1 | /* |
2 | * |
3 | * Copyright (c) 2004 |
4 | * John Maddock |
5 | * |
6 | * Use, modification and distribution are subject to the |
7 | * Boost Software License, Version 1.0. (See accompanying file |
8 | * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) |
9 | * |
10 | */ |
11 | |
12 | /* |
13 | * LOCATION: see http://www.boost.org for most recent version. |
14 | * FILE basic_regex_parser.cpp |
15 | * VERSION see <boost/version.hpp> |
16 | * DESCRIPTION: Declares template class basic_regex_parser. |
17 | */ |
18 | |
19 | #ifndef BOOST_REGEX_V4_BASIC_REGEX_PARSER_HPP |
20 | #define BOOST_REGEX_V4_BASIC_REGEX_PARSER_HPP |
21 | |
22 | #ifdef BOOST_MSVC |
23 | #pragma warning(push) |
24 | #pragma warning(disable: 4103) |
25 | #endif |
26 | #ifdef BOOST_HAS_ABI_HEADERS |
27 | # include BOOST_ABI_PREFIX |
28 | #endif |
29 | #ifdef BOOST_MSVC |
30 | #pragma warning(pop) |
31 | #endif |
32 | |
33 | namespace boost{ |
34 | namespace BOOST_REGEX_DETAIL_NS{ |
35 | |
36 | #ifdef BOOST_MSVC |
37 | #pragma warning(push) |
38 | #pragma warning(disable:4244 4800) |
39 | #endif |
40 | |
41 | template <class charT, class traits> |
42 | class basic_regex_parser : public basic_regex_creator<charT, traits> |
43 | { |
44 | public: |
45 | basic_regex_parser(regex_data<charT, traits>* data); |
46 | void parse(const charT* p1, const charT* p2, unsigned flags); |
47 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position); |
48 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position, std::string message, std::ptrdiff_t start_pos); |
49 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position, const std::string& message) |
50 | { |
51 | fail(error_code, position, message, position); |
52 | } |
53 | |
54 | bool parse_all(); |
55 | bool parse_basic(); |
56 | bool parse_extended(); |
57 | bool parse_literal(); |
58 | bool parse_open_paren(); |
59 | bool parse_basic_escape(); |
60 | bool parse_extended_escape(); |
61 | bool parse_match_any(); |
62 | bool parse_repeat(std::size_t low = 0, std::size_t high = (std::numeric_limits<std::size_t>::max)()); |
63 | bool parse_repeat_range(bool isbasic); |
64 | bool parse_alt(); |
65 | bool parse_set(); |
66 | bool parse_backref(); |
67 | void parse_set_literal(basic_char_set<charT, traits>& char_set); |
68 | bool parse_inner_set(basic_char_set<charT, traits>& char_set); |
69 | bool parse_QE(); |
70 | bool parse_perl_extension(); |
71 | bool parse_perl_verb(); |
72 | bool match_verb(const char*); |
73 | bool add_emacs_code(bool negate); |
74 | bool unwind_alts(std::ptrdiff_t last_paren_start); |
75 | digraph<charT> get_next_set_literal(basic_char_set<charT, traits>& char_set); |
76 | charT unescape_character(); |
77 | regex_constants::syntax_option_type parse_options(); |
78 | |
79 | private: |
80 | typedef bool (basic_regex_parser::*parser_proc_type)(); |
81 | typedef typename traits::string_type string_type; |
82 | typedef typename traits::char_class_type char_class_type; |
83 | parser_proc_type m_parser_proc; // the main parser to use |
84 | const charT* m_base; // the start of the string being parsed |
85 | const charT* m_end; // the end of the string being parsed |
86 | const charT* m_position; // our current parser position |
87 | unsigned m_mark_count; // how many sub-expressions we have |
88 | int m_mark_reset; // used to indicate that we're inside a (?|...) block. |
89 | unsigned m_max_mark; // largest mark count seen inside a (?|...) block. |
90 | std::ptrdiff_t m_paren_start; // where the last seen ')' began (where repeats are inserted). |
91 | std::ptrdiff_t m_alt_insert_point; // where to insert the next alternative |
92 | bool m_has_case_change; // true if somewhere in the current block the case has changed |
93 | #if defined(BOOST_MSVC) && defined(_M_IX86) |
94 | // This is an ugly warning suppression workaround (for warnings *inside* std::vector |
95 | // that can not otherwise be suppressed)... |
96 | BOOST_STATIC_ASSERT(sizeof(long) >= sizeof(void*)); |
97 | std::vector<long> m_alt_jumps; // list of alternative in the current scope. |
98 | #else |
99 | std::vector<std::ptrdiff_t> m_alt_jumps; // list of alternative in the current scope. |
100 | #endif |
101 | |
102 | basic_regex_parser& operator=(const basic_regex_parser&); |
103 | basic_regex_parser(const basic_regex_parser&); |
104 | }; |
105 | |
106 | template <class charT, class traits> |
107 | basic_regex_parser<charT, traits>::basic_regex_parser(regex_data<charT, traits>* data) |
108 | : basic_regex_creator<charT, traits>(data), m_mark_count(0), m_mark_reset(-1), m_max_mark(0), m_paren_start(0), m_alt_insert_point(0), m_has_case_change(false) |
109 | { |
110 | } |
111 | |
112 | template <class charT, class traits> |
113 | void basic_regex_parser<charT, traits>::parse(const charT* p1, const charT* p2, unsigned l_flags) |
114 | { |
115 | // pass l_flags on to base class: |
116 | this->init(l_flags); |
117 | // set up pointers: |
118 | m_position = m_base = p1; |
119 | m_end = p2; |
120 | // empty strings are errors: |
121 | if((p1 == p2) && |
122 | ( |
123 | ((l_flags & regbase::main_option_type) != regbase::perl_syntax_group) |
124 | || (l_flags & regbase::no_empty_expressions) |
125 | ) |
126 | ) |
127 | { |
128 | fail(regex_constants::error_empty, 0); |
129 | return; |
130 | } |
131 | // select which parser to use: |
132 | switch(l_flags & regbase::main_option_type) |
133 | { |
134 | case regbase::perl_syntax_group: |
135 | { |
136 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_extended; |
137 | // |
138 | // Add a leading paren with index zero to give recursions a target: |
139 | // |
140 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); |
141 | br->index = 0; |
142 | br->icase = this->flags() & regbase::icase; |
143 | break; |
144 | } |
145 | case regbase::basic_syntax_group: |
146 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_basic; |
147 | break; |
148 | case regbase::literal: |
149 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_literal; |
150 | break; |
151 | default: |
152 | // Ooops, someone has managed to set more than one of the main option flags, |
153 | // so this must be an error: |
154 | fail(regex_constants::error_unknown, 0, "An invalid combination of regular expression syntax flags was used." ); |
155 | return; |
156 | } |
157 | |
158 | // parse all our characters: |
159 | bool result = parse_all(); |
160 | // |
161 | // Unwind our alternatives: |
162 | // |
163 | unwind_alts(last_paren_start: -1); |
164 | // reset l_flags as a global scope (?imsx) may have altered them: |
165 | this->flags(l_flags); |
166 | // if we haven't gobbled up all the characters then we must |
167 | // have had an unexpected ')' : |
168 | if(!result) |
169 | { |
170 | fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_position), "Found a closing ) with no corresponding openening parenthesis." ); |
171 | return; |
172 | } |
173 | // if an error has been set then give up now: |
174 | if(this->m_pdata->m_status) |
175 | return; |
176 | // fill in our sub-expression count: |
177 | this->m_pdata->m_mark_count = 1 + m_mark_count; |
178 | this->finalize(p1, p2); |
179 | } |
180 | |
181 | template <class charT, class traits> |
182 | void basic_regex_parser<charT, traits>::fail(regex_constants::error_type error_code, std::ptrdiff_t position) |
183 | { |
184 | // get the error message: |
185 | std::string message = this->m_pdata->m_ptraits->error_string(error_code); |
186 | fail(error_code, position, message); |
187 | } |
188 | |
189 | template <class charT, class traits> |
190 | void basic_regex_parser<charT, traits>::fail(regex_constants::error_type error_code, std::ptrdiff_t position, std::string message, std::ptrdiff_t start_pos) |
191 | { |
192 | if(0 == this->m_pdata->m_status) // update the error code if not already set |
193 | this->m_pdata->m_status = error_code; |
194 | m_position = m_end; // don't bother parsing anything else |
195 | |
196 | #ifndef BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS |
197 | // |
198 | // Augment error message with the regular expression text: |
199 | // |
200 | if(start_pos == position) |
201 | start_pos = (std::max)(a: static_cast<std::ptrdiff_t>(0), b: position - static_cast<std::ptrdiff_t>(10)); |
202 | std::ptrdiff_t end_pos = (std::min)(a: position + static_cast<std::ptrdiff_t>(10), b: static_cast<std::ptrdiff_t>(m_end - m_base)); |
203 | if(error_code != regex_constants::error_empty) |
204 | { |
205 | if((start_pos != 0) || (end_pos != (m_end - m_base))) |
206 | message += " The error occurred while parsing the regular expression fragment: '" ; |
207 | else |
208 | message += " The error occurred while parsing the regular expression: '" ; |
209 | if(start_pos != end_pos) |
210 | { |
211 | message += std::string(m_base + start_pos, m_base + position); |
212 | message += ">>>HERE>>>" ; |
