]>
Commit | Line | Data |
---|---|---|
6cb784b6 TS |
1 | // class template regex -*- C++ -*- |
2 | ||
3 | // Copyright (C) 2013 Free Software Foundation, Inc. | |
4 | // | |
5 | // This file is part of the GNU ISO C++ Library. This library is free | |
6 | // software; you can redistribute it and/or modify it under the | |
7 | // terms of the GNU General Public License as published by the | |
8 | // Free Software Foundation; either version 3, or (at your option) | |
9 | // any later version. | |
10 | ||
11 | // This library is distributed in the hope that it will be useful, | |
12 | // but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | // GNU General Public License for more details. | |
15 | ||
16 | // Under Section 7 of GPL version 3, you are granted additional | |
17 | // permissions described in the GCC Runtime Library Exception, version | |
18 | // 3.1, as published by the Free Software Foundation. | |
19 | ||
20 | // You should have received a copy of the GNU General Public License and | |
21 | // a copy of the GCC Runtime Library Exception along with this program; | |
22 | // see the files COPYING3 and COPYING.RUNTIME respectively. If not, see | |
23 | // <http://www.gnu.org/licenses/>. | |
24 | ||
25 | /** | |
26 | * @file bits/regex_compiler.tcc | |
27 | * This is an internal header file, included by other library headers. | |
28 | * Do not attempt to use it directly. @headername{regex} | |
29 | */ | |
30 | ||
31 | namespace std _GLIBCXX_VISIBILITY(default) | |
32 | { | |
33 | namespace __detail | |
34 | { | |
35 | _GLIBCXX_BEGIN_NAMESPACE_VERSION | |
36 | ||
37 | template<typename _BiIter> | |
38 | void | |
39 | _Scanner<_BiIter>:: | |
40 | _M_advance() | |
41 | { | |
42 | if (_M_current == _M_end) | |
43 | { | |
44 | _M_curToken = _S_token_eof; | |
45 | return; | |
46 | } | |
47 | ||
48 | _CharT __c = *_M_current; | |
49 | if (_M_state & _S_state_in_bracket) | |
50 | { | |
51 | _M_scan_in_bracket(); | |
52 | return; | |
53 | } | |
54 | if (_M_state & _S_state_in_brace) | |
55 | { | |
56 | _M_scan_in_brace(); | |
57 | return; | |
58 | } | |
59 | #if 0 | |
60 | // TODO: re-enable line anchors when _M_assertion is implemented. | |
61 | // See PR libstdc++/47724 | |
62 | else if (_M_state & _S_state_at_start && __c == _M_ctype.widen('^')) | |
63 | { | |
64 | _M_curToken = _S_token_line_begin; | |
65 | ++_M_current; | |
66 | return; | |
67 | } | |
68 | else if (__c == _M_ctype.widen('$')) | |
69 | { | |
70 | _M_curToken = _S_token_line_end; | |
71 | ++_M_current; | |
72 | return; | |
73 | } | |
74 | #endif | |
75 | else if (__c == _M_ctype.widen('.')) | |
76 | { | |
77 | _M_curToken = _S_token_anychar; | |
78 | ++_M_current; | |
79 | return; | |
80 | } | |
81 | else if (__c == _M_ctype.widen('*')) | |
82 | { | |
83 | _M_curToken = _S_token_closure0; | |
84 | ++_M_current; | |
85 | return; | |
86 | } | |
87 | else if (__c == _M_ctype.widen('+')) | |
88 | { | |
89 | _M_curToken = _S_token_closure1; | |
90 | ++_M_current; | |
91 | return; | |
92 | } | |
93 | else if (__c == _M_ctype.widen('|')) | |
94 | { | |
95 | _M_curToken = _S_token_or; | |
96 | ++_M_current; | |
97 | return; | |
98 | } | |
99 | else if (__c == _M_ctype.widen('[')) | |
100 | { | |
101 | if (*++_M_current == _M_ctype.widen('^')) | |
102 | { | |
103 | _M_curToken = _S_token_bracket_inverse_begin; | |
104 | ++_M_current; | |
105 | } | |
106 | else | |
107 | _M_curToken = _S_token_bracket_begin; | |
108 | _M_state |= _S_state_in_bracket; | |
109 | return; | |
110 | } | |
111 | else if (__c == _M_ctype.widen('\\')) | |
112 | { | |
