )* */
|
---|
5996 |
|
---|
5997 | /* ZWJ (Glue_After_Zwj | E_Base_GAZ Extend* E_Modifier?) */
|
---|
5998 | np1 = node_new_cclass();
|
---|
5999 | if (IS_NULL(np1)) goto err;
|
---|
6000 | cc = NCCLASS(np1);
|
---|
6001 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=E_Modifier"), 0, 0, env);
|
---|
6002 | if (r != 0) goto err;
|
---|
6003 |
|
---|
6004 | tmp = node_new_quantifier(0, 1, 0);
|
---|
6005 | if (IS_NULL(tmp)) goto err;
|
---|
6006 | NQTFR(tmp)->target = np1;
|
---|
6007 | np1 = tmp;
|
---|
6008 |
|
---|
6009 | tmp = node_new_list(np1, NULL_NODE);
|
---|
6010 | if (IS_NULL(tmp)) goto err;
|
---|
6011 | list2 = tmp;
|
---|
6012 | np1 = NULL;
|
---|
6013 |
|
---|
6014 | np1 = node_new_cclass();
|
---|
6015 | if (IS_NULL(np1)) goto err;
|
---|
6016 | cc = NCCLASS(np1);
|
---|
6017 | r = add_ctype_to_cc(cc, extend, 0, 0, env);
|
---|
6018 | if (r != 0) goto err;
|
---|
6019 |
|
---|
6020 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
6021 | if (IS_NULL(tmp)) goto err;
|
---|
6022 | NQTFR(tmp)->target = np1;
|
---|
6023 | np1 = tmp;
|
---|
6024 |
|
---|
6025 | tmp = node_new_list(np1, list2);
|
---|
6026 | if (IS_NULL(tmp)) goto err;
|
---|
6027 | list2 = tmp;
|
---|
6028 | np1 = NULL;
|
---|
6029 |
|
---|
6030 | np1 = node_new_cclass();
|
---|
6031 | if (IS_NULL(np1)) goto err;
|
---|
6032 | cc = NCCLASS(np1);
|
---|
6033 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=E_Base_GAZ"), 0, 0, env);
|
---|
6034 | if (r != 0) goto err;
|
---|
6035 |
|
---|
6036 | tmp = node_new_list(np1, list2);
|
---|
6037 | if (IS_NULL(tmp)) goto err;
|
---|
6038 | list2 = tmp;
|
---|
6039 | np1 = NULL;
|
---|
6040 |
|
---|
6041 | tmp = onig_node_new_alt(list2, NULL_NODE);
|
---|
6042 | if (IS_NULL(tmp)) goto err;
|
---|
6043 | alt2 = tmp;
|
---|
6044 | list2 = NULL;
|
---|
6045 |
|
---|
6046 | /* Glue_After_Zwj */
|
---|
6047 | np1 = node_new_cclass();
|
---|
6048 | if (IS_NULL(np1)) goto err;
|
---|
6049 | cc = NCCLASS(np1);
|
---|
6050 | r = add_ctype_to_cc(cc, extend, 0, 0, env);
|
---|
6051 | if (r != 0) goto err;
|
---|
6052 |
|
---|
6053 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
6054 | if (IS_NULL(tmp)) goto err;
|
---|
6055 | NQTFR(tmp)->target = np1;
|
---|
6056 | np1 = tmp;
|
---|
6057 |
|
---|
6058 | tmp = node_new_list(np1, NULL_NODE);
|
---|
6059 | if (IS_NULL(tmp)) goto err;
|
---|
6060 | list2 = tmp;
|
---|
6061 | np1 = NULL;
|
---|
6062 |
|
---|
6063 | np1 = node_new_cclass();
|
---|
6064 | if (IS_NULL(np1)) goto err;
|
---|
6065 | cc = NCCLASS(np1);
|
---|
6066 | {
|
---|
6067 | static const OnigCodePoint ranges[] = {
|
---|
6068 | 13,
|
---|
6069 | 0x1F308, 0x1F308,
|
---|
6070 | 0x1F33E, 0x1F33E,
|
---|
6071 | 0x1F373, 0x1F373,
|
---|
6072 | 0x1F393, 0x1F393,
|
---|
6073 | 0x1F3A4, 0x1F3A4,
|
---|
6074 | 0x1F3A8, 0x1F3A8,
|
---|
6075 | 0x1F3EB, 0x1F3EB,
|
---|
6076 | 0x1F3ED, 0x1F3ED,
|
---|
6077 | 0x1F4BB, 0x1F4BC,
|
---|
6078 | 0x1F527, 0x1F527,
|
---|
6079 | 0x1F52C, 0x1F52C,
|
---|
6080 | 0x1F680, 0x1F680,
|
---|
6081 | 0x1F692, 0x1F692,
|
---|
6082 | };
|
---|
6083 | r = add_ctype_to_cc_by_range(cc, -1, 0, env, sb_out, ranges);
|
---|
6084 | if (r != 0) goto err;
|
---|
6085 | }
|
---|
6086 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=Glue_After_Zwj"), 0, 0, env);
|
---|
6087 | if (r != 0) goto err;
|
---|
6088 |
|
---|
6089 | tmp = node_new_list(np1, list2);
|
---|
6090 | if (IS_NULL(tmp)) goto err;
|
---|
6091 | list2 = tmp;
|
---|
6092 | np1 = NULL;
|
---|
6093 |
|
---|
6094 | tmp = onig_node_new_alt(list2, alt2);
|
---|
6095 | if (IS_NULL(tmp)) goto err;
|
---|
6096 | alt2 = tmp;
|
---|
6097 | list2 = NULL;
|
---|
6098 |
|
---|
6099 | /* Emoji variation sequence
|
---|
6100 | * http://unicode.org/Public/emoji/4.0/emoji-zwj-sequences.txt
|
---|
6101 | */
|
---|
6102 | r = ONIGENC_CODE_TO_MBC(env->enc, 0xfe0f, buf);
|
---|
6103 | if (r < 0) goto err;
|
---|
6104 | np1 = node_new_str_raw(buf, buf + r);
|
---|
6105 | if (IS_NULL(np1)) goto err;
|
---|
6106 |
|
---|
6107 | tmp = node_new_quantifier(0, 1, 0);
|
---|
6108 | if (IS_NULL(tmp)) goto err;
|
---|
6109 | NQTFR(tmp)->target = np1;
|
---|
6110 | np1 = tmp;
|
---|
6111 |
|
---|
6112 | tmp = node_new_list(np1, NULL_NODE);
|
---|
6113 | if (IS_NULL(tmp)) goto err;
|
---|
6114 | list2 = tmp;
|
---|
6115 | np1 = NULL;
|
---|
6116 |
|