213 | message += std::string(m_base + position, m_base + end_pos); |
214 | } |
215 | message += "'." ; |
216 | } |
217 | #endif |
218 | |
219 | #ifndef BOOST_NO_EXCEPTIONS |
220 | if(0 == (this->flags() & regex_constants::no_except)) |
221 | { |
222 | boost::regex_error e(message, error_code, position); |
223 | e.raise(); |
224 | } |
225 | #else |
226 | (void)position; // suppress warnings. |
227 | #endif |
228 | } |
229 | |
230 | template <class charT, class traits> |
231 | bool basic_regex_parser<charT, traits>::parse_all() |
232 | { |
233 | bool result = true; |
234 | while(result && (m_position != m_end)) |
235 | { |
236 | result = (this->*m_parser_proc)(); |
237 | } |
238 | return result; |
239 | } |
240 | |
241 | #ifdef BOOST_MSVC |
242 | #pragma warning(push) |
243 | #pragma warning(disable:4702) |
244 | #endif |
245 | template <class charT, class traits> |
246 | bool basic_regex_parser<charT, traits>::parse_basic() |
247 | { |
248 | switch(this->m_traits.syntax_type(*m_position)) |
249 | { |
250 | case regex_constants::syntax_escape: |
251 | return parse_basic_escape(); |
252 | case regex_constants::syntax_dot: |
253 | return parse_match_any(); |
254 | case regex_constants::syntax_caret: |
255 | ++m_position; |
256 | this->append_state(syntax_element_start_line); |
257 | break; |
258 | case regex_constants::syntax_dollar: |
259 | ++m_position; |
260 | this->append_state(syntax_element_end_line); |
261 | break; |
262 | case regex_constants::syntax_star: |
263 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line)) |
264 | return parse_literal(); |
265 | else |
266 | { |
267 | ++m_position; |
268 | return parse_repeat(); |
269 | } |
270 | case regex_constants::syntax_plus: |
271 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line) || !(this->flags() & regbase::emacs_ex)) |
272 | return parse_literal(); |
273 | else |
274 | { |
275 | ++m_position; |
276 | return parse_repeat(low: 1); |
277 | } |
278 | case regex_constants::syntax_question: |
279 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line) || !(this->flags() & regbase::emacs_ex)) |
280 | return parse_literal(); |
281 | else |
282 | { |
283 | ++m_position; |
284 | return parse_repeat(low: 0, high: 1); |
285 | } |
286 | case regex_constants::syntax_open_set: |
287 | return parse_set(); |
288 | case regex_constants::syntax_newline: |
289 | if(this->flags() & regbase::newline_alt) |
290 | return parse_alt(); |
291 | else |
292 | return parse_literal(); |
293 | default: |
294 | return parse_literal(); |
295 | } |
296 | return true; |
297 | } |
298 | |
299 | template <class charT, class traits> |
300 | bool basic_regex_parser<charT, traits>::parse_extended() |
301 | { |
302 | bool result = true; |
303 | switch(this->m_traits.syntax_type(*m_position)) |
304 | { |
305 | case regex_constants::syntax_open_mark: |
306 | return parse_open_paren(); |
307 | case regex_constants::syntax_close_mark: |
308 | return false; |
309 | case regex_constants::syntax_escape: |
310 | return parse_extended_escape(); |
311 | case regex_constants::syntax_dot: |
312 | return parse_match_any(); |
313 | case regex_constants::syntax_caret: |
314 | ++m_position; |
315 | this->append_state( |
316 | (this->flags() & regex_constants::no_mod_m ? syntax_element_buffer_start : syntax_element_start_line)); |
317 | break; |
318 | case regex_constants::syntax_dollar: |
319 | ++m_position; |
320 | this->append_state( |
321 | (this->flags() & regex_constants::no_mod_m ? syntax_element_buffer_end : syntax_element_end_line)); |
322 | break; |
323 | case regex_constants::syntax_star: |
324 | if(m_position == this->m_base) |
325 | { |
326 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"*\" cannot start a regular expression." ); |
327 | return false; |
328 | } |
329 | ++m_position; |
330 | return parse_repeat(); |
331 | case regex_constants::syntax_question: |
332 | if(m_position == this->m_base) |
333 | { |
334 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"?\" cannot start a regular expression." ); |
335 | return false; |
336 | } |
337 | ++m_position; |
338 | return parse_repeat(low: 0,high: 1); |
339 | case regex_constants::syntax_plus: |
340 | if(m_position == this->m_base) |
341 | { |
342 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"+\" cannot start a regular expression." ); |
343 | return false; |
344 | } |
345 | ++m_position; |
346 | return parse_repeat(low: 1); |
347 | case regex_constants::syntax_open_brace: |
348 | ++m_position; |
349 | return parse_repeat_range(isbasic: false); |
350 | case regex_constants::syntax_close_brace: |
351 | if((this->flags() & regbase::no_perl_ex) == regbase::no_perl_ex) |
352 | { |
353 | fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {." ); |
354 | return false; |
355 | } |
356 | result = parse_literal(); |
357 | break; |
358 | case regex_constants::syntax_or: |
359 | return parse_alt(); |
360 | case regex_constants::syntax_open_set: |
361 | return parse_set(); |
362 | case regex_constants::syntax_newline: |
363 | if(this->flags() & regbase::newline_alt) |
364 | return parse_alt(); |
365 | else |
366 | return parse_literal(); |
367 | case regex_constants::syntax_hash: |
368 | // |
369 | // If we have a mod_x flag set, then skip until |
370 | // we get to a newline character: |
371 | // |
372 | if((this->flags() |
373 | & (regbase::no_perl_ex|regbase::mod_x)) |
374 | == regbase::mod_x) |
375 | { |
376 | while((m_position != m_end) && !is_separator(*m_position++)){} |
377 | return true; |
378 | } |
379 | BOOST_FALLTHROUGH; |
380 | default: |
381 | result = parse_literal(); |
382 | break; |
383 | } |
384 | return result; |
385 | } |
386 | #ifdef BOOST_MSVC |
387 | #pragma warning(pop) |
388 | #endif |
389 | |
390 | template <class charT, class traits> |
391 | bool basic_regex_parser<charT, traits>::parse_literal() |
392 | { |
393 | // append this as a literal provided it's not a space character |
394 | // or the perl option regbase::mod_x is not set: |
395 | if( |
396 | ((this->flags() |
397 | & (regbase::main_option_type|regbase::mod_x|regbase::no_perl_ex)) |
398 | != regbase::mod_x) |
399 | || !this->m_traits.isctype(*m_position, this->m_mask_space)) |
400 | this->append_literal(*m_position); |
401 | ++m_position; |
402 | return true; |
403 | } |
404 | |
405 | template <class charT, class traits> |
406 | bool basic_regex_parser<charT, traits>::parse_open_paren() |
407 | { |
408 | // |
409 | // skip the '(' and error check: |
410 | // |
411 | if(++m_position == m_end) |
412 | { |
413 | fail(regex_constants::error_paren, m_position - m_base); |
414 | return false; |
415 | } |
416 | // |
417 | // begin by checking for a perl-style (?...) extension: |
418 | // |
419 | if( |
420 | ((this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) == 0) |
421 | || ((this->flags() & (regbase::main_option_type | regbase::emacs_ex)) == (regbase::basic_syntax_group|regbase::emacs_ex)) |
422 | ) |
423 | { |
424 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question) |
425 | return parse_perl_extension(); |
426 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_star) |
427 | return parse_perl_verb(); |
428 | } |
429 | // |
430 | // update our mark count, and append the required state: |
431 | // |
432 | unsigned markid = 0; |
433 | if(0 == (this->flags() & regbase::nosubs)) |
434 | { |
435 | markid = ++m_mark_count; |
436 | #ifndef BOOST_NO_STD_DISTANCE |
437 | if(this->flags() & regbase::save_subexpression_location) |
438 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>(std::distance(m_base, m_position) - 1, 0)); |
439 | #else |
440 | if(this->flags() & regbase::save_subexpression_location) |
441 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>((m_position - m_base) - 1, 0)); |
442 | #endif |
443 | } |
444 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); |
445 | pb->index = markid; |
446 | pb->icase = this->flags() & regbase::icase; |
447 | std::ptrdiff_t last_paren_start = this->getoffset(pb); |
448 | // back up insertion point for alternations, and set new point: |
449 | std::ptrdiff_t last_alt_point = m_alt_insert_point; |
450 | this->m_pdata->m_data.align(); |
451 | m_alt_insert_point = this->m_pdata->m_data.size(); |
452 | // |
453 | // back up the current flags in case we have a nested (?imsx) group: |
454 | // |
455 | regex_constants::syntax_option_type opts = this->flags(); |
456 | bool old_case_change = m_has_case_change; |
457 | m_has_case_change = false; // no changes to this scope as yet... |
458 | // |
459 | // Back up branch reset data in case we have a nested (?|...) |
460 | // |
461 | int mark_reset = m_mark_reset; |
462 | m_mark_reset = -1; |
463 | // |
464 | // now recursively add more states, this will terminate when we get to a |
465 | // matching ')' : |
466 | // |
467 | parse_all(); |
468 | // |
469 | // Unwind pushed alternatives: |
470 | // |
471 | if(0 == unwind_alts(last_paren_start)) |