113 | _M_eat_escape(); | |
114 | return; | |
115 | } | |
116 | else if (!(_M_flags & (regex_constants::basic | regex_constants::grep))) | |
117 | { | |
118 | if (__c == _M_ctype.widen('(')) | |
119 | { | |
120 | _M_curToken = _S_token_subexpr_begin; | |
121 | ++_M_current; | |
122 | return; | |
123 | } | |
124 | else if (__c == _M_ctype.widen(')')) | |
125 | { | |
126 | _M_curToken = _S_token_subexpr_end; | |
127 | ++_M_current; | |
128 | return; | |
129 | } | |
130 | else if (__c == _M_ctype.widen('{')) | |
131 | { | |
132 | _M_curToken = _S_token_interval_begin; | |
133 | _M_state |= _S_state_in_brace; | |
134 | ++_M_current; | |
135 | return; | |
136 | } | |
137 | } | |
138 | ||
139 | _M_curToken = _S_token_ord_char; | |
140 | _M_curValue.assign(1, __c); | |
141 | ++_M_current; | |
142 | } | |
143 | ||
144 | template<typename _BiIter> | |
145 | void | |
146 | _Scanner<_BiIter>:: | |
147 | _M_scan_in_brace() | |
148 | { | |
149 | if (_M_ctype.is(_CtypeT::digit, *_M_current)) | |
150 | { | |
151 | _M_curToken = _S_token_dup_count; | |
152 | _M_curValue.assign(1, *_M_current); | |
153 | ++_M_current; | |
154 | while (_M_current != _M_end | |
155 | && _M_ctype.is(_CtypeT::digit, *_M_current)) | |
156 | { | |
157 | _M_curValue += *_M_current; | |
158 | ++_M_current; | |
159 | } | |
160 | return; | |
161 | } | |
162 | else if (*_M_current == _M_ctype.widen(',')) | |
163 | { | |
164 | _M_curToken = _S_token_comma; | |
165 | ++_M_current; | |
166 | return; | |
167 | } | |
168 | if (_M_flags & (regex_constants::basic | regex_constants::grep)) | |
169 | { | |
170 | if (*_M_current == _M_ctype.widen('\\')) | |
171 | _M_eat_escape(); | |
172 | } | |
173 | else | |
174 | { | |
175 | if (*_M_current == _M_ctype.widen('}')) | |
176 | { | |
177 | _M_curToken = _S_token_interval_end; | |
178 | _M_state &= ~_S_state_in_brace; | |
179 | ++_M_current; | |
180 | return; | |
181 | } | |
182 | } | |
183 | } | |
184 | ||
185 | template<typename _BiIter> | |
186 | void | |
187 | _Scanner<_BiIter>:: | |
188 | _M_scan_in_bracket() | |
189 | { | |
190 | if (*_M_current == _M_ctype.widen('[')) | |
191 | { | |
192 | ++_M_current; | |
193 | if (_M_current == _M_end) | |
194 | { | |
195 | _M_curToken = _S_token_eof; | |
196 | return; | |
197 | } | |
198 | ||
199 | if (*_M_current == _M_ctype.widen('.')) | |
200 | { | |
201 | _M_curToken = _S_token_collsymbol; | |
202 | _M_eat_collsymbol(); | |
203 | return; | |
204 | } | |
205 | else if (*_M_current == _M_ctype.widen(':')) | |
206 | { | |
207 | _M_curToken = _S_token_char_class_name; | |
208 | _M_eat_charclass(); | |
209 | return; | |
210 | } | |
211 | else if (*_M_current == _M_ctype.widen('=')) | |
212 | { | |
213 | _M_curToken = _S_token_equiv_class_name; | |
214 | _M_eat_equivclass(); | |
215 | return; | |
216 | } | |
217 | } | |
218 | else if (*_M_current == _M_ctype.widen('-')) | |
219 | { | |
220 | _M_curToken = _S_token_dash; | |
221 | ++_M_current; | |
222 | return; | |
223 | } | |
224 | else if (*_M_current == _M_ctype.widen(']')) | |
225 | { | |
226 | _M_curToken = _S_token_bracket_end; | |
227 | _M_state &= ~_S_state_in_bracket; | |
228 | ++_M_current; | |
229 | return; | |
230 | } | |
231 | else if (*_M_current == _M_ctype.widen('\\')) | |
232 | { | |
233 | _M_eat_escape(); | |
234 | return; | |
235 | } | |
236 | _M_curToken = _S_token_collelem_single; | |
237 | _M_curValue.assign(1, *_M_current); | |
238 | ++_M_current; | |
239 | } | |
240 | ||
241 | // TODO Complete it. | |
242 | template<typename _BiIter> | |
243 | void | |
244 | _Scanner<_BiIter>:: | |
245 | _M_eat_escape() | |