---|
6117 | np1 = node_new_cclass();
|
---|
6118 | if (IS_NULL(np1)) goto err;
|
---|
6119 | cc = NCCLASS(np1);
|
---|
6120 | {
|
---|
6121 | static const OnigCodePoint ranges[] = {
|
---|
6122 | 4,
|
---|
6123 | 0x2640, 0x2640,
|
---|
6124 | 0x2642, 0x2642,
|
---|
6125 | 0x2695, 0x2696,
|
---|
6126 | 0x2708, 0x2708,
|
---|
6127 | };
|
---|
6128 | r = add_ctype_to_cc_by_range(cc, -1, 0, env, sb_out, ranges);
|
---|
6129 | if (r != 0) goto err;
|
---|
6130 | }
|
---|
6131 |
|
---|
6132 | tmp = node_new_list(np1, list2);
|
---|
6133 | if (IS_NULL(tmp)) goto err;
|
---|
6134 | list2 = tmp;
|
---|
6135 | np1 = NULL;
|
---|
6136 |
|
---|
6137 | tmp = onig_node_new_alt(list2, alt2);
|
---|
6138 | if (IS_NULL(tmp)) goto err;
|
---|
6139 | alt2 = tmp;
|
---|
6140 | list2 = NULL;
|
---|
6141 |
|
---|
6142 | tmp = node_new_list(alt2, NULL_NODE);
|
---|
6143 | if (IS_NULL(tmp)) goto err;
|
---|
6144 | list2 = tmp;
|
---|
6145 | alt2 = NULL;
|
---|
6146 |
|
---|
6147 | /* ZWJ */
|
---|
6148 | r = ONIGENC_CODE_TO_MBC(env->enc, 0x200D, buf);
|
---|
6149 | if (r < 0) goto err;
|
---|
6150 | np1 = node_new_str_raw(buf, buf + r);
|
---|
6151 | if (IS_NULL(np1)) goto err;
|
---|
6152 |
|
---|
6153 | tmp = node_new_list(np1, list2);
|
---|
6154 | if (IS_NULL(tmp)) goto err;
|
---|
6155 | list2 = tmp;
|
---|
6156 | np1 = NULL;
|
---|
6157 |
|
---|
6158 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
6159 | if (IS_NULL(tmp)) goto err;
|
---|
6160 | NQTFR(tmp)->target = list2;
|
---|
6161 | np1 = tmp;
|
---|
6162 | list2 = NULL;
|
---|
6163 |
|
---|
6164 | tmp = node_new_list(np1, NULL_NODE);
|
---|
6165 | if (IS_NULL(tmp)) goto err;
|
---|
6166 | list2 = tmp;
|
---|
6167 | np1 = NULL;
|
---|
6168 |
|
---|
6169 | /* E_Modifier? */
|
---|
6170 | np1 = node_new_cclass();
|
---|
6171 | if (IS_NULL(np1)) goto err;
|
---|
6172 | cc = NCCLASS(np1);
|
---|
6173 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=E_Modifier"), 0, 0, env);
|
---|
6174 | if (r != 0) goto err;
|
---|
6175 |
|
---|
6176 | tmp = node_new_quantifier(0, 1, 0);
|
---|
6177 | if (IS_NULL(tmp)) goto err;
|
---|
6178 | NQTFR(tmp)->target = np1;
|
---|
6179 | np1 = tmp;
|
---|
6180 |
|
---|
6181 | tmp = node_new_list(np1, list2);
|
---|
6182 | if (IS_NULL(tmp)) goto err;
|
---|
6183 | list2 = tmp;
|
---|
6184 | np1 = NULL;
|
---|
6185 |
|
---|
6186 | /* Extend* */
|
---|
6187 | np1 = node_new_cclass();
|
---|
6188 | if (IS_NULL(np1)) goto err;
|
---|
6189 | cc = NCCLASS(np1);
|
---|
6190 | r = add_ctype_to_cc(cc, extend, 0, 0, env);
|
---|
6191 | if (r != 0) goto err;
|
---|
6192 |
|
---|
6193 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
6194 | if (IS_NULL(tmp)) goto err;
|
---|
6195 | NQTFR(tmp)->target = np1;
|
---|
6196 | np1 = tmp;
|
---|
6197 |
|
---|
6198 | tmp = node_new_list(np1, list2);
|
---|
6199 | if (IS_NULL(tmp)) goto err;
|
---|
6200 | list2 = tmp;
|
---|
6201 | np1 = NULL;
|
---|
6202 |
|
---|
6203 | /* (E_Base | EBG) */
|
---|
6204 | np1 = node_new_cclass();
|
---|
6205 | if (IS_NULL(np1)) goto err;
|
---|
6206 | cc = NCCLASS(np1);
|
---|
6207 | {
|
---|
6208 | static const OnigCodePoint ranges[] = {
|
---|
6209 | 8,
|
---|
6210 | 0x1F3C2, 0x1F3C2,
|
---|
6211 | 0x1F3C7, 0x1F3C7,
|
---|
6212 | 0x1F3CC, 0x1F3CC,
|
---|
6213 | 0x1F3F3, 0x1F3F3,
|
---|
6214 | 0x1F441, 0x1F441,
|
---|
6215 | 0x1F46F, 0x1F46F,
|
---|
6216 | 0x1F574, 0x1F574,
|
---|
6217 | 0x1F6CC, 0x1F6CC,
|
---|
6218 | };
|
---|
6219 | r = add_ctype_to_cc_by_range(cc, -1, 0, env, sb_out, ranges);
|
---|
6220 | if (r != 0) goto err;
|
---|
6221 | }
|
---|
6222 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=E_Base"), 0, 0, env);
|
---|
6223 | if (r != 0) goto err;
|
---|
6224 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=E_Base_GAZ"), 0, 0, env);
|
---|
6225 | if (r != 0) goto err;
|
---|
6226 |
|
---|
6227 | tmp = node_new_list(np1, list2);
|
---|
6228 | if (IS_NULL(tmp)) goto err;
|
---|
6229 | list2 = tmp;
|
---|
6230 | np1 = NULL;
|
---|
6231 |
|
---|
6232 | tmp = onig_node_new_alt(list2, alt);
|
---|
6233 | if (IS_NULL(tmp)) goto err;
|
---|
6234 | alt = tmp;
|
---|
6235 | list2 = NULL;
|
---|
6236 |
|
---|
6237 | /* ZWJ (E_Base_GAZ | Glue_After_Zwj) E_Modifier? */
|
---|
6238 | /* a sequence starting with ZWJ seems artificial, but GraphemeBreakTest
|
---|
6239 | * has such examples.