472 | return false; |
473 | // |
474 | // restore flags: |
475 | // |
476 | if(m_has_case_change) |
477 | { |
478 | // the case has changed in one or more of the alternatives |
479 | // within the scoped (...) block: we have to add a state |
480 | // to reset the case sensitivity: |
481 | static_cast<re_case*>( |
482 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) |
483 | )->icase = opts & regbase::icase; |
484 | } |
485 | this->flags(opts); |
486 | m_has_case_change = old_case_change; |
487 | // |
488 | // restore branch reset: |
489 | // |
490 | m_mark_reset = mark_reset; |
491 | // |
492 | // we either have a ')' or we have run out of characters prematurely: |
493 | // |
494 | if(m_position == m_end) |
495 | { |
496 | this->fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_end)); |
497 | return false; |
498 | } |
499 | BOOST_ASSERT(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark); |
500 | #ifndef BOOST_NO_STD_DISTANCE |
501 | if(markid && (this->flags() & regbase::save_subexpression_location)) |
502 | this->m_pdata->m_subs.at(markid - 1).second = std::distance(m_base, m_position); |
503 | #else |
504 | if(markid && (this->flags() & regbase::save_subexpression_location)) |
505 | this->m_pdata->m_subs.at(markid - 1).second = (m_position - m_base); |
506 | #endif |
507 | ++m_position; |
508 | // |
509 | // append closing parenthesis state: |
510 | // |
511 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); |
512 | pb->index = markid; |
513 | pb->icase = this->flags() & regbase::icase; |
514 | this->m_paren_start = last_paren_start; |
515 | // |
516 | // restore the alternate insertion point: |
517 | // |
518 | this->m_alt_insert_point = last_alt_point; |
519 | // |
520 | // allow backrefs to this mark: |
521 | // |
522 | if((markid > 0) && (markid < sizeof(unsigned) * CHAR_BIT)) |
523 | this->m_backrefs |= 1u << (markid - 1); |
524 | |
525 | return true; |
526 | } |
527 | |
528 | template <class charT, class traits> |
529 | bool basic_regex_parser<charT, traits>::parse_basic_escape() |
530 | { |
531 | ++m_position; |
532 | bool result = true; |
533 | switch(this->m_traits.escape_syntax_type(*m_position)) |
534 | { |
535 | case regex_constants::syntax_open_mark: |
536 | return parse_open_paren(); |
537 | case regex_constants::syntax_close_mark: |
538 | return false; |
539 | case regex_constants::syntax_plus: |
540 | if(this->flags() & regex_constants::bk_plus_qm) |
541 | { |
542 | ++m_position; |
543 | return parse_repeat(low: 1); |
544 | } |
545 | else |
546 | return parse_literal(); |
547 | case regex_constants::syntax_question: |
548 | if(this->flags() & regex_constants::bk_plus_qm) |
549 | { |
550 | ++m_position; |
551 | return parse_repeat(low: 0, high: 1); |
552 | } |
553 | else |
554 | return parse_literal(); |
555 | case regex_constants::syntax_open_brace: |
556 | if(this->flags() & regbase::no_intervals) |
557 | return parse_literal(); |
558 | ++m_position; |
559 | return parse_repeat_range(isbasic: true); |
560 | case regex_constants::syntax_close_brace: |
561 | if(this->flags() & regbase::no_intervals) |
562 | return parse_literal(); |
563 | fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {." ); |
564 | return false; |
565 | case regex_constants::syntax_or: |
566 | if(this->flags() & regbase::bk_vbar) |
567 | return parse_alt(); |
568 | else |
569 | result = parse_literal(); |
570 | break; |
571 | case regex_constants::syntax_digit: |
572 | return parse_backref(); |
573 | case regex_constants::escape_type_start_buffer: |
574 | if(this->flags() & regbase::emacs_ex) |
575 | { |
576 | ++m_position; |
577 | this->append_state(syntax_element_buffer_start); |
578 | } |
579 | else |
580 | result = parse_literal(); |
581 | break; |
582 | case regex_constants::escape_type_end_buffer: |
583 | if(this->flags() & regbase::emacs_ex) |
584 | { |
585 | ++m_position; |
586 | this->append_state(syntax_element_buffer_end); |
587 | } |
588 | else |
589 | result = parse_literal(); |
590 | break; |
591 | case regex_constants::escape_type_word_assert: |
592 | if(this->flags() & regbase::emacs_ex) |
593 | { |
594 | ++m_position; |
595 | this->append_state(syntax_element_word_boundary); |
596 | } |
597 | else |
598 | result = parse_literal(); |
599 | break; |
600 | case regex_constants::escape_type_not_word_assert: |
601 | if(this->flags() & regbase::emacs_ex) |
602 | { |
603 | ++m_position; |
604 | this->append_state(syntax_element_within_word); |
605 | } |
606 | else |
607 | result = parse_literal(); |
608 | break; |
609 | case regex_constants::escape_type_left_word: |
610 | if(this->flags() & regbase::emacs_ex) |
611 | { |
612 | ++m_position; |
613 | this->append_state(syntax_element_word_start); |
614 | } |
615 | else |
616 | result = parse_literal(); |
617 | break; |
618 | case regex_constants::escape_type_right_word: |
619 | if(this->flags() & regbase::emacs_ex) |
620 | { |
621 | ++m_position; |
622 | this->append_state(syntax_element_word_end); |
623 | } |
624 | else |
625 | result = parse_literal(); |
626 | break; |
627 | default: |
628 | if(this->flags() & regbase::emacs_ex) |
629 | { |
630 | bool negate = true; |
631 | switch(*m_position) |
632 | { |
633 | case 'w': |
634 | negate = false; |
635 | BOOST_FALLTHROUGH; |
636 | case 'W': |
637 | { |
638 | basic_char_set<charT, traits> char_set; |
639 | if(negate) |
640 | char_set.negate(); |
641 | char_set.add_class(this->m_word_mask); |
642 | if(0 == this->append_set(char_set)) |
643 | { |
644 | fail(regex_constants::error_ctype, m_position - m_base); |
645 | return false; |
646 | } |
647 | ++m_position; |
648 | return true; |
649 | } |
650 | case 's': |
651 | negate = false; |
652 | BOOST_FALLTHROUGH; |
653 | case 'S': |
654 | return add_emacs_code(negate); |
655 | case 'c': |
656 | case 'C': |
657 | // not supported yet: |
658 | fail(regex_constants::error_escape, m_position - m_base, "The \\c and \\C escape sequences are not supported by POSIX basic regular expressions: try the Perl syntax instead." ); |
659 | return false; |
660 | default: |
661 | break; |
662 | } |
663 | } |
664 | result = parse_literal(); |
665 | break; |
666 | } |
667 | return result; |
668 | } |
669 | |
670 | template <class charT, class traits> |
671 | bool basic_regex_parser<charT, traits>::parse_extended_escape() |
672 | { |
673 | ++m_position; |
674 | if(m_position == m_end) |
675 | { |
676 | fail(regex_constants::error_escape, m_position - m_base, "Incomplete escape sequence found." ); |
677 | return false; |
678 | } |
679 | bool negate = false; // in case this is a character class escape: \w \d etc |
680 | switch(this->m_traits.escape_syntax_type(*m_position)) |
681 | { |
682 | case regex_constants::escape_type_not_class: |
683 | negate = true; |
684 | BOOST_FALLTHROUGH; |
685 | case regex_constants::escape_type_class: |
686 | { |
687 | escape_type_class_jump: |
688 | typedef typename traits::char_class_type m_type; |
689 | m_type m = this->m_traits.lookup_classname(m_position, m_position+1); |
690 | if(m != 0) |
691 | { |
692 | basic_char_set<charT, traits> char_set; |
693 | if(negate) |
694 | char_set.negate(); |
695 | char_set.add_class(m); |
696 | if(0 == this->append_set(char_set)) |
697 | { |
698 | fail(regex_constants::error_ctype, m_position - m_base); |
699 | return false; |
700 | } |
701 | ++m_position; |
702 | return true; |
703 | } |
704 | // |
705 | // not a class, just a regular unknown escape: |
706 | // |
707 | this->append_literal(unescape_character()); |
708 | break; |
709 | } |
710 | case regex_constants::syntax_digit: |
711 | return parse_backref(); |
712 | case regex_constants::escape_type_left_word: |
713 | ++m_position; |
714 | this->append_state(syntax_element_word_start); |
715 | break; |
716 | case regex_constants::escape_type_right_word: |
717 | ++m_position; |
718 | this->append_state(syntax_element_word_end); |
719 | break; |
720 | case regex_constants::escape_type_start_buffer: |
721 | ++m_position; |
722 | this->append_state(syntax_element_buffer_start); |
723 | break; |
724 | case regex_constants::escape_type_end_buffer: |
725 | ++m_position; |
726 | this->append_state(syntax_element_buffer_end); |
727 | break; |
728 | case regex_constants::escape_type_word_assert: |
729 | ++m_position; |
730 | this->append_state(syntax_element_word_boundary); |
731 | break; |
732 | case regex_constants::escape_type_not_word_assert: |
733 | ++m_position; |
734 | this->append_state(syntax_element_within_word); |
735 | break; |
736 | case regex_constants::escape_type_Z: |
737 | ++m_position; |
738 | this->append_state(syntax_element_soft_buffer_end); |
739 | break; |
740 | case regex_constants::escape_type_Q: |
741 | return parse_QE(); |
742 | case regex_constants::escape_type_C: |
743 | return parse_match_any(); |
744 | case regex_constants::escape_type_X: |
745 | ++m_position; |
746 | this->append_state(syntax_element_combining); |