246 | { | |
247 | ++_M_current; | |
248 | if (_M_current == _M_end) | |
249 | { | |
250 | _M_curToken = _S_token_eof; | |
251 | return; | |
252 | } | |
253 | _CharT __c = *_M_current; | |
254 | ++_M_current; | |
255 | ||
256 | if (__c == _M_ctype.widen('(')) | |
257 | { | |
258 | if (!(_M_flags & (regex_constants::basic | regex_constants::grep))) | |
259 | { | |
260 | _M_curToken = _S_token_ord_char; | |
261 | _M_curValue.assign(1, __c); | |
262 | } | |
263 | else | |
264 | _M_curToken = _S_token_subexpr_begin; | |
265 | } | |
266 | else if (__c == _M_ctype.widen(')')) | |
267 | { | |
268 | if (!(_M_flags & (regex_constants::basic | regex_constants::grep))) | |
269 | { | |
270 | _M_curToken = _S_token_ord_char; | |
271 | _M_curValue.assign(1, __c); | |
272 | } | |
273 | else | |
274 | _M_curToken = _S_token_subexpr_end; | |
275 | } | |
276 | else if (__c == _M_ctype.widen('{')) | |
277 | { | |
278 | if (!(_M_flags & (regex_constants::basic | regex_constants::grep))) | |
279 | { | |
280 | _M_curToken = _S_token_ord_char; | |
281 | _M_curValue.assign(1, __c); | |
282 | } | |
283 | else | |
284 | { | |
285 | _M_curToken = _S_token_interval_begin; | |
286 | _M_state |= _S_state_in_brace; | |
287 | } | |
288 | } | |
289 | else if (__c == _M_ctype.widen('}')) | |
290 | { | |
291 | if (!(_M_flags & (regex_constants::basic | regex_constants::grep))) | |
292 | { | |
293 | _M_curToken = _S_token_ord_char; | |
294 | _M_curValue.assign(1, __c); | |
295 | } | |
296 | else | |
297 | { | |
298 | if (!(_M_state && _S_state_in_brace)) | |
299 | __throw_regex_error(regex_constants::error_badbrace); | |
300 | _M_state &= ~_S_state_in_brace; | |
301 | _M_curToken = _S_token_interval_end; | |
302 | } | |
303 | } | |
304 | else if (__c == _M_ctype.widen('x')) | |
305 | { | |
306 | ++_M_current; | |
307 | if (_M_current == _M_end) | |
308 | { | |
309 | _M_curToken = _S_token_eof; | |
310 | return; | |
311 | } | |
312 | if (_M_ctype.is(_CtypeT::digit, *_M_current)) | |
313 | { | |
314 | _M_curValue.assign(1, *_M_current); | |
315 | ++_M_current; | |
316 | if (_M_current == _M_end) | |
317 | { | |
318 | _M_curToken = _S_token_eof; | |
319 | return; | |
320 | } | |
321 | if (_M_ctype.is(_CtypeT::digit, *_M_current)) | |
322 | { | |
323 | _M_curValue += *_M_current; | |
324 | ++_M_current; | |
325 | return; | |
326 | } | |
327 | } | |
328 | } | |
329 | else if (__c == _M_ctype.widen('^') | |
330 | || __c == _M_ctype.widen('.') | |
331 | || __c == _M_ctype.widen('*') | |
332 | || __c == _M_ctype.widen('$') | |
333 | || __c == _M_ctype.widen('\\')) | |
334 | { | |
335 | _M_curToken = _S_token_ord_char; | |
336 | _M_curValue.assign(1, __c); | |
337 | } | |
338 | else if (_M_ctype.is(_CtypeT::digit, __c)) | |
339 | { | |
340 | _M_curToken = _S_token_backref; | |
341 | _M_curValue.assign(1, __c); | |
342 | } | |
343 | else if (_M_state & _S_state_in_bracket) | |
344 | { | |
345 | if (__c == _M_ctype.widen('-') | |
346 | || __c == _M_ctype.widen('[') | |
347 | || __c == _M_ctype.widen(']')) | |
348 | { | |
349 | _M_curToken = _S_token_ord_char; | |
350 | _M_curValue.assign(1, __c); | |
351 | } | |
352 | else if ((_M_flags & regex_constants::ECMAScript) | |
353 | && __c == _M_ctype.widen('b')) | |
354 | { | |
355 | _M_curToken = _S_token_ord_char; | |
356 | _M_curValue.assign(1, _M_ctype.widen(' ')); | |
357 | } | |
358 | else | |
359 | __throw_regex_error(regex_constants::error_escape); | |
360 | } | |
361 | else | |
362 | __throw_regex_error(regex_constants::error_escape); | |
363 | } | |
364 | ||
365 | // Eats a character class or throwns an exception. | |