|
---|
6240 | * http://www.unicode.org/Public/9.0.0/ucd/auxiliary/GraphemeBreakTest.html
|
---|
6241 | */
|
---|
6242 | np1 = node_new_cclass();
|
---|
6243 | if (IS_NULL(np1)) goto err;
|
---|
6244 | cc = NCCLASS(np1);
|
---|
6245 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=E_Modifier"), 0, 0, env);
|
---|
6246 | if (r != 0) goto err;
|
---|
6247 |
|
---|
6248 | tmp = node_new_quantifier(0, 1, 0);
|
---|
6249 | if (IS_NULL(tmp)) goto err;
|
---|
6250 | NQTFR(tmp)->target = np1;
|
---|
6251 | np1 = tmp;
|
---|
6252 |
|
---|
6253 | tmp = node_new_list(np1, NULL_NODE);
|
---|
6254 | if (IS_NULL(tmp)) goto err;
|
---|
6255 | list2 = tmp;
|
---|
6256 | np1 = NULL;
|
---|
6257 |
|
---|
6258 | np1 = node_new_cclass();
|
---|
6259 | if (IS_NULL(np1)) goto err;
|
---|
6260 | cc = NCCLASS(np1);
|
---|
6261 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=Glue_After_Zwj"), 0, 0, env);
|
---|
6262 | if (r != 0) goto err;
|
---|
6263 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=E_Base_GAZ"), 0, 0, env);
|
---|
6264 | if (r != 0) goto err;
|
---|
6265 |
|
---|
6266 | tmp = node_new_list(np1, list2);
|
---|
6267 | if (IS_NULL(tmp)) goto err;
|
---|
6268 | list2 = tmp;
|
---|
6269 | np1 = NULL;
|
---|
6270 |
|
---|
6271 | r = ONIGENC_CODE_TO_MBC(env->enc, 0x200D, buf);
|
---|
6272 | if (r < 0) goto err;
|
---|
6273 | np1 = node_new_str_raw(buf, buf + r);
|
---|
6274 | if (IS_NULL(np1)) goto err;
|
---|
6275 |
|
---|
6276 | tmp = node_new_list(np1, list2);
|
---|
6277 | if (IS_NULL(tmp)) goto err;
|
---|
6278 | list2 = tmp;
|
---|
6279 | np1 = NULL;
|
---|
6280 |
|
---|
6281 | tmp = onig_node_new_alt(list2, alt);
|
---|
6282 | if (IS_NULL(tmp)) goto err;
|
---|
6283 | alt = tmp;
|
---|
6284 | list2 = NULL;
|
---|
6285 |
|
---|
6286 | /* RI-Sequence := Regional_Indicator{2} */
|
---|
6287 | np1 = node_new_cclass();
|
---|
6288 | if (IS_NULL(np1)) goto err;
|
---|
6289 | cc = NCCLASS(np1);
|
---|
6290 | r = add_code_range(&(cc->mbuf), env, 0x1F1E6, 0x1F1FF);
|
---|
6291 | if (r != 0) goto err;
|
---|
6292 |
|
---|
6293 | tmp = node_new_quantifier(2, 2, 0);
|
---|
6294 | if (IS_NULL(tmp)) goto err;
|
---|
6295 | NQTFR(tmp)->target = np1;
|
---|
6296 | np1 = tmp;
|
---|
6297 |
|
---|
6298 | tmp = node_new_list(np1, list2);
|
---|
6299 | if (IS_NULL(tmp)) goto err;
|
---|
6300 | list2 = tmp;
|
---|
6301 | np1 = NULL;
|
---|
6302 |
|
---|
6303 | tmp = onig_node_new_alt(list2, alt);
|
---|
6304 | if (IS_NULL(tmp)) goto err;
|
---|
6305 | alt = tmp;
|
---|
6306 | list2 = NULL;
|
---|
6307 |
|
---|
6308 | tmp = node_new_list(alt, list);
|
---|
6309 | if (IS_NULL(tmp)) goto err;
|
---|
6310 | list = tmp;
|
---|
6311 | alt = NULL;
|
---|
6312 |
|
---|
6313 | /* Prepend* */
|
---|
6314 | np1 = node_new_cclass();
|
---|
6315 | if (IS_NULL(np1)) goto err;
|
---|
6316 | cc = NCCLASS(np1);
|
---|
6317 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=Prepend"), 0, 0, env);
|
---|
6318 | if (r != 0) goto err;
|
---|
6319 |
|
---|
6320 | tmp = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
6321 | if (IS_NULL(tmp)) goto err;
|
---|
6322 | NQTFR(tmp)->target = np1;
|
---|
6323 | np1 = tmp;
|
---|
6324 |
|
---|
6325 | tmp = node_new_list(np1, list);
|
---|
6326 | if (IS_NULL(tmp)) goto err;
|
---|
6327 | list = tmp;
|
---|
6328 | np1 = NULL;
|
---|
6329 |
|
---|
6330 | /* PerlSyntax: (?s:.), RubySyntax: (?m:.) */
|
---|
6331 | np1 = node_new_anychar();
|
---|
6332 | if (IS_NULL(np1)) goto err;
|
---|
6333 |
|
---|
6334 | option = env->option;
|
---|
6335 | ONOFF(option, ONIG_OPTION_MULTILINE, 0);
|
---|
6336 | tmp = node_new_option(option);
|
---|
6337 | if (IS_NULL(tmp)) goto err;
|
---|
6338 | NENCLOSE(tmp)->target = np1;
|
---|
6339 | np1 = tmp;
|
---|
6340 |
|
---|
6341 | tmp = onig_node_new_alt(np1, NULL_NODE);
|
---|
6342 | if (IS_NULL(tmp)) goto err;
|
---|
6343 | alt = tmp;
|
---|
6344 | np1 = NULL;
|
---|
6345 |
|
---|
6346 | /* Prepend+ */
|
---|
6347 | r = ONIGENC_CODE_TO_MBC(env->enc, 0x200D, buf);
|
---|
6348 | if (r < 0) goto err;
|
---|
6349 | np1 = node_new_str_raw(buf, buf + r);
|
---|
6350 | if (IS_NULL(np1)) goto err;
|
---|
6351 |
|
---|
6352 | tmp = node_new_quantifier(0, 1, 0);
|
---|
6353 | if (IS_NULL(tmp)) goto err;
|
---|
6354 | NQTFR(tmp)->target = np1;
|
---|
6355 | np1 = tmp;
|
---|
6356 |
|
---|
6357 | tmp = node_new_list(np1, NULL_NODE);
|
---|
6358 | if (IS_NULL(tmp)) goto err;
|
---|
6359 | list2 = tmp;
|
---|
6360 | np1 = NULL;
|
---|
6361 |
|
---|
6362 | np1 = node_new_cclass();
|
---|
6363 | if (IS_NULL(np1)) goto err;
|
---|
6364 | cc = NCCLASS(np1);
|
---|