747 | break; |
748 | case regex_constants::escape_type_G: |
749 | ++m_position; |
750 | this->append_state(syntax_element_restart_continue); |
751 | break; |
752 | case regex_constants::escape_type_not_property: |
753 | negate = true; |
754 | BOOST_FALLTHROUGH; |
755 | case regex_constants::escape_type_property: |
756 | { |
757 | ++m_position; |
758 | char_class_type m; |
759 | if(m_position == m_end) |
760 | { |
761 | fail(regex_constants::error_escape, m_position - m_base, "Incomplete property escape found." ); |
762 | return false; |
763 | } |
764 | // maybe have \p{ddd} |
765 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) |
766 | { |
767 | const charT* base = m_position; |
768 | // skip forward until we find enclosing brace: |
769 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) |
770 | ++m_position; |
771 | if(m_position == m_end) |
772 | { |
773 | fail(regex_constants::error_escape, m_position - m_base, "Closing } missing from property escape sequence." ); |
774 | return false; |
775 | } |
776 | m = this->m_traits.lookup_classname(++base, m_position++); |
777 | } |
778 | else |
779 | { |
780 | m = this->m_traits.lookup_classname(m_position, m_position+1); |
781 | ++m_position; |
782 | } |
783 | if(m != 0) |
784 | { |
785 | basic_char_set<charT, traits> char_set; |
786 | if(negate) |
787 | char_set.negate(); |
788 | char_set.add_class(m); |
789 | if(0 == this->append_set(char_set)) |
790 | { |
791 | fail(regex_constants::error_ctype, m_position - m_base); |
792 | return false; |
793 | } |
794 | return true; |
795 | } |
796 | fail(regex_constants::error_ctype, m_position - m_base, "Escape sequence was neither a valid property nor a valid character class name." ); |
797 | return false; |
798 | } |
799 | case regex_constants::escape_type_reset_start_mark: |
800 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) |
801 | { |
802 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); |
803 | pb->index = -5; |
804 | pb->icase = this->flags() & regbase::icase; |
805 | this->m_pdata->m_data.align(); |
806 | ++m_position; |
807 | return true; |
808 | } |
809 | goto escape_type_class_jump; |
810 | case regex_constants::escape_type_line_ending: |
811 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) |
812 | { |
813 | const charT* e = get_escape_R_string<charT>(); |
814 | const charT* old_position = m_position; |
815 | const charT* old_end = m_end; |
816 | const charT* old_base = m_base; |
817 | m_position = e; |
818 | m_base = e; |
819 | m_end = e + traits::length(e); |
820 | bool r = parse_all(); |
821 | m_position = ++old_position; |
822 | m_end = old_end; |
823 | m_base = old_base; |
824 | return r; |
825 | } |
826 | goto escape_type_class_jump; |
827 | case regex_constants::escape_type_extended_backref: |
828 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) |
829 | { |
830 | bool have_brace = false; |
831 | bool negative = false; |
832 | static const char* incomplete_message = "Incomplete \\g escape found." ; |
833 | if(++m_position == m_end) |
834 | { |
835 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); |
836 | return false; |
837 | } |
838 | // maybe have \g{ddd} |
839 | regex_constants::syntax_type syn = this->m_traits.syntax_type(*m_position); |
840 | regex_constants::syntax_type syn_end = 0; |
841 | if((syn == regex_constants::syntax_open_brace) |
842 | || (syn == regex_constants::escape_type_left_word) |
843 | || (syn == regex_constants::escape_type_end_buffer)) |
844 | { |
845 | if(++m_position == m_end) |
846 | { |
847 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); |
848 | return false; |
849 | } |
850 | have_brace = true; |
851 | switch(syn) |
852 | { |
853 | case regex_constants::syntax_open_brace: |
854 | syn_end = regex_constants::syntax_close_brace; |
855 | break; |
856 | case regex_constants::escape_type_left_word: |
857 | syn_end = regex_constants::escape_type_right_word; |
858 | break; |
859 | default: |
860 | syn_end = regex_constants::escape_type_end_buffer; |
861 | break; |
862 | } |
863 | } |
864 | negative = (*m_position == static_cast<charT>('-')); |
865 | if((negative) && (++m_position == m_end)) |
866 | { |
867 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); |
868 | return false; |
869 | } |
870 | const charT* pc = m_position; |
871 | int i = this->m_traits.toi(pc, m_end, 10); |
872 | if((i < 0) && syn_end) |
873 | { |
874 | // Check for a named capture, get the leftmost one if there is more than one: |
875 | const charT* base = m_position; |
876 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != syn_end)) |
877 | { |
878 | ++m_position; |
879 | } |
880 | i = hash_value_from_capture_name(base, m_position); |
881 | pc = m_position; |
882 | } |
883 | if(negative) |
884 | i = 1 + m_mark_count - i; |
885 | if(((i > 0) && (this->m_backrefs & (1u << (i-1)))) || ((i > 10000) && (this->m_pdata->get_id(i) > 0) && (this->m_backrefs & (1u << (this->m_pdata->get_id(i)-1))))) |
886 | { |
887 | m_position = pc; |
888 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace))); |
889 | pb->index = i; |
890 | pb->icase = this->flags() & regbase::icase; |
891 | } |
892 | else |
893 | { |
894 | fail(regex_constants::error_backref, m_position - m_base); |
895 | return false; |
896 | } |
897 | m_position = pc; |
898 | if(have_brace) |
899 | { |
900 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != syn_end)) |
901 | { |
902 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); |
903 | return false; |
904 | } |
905 | ++m_position; |
906 | } |
907 | return true; |
908 | } |
909 | goto escape_type_class_jump; |
910 | case regex_constants::escape_type_control_v: |
911 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) |
912 | goto escape_type_class_jump; |
913 | BOOST_FALLTHROUGH; |
914 | default: |
915 | this->append_literal(unescape_character()); |
916 | break; |
917 | } |
918 | return true; |
919 | } |
920 | |
921 | template <class charT, class traits> |
922 | bool basic_regex_parser<charT, traits>::parse_match_any() |
923 | { |
924 | // |
925 | // we have a '.' that can match any character: |
926 | // |
927 | ++m_position; |
928 | static_cast<re_dot*>( |
929 | this->append_state(syntax_element_wild, sizeof(re_dot)) |
930 | )->mask = static_cast<unsigned char>(this->flags() & regbase::no_mod_s |
931 | ? BOOST_REGEX_DETAIL_NS::force_not_newline |
932 | : this->flags() & regbase::mod_s ? |
933 | BOOST_REGEX_DETAIL_NS::force_newline : BOOST_REGEX_DETAIL_NS::dont_care); |
934 | return true; |
935 | } |
936 | |
937 | template <class charT, class traits> |
938 | bool basic_regex_parser<charT, traits>::parse_repeat(std::size_t low, std::size_t high) |
939 | { |
940 | bool greedy = true; |
941 | bool pocessive = false; |
942 | std::size_t insert_point; |
943 | // |
944 | // when we get to here we may have a non-greedy ? mark still to come: |
945 | // |
946 | if((m_position != m_end) |
947 | && ( |
948 | (0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) |
949 | || ((regbase::basic_syntax_group|regbase::emacs_ex) == (this->flags() & (regbase::main_option_type | regbase::emacs_ex))) |
950 | ) |
951 | ) |
952 | { |
953 | // OK we have a perl or emacs regex, check for a '?': |
954 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question) |
955 | { |
956 | greedy = false; |
957 | ++m_position; |
958 | } |
959 | // for perl regexes only check for pocessive ++ repeats. |
960 | if((m_position != m_end) |
961 | && (0 == (this->flags() & regbase::main_option_type)) |
962 | && (this->m_traits.syntax_type(*m_position) == regex_constants::syntax_plus)) |
963 | { |
964 | pocessive = true; |
965 | ++m_position; |
966 | } |
967 | } |
968 | if(0 == this->m_last_state) |
969 | { |
970 | fail(regex_constants::error_badrepeat, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_position), "Nothing to repeat." ); |
971 | return false; |
972 | } |
973 | if(this->m_last_state->type == syntax_element_endmark) |
974 | { |
975 | // insert a repeat before the '(' matching the last ')': |
976 | insert_point = this->m_paren_start; |
977 | } |
978 | else if((this->m_last_state->type == syntax_element_literal) && (static_cast<re_literal*>(this->m_last_state)->length > 1)) |
979 | { |
980 | // the last state was a literal with more than one character, split it in two: |
981 | re_literal* lit = static_cast<re_literal*>(this->m_last_state); |
982 | charT c = (static_cast<charT*>(static_cast<void*>(lit+1)))[lit->length - 1]; |
983 | lit->length -= 1; |
984 | // now append new state: |
985 | lit = static_cast<re_literal*>(this->append_state(syntax_element_literal, sizeof(re_literal) + sizeof(charT))); |
986 | lit->length = 1; |
987 | (static_cast<charT*>(static_cast<void*>(lit+1)))[0] = c; |
988 | insert_point = this->getoffset(this->m_last_state); |
989 | } |
990 | else |
991 | { |
992 | // repeat the last state whatever it was, need to add some error checking here: |