366 | // current point to ':' delimiter on entry, char after ']' on return | |
367 | template<typename _BiIter> | |
368 | void | |
369 | _Scanner<_BiIter>:: | |
370 | _M_eat_charclass() | |
371 | { | |
372 | ++_M_current; // skip ':' | |
373 | if (_M_current == _M_end) | |
374 | __throw_regex_error(regex_constants::error_ctype); | |
375 | for (_M_curValue.clear(); | |
376 | _M_current != _M_end && *_M_current != _M_ctype.widen(':'); | |
377 | ++_M_current) | |
378 | _M_curValue += *_M_current; | |
379 | if (_M_current == _M_end) | |
380 | __throw_regex_error(regex_constants::error_ctype); | |
381 | ++_M_current; // skip ':' | |
382 | if (*_M_current != _M_ctype.widen(']')) | |
383 | __throw_regex_error(regex_constants::error_ctype); | |
384 | ++_M_current; // skip ']' | |
385 | } | |
386 | ||
387 | ||
388 | template<typename _BiIter> | |
389 | void | |
390 | _Scanner<_BiIter>:: | |
391 | _M_eat_equivclass() | |
392 | { | |
393 | ++_M_current; // skip '=' | |
394 | if (_M_current == _M_end) | |
395 | __throw_regex_error(regex_constants::error_collate); | |
396 | for (_M_curValue.clear(); | |
397 | _M_current != _M_end && *_M_current != _M_ctype.widen('='); | |
398 | ++_M_current) | |
399 | _M_curValue += *_M_current; | |
400 | if (_M_current == _M_end) | |
401 | __throw_regex_error(regex_constants::error_collate); | |
402 | ++_M_current; // skip '=' | |
403 | if (*_M_current != _M_ctype.widen(']')) | |
404 | __throw_regex_error(regex_constants::error_collate); | |
405 | ++_M_current; // skip ']' | |
406 | } | |
407 | ||
408 | ||
409 | template<typename _BiIter> | |
410 | void | |
411 | _Scanner<_BiIter>:: | |
412 | _M_eat_collsymbol() | |
413 | { | |
414 | ++_M_current; // skip '.' | |
415 | if (_M_current == _M_end) | |
416 | __throw_regex_error(regex_constants::error_collate); | |
417 | for (_M_curValue.clear(); | |
418 | _M_current != _M_end && *_M_current != _M_ctype.widen('.'); | |
419 | ++_M_current) | |
420 | _M_curValue += *_M_current; | |
421 | if (_M_current == _M_end) | |
422 | __throw_regex_error(regex_constants::error_collate); | |
423 | ++_M_current; // skip '.' | |
424 | if (*_M_current != _M_ctype.widen(']')) | |
425 | __throw_regex_error(regex_constants::error_collate); | |
426 | ++_M_current; // skip ']' | |
427 | } | |
428 | ||
429 | #ifdef _GLIBCXX_DEBUG | |
430 | template<typename _BiIter> | |
431 | std::ostream& | |
432 | _Scanner<_BiIter>:: | |
433 | _M_print(std::ostream& ostr) | |
434 | { | |
435 | switch (_M_curToken) | |
436 | { | |
437 | case _S_token_anychar: | |
438 | ostr << "any-character\n"; | |
439 | break; | |
440 | case _S_token_backref: | |
441 | ostr << "backref\n"; | |
442 | break; | |
443 | case _S_token_bracket_begin: | |
444 | ostr << "bracket-begin\n"; | |
445 | break; | |
446 | case _S_token_bracket_inverse_begin: | |
447 | ostr << "bracket-inverse-begin\n"; | |
448 | break; | |
449 | case _S_token_bracket_end: | |
450 | ostr << "bracket-end\n"; | |
451 | break; | |
452 | case _S_token_char_class_name: | |
453 | ostr << "char-class-name \"" << _M_curValue << "\"\n"; | |
454 | break; | |
455 | case _S_token_closure0: | |
456 | ostr << "closure0\n"; | |
457 | break; | |
458 | case _S_token_closure1: | |
459 | ostr << "closure1\n"; | |
460 | break; | |
461 | case _S_token_collelem_multi: | |
462 | ostr << "coll-elem-multi \"" << _M_curValue << "\"\n"; | |
463 | break; | |
464 | case _S_token_collelem_single: | |
465 | ostr << "coll-elem-single \"" << _M_curValue << "\"\n"; | |
466 | break; | |
467 | case _S_token_collsymbol: | |
468 | ostr << "collsymbol \"" << _M_curValue << "\"\n"; | |