6365 | r = add_ctype_to_cc(cc, propname2ctype(env, "Grapheme_Cluster_Break=Prepend"), 0, 0, env);
|
---|
6366 | if (r != 0) goto err;
|
---|
6367 |
|
---|
6368 | tmp = node_new_quantifier(1, REPEAT_INFINITE, 0);
|
---|
6369 | if (IS_NULL(tmp)) goto err;
|
---|
6370 | NQTFR(tmp)->target = np1;
|
---|
6371 | np1 = tmp;
|
---|
6372 |
|
---|
6373 | tmp = node_new_list(np1, list2);
|
---|
6374 | if (IS_NULL(tmp)) goto err;
|
---|
6375 | list2 = tmp;
|
---|
6376 | np1 = NULL;
|
---|
6377 |
|
---|
6378 | tmp = onig_node_new_alt(list2, alt);
|
---|
6379 | if (IS_NULL(tmp)) goto err;
|
---|
6380 | alt = tmp;
|
---|
6381 | list2 = NULL;
|
---|
6382 |
|
---|
6383 | tmp = onig_node_new_alt(list, alt);
|
---|
6384 | if (IS_NULL(tmp)) goto err;
|
---|
6385 | alt = tmp;
|
---|
6386 | list = NULL;
|
---|
6387 | }
|
---|
6388 | else
|
---|
6389 | #endif /* USE_UNICODE_PROPERTIES */
|
---|
6390 | {
|
---|
6391 | /* PerlSyntax: (?s:.), RubySyntax: (?m:.) */
|
---|
6392 | np1 = node_new_anychar();
|
---|
6393 | if (IS_NULL(np1)) goto err;
|
---|
6394 |
|
---|
6395 | option = env->option;
|
---|
6396 | ONOFF(option, ONIG_OPTION_MULTILINE, 0);
|
---|
6397 | tmp = node_new_option(option);
|
---|
6398 | if (IS_NULL(tmp)) goto err;
|
---|
6399 | NENCLOSE(tmp)->target = np1;
|
---|
6400 | np1 = tmp;
|
---|
6401 |
|
---|
6402 | alt = onig_node_new_alt(np1, NULL_NODE);
|
---|
6403 | if (IS_NULL(alt)) goto err;
|
---|
6404 | np1 = NULL;
|
---|
6405 | }
|
---|
6406 |
|
---|
6407 | /* \x0D\x0A */
|
---|
6408 | r = ONIGENC_CODE_TO_MBC(env->enc, 0x0D, buf);
|
---|
6409 | if (r < 0) goto err;
|
---|
6410 | num1 = r;
|
---|
6411 | r = ONIGENC_CODE_TO_MBC(env->enc, 0x0A, buf + num1);
|
---|
6412 | if (r < 0) goto err;
|
---|
6413 | np1 = node_new_str_raw(buf, buf + num1 + r);
|
---|
6414 | if (IS_NULL(np1)) goto err;
|
---|
6415 |
|
---|
6416 | tmp = onig_node_new_alt(np1, alt);
|
---|
6417 | if (IS_NULL(tmp)) goto err;
|
---|
6418 | alt = tmp;
|
---|
6419 | np1 = NULL;
|
---|
6420 |
|
---|
6421 | /* (?>\x0D\x0A|...) */
|
---|
6422 | tmp = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
|
---|
6423 | if (IS_NULL(tmp)) goto err;
|
---|
6424 | NENCLOSE(tmp)->target = alt;
|
---|
6425 | np1 = tmp;
|
---|
6426 |
|
---|
6427 | #ifdef USE_UNICODE_PROPERTIES
|
---|
6428 | if (ONIGENC_IS_UNICODE(env->enc)) {
|
---|
6429 | /* Don't ignore case. */
|
---|
6430 | option = env->option;
|
---|
6431 | ONOFF(option, ONIG_OPTION_IGNORECASE, 1);
|
---|
6432 | *np = node_new_option(option);
|
---|
6433 | if (IS_NULL(*np)) goto err;
|
---|
6434 | NENCLOSE(*np)->target = np1;
|
---|
6435 | }
|
---|
6436 | else
|
---|
6437 | #endif
|
---|
6438 | {
|
---|
6439 | *np = np1;
|
---|
6440 | }
|
---|
6441 | return ONIG_NORMAL;
|
---|
6442 |
|
---|
6443 | err:
|
---|
6444 | onig_node_free(np1);
|
---|
6445 | onig_node_free(list);
|
---|
6446 | onig_node_free(list2);
|
---|
6447 | onig_node_free(alt);
|
---|
6448 | onig_node_free(alt2);
|
---|
6449 | bbuf_free(pbuf1);
|
---|
6450 | return (r == 0) ? ONIGERR_MEMORY : r;
|
---|
6451 | }
|
---|
6452 |
|
---|
6453 | static int
|
---|
6454 | countbits(unsigned int bits)
|
---|
6455 | {
|
---|
6456 | bits = (bits & 0x55555555) + ((bits >> 1) & 0x55555555);
|
---|
6457 | bits = (bits & 0x33333333) + ((bits >> 2) & 0x33333333);
|
---|
6458 | bits = (bits & 0x0f0f0f0f) + ((bits >> 4) & 0x0f0f0f0f);
|
---|
6459 | bits = (bits & 0x00ff00ff) + ((bits >> 8) & 0x00ff00ff);
|
---|
6460 | return (bits & 0x0000ffff) + ((bits >>16) & 0x0000ffff);
|
---|
6461 | }
|
---|
6462 |
|
---|
6463 | static int
|
---|
6464 | is_onechar_cclass(CClassNode* cc, OnigCodePoint* code)
|
---|
6465 | {
|
---|
6466 | const OnigCodePoint not_found = ONIG_LAST_CODE_POINT;
|
---|
6467 | OnigCodePoint c = not_found;
|
---|
6468 | int i;
|
---|
6469 | BBuf *bbuf = cc->mbuf;
|
---|
6470 |
|
---|
6471 | if (IS_NCCLASS_NOT(cc)) return 0;
|
---|
6472 |
|
---|
6473 | /* check bbuf */
|
---|
6474 | if (IS_NOT_NULL(bbuf)) {
|
---|
6475 | OnigCodePoint n, *data;
|
---|
6476 | GET_CODE_POINT(n, bbuf->p);
|
---|
6477 | data = (OnigCodePoint* )(bbuf->p) + 1;
|
---|
6478 | if ((n == 1) && (data[0] == data[1])) {
|
---|
6479 | /* only one char found in the bbuf, save the code point. */
|
---|
6480 | c = data[0];
|
---|
6481 | if (((c < SINGLE_BYTE_SIZE) && BITSET_AT(cc->bs, c))) {
|
---|
6482 | /* skip if c is included in the bitset */
|
---|
6483 | c = not_found;
|
---|
6484 | }
|
---|
6485 | }
|
---|
6486 | else {
|
---|
6487 | return 0; /* the bbuf contains multiple chars */
|
---|
6488 | }
|
---|
6489 | }
|
---|
6490 |
|
---|
6491 | /* check bitset */
|
---|