993 | switch(this->m_last_state->type) |
994 | { |
995 | case syntax_element_start_line: |
996 | case syntax_element_end_line: |
997 | case syntax_element_word_boundary: |
998 | case syntax_element_within_word: |
999 | case syntax_element_word_start: |
1000 | case syntax_element_word_end: |
1001 | case syntax_element_buffer_start: |
1002 | case syntax_element_buffer_end: |
1003 | case syntax_element_alt: |
1004 | case syntax_element_soft_buffer_end: |
1005 | case syntax_element_restart_continue: |
1006 | case syntax_element_jump: |
1007 | case syntax_element_startmark: |
1008 | case syntax_element_backstep: |
1009 | // can't legally repeat any of the above: |
1010 | fail(regex_constants::error_badrepeat, m_position - m_base); |
1011 | return false; |
1012 | default: |
1013 | // do nothing... |
1014 | break; |
1015 | } |
1016 | insert_point = this->getoffset(this->m_last_state); |
1017 | } |
1018 | // |
1019 | // OK we now know what to repeat, so insert the repeat around it: |
1020 | // |
1021 | re_repeat* rep = static_cast<re_repeat*>(this->insert_state(insert_point, syntax_element_rep, re_repeater_size)); |
1022 | rep->min = low; |
1023 | rep->max = high; |
1024 | rep->greedy = greedy; |
1025 | rep->leading = false; |
1026 | // store our repeater position for later: |
1027 | std::ptrdiff_t rep_off = this->getoffset(rep); |
1028 | // and append a back jump to the repeat: |
1029 | re_jump* jmp = static_cast<re_jump*>(this->append_state(syntax_element_jump, sizeof(re_jump))); |
1030 | jmp->alt.i = rep_off - this->getoffset(jmp); |
1031 | this->m_pdata->m_data.align(); |
1032 | // now fill in the alt jump for the repeat: |
1033 | rep = static_cast<re_repeat*>(this->getaddress(rep_off)); |
1034 | rep->alt.i = this->m_pdata->m_data.size() - rep_off; |
1035 | // |
1036 | // If the repeat is pocessive then bracket the repeat with a (?>...) |
1037 | // independent sub-expression construct: |
1038 | // |
1039 | if(pocessive) |
1040 | { |
1041 | if(m_position != m_end) |
1042 | { |
1043 | // |
1044 | // Check for illegal following quantifier, we have to do this here, because |
1045 | // the extra states we insert below circumvents our usual error checking :-( |
1046 | // |
1047 | switch(this->m_traits.syntax_type(*m_position)) |
1048 | { |
1049 | case regex_constants::syntax_star: |
1050 | case regex_constants::syntax_plus: |
1051 | case regex_constants::syntax_question: |
1052 | case regex_constants::syntax_open_brace: |
1053 | fail(regex_constants::error_badrepeat, m_position - m_base); |
1054 | return false; |
1055 | } |
1056 | } |
1057 | re_brace* pb = static_cast<re_brace*>(this->insert_state(insert_point, syntax_element_startmark, sizeof(re_brace))); |
1058 | pb->index = -3; |
1059 | pb->icase = this->flags() & regbase::icase; |
1060 | jmp = static_cast<re_jump*>(this->insert_state(insert_point + sizeof(re_brace), syntax_element_jump, sizeof(re_jump))); |
1061 | this->m_pdata->m_data.align(); |
1062 | jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp); |
1063 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); |
1064 | pb->index = -3; |
1065 | pb->icase = this->flags() & regbase::icase; |
1066 | } |
1067 | return true; |
1068 | } |
1069 | |
1070 | template <class charT, class traits> |
1071 | bool basic_regex_parser<charT, traits>::parse_repeat_range(bool isbasic) |
1072 | { |
1073 | static const char* incomplete_message = "Missing } in quantified repetition." ; |
1074 | // |
1075 | // parse a repeat-range: |
1076 | // |
1077 | std::size_t min, max; |
1078 | int v; |
1079 | // skip whitespace: |
1080 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) |
1081 | ++m_position; |
1082 | if(this->m_position == this->m_end) |
1083 | { |
1084 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) |
1085 | { |
1086 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); |
1087 | return false; |
1088 | } |
1089 | // Treat the opening '{' as a literal character, rewind to start of error: |
1090 | --m_position; |
1091 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; |
1092 | return parse_literal(); |
1093 | } |
1094 | // get min: |
1095 | v = this->m_traits.toi(m_position, m_end, 10); |
1096 | // skip whitespace: |
1097 | if(v < 0) |
1098 | { |
1099 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) |
1100 | { |
1101 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); |
1102 | return false; |
1103 | } |
1104 | // Treat the opening '{' as a literal character, rewind to start of error: |
1105 | --m_position; |
1106 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; |
1107 | return parse_literal(); |
1108 | } |
1109 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) |
1110 | ++m_position; |
1111 | if(this->m_position == this->m_end) |
1112 | { |
1113 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) |
1114 | { |
1115 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); |
1116 | return false; |
1117 | } |
1118 | // Treat the opening '{' as a literal character, rewind to start of error: |
1119 | --m_position; |
1120 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; |
1121 | return parse_literal(); |
1122 | } |
1123 | min = v; |
1124 | // see if we have a comma: |
1125 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_comma) |
1126 | { |
1127 | // move on and error check: |
1128 | ++m_position; |
1129 | // skip whitespace: |
1130 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) |
1131 | ++m_position; |
1132 | if(this->m_position == this->m_end) |
1133 | { |
1134 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) |
1135 | { |
1136 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); |
1137 | return false; |
1138 | } |
1139 | // Treat the opening '{' as a literal character, rewind to start of error: |
1140 | --m_position; |
1141 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; |
1142 | return parse_literal(); |
1143 | } |
1144 | // get the value if any: |
1145 | v = this->m_traits.toi(m_position, m_end, 10); |
1146 | max = (v >= 0) ? (std::size_t)v : (std::numeric_limits<std::size_t>::max)(); |
1147 | } |
1148 | else |
1149 | { |
1150 | // no comma, max = min: |
1151 | max = min; |
1152 | } |
1153 | // skip whitespace: |
1154 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) |
1155 | ++m_position; |
1156 | // OK now check trailing }: |
1157 | if(this->m_position == this->m_end) |
1158 | { |
1159 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) |
1160 | { |
1161 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); |
1162 | return false; |
1163 | } |
1164 | // Treat the opening '{' as a literal character, rewind to start of error: |
1165 | --m_position; |
1166 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; |
1167 | return parse_literal(); |
1168 | } |
1169 | if(isbasic) |
1170 | { |
1171 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_escape) |
1172 | { |
1173 | ++m_position; |
1174 | if(this->m_position == this->m_end) |
1175 | { |
1176 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); |
1177 | return false; |
1178 | } |
1179 | } |
1180 | else |
1181 | { |
1182 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); |
1183 | return false; |
1184 | } |
1185 | } |
1186 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_brace) |
1187 | ++m_position; |
1188 | else |
1189 | { |
1190 | // Treat the opening '{' as a literal character, rewind to start of error: |
1191 | --m_position; |
1192 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; |
1193 | return parse_literal(); |
1194 | } |
1195 | // |
1196 | // finally go and add the repeat, unless error: |
1197 | // |
1198 | if(min > max) |
1199 | { |
1200 | // Backtrack to error location: |
1201 | m_position -= 2; |
1202 | while(this->m_traits.isctype(*m_position, this->m_word_mask)) --m_position; |
1203 | ++m_position; |
1204 | fail(regex_constants::error_badbrace, m_position - m_base); |
1205 | return false; |
1206 | } |
1207 | return parse_repeat(low: min, high: max); |
1208 | } |
1209 | |
1210 | template <class charT, class traits> |
1211 | bool basic_regex_parser<charT, traits>::parse_alt() |
1212 | { |
1213 | // |
1214 | // error check: if there have been no previous states, |
1215 | // or if the last state was a '(' then error: |
1216 | // |
1217 | if( |
1218 | ((this->m_last_state == 0) || (this->m_last_state->type == syntax_element_startmark)) |
1219 | && |
1220 | !( |
1221 | ((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) |
1222 | && |
1223 | ((this->flags() & regbase::no_empty_expressions) == 0) |
1224 | ) |
1225 | ) |
1226 | { |
1227 | fail(regex_constants::error_empty, this->m_position - this->m_base, "A regular expression cannot start with the alternation operator |." ); |
1228 | return false; |
1229 | } |
1230 | // |
1231 | // Reset mark count if required: |
1232 | // |
1233 | if(m_max_mark < m_mark_count) |
1234 | m_max_mark = m_mark_count; |