469 | break; | |
470 | case _S_token_comma: | |
471 | ostr << "comma\n"; | |
472 | break; | |
473 | case _S_token_dash: | |
474 | ostr << "dash\n"; | |
475 | break; | |
476 | case _S_token_dup_count: | |
477 | ostr << "dup count: " << _M_curValue << "\n"; | |
478 | break; | |
479 | case _S_token_eof: | |
480 | ostr << "EOF\n"; | |
481 | break; | |
482 | case _S_token_equiv_class_name: | |
483 | ostr << "equiv-class-name \"" << _M_curValue << "\"\n"; | |
484 | break; | |
485 | case _S_token_interval_begin: | |
486 | ostr << "interval begin\n"; | |
487 | break; | |
488 | case _S_token_interval_end: | |
489 | ostr << "interval end\n"; | |
490 | break; | |
491 | case _S_token_line_begin: | |
492 | ostr << "line begin\n"; | |
493 | break; | |
494 | case _S_token_line_end: | |
495 | ostr << "line end\n"; | |
496 | break; | |
497 | case _S_token_opt: | |
498 | ostr << "opt\n"; | |
499 | break; | |
500 | case _S_token_or: | |
501 | ostr << "or\n"; | |
502 | break; | |
503 | case _S_token_ord_char: | |
504 | ostr << "ordinary character: \"" << _M_value() << "\"\n"; | |
505 | break; | |
506 | case _S_token_subexpr_begin: | |
507 | ostr << "subexpr begin\n"; | |
508 | break; | |
509 | case _S_token_subexpr_end: | |
510 | ostr << "subexpr end\n"; | |
511 | break; | |
512 | case _S_token_word_begin: | |
513 | ostr << "word begin\n"; | |
514 | break; | |
515 | case _S_token_word_end: | |
516 | ostr << "word end\n"; | |
517 | break; | |
518 | case _S_token_unknown: | |
519 | ostr << "-- unknown token --\n"; | |
520 | break; | |
521 | default: | |
522 | _GLIBCXX_DEBUG_ASSERT(false); | |
523 | } | |
524 | return ostr; | |
525 | } | |
526 | #endif | |
527 | ||
528 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
529 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
530 | _Compiler(_InputIter __b, _InputIter __e, | |
531 | const _TraitsT& __traits, _FlagT __flags) | |
532 | : _M_traits(__traits), _M_scanner(__b, __e, __flags, _M_traits.getloc()), | |
533 | _M_state_store(__flags), _M_flags(__flags) | |
534 | { | |
535 | _StateSeqT __r(_M_state_store, | |
536 | _M_state_store._M_insert_subexpr_begin()); | |
537 | _M_disjunction(); | |
538 | if (!_M_stack.empty()) | |
539 | { | |
540 | __r._M_append(_M_stack.top()); | |
541 | _M_stack.pop(); | |
542 | } | |
543 | __r._M_append(_M_state_store._M_insert_subexpr_end()); | |
544 | __r._M_append(_M_state_store._M_insert_accept()); | |
545 | } | |
546 | ||
547 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
548 | bool | |
549 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
550 | _M_match_token(_Compiler<_InputIter, _CharT, _TraitsT>::_TokenT token) | |
551 | { | |
552 | if (token == _M_scanner._M_token()) | |
553 | { | |
554 | _M_cur_value = _M_scanner._M_value(); | |
555 | _M_scanner._M_advance(); | |
556 | return true; | |
557 | } | |
558 | return false; | |
559 | } | |
560 | ||
561 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
562 | void | |
563 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
564 | _M_disjunction() | |
565 | { | |
566 | this->_M_alternative(); | |
567 | if (_M_match_token(_ScannerT::_S_token_or)) | |
568 | { | |
569 | _StateSeqT __alt1 = _M_stack.top(); _M_stack.pop(); | |
570 | this->_M_disjunction(); | |
571 | _StateSeqT __alt2 = _M_stack.top(); _M_stack.pop(); | |
572 | _M_stack.push(_StateSeqT(__alt1, __alt2)); | |
573 | } | |
574 | } | |
575 | ||
576 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
577 | void | |
578 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
579 | _M_alternative() | |
580 | { | |