6492 | for (i = 0; i < BITSET_SIZE; i++) {
|
---|
6493 | Bits b1 = cc->bs[i];
|
---|
6494 | if (b1 != 0) {
|
---|
6495 | if (((b1 & (b1 - 1)) == 0) && (c == not_found)) {
|
---|
6496 | c = BITS_IN_ROOM * i + countbits(b1 - 1);
|
---|
6497 | } else {
|
---|
6498 | return 0; /* the character class contains multiple chars */
|
---|
6499 | }
|
---|
6500 | }
|
---|
6501 | }
|
---|
6502 |
|
---|
6503 | if (c != not_found) {
|
---|
6504 | *code = c;
|
---|
6505 | return 1;
|
---|
6506 | }
|
---|
6507 |
|
---|
6508 | /* the character class contains no char. */
|
---|
6509 | return 0;
|
---|
6510 | }
|
---|
6511 |
|
---|
6512 |
|
---|
6513 | static int
|
---|
6514 | parse_exp(Node** np, OnigToken* tok, int term,
|
---|
6515 | UChar** src, UChar* end, ScanEnv* env)
|
---|
6516 | {
|
---|
6517 | int r, len, group = 0;
|
---|
6518 | Node* qn;
|
---|
6519 | Node** targetp;
|
---|
6520 |
|
---|
6521 | *np = NULL;
|
---|
6522 | if (tok->type == (enum TokenSyms )term)
|
---|
6523 | goto end_of_token;
|
---|
6524 |
|
---|
6525 | switch (tok->type) {
|
---|
6526 | case TK_ALT:
|
---|
6527 | case TK_EOT:
|
---|
6528 | end_of_token:
|
---|
6529 | *np = node_new_empty();
|
---|
6530 | return tok->type;
|
---|
6531 | break;
|
---|
6532 |
|
---|
6533 | case TK_SUBEXP_OPEN:
|
---|
6534 | r = parse_enclose(np, tok, TK_SUBEXP_CLOSE, src, end, env);
|
---|
6535 | if (r < 0) return r;
|
---|
6536 | if (r == 1) group = 1;
|
---|
6537 | else if (r == 2) { /* option only */
|
---|
6538 | Node* target;
|
---|
6539 | OnigOptionType prev = env->option;
|
---|
6540 |
|
---|
6541 | env->option = NENCLOSE(*np)->option;
|
---|
6542 | r = fetch_token(tok, src, end, env);
|
---|
6543 | if (r < 0) {
|
---|
6544 | env->option = prev;
|
---|
6545 | return r;
|
---|
6546 | }
|
---|
6547 | r = parse_subexp(&target, tok, term, src, end, env);
|
---|
6548 | env->option = prev;
|
---|
6549 | if (r < 0) {
|
---|
6550 | onig_node_free(target);
|
---|
6551 | return r;
|
---|
6552 | }
|
---|
6553 | NENCLOSE(*np)->target = target;
|
---|
6554 | return tok->type;
|
---|
6555 | }
|
---|
6556 | break;
|
---|
6557 |
|
---|
6558 | case TK_SUBEXP_CLOSE:
|
---|
6559 | if (! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP))
|
---|
6560 | return ONIGERR_UNMATCHED_CLOSE_PARENTHESIS;
|
---|
6561 |
|
---|
6562 | if (tok->escaped) goto tk_raw_byte;
|
---|
6563 | else goto tk_byte;
|
---|
6564 | break;
|
---|
6565 |
|
---|
6566 | case TK_LINEBREAK:
|
---|
6567 | r = node_linebreak(np, env);
|
---|
6568 | if (r < 0) return r;
|
---|
6569 | break;
|
---|
6570 |
|
---|
6571 | case TK_EXTENDED_GRAPHEME_CLUSTER:
|
---|
6572 | r = node_extended_grapheme_cluster(np, env);
|
---|
6573 | if (r < 0) return r;
|
---|
6574 | break;
|
---|
6575 |
|
---|
6576 | case TK_KEEP:
|
---|
6577 | *np = onig_node_new_anchor(ANCHOR_KEEP);
|
---|
6578 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6579 | break;
|
---|
6580 |
|
---|
6581 | case TK_STRING:
|
---|
6582 | tk_byte:
|
---|
6583 | {
|
---|
6584 | *np = node_new_str(tok->backp, *src);
|
---|
6585 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6586 |
|
---|
6587 | string_loop:
|
---|
6588 | while (1) {
|
---|
6589 | r = fetch_token(tok, src, end, env);
|
---|
6590 | if (r < 0) return r;
|
---|
6591 | if (r == TK_STRING) {
|
---|
6592 | r = onig_node_str_cat(*np, tok->backp, *src);
|
---|
6593 | }
|
---|
6594 | #ifndef NUMBERED_CHAR_IS_NOT_CASE_AMBIG
|
---|
6595 | else if (r == TK_CODE_POINT) {
|
---|
6596 | r = node_str_cat_codepoint(*np, env->enc, tok->u.code);
|
---|
6597 | }
|
---|
6598 | #endif
|
---|
6599 | else {
|
---|
6600 | break;
|
---|
6601 | }
|
---|
6602 | if (r < 0) return r;
|
---|
6603 | }
|
---|
6604 |
|
---|
6605 | string_end:
|
---|
6606 | targetp = np;
|
---|
6607 | goto repeat;
|
---|
6608 | }
|
---|
6609 | break;
|
---|
6610 |
|
---|
6611 | case TK_RAW_BYTE:
|
---|
6612 | tk_raw_byte:
|
---|
6613 | {
|
---|
6614 | *np = node_new_str_raw_char((UChar )tok->u.c);
|
---|
6615 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6616 | len = 1;
|
---|
6617 | while (1) {
|
---|
6618 | if (len >= ONIGENC_MBC_MINLEN(env->enc)) {
|
---|
6619 | if (len == enclen(env->enc, NSTR(*np)->s, NSTR(*np)->end)) {
|
---|
6620 | r = fetch_token(tok, src, end, env);
|
---|
6621 | NSTRING_CLEAR_RAW(*np);
|
---|
6622 | goto string_end;
|
---|
6623 | }
|
---|
6624 | }
|
---|
6625 |
|
---|
6626 | r = fetch_token(tok, src, end, env);
|
---|
6627 | if (r < 0) return r;
|
---|
6628 | if (r != TK_RAW_BYTE) {
|
---|
6629 | /* Don't use this, it is wrong for little endian encodings. */
|
---|
6630 | #ifdef USE_PAD_TO_SHORT_BYTE_CHAR
|
---|