1235 | if(m_mark_reset >= 0) |
1236 | m_mark_count = m_mark_reset; |
1237 | |
1238 | ++m_position; |
1239 | // |
1240 | // we need to append a trailing jump: |
1241 | // |
1242 | re_syntax_base* pj = this->append_state(BOOST_REGEX_DETAIL_NS::syntax_element_jump, sizeof(re_jump)); |
1243 | std::ptrdiff_t jump_offset = this->getoffset(pj); |
1244 | // |
1245 | // now insert the alternative: |
1246 | // |
1247 | re_alt* palt = static_cast<re_alt*>(this->insert_state(this->m_alt_insert_point, syntax_element_alt, re_alt_size)); |
1248 | jump_offset += re_alt_size; |
1249 | this->m_pdata->m_data.align(); |
1250 | palt->alt.i = this->m_pdata->m_data.size() - this->getoffset(palt); |
1251 | // |
1252 | // update m_alt_insert_point so that the next alternate gets |
1253 | // inserted at the start of the second of the two we've just created: |
1254 | // |
1255 | this->m_alt_insert_point = this->m_pdata->m_data.size(); |
1256 | // |
1257 | // the start of this alternative must have a case changes state |
1258 | // if the current block has messed around with case changes: |
1259 | // |
1260 | if(m_has_case_change) |
1261 | { |
1262 | static_cast<re_case*>( |
1263 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) |
1264 | )->icase = this->m_icase; |
1265 | } |
1266 | // |
1267 | // push the alternative onto our stack, a recursive |
1268 | // implementation here is easier to understand (and faster |
1269 | // as it happens), but causes all kinds of stack overflow problems |
1270 | // on programs with small stacks (COM+). |
1271 | // |
1272 | m_alt_jumps.push_back(x: jump_offset); |
1273 | return true; |
1274 | } |
1275 | |
1276 | template <class charT, class traits> |
1277 | bool basic_regex_parser<charT, traits>::parse_set() |
1278 | { |
1279 | static const char* incomplete_message = "Character set declaration starting with [ terminated prematurely - either no ] was found or the set had no content." ; |
1280 | ++m_position; |
1281 | if(m_position == m_end) |
1282 | { |
1283 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1284 | return false; |
1285 | } |
1286 | basic_char_set<charT, traits> char_set; |
1287 | |
1288 | const charT* base = m_position; // where the '[' was |
1289 | const charT* item_base = m_position; // where the '[' or '^' was |
1290 | |
1291 | while(m_position != m_end) |
1292 | { |
1293 | switch(this->m_traits.syntax_type(*m_position)) |
1294 | { |
1295 | case regex_constants::syntax_caret: |
1296 | if(m_position == base) |
1297 | { |
1298 | char_set.negate(); |
1299 | ++m_position; |
1300 | item_base = m_position; |
1301 | } |
1302 | else |
1303 | parse_set_literal(char_set); |
1304 | break; |
1305 | case regex_constants::syntax_close_set: |
1306 | if(m_position == item_base) |
1307 | { |
1308 | parse_set_literal(char_set); |
1309 | break; |
1310 | } |
1311 | else |
1312 | { |
1313 | ++m_position; |
1314 | if(0 == this->append_set(char_set)) |
1315 | { |
1316 | fail(regex_constants::error_ctype, m_position - m_base); |
1317 | return false; |
1318 | } |
1319 | } |
1320 | return true; |
1321 | case regex_constants::syntax_open_set: |
1322 | if(parse_inner_set(char_set)) |
1323 | break; |
1324 | return true; |
1325 | case regex_constants::syntax_escape: |
1326 | { |
1327 | // |
1328 | // look ahead and see if this is a character class shortcut |
1329 | // \d \w \s etc... |
1330 | // |
1331 | ++m_position; |
1332 | if(this->m_traits.escape_syntax_type(*m_position) |
1333 | == regex_constants::escape_type_class) |
1334 | { |
1335 | char_class_type m = this->m_traits.lookup_classname(m_position, m_position+1); |
1336 | if(m != 0) |
1337 | { |
1338 | char_set.add_class(m); |
1339 | ++m_position; |
1340 | break; |
1341 | } |
1342 | } |
1343 | else if(this->m_traits.escape_syntax_type(*m_position) |
1344 | == regex_constants::escape_type_not_class) |
1345 | { |
1346 | // negated character class: |
1347 | char_class_type m = this->m_traits.lookup_classname(m_position, m_position+1); |
1348 | if(m != 0) |
1349 | { |
1350 | char_set.add_negated_class(m); |
1351 | ++m_position; |
1352 | break; |
1353 | } |
1354 | } |
1355 | // not a character class, just a regular escape: |
1356 | --m_position; |
1357 | parse_set_literal(char_set); |
1358 | break; |
1359 | } |
1360 | default: |
1361 | parse_set_literal(char_set); |
1362 | break; |
1363 | } |
1364 | } |
1365 | return m_position != m_end; |
1366 | } |
1367 | |
1368 | template <class charT, class traits> |
1369 | bool basic_regex_parser<charT, traits>::parse_inner_set(basic_char_set<charT, traits>& char_set) |
1370 | { |
1371 | static const char* incomplete_message = "Character class declaration starting with [ terminated prematurely - either no ] was found or the set had no content." ; |
1372 | // |
1373 | // we have either a character class [:name:] |
1374 | // a collating element [.name.] |
1375 | // or an equivalence class [=name=] |
1376 | // |
1377 | if(m_end == ++m_position) |
1378 | { |
1379 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1380 | return false; |
1381 | } |
1382 | switch(this->m_traits.syntax_type(*m_position)) |
1383 | { |
1384 | case regex_constants::syntax_dot: |
1385 | // |
1386 | // a collating element is treated as a literal: |
1387 | // |
1388 | --m_position; |
1389 | parse_set_literal(char_set); |
1390 | return true; |
1391 | case regex_constants::syntax_colon: |
1392 | { |
1393 | // check that character classes are actually enabled: |
1394 | if((this->flags() & (regbase::main_option_type | regbase::no_char_classes)) |
1395 | == (regbase::basic_syntax_group | regbase::no_char_classes)) |
1396 | { |
1397 | --m_position; |
1398 | parse_set_literal(char_set); |
1399 | return true; |
1400 | } |
1401 | // skip the ':' |
1402 | if(m_end == ++m_position) |
1403 | { |
1404 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1405 | return false; |
1406 | } |
1407 | const charT* name_first = m_position; |
1408 | // skip at least one character, then find the matching ':]' |
1409 | if(m_end == ++m_position) |
1410 | { |
1411 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1412 | return false; |
1413 | } |
1414 | while((m_position != m_end) |
1415 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_colon)) |
1416 | ++m_position; |
1417 | const charT* name_last = m_position; |
1418 | if(m_end == m_position) |
1419 | { |
1420 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1421 | return false; |
1422 | } |
1423 | if((m_end == ++m_position) |
1424 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) |
1425 | { |
1426 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1427 | return false; |
1428 | } |
1429 | // |
1430 | // check for negated class: |
1431 | // |
1432 | bool negated = false; |
1433 | if(this->m_traits.syntax_type(*name_first) == regex_constants::syntax_caret) |
1434 | { |
1435 | ++name_first; |
1436 | negated = true; |
1437 | } |
1438 | typedef typename traits::char_class_type m_type; |
1439 | m_type m = this->m_traits.lookup_classname(name_first, name_last); |
1440 | if(m == 0) |
1441 | { |
1442 | if(char_set.empty() && (name_last - name_first == 1)) |
1443 | { |
1444 | // maybe a special case: |
1445 | ++m_position; |
1446 | if( (m_position != m_end) |
1447 | && (this->m_traits.syntax_type(*m_position) |
1448 | == regex_constants::syntax_close_set)) |
1449 | { |
1450 | if(this->m_traits.escape_syntax_type(*name_first) |
1451 | == regex_constants::escape_type_left_word) |
1452 | { |
1453 | ++m_position; |
1454 | this->append_state(syntax_element_word_start); |
1455 | return false; |
1456 | } |
1457 | if(this->m_traits.escape_syntax_type(*name_first) |
1458 | == regex_constants::escape_type_right_word) |
1459 | { |
1460 | ++m_position; |
1461 | this->append_state(syntax_element_word_end); |
1462 | return false; |
1463 | } |
1464 | } |
1465 | } |
1466 | fail(regex_constants::error_ctype, name_first - m_base); |
1467 | return false; |
1468 | } |
1469 | if(negated == false) |
1470 | char_set.add_class(m); |
1471 | else |
1472 | char_set.add_negated_class(m); |
1473 | ++m_position; |
1474 | break; |
1475 | } |
1476 | case regex_constants::syntax_equal: |
1477 | { |
1478 | // skip the '=' |
1479 | if(m_end == ++m_position) |
1480 | { |
1481 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1482 | return false; |
1483 | } |
1484 | const charT* name_first = m_position; |
1485 | // skip at least one character, then find the matching '=]' |
1486 | if(m_end == ++m_position) |
1487 | { |
1488 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1489 | return false; |
1490 | } |
1491 | while((m_position != m_end) |
1492 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal)) |
1493 | ++m_position; |
1494 | const charT* name_last = m_position; |
1495 | if(m_end == m_position) |
1496 | { |
1497 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1498 | return false; |
1499 | } |
1500 | if((m_end == ++m_position) |