581 | if (this->_M_term()) | |
582 | { | |
583 | _StateSeqT __re = _M_stack.top(); _M_stack.pop(); | |
584 | this->_M_alternative(); | |
585 | if (!_M_stack.empty()) | |
586 | { | |
587 | __re._M_append(_M_stack.top()); | |
588 | _M_stack.pop(); | |
589 | } | |
590 | _M_stack.push(__re); | |
591 | } | |
592 | } | |
593 | ||
594 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
595 | bool | |
596 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
597 | _M_term() | |
598 | { | |
599 | if (this->_M_assertion()) | |
600 | return true; | |
601 | if (this->_M_atom()) | |
602 | { | |
603 | this->_M_quantifier(); | |
604 | return true; | |
605 | } | |
606 | return false; | |
607 | } | |
608 | ||
609 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
610 | bool | |
611 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
612 | _M_assertion() | |
613 | { | |
614 | if (_M_match_token(_ScannerT::_S_token_line_begin)) | |
615 | { | |
616 | // __m.push(_Matcher::_S_opcode_line_begin); | |
617 | return true; | |
618 | } | |
619 | if (_M_match_token(_ScannerT::_S_token_line_end)) | |
620 | { | |
621 | // __m.push(_Matcher::_S_opcode_line_end); | |
622 | return true; | |
623 | } | |
624 | if (_M_match_token(_ScannerT::_S_token_word_begin)) | |
625 | { | |
626 | // __m.push(_Matcher::_S_opcode_word_begin); | |
627 | return true; | |
628 | } | |
629 | if (_M_match_token(_ScannerT::_S_token_word_end)) | |
630 | { | |
631 | // __m.push(_Matcher::_S_opcode_word_end); | |
632 | return true; | |
633 | } | |
634 | return false; | |
635 | } | |
636 | ||
637 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
638 | void | |
639 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
640 | _M_quantifier() | |
641 | { | |
642 | if (_M_match_token(_ScannerT::_S_token_closure0)) | |
643 | { | |
644 | if (_M_stack.empty()) | |
645 | __throw_regex_error(regex_constants::error_badrepeat); | |
646 | _StateSeqT __r(_M_stack.top(), -1); | |
647 | __r._M_append(__r._M_front()); | |
648 | _M_stack.pop(); | |
649 | _M_stack.push(__r); | |
650 | return; | |
651 | } | |
652 | if (_M_match_token(_ScannerT::_S_token_closure1)) | |
653 | { | |
654 | if (_M_stack.empty()) | |
655 | __throw_regex_error(regex_constants::error_badrepeat); | |
656 | _StateSeqT __r(_M_state_store, | |
657 | _M_state_store. | |
658 | _M_insert_alt(_S_invalid_state_id, | |
659 | _M_stack.top()._M_front())); | |
660 | _M_stack.top()._M_append(__r); | |
661 | return; | |
662 | } | |
663 | if (_M_match_token(_ScannerT::_S_token_opt)) | |
664 | { | |
665 | if (_M_stack.empty()) | |
666 | __throw_regex_error(regex_constants::error_badrepeat); | |
667 | _StateSeqT __r(_M_stack.top(), -1); | |
668 | _M_stack.pop(); | |
669 | _M_stack.push(__r); | |
670 | return; | |
671 | } | |
672 | if (_M_match_token(_ScannerT::_S_token_interval_begin)) | |
673 | { | |
674 | if (_M_stack.empty()) | |
675 | __throw_regex_error(regex_constants::error_badrepeat); | |
676 | if (!_M_match_token(_ScannerT::_S_token_dup_count)) | |
677 | __throw_regex_error(regex_constants::error_badbrace); | |
678 | _StateSeqT __r(_M_stack.top()); | |
679 | int __min_rep = _M_cur_int_value(10); | |
680 | for (int __i = 1; __i < __min_rep; ++__i) | |
681 | _M_stack.top()._M_append(__r._M_clone()); | |
682 | if (_M_match_token(_ScannerT::_S_token_comma)) | |
683 | if (_M_match_token(_ScannerT::_S_token_dup_count)) | |
684 | { | |
685 | int __n = _M_cur_int_value(10) - __min_rep; | |
686 | if (__n < 0) | |
687 | __throw_regex_error(regex_constants::error_badbrace); | |
688 | for (int __i = 0; __i < __n; ++__i) | |