6631 | int rem;
|
---|
6632 | if (len < ONIGENC_MBC_MINLEN(env->enc)) {
|
---|
6633 | rem = ONIGENC_MBC_MINLEN(env->enc) - len;
|
---|
6634 | (void )node_str_head_pad(NSTR(*np), rem, (UChar )0);
|
---|
6635 | if (len + rem == enclen(env->enc, NSTR(*np)->s)) {
|
---|
6636 | NSTRING_CLEAR_RAW(*np);
|
---|
6637 | goto string_end;
|
---|
6638 | }
|
---|
6639 | }
|
---|
6640 | #endif
|
---|
6641 | return ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
|
---|
6642 | }
|
---|
6643 |
|
---|
6644 | r = node_str_cat_char(*np, (UChar )tok->u.c);
|
---|
6645 | if (r < 0) return r;
|
---|
6646 |
|
---|
6647 | len++;
|
---|
6648 | }
|
---|
6649 | }
|
---|
6650 | break;
|
---|
6651 |
|
---|
6652 | case TK_CODE_POINT:
|
---|
6653 | {
|
---|
6654 | *np = node_new_empty();
|
---|
6655 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6656 | r = node_str_cat_codepoint(*np, env->enc, tok->u.code);
|
---|
6657 | if (r != 0) return r;
|
---|
6658 | #ifdef NUMBERED_CHAR_IS_NOT_CASE_AMBIG
|
---|
6659 | NSTRING_SET_RAW(*np);
|
---|
6660 | #else
|
---|
6661 | goto string_loop;
|
---|
6662 | #endif
|
---|
6663 | }
|
---|
6664 | break;
|
---|
6665 |
|
---|
6666 | case TK_QUOTE_OPEN:
|
---|
6667 | {
|
---|
6668 | OnigCodePoint end_op[2];
|
---|
6669 | UChar *qstart, *qend, *nextp;
|
---|
6670 |
|
---|
6671 | end_op[0] = (OnigCodePoint )MC_ESC(env->syntax);
|
---|
6672 | end_op[1] = (OnigCodePoint )'E';
|
---|
6673 | qstart = *src;
|
---|
6674 | qend = find_str_position(end_op, 2, qstart, end, &nextp, env->enc);
|
---|
6675 | if (IS_NULL(qend)) {
|
---|
6676 | nextp = qend = end;
|
---|
6677 | }
|
---|
6678 | *np = node_new_str(qstart, qend);
|
---|
6679 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6680 | *src = nextp;
|
---|
6681 | }
|
---|
6682 | break;
|
---|
6683 |
|
---|
6684 | case TK_CHAR_TYPE:
|
---|
6685 | {
|
---|
6686 | switch (tok->u.prop.ctype) {
|
---|
6687 | case ONIGENC_CTYPE_WORD:
|
---|
6688 | *np = node_new_ctype(tok->u.prop.ctype, tok->u.prop.not,
|
---|
6689 | IS_ASCII_RANGE(env->option));
|
---|
6690 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6691 | break;
|
---|
6692 |
|
---|
6693 | case ONIGENC_CTYPE_SPACE:
|
---|
6694 | case ONIGENC_CTYPE_DIGIT:
|
---|
6695 | case ONIGENC_CTYPE_XDIGIT:
|
---|
6696 | {
|
---|
6697 | CClassNode* cc;
|
---|
6698 |
|
---|
6699 | *np = node_new_cclass();
|
---|
6700 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6701 | cc = NCCLASS(*np);
|
---|
6702 | r = add_ctype_to_cc(cc, tok->u.prop.ctype, 0,
|
---|
6703 | IS_ASCII_RANGE(env->option), env);
|
---|
6704 | if (r != 0) return r;
|
---|
6705 | if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
|
---|
6706 | }
|
---|
6707 | break;
|
---|
6708 |
|
---|
6709 | default:
|
---|
6710 | return ONIGERR_PARSER_BUG;
|
---|
6711 | break;
|
---|
6712 | }
|
---|
6713 | }
|
---|
6714 | break;
|
---|
6715 |
|
---|
6716 | case TK_CHAR_PROPERTY:
|
---|
6717 | r = parse_char_property(np, tok, src, end, env);
|
---|
6718 | if (r != 0) return r;
|
---|
6719 | break;
|
---|
6720 |
|
---|
6721 | case TK_CC_OPEN:
|
---|
6722 | {
|
---|
6723 | Node *asc_node;
|
---|
6724 | CClassNode* cc;
|
---|
6725 | OnigCodePoint code;
|
---|
6726 |
|
---|
6727 | r = parse_char_class(np, &asc_node, tok, src, end, env);
|
---|
6728 | if (r != 0) {
|
---|
6729 | onig_node_free(asc_node);
|
---|
6730 | return r;
|
---|
6731 | }
|
---|
6732 |
|
---|
6733 | cc = NCCLASS(*np);
|
---|
6734 | if (is_onechar_cclass(cc, &code)) {
|
---|
6735 | onig_node_free(*np);
|
---|
6736 | onig_node_free(asc_node);
|
---|
6737 | *np = node_new_empty();
|
---|
6738 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6739 | r = node_str_cat_codepoint(*np, env->enc, code);
|
---|
6740 | if (r != 0) return r;
|
---|
6741 | goto string_loop;
|
---|
6742 | }
|
---|
6743 | if (IS_IGNORECASE(env->option)) {
|
---|
6744 | r = cclass_case_fold(np, cc, NCCLASS(asc_node), env);
|
---|
6745 | if (r != 0) {
|
---|
6746 | onig_node_free(asc_node);
|
---|
6747 | return r;
|
---|
6748 | }
|
---|
6749 | }
|
---|
6750 | onig_node_free(asc_node);
|
---|
6751 | }
|
---|
6752 | break;
|
---|
6753 |
|
---|
6754 | case TK_ANYCHAR:
|
---|
6755 | *np = node_new_anychar();
|
---|
6756 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6757 | break;
|
---|
6758 |
|
---|
6759 | case TK_ANYCHAR_ANYTIME:
|
---|
6760 | *np = node_new_anychar();
|
---|
6761 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6762 | qn = node_new_quantifier(0, REPEAT_INFINITE, 0);
|
---|
6763 | CHECK_NULL_RETURN_MEMERR(qn);
|
---|
6764 | NQTFR(qn)->target = *np;
|
---|
6765 | *np = qn;
|
---|
6766 | break;