1501 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) |
1502 | { |
1503 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); |
1504 | return false; |
1505 | } |
1506 | string_type m = this->m_traits.lookup_collatename(name_first, name_last); |
1507 | if((0 == m.size()) || (m.size() > 2)) |
1508 | { |
1509 | fail(regex_constants::error_collate, name_first - m_base); |
1510 | return false; |
1511 | } |
1512 | digraph<charT> d; |
1513 | d.first = m[0]; |
1514 | if(m.size() > 1) |
1515 | d.second = m[1]; |
1516 | else |
1517 | d.second = 0; |
1518 | char_set.add_equivalent(d); |
1519 | ++m_position; |
1520 | break; |
1521 | } |
1522 | default: |
1523 | --m_position; |
1524 | parse_set_literal(char_set); |
1525 | break; |
1526 | } |
1527 | return true; |
1528 | } |
1529 | |
1530 | template <class charT, class traits> |
1531 | void basic_regex_parser<charT, traits>::parse_set_literal(basic_char_set<charT, traits>& char_set) |
1532 | { |
1533 | digraph<charT> start_range(get_next_set_literal(char_set)); |
1534 | if(m_end == m_position) |
1535 | { |
1536 | fail(regex_constants::error_brack, m_position - m_base); |
1537 | return; |
1538 | } |
1539 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_dash) |
1540 | { |
1541 | // we have a range: |
1542 | if(m_end == ++m_position) |
1543 | { |
1544 | fail(regex_constants::error_brack, m_position - m_base); |
1545 | return; |
1546 | } |
1547 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set) |
1548 | { |
1549 | digraph<charT> end_range = get_next_set_literal(char_set); |
1550 | char_set.add_range(start_range, end_range); |
1551 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_dash) |
1552 | { |
1553 | if(m_end == ++m_position) |
1554 | { |
1555 | fail(regex_constants::error_brack, m_position - m_base); |
1556 | return; |
1557 | } |
1558 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_set) |
1559 | { |
1560 | // trailing - : |
1561 | --m_position; |
1562 | return; |
1563 | } |
1564 | fail(regex_constants::error_range, m_position - m_base); |
1565 | return; |
1566 | } |
1567 | return; |
1568 | } |
1569 | --m_position; |
1570 | } |
1571 | char_set.add_single(start_range); |
1572 | } |
1573 | |
1574 | template <class charT, class traits> |
1575 | digraph<charT> basic_regex_parser<charT, traits>::get_next_set_literal(basic_char_set<charT, traits>& char_set) |
1576 | { |
1577 | digraph<charT> result; |
1578 | switch(this->m_traits.syntax_type(*m_position)) |
1579 | { |
1580 | case regex_constants::syntax_dash: |
1581 | if(!char_set.empty()) |
1582 | { |
1583 | // see if we are at the end of the set: |
1584 | if((++m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) |
1585 | { |
1586 | fail(regex_constants::error_range, m_position - m_base); |
1587 | return result; |
1588 | } |
1589 | --m_position; |
1590 | } |
1591 | result.first = *m_position++; |
1592 | return result; |
1593 | case regex_constants::syntax_escape: |
1594 | // check to see if escapes are supported first: |
1595 | if(this->flags() & regex_constants::no_escape_in_lists) |
1596 | { |
1597 | result = *m_position++; |
1598 | break; |
1599 | } |
1600 | ++m_position; |
1601 | result = unescape_character(); |
1602 | break; |
1603 | case regex_constants::syntax_open_set: |
1604 | { |
1605 | if(m_end == ++m_position) |
1606 | { |
1607 | fail(regex_constants::error_collate, m_position - m_base); |
1608 | return result; |
1609 | } |
1610 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_dot) |
1611 | { |
1612 | --m_position; |
1613 | result.first = *m_position; |
1614 | ++m_position; |
1615 | return result; |
1616 | } |
1617 | if(m_end == ++m_position) |
1618 | { |
1619 | fail(regex_constants::error_collate, m_position - m_base); |
1620 | return result; |
1621 | } |
1622 | const charT* name_first = m_position; |
1623 | // skip at least one character, then find the matching ':]' |
1624 | if(m_end == ++m_position) |
1625 | { |
1626 | fail(regex_constants::error_collate, name_first - m_base); |
1627 | return result; |
1628 | } |
1629 | while((m_position != m_end) |
1630 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_dot)) |
1631 | ++m_position; |
1632 | const charT* name_last = m_position; |
1633 | if(m_end == m_position) |
1634 | { |
1635 | fail(regex_constants::error_collate, name_first - m_base); |
1636 | return result; |
1637 | } |
1638 | if((m_end == ++m_position) |
1639 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) |
1640 | { |
1641 | fail(regex_constants::error_collate, name_first - m_base); |
1642 | return result; |
1643 | } |
1644 | ++m_position; |
1645 | string_type s = this->m_traits.lookup_collatename(name_first, name_last); |
1646 | if(s.empty() || (s.size() > 2)) |
1647 | { |
1648 | fail(regex_constants::error_collate, name_first - m_base); |
1649 | return result; |
1650 | } |
1651 | result.first = s[0]; |
1652 | if(s.size() > 1) |
1653 | result.second = s[1]; |
1654 | else |
1655 | result.second = 0; |
1656 | return result; |
1657 | } |
1658 | default: |
1659 | result = *m_position++; |
1660 | } |
1661 | return result; |
1662 | } |
1663 | |
1664 | // |
1665 | // does a value fit in the specified charT type? |
1666 | // |
1667 | template <class charT> |
1668 | bool valid_value(charT, int v, const mpl::true_&) |
1669 | { |
1670 | return (v >> (sizeof(charT) * CHAR_BIT)) == 0; |
1671 | } |
1672 | template <class charT> |
1673 | bool valid_value(charT, int, const mpl::false_&) |
1674 | { |
1675 | return true; // v will alsways fit in a charT |
1676 | } |
1677 | template <class charT> |
1678 | bool valid_value(charT c, int v) |
1679 | { |
1680 | return valid_value(c, v, mpl::bool_<(sizeof(charT) < sizeof(int))>()); |
1681 | } |
1682 | |
1683 | template <class charT, class traits> |
1684 | charT basic_regex_parser<charT, traits>::unescape_character() |
1685 | { |
1686 | #ifdef BOOST_MSVC |
1687 | #pragma warning(push) |
1688 | #pragma warning(disable:4127) |
1689 | #endif |
1690 | charT result(0); |
1691 | if(m_position == m_end) |
1692 | { |
1693 | fail(regex_constants::error_escape, m_position - m_base, "Escape sequence terminated prematurely." ); |
1694 | return false; |
1695 | } |
1696 | switch(this->m_traits.escape_syntax_type(*m_position)) |
1697 | { |
1698 | case regex_constants::escape_type_control_a: |
1699 | result = charT('\a'); |
1700 | break; |
1701 | case regex_constants::escape_type_e: |
1702 | result = charT(27); |
1703 | break; |
1704 | case regex_constants::escape_type_control_f: |
1705 | result = charT('\f'); |
1706 | break; |
1707 | case regex_constants::escape_type_control_n: |
1708 | result = charT('\n'); |
1709 | break; |
1710 | case regex_constants::escape_type_control_r: |
1711 | result = charT('\r'); |
1712 | break; |
1713 | case regex_constants::escape_type_control_t: |
1714 | result = charT('\t'); |
1715 | break; |
1716 | case regex_constants::escape_type_control_v: |
1717 | result = charT('\v'); |
1718 | break; |
1719 | case regex_constants::escape_type_word_assert: |
1720 | result = charT('\b'); |
1721 | break; |
1722 | case regex_constants::escape_type_ascii_control: |
1723 | ++m_position; |
1724 | if(m_position == m_end) |
1725 | { |
1726 | // Rewind to start of escape: |
1727 | --m_position; |
1728 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1729 | fail(regex_constants::error_escape, m_position - m_base, "ASCII escape sequence terminated prematurely." ); |
1730 | return result; |
1731 | } |
1732 | result = static_cast<charT>(*m_position % 32); |
1733 | break; |
1734 | case regex_constants::escape_type_hex: |
1735 | ++m_position; |
1736 | if(m_position == m_end) |
1737 | { |
1738 | // Rewind to start of escape: |
1739 | --m_position; |
1740 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1741 | fail(regex_constants::error_escape, m_position - m_base, "Hexadecimal escape sequence terminated prematurely." ); |
1742 | return result; |
1743 | } |
1744 | // maybe have \x{ddd} |
1745 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) |
1746 | { |
1747 | ++m_position; |
1748 | if(m_position == m_end) |
1749 | { |
1750 | // Rewind to start of escape: |
1751 | --m_position; |
1752 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1753 | fail(regex_constants::error_escape, m_position - m_base, "Missing } in hexadecimal escape sequence." ); |
1754 | return result; |
1755 | } |
1756 | int i = this->m_traits.toi(m_position, m_end, 16); |
1757 | if((m_position == m_end) |
1758 | || (i < 0) |
1759 | || ((std::numeric_limits<charT>::is_specialized) && (i > (int)(std::numeric_limits<charT>::max)())) |
1760 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) |
1761 | { |
1762 | // Rewind to start of escape: |
1763 | --m_position; |
1764 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1765 | fail(regex_constants::error_badbrace, m_position - m_base, "Hexadecimal escape sequence was invalid." ); |