689 | { | |
690 | _StateSeqT __r(_M_state_store, | |
691 | _M_state_store. | |
692 | _M_insert_alt(_S_invalid_state_id, | |
693 | _M_stack.top()._M_front())); | |
694 | _M_stack.top()._M_append(__r); | |
695 | } | |
696 | } | |
697 | else | |
698 | { | |
699 | _StateSeqT __r(_M_stack.top(), -1); | |
700 | __r._M_push_back(__r._M_front()); | |
701 | _M_stack.pop(); | |
702 | _M_stack.push(__r); | |
703 | } | |
704 | if (!_M_match_token(_ScannerT::_S_token_interval_end)) | |
705 | __throw_regex_error(regex_constants::error_brace); | |
706 | return; | |
707 | } | |
708 | } | |
709 | ||
710 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
711 | bool | |
712 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
713 | _M_atom() | |
714 | { | |
715 | if (_M_match_token(_ScannerT::_S_token_anychar)) | |
716 | { | |
717 | const static auto& | |
718 | __any_matcher = [](_CharT) -> bool | |
719 | { return true; }; | |
720 | ||
721 | _M_stack.push(_StateSeqT(_M_state_store, | |
722 | _M_state_store._M_insert_matcher | |
723 | (__any_matcher))); | |
724 | return true; | |
725 | } | |
726 | if (_M_match_token(_ScannerT::_S_token_ord_char)) | |
727 | { | |
728 | auto __c = _M_cur_value[0]; | |
729 | __detail::_Matcher<_CharT> f; | |
730 | if (_M_flags & regex_constants::icase) | |
731 | { | |
732 | auto __traits = this->_M_traits; | |
733 | __c = __traits.translate_nocase(__c); | |
734 | f = [__traits, __c](_CharT __ch) -> bool | |
735 | { return __traits.translate_nocase(__ch) == __c; }; | |
736 | } | |
737 | else | |
738 | f = [__c](_CharT __ch) -> bool | |
739 | { return __ch == __c; }; | |
740 | ||
741 | _M_stack.push(_StateSeqT(_M_state_store, | |
742 | _M_state_store._M_insert_matcher(f))); | |
743 | return true; | |
744 | } | |
745 | if (_M_match_token(_ScannerT::_S_token_backref)) | |
746 | { | |
747 | // __m.push(_Matcher::_S_opcode_ordchar, _M_cur_value); | |
ce645eb0 TS |
748 | _M_stack.push(_StateSeqT(_M_state_store, _M_state_store. |
749 | _M_insert_backref(_M_cur_int_value(10)))); | |
750 | return true; | |
6cb784b6 TS |
751 | } |
752 | if (_M_match_token(_ScannerT::_S_token_subexpr_begin)) | |
753 | { | |
754 | int __mark = _M_state_store._M_sub_count(); | |
755 | _StateSeqT __r(_M_state_store, | |
756 | _M_state_store. | |
757 | _M_insert_subexpr_begin()); | |
758 | this->_M_disjunction(); | |
759 | if (!_M_match_token(_ScannerT::_S_token_subexpr_end)) | |
760 | __throw_regex_error(regex_constants::error_paren); | |
761 | if (!_M_stack.empty()) | |
762 | { | |
763 | __r._M_append(_M_stack.top()); | |
764 | _M_stack.pop(); | |
765 | } | |
766 | __r._M_append(_M_state_store._M_insert_subexpr_end()); | |
767 | _M_stack.push(__r); | |
768 | return true; | |
769 | } | |
770 | return _M_bracket_expression(); | |
771 | } | |
772 | ||
773 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
774 | bool | |
775 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
776 | _M_bracket_expression() | |
777 | { | |
778 | bool __inverse = | |
779 | _M_match_token(_ScannerT::_S_token_bracket_inverse_begin); | |
780 | if (!(__inverse || _M_match_token(_ScannerT::_S_token_bracket_begin))) | |
781 | return false; | |
782 | _BMatcherT __matcher( __inverse, _M_traits, _M_flags); | |
783 | // special case: only if _not_ chr first after | |
784 | // '[' or '[^' or if ECMAscript | |
785 | if (!_M_bracket_list(__matcher) // list is empty | |
786 | && !(_M_flags & regex_constants::ECMAScript)) | |
787 | __throw_regex_error(regex_constants::error_brack); | |
788 | _M_stack.push(_StateSeqT(_M_state_store, | |