|
---|
6767 |
|
---|
6768 | case TK_BACKREF:
|
---|
6769 | len = tok->u.backref.num;
|
---|
6770 | *np = node_new_backref(len,
|
---|
6771 | (len > 1 ? tok->u.backref.refs : &(tok->u.backref.ref1)),
|
---|
6772 | tok->u.backref.by_name,
|
---|
6773 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
6774 | tok->u.backref.exist_level,
|
---|
6775 | tok->u.backref.level,
|
---|
6776 | #endif
|
---|
6777 | env);
|
---|
6778 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6779 | break;
|
---|
6780 |
|
---|
6781 | #ifdef USE_SUBEXP_CALL
|
---|
6782 | case TK_CALL:
|
---|
6783 | {
|
---|
6784 | int gnum = tok->u.call.gnum;
|
---|
6785 |
|
---|
6786 | if (gnum < 0 || tok->u.call.rel != 0) {
|
---|
6787 | if (gnum > 0) gnum--;
|
---|
6788 | gnum = BACKREF_REL_TO_ABS(gnum, env);
|
---|
6789 | if (gnum <= 0)
|
---|
6790 | return ONIGERR_INVALID_BACKREF;
|
---|
6791 | }
|
---|
6792 | *np = node_new_call(tok->u.call.name, tok->u.call.name_end, gnum);
|
---|
6793 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6794 | env->num_call++;
|
---|
6795 | }
|
---|
6796 | break;
|
---|
6797 | #endif
|
---|
6798 |
|
---|
6799 | case TK_ANCHOR:
|
---|
6800 | *np = onig_node_new_anchor(tok->u.anchor.subtype);
|
---|
6801 | CHECK_NULL_RETURN_MEMERR(*np);
|
---|
6802 | NANCHOR(*np)->ascii_range = tok->u.anchor.ascii_range;
|
---|
6803 | break;
|
---|
6804 |
|
---|
6805 | case TK_OP_REPEAT:
|
---|
6806 | case TK_INTERVAL:
|
---|
6807 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS)) {
|
---|
6808 | if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS))
|
---|
6809 | return ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED;
|
---|
6810 | else
|
---|
6811 | *np = node_new_empty();
|
---|
6812 | }
|
---|
6813 | else {
|
---|
6814 | goto tk_byte;
|
---|
6815 | }
|
---|
6816 | break;
|
---|
6817 |
|
---|
6818 | default:
|
---|
6819 | return ONIGERR_PARSER_BUG;
|
---|
6820 | break;
|
---|
6821 | }
|
---|
6822 |
|
---|
6823 | {
|
---|
6824 | targetp = np;
|
---|
6825 |
|
---|
6826 | re_entry:
|
---|
6827 | r = fetch_token(tok, src, end, env);
|
---|
6828 | if (r < 0) return r;
|
---|
6829 |
|
---|
6830 | repeat:
|
---|
6831 | if (r == TK_OP_REPEAT || r == TK_INTERVAL) {
|
---|
6832 | if (is_invalid_quantifier_target(*targetp))
|
---|
6833 | return ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID;
|
---|
6834 |
|
---|
6835 | qn = node_new_quantifier(tok->u.repeat.lower, tok->u.repeat.upper,
|
---|
6836 | (r == TK_INTERVAL ? 1 : 0));
|
---|
6837 | CHECK_NULL_RETURN_MEMERR(qn);
|
---|
6838 | NQTFR(qn)->greedy = tok->u.repeat.greedy;
|
---|
6839 | r = set_quantifier(qn, *targetp, group, env);
|
---|
6840 | if (r < 0) {
|
---|
6841 | onig_node_free(qn);
|
---|
6842 | return r;
|
---|
6843 | }
|
---|
6844 |
|
---|
6845 | if (tok->u.repeat.possessive != 0) {
|
---|
6846 | Node* en;
|
---|
6847 | en = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
|
---|
6848 | if (IS_NULL(en)) {
|
---|
6849 | onig_node_free(qn);
|
---|
6850 | return ONIGERR_MEMORY;
|
---|
6851 | }
|
---|
6852 | NENCLOSE(en)->target = qn;
|
---|
6853 | qn = en;
|
---|
6854 | }
|
---|
6855 |
|
---|
6856 | if (r == 0) {
|
---|
6857 | *targetp = qn;
|
---|
6858 | }
|
---|
6859 | else if (r == 1) {
|
---|
6860 | onig_node_free(qn);
|
---|
6861 | }
|
---|
6862 | else if (r == 2) { /* split case: /abc+/ */
|
---|
6863 | Node *tmp;
|
---|
6864 |
|
---|
6865 | *targetp = node_new_list(*targetp, NULL);
|
---|
6866 | if (IS_NULL(*targetp)) {
|
---|
6867 | onig_node_free(qn);
|
---|
6868 | return ONIGERR_MEMORY;
|
---|
6869 | }
|
---|
6870 | tmp = NCDR(*targetp) = node_new_list(qn, NULL);
|
---|
6871 | if (IS_NULL(tmp)) {
|
---|
6872 | onig_node_free(qn);
|
---|
6873 | return ONIGERR_MEMORY;
|
---|
6874 | }
|
---|
6875 | targetp = &(NCAR(tmp));
|
---|
6876 | }
|
---|
6877 | goto re_entry;
|
---|
6878 | }
|
---|
6879 | }
|
---|
6880 |
|
---|
6881 | return r;
|
---|
6882 | }
|
---|
6883 |
|
---|
6884 | static int
|
---|
6885 | parse_branch(Node** top, OnigToken* tok, int term,
|
---|
6886 | UChar** src, UChar* end, ScanEnv* env)
|
---|
6887 | {
|
---|
6888 | int r;
|
---|
6889 | Node *node, **headp;
|
---|
6890 |
|
---|
6891 | *top = NULL;
|
---|
6892 | r = parse_exp(&node, tok, term, src, end, env);
|
---|
6893 | if (r < 0) {
|
---|
6894 | onig_node_free(node);
|
---|
6895 | return r;
|
---|
6896 | }
|
---|
6897 |
|
---|
6898 | if (r == TK_EOT || r == term || r == TK_ALT) {
|
---|
6899 | *top = node;
|
---|
6900 | }
|
---|
6901 | else {
|
---|
6902 | *top = node_new_list(node, NULL);
|
---|
6903 | headp = &(NCDR(*top));
|
---|