789 | _M_state_store._M_insert_matcher(__matcher))); | |
790 | return true; | |
791 | } | |
792 | ||
793 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
794 | bool // list is non-empty | |
795 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
796 | _M_bracket_list(_BMatcherT& __matcher) | |
797 | { | |
798 | if (_M_match_token(_ScannerT::_S_token_bracket_end)) | |
799 | return false; | |
800 | _M_expression_term(__matcher); | |
801 | _M_bracket_list(__matcher); | |
802 | return true; | |
803 | } | |
804 | ||
805 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
806 | void | |
807 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
808 | _M_expression_term(_BMatcherT& __matcher) | |
809 | { | |
810 | if (_M_match_token(_ScannerT::_S_token_collsymbol)) | |
811 | { | |
812 | __matcher._M_add_collating_element(_M_cur_value); | |
813 | return; | |
814 | } | |
815 | if (_M_match_token(_ScannerT::_S_token_equiv_class_name)) | |
816 | { | |
817 | __matcher._M_add_equivalence_class(_M_cur_value); | |
818 | return; | |
819 | } | |
820 | if (_M_match_token(_ScannerT::_S_token_char_class_name)) | |
821 | { | |
822 | __matcher._M_add_character_class(_M_cur_value); | |
823 | return; | |
824 | } | |
825 | if (_M_match_token(_ScannerT::_S_token_collelem_single)) // [a | |
826 | { | |
827 | auto __ch = _M_cur_value[0]; | |
828 | if (_M_match_token(_ScannerT::_S_token_dash)) // [a- | |
829 | { | |
830 | // If the dash is the last character in the bracket expression, | |
831 | // it is not special. | |
832 | if (_M_scanner._M_token() == _ScannerT::_S_token_bracket_end) | |
833 | __matcher._M_add_char(_M_cur_value[0]); // [a-] <=> [a\-] | |
834 | else // [a-z] | |
835 | { | |
836 | if (!_M_match_token(_ScannerT::_S_token_collelem_single)) | |
837 | __throw_regex_error(regex_constants::error_range); | |
838 | __matcher._M_make_range(__ch, _M_cur_value[0]); | |
839 | } | |
840 | } | |
841 | else // [a] | |
842 | __matcher._M_add_char(__ch); | |
843 | return; | |
844 | } | |
845 | __throw_regex_error(regex_constants::error_brack); | |
846 | } | |
847 | ||
848 | template<typename _InputIter, typename _CharT, typename _TraitsT> | |
849 | int | |
850 | _Compiler<_InputIter, _CharT, _TraitsT>:: | |
851 | _M_cur_int_value(int __radix) | |
852 | { | |
853 | int __v = 0; | |
854 | for (typename _StringT::size_type __i = 0; | |
855 | __i < _M_cur_value.length(); ++__i) | |
856 | __v =__v * __radix + _M_traits.value(_M_cur_value[__i], __radix); | |
857 | return __v; | |
858 | } | |
859 | ||
860 | template<typename _CharT, typename _TraitsT> | |
861 | bool _BracketMatcher<_CharT, _TraitsT>:: | |
862 | operator()(_CharT __ch) const | |
863 | { | |
864 | auto __oldch = __ch; | |
865 | if (_M_flags & regex_constants::collate) | |
866 | if (_M_is_icase()) | |
867 | __ch = _M_traits.translate_nocase(__ch); | |
868 | else | |
869 | __ch = _M_traits.translate(__ch); | |
870 | ||
871 | bool __ret = false; | |
872 | for (auto __c : _M_char_set) | |
873 | if (__c == __ch) | |
874 | { | |
875 | __ret = true; | |
876 | break; | |
877 | } | |
878 | if (!__ret && _M_traits.isctype(__oldch, _M_class_set)) | |
879 | __ret = true; | |
880 | else | |
881 | { | |
882 | _StringT __s = _M_get_str(__ch); | |
883 | for (auto& __it : _M_range_set) | |
884 | if (__it.first <= __s && __s <= __it.second) | |
885 | { | |
886 | __ret = true; | |
887 | break; | |
888 | } | |
889 | } | |
890 | if (_M_is_non_matching) | |
891 | __ret = !__ret; | |
892 | return __ret; | |
893 | } | |
894 | ||
895 | _GLIBCXX_END_NAMESPACE_VERSION | |
896 | } // namespace __detail | |
897 | } // namespace |