6904 | while (r != TK_EOT && r != term && r != TK_ALT) {
|
---|
6905 | r = parse_exp(&node, tok, term, src, end, env);
|
---|
6906 | if (r < 0) {
|
---|
6907 | onig_node_free(node);
|
---|
6908 | return r;
|
---|
6909 | }
|
---|
6910 |
|
---|
6911 | if (NTYPE(node) == NT_LIST) {
|
---|
6912 | *headp = node;
|
---|
6913 | while (IS_NOT_NULL(NCDR(node))) node = NCDR(node);
|
---|
6914 | headp = &(NCDR(node));
|
---|
6915 | }
|
---|
6916 | else {
|
---|
6917 | *headp = node_new_list(node, NULL);
|
---|
6918 | headp = &(NCDR(*headp));
|
---|
6919 | }
|
---|
6920 | }
|
---|
6921 | }
|
---|
6922 |
|
---|
6923 | return r;
|
---|
6924 | }
|
---|
6925 |
|
---|
6926 | /* term_tok: TK_EOT or TK_SUBEXP_CLOSE */
|
---|
6927 | static int
|
---|
6928 | parse_subexp(Node** top, OnigToken* tok, int term,
|
---|
6929 | UChar** src, UChar* end, ScanEnv* env)
|
---|
6930 | {
|
---|
6931 | int r;
|
---|
6932 | Node *node, **headp;
|
---|
6933 |
|
---|
6934 | *top = NULL;
|
---|
6935 | env->parse_depth++;
|
---|
6936 | if (env->parse_depth > ParseDepthLimit)
|
---|
6937 | return ONIGERR_PARSE_DEPTH_LIMIT_OVER;
|
---|
6938 | r = parse_branch(&node, tok, term, src, end, env);
|
---|
6939 | if (r < 0) {
|
---|
6940 | onig_node_free(node);
|
---|
6941 | return r;
|
---|
6942 | }
|
---|
6943 |
|
---|
6944 | if (r == term) {
|
---|
6945 | *top = node;
|
---|
6946 | }
|
---|
6947 | else if (r == TK_ALT) {
|
---|
6948 | *top = onig_node_new_alt(node, NULL);
|
---|
6949 | headp = &(NCDR(*top));
|
---|
6950 | while (r == TK_ALT) {
|
---|
6951 | r = fetch_token(tok, src, end, env);
|
---|
6952 | if (r < 0) return r;
|
---|
6953 | r = parse_branch(&node, tok, term, src, end, env);
|
---|
6954 | if (r < 0) {
|
---|
6955 | onig_node_free(node);
|
---|
6956 | return r;
|
---|
6957 | }
|
---|
6958 |
|
---|
6959 | *headp = onig_node_new_alt(node, NULL);
|
---|
6960 | headp = &(NCDR(*headp));
|
---|
6961 | }
|
---|
6962 |
|
---|
6963 | if (tok->type != (enum TokenSyms )term)
|
---|
6964 | goto err;
|
---|
6965 | }
|
---|
6966 | else {
|
---|
6967 | onig_node_free(node);
|
---|
6968 | err:
|
---|
6969 | if (term == TK_SUBEXP_CLOSE)
|
---|
6970 | return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
|
---|
6971 | else
|
---|
6972 | return ONIGERR_PARSER_BUG;
|
---|
6973 | }
|
---|
6974 |
|
---|
6975 | env->parse_depth--;
|
---|
6976 | return r;
|
---|
6977 | }
|
---|
6978 |
|
---|
6979 | static int
|
---|
6980 | parse_regexp(Node** top, UChar** src, UChar* end, ScanEnv* env)
|
---|
6981 | {
|
---|
6982 | int r;
|
---|
6983 | OnigToken tok;
|
---|
6984 |
|
---|
6985 | r = fetch_token(&tok, src, end, env);
|
---|
6986 | if (r < 0) return r;
|
---|
6987 | r = parse_subexp(top, &tok, TK_EOT, src, end, env);
|
---|
6988 | if (r < 0) return r;
|
---|
6989 |
|
---|
6990 | #ifdef USE_SUBEXP_CALL
|
---|
6991 | if (env->num_call > 0) {
|
---|
6992 | /* Capture the pattern itself. It is used for (?R), (?0) and \g<0>. */
|
---|
6993 | const int num = 0;
|
---|
6994 | Node* np;
|
---|
6995 | np = node_new_enclose_memory(env->option, 0);
|
---|
6996 | CHECK_NULL_RETURN_MEMERR(np);
|
---|
6997 | NENCLOSE(np)->regnum = num;
|
---|
6998 | NENCLOSE(np)->target = *top;
|
---|
6999 | r = scan_env_set_mem_node(env, num, np);
|
---|
7000 | if (r != 0) {
|
---|
7001 | onig_node_free(np);
|
---|
7002 | return r;
|
---|
7003 | }
|
---|
7004 | *top = np;
|
---|
7005 | }
|
---|
7006 | #endif
|
---|
7007 | return 0;
|
---|
7008 | }
|
---|
7009 |
|
---|
7010 | extern int
|
---|
7011 | onig_parse_make_tree(Node** root, const UChar* pattern, const UChar* end,
|
---|
7012 | regex_t* reg, ScanEnv* env)
|
---|
7013 | {
|
---|
7014 | int r;
|
---|
7015 | UChar* p;
|
---|
7016 |
|
---|
7017 | #ifdef USE_NAMED_GROUP
|
---|
7018 | names_clear(reg);
|
---|
7019 | #endif
|
---|
7020 |
|
---|
7021 | scan_env_clear(env);
|
---|
7022 | env->option = reg->options;
|
---|
7023 | env->case_fold_flag = reg->case_fold_flag;
|
---|
7024 | env->enc = reg->enc;
|
---|
7025 | env->syntax = reg->syntax;
|
---|
7026 | env->pattern = (UChar* )pattern;
|
---|
7027 | env->pattern_end = (UChar* )end;
|
---|
7028 | env->reg = reg;
|
---|
7029 |
|
---|
7030 | *root = NULL;
|
---|
7031 | p = (UChar* )pattern;
|
---|
7032 | r = parse_regexp(root, &p, (UChar* )end, env);
|
---|
7033 | reg->num_mem = env->num_mem;
|
---|
7034 | return r;
|
---|
7035 | }
|
---|
7036 |
|
---|
7037 | extern void
|
---|
7038 | onig_scan_env_set_error_string(ScanEnv* env, int ecode ARG_UNUSED,
|
---|
7039 | UChar* arg, UChar* arg_end)
|
---|
7040 | {
|
---|
7041 | env->error = arg;
|
---|
7042 | env->error_end = arg_end;
|
---|
7043 | }
|
---|