1766 | return result; |
1767 | } |
1768 | ++m_position; |
1769 | result = charT(i); |
1770 | } |
1771 | else |
1772 | { |
1773 | std::ptrdiff_t len = (std::min)(a: static_cast<std::ptrdiff_t>(2), b: static_cast<std::ptrdiff_t>(m_end - m_position)); |
1774 | int i = this->m_traits.toi(m_position, m_position + len, 16); |
1775 | if((i < 0) |
1776 | || !valid_value(charT(0), i)) |
1777 | { |
1778 | // Rewind to start of escape: |
1779 | --m_position; |
1780 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1781 | fail(regex_constants::error_escape, m_position - m_base, "Escape sequence did not encode a valid character." ); |
1782 | return result; |
1783 | } |
1784 | result = charT(i); |
1785 | } |
1786 | return result; |
1787 | case regex_constants::syntax_digit: |
1788 | { |
1789 | // an octal escape sequence, the first character must be a zero |
1790 | // followed by up to 3 octal digits: |
1791 | std::ptrdiff_t len = (std::min)(::boost::BOOST_REGEX_DETAIL_NS::distance(m_position, m_end), static_cast<std::ptrdiff_t>(4)); |
1792 | const charT* bp = m_position; |
1793 | int val = this->m_traits.toi(bp, bp + 1, 8); |
1794 | if(val != 0) |
1795 | { |
1796 | // Rewind to start of escape: |
1797 | --m_position; |
1798 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1799 | // Oops not an octal escape after all: |
1800 | fail(regex_constants::error_escape, m_position - m_base, "Invalid octal escape sequence." ); |
1801 | return result; |
1802 | } |
1803 | val = this->m_traits.toi(m_position, m_position + len, 8); |
1804 | if(val < 0) |
1805 | { |
1806 | // Rewind to start of escape: |
1807 | --m_position; |
1808 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1809 | fail(regex_constants::error_escape, m_position - m_base, "Octal escape sequence is invalid." ); |
1810 | return result; |
1811 | } |
1812 | return static_cast<charT>(val); |
1813 | } |
1814 | case regex_constants::escape_type_named_char: |
1815 | { |
1816 | ++m_position; |
1817 | if(m_position == m_end) |
1818 | { |
1819 | // Rewind to start of escape: |
1820 | --m_position; |
1821 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1822 | fail(regex_constants::error_escape, m_position - m_base); |
1823 | return false; |
1824 | } |
1825 | // maybe have \N{name} |
1826 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) |
1827 | { |
1828 | const charT* base = m_position; |
1829 | // skip forward until we find enclosing brace: |
1830 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) |
1831 | ++m_position; |
1832 | if(m_position == m_end) |
1833 | { |
1834 | // Rewind to start of escape: |
1835 | --m_position; |
1836 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1837 | fail(regex_constants::error_escape, m_position - m_base); |
1838 | return false; |
1839 | } |
1840 | string_type s = this->m_traits.lookup_collatename(++base, m_position++); |
1841 | if(s.empty()) |
1842 | { |
1843 | // Rewind to start of escape: |
1844 | --m_position; |
1845 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1846 | fail(regex_constants::error_collate, m_position - m_base); |
1847 | return false; |
1848 | } |
1849 | if(s.size() == 1) |
1850 | { |
1851 | return s[0]; |
1852 | } |
1853 | } |
1854 | // fall through is a failure: |
1855 | // Rewind to start of escape: |
1856 | --m_position; |
1857 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1858 | fail(regex_constants::error_escape, m_position - m_base); |
1859 | return false; |
1860 | } |
1861 | default: |
1862 | result = *m_position; |
1863 | break; |
1864 | } |
1865 | ++m_position; |
1866 | return result; |
1867 | #ifdef BOOST_MSVC |
1868 | #pragma warning(pop) |
1869 | #endif |
1870 | } |
1871 | |
1872 | template <class charT, class traits> |
1873 | bool basic_regex_parser<charT, traits>::parse_backref() |
1874 | { |
1875 | BOOST_ASSERT(m_position != m_end); |
1876 | const charT* pc = m_position; |
1877 | int i = this->m_traits.toi(pc, pc + 1, 10); |
1878 | if((i == 0) || (((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) && (this->flags() & regbase::no_bk_refs))) |
1879 | { |
1880 | // not a backref at all but an octal escape sequence: |
1881 | charT c = unescape_character(); |
1882 | this->append_literal(c); |
1883 | } |
1884 | else if((i > 0) && (this->m_backrefs & (1u << (i-1)))) |
1885 | { |
1886 | m_position = pc; |
1887 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace))); |
1888 | pb->index = i; |
1889 | pb->icase = this->flags() & regbase::icase; |
1890 | } |
1891 | else |
1892 | { |
1893 | // Rewind to start of escape: |
1894 | --m_position; |
1895 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
1896 | fail(regex_constants::error_backref, m_position - m_base); |
1897 | return false; |
1898 | } |
1899 | return true; |
1900 | } |
1901 | |
1902 | template <class charT, class traits> |
1903 | bool basic_regex_parser<charT, traits>::parse_QE() |
1904 | { |
1905 | #ifdef BOOST_MSVC |
1906 | #pragma warning(push) |
1907 | #pragma warning(disable:4127) |
1908 | #endif |
1909 | // |
1910 | // parse a \Q...\E sequence: |
1911 | // |
1912 | ++m_position; // skip the Q |
1913 | const charT* start = m_position; |
1914 | const charT* end; |
1915 | do |
1916 | { |
1917 | while((m_position != m_end) |
1918 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape)) |
1919 | ++m_position; |
1920 | if(m_position == m_end) |
1921 | { |
1922 | // a \Q...\E sequence may terminate with the end of the expression: |
1923 | end = m_position; |
1924 | break; |
1925 | } |
1926 | if(++m_position == m_end) // skip the escape |
1927 | { |
1928 | fail(regex_constants::error_escape, m_position - m_base, "Unterminated \\Q...\\E sequence." ); |
1929 | return false; |
1930 | } |
1931 | // check to see if it's a \E: |
1932 | if(this->m_traits.escape_syntax_type(*m_position) == regex_constants::escape_type_E) |
1933 | { |
1934 | ++m_position; |
1935 | end = m_position - 2; |
1936 | break; |
1937 | } |
1938 | // otherwise go round again: |
1939 | }while(true); |
1940 | // |
1941 | // now add all the character between the two escapes as literals: |
1942 | // |
1943 | while(start != end) |
1944 | { |
1945 | this->append_literal(*start); |
1946 | ++start; |
1947 | } |
1948 | return true; |
1949 | #ifdef BOOST_MSVC |
1950 | #pragma warning(pop) |
1951 | #endif |
1952 | } |
1953 | |
1954 | template <class charT, class traits> |
1955 | bool basic_regex_parser<charT, traits>::parse_perl_extension() |
1956 | { |
1957 | if(++m_position == m_end) |
1958 | { |
1959 | // Rewind to start of (? sequence: |
1960 | --m_position; |
1961 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
1962 | fail(regex_constants::error_perl_extension, m_position - m_base); |
1963 | return false; |
1964 | } |
1965 | // |
1966 | // treat comments as a special case, as these |
1967 | // are the only ones that don't start with a leading |
1968 | // startmark state: |
1969 | // |
1970 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_hash) |
1971 | { |
1972 | while((m_position != m_end) |
1973 | && (this->m_traits.syntax_type(*m_position++) != regex_constants::syntax_close_mark)) |
1974 | {} |
1975 | return true; |
1976 | } |
1977 | // |
1978 | // backup some state, and prepare the way: |
1979 | // |
1980 | int markid = 0; |
1981 | std::ptrdiff_t jump_offset = 0; |
1982 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); |
1983 | pb->icase = this->flags() & regbase::icase; |
1984 | std::ptrdiff_t last_paren_start = this->getoffset(pb); |
1985 | // back up insertion point for alternations, and set new point: |
1986 | std::ptrdiff_t last_alt_point = m_alt_insert_point; |
1987 | this->m_pdata->m_data.align(); |
1988 | m_alt_insert_point = this->m_pdata->m_data.size(); |
1989 | std::ptrdiff_t expected_alt_point = m_alt_insert_point; |
1990 | bool restore_flags = true; |
1991 | regex_constants::syntax_option_type old_flags = this->flags(); |
1992 | bool old_case_change = m_has_case_change; |
1993 | m_has_case_change = false; |
1994 | charT name_delim; |
1995 | int mark_reset = m_mark_reset; |
1996 | int max_mark = m_max_mark; |
1997 | m_mark_reset = -1; |
1998 | m_max_mark = m_mark_count; |
1999 | int v; |
2000 | // |
2001 | // select the actual extension used: |
2002 | // |
2003 | switch(this->m_traits.syntax_type(*m_position)) |
2004 | { |
2005 | case regex_constants::syntax_or: |
2006 | m_mark_reset = m_mark_count; |
2007 | BOOST_FALLTHROUGH; |
2008 | case regex_constants::syntax_colon: |
2009 | // |
2010 | // a non-capturing mark: |
2011 | // |
2012 | pb->index = markid = 0; |
2013 | ++m_position; |
2014 | break; |
2015 | case regex_constants::syntax_digit: |
2016 | { |
2017 | // |
2018 | // a recursive subexpression: |
2019 | // |
2020 | v = this->m_traits.toi(m_position, m_end, 10); |
2021 | if((v < 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2022 | { |
2023 | // Rewind to start of (? sequence: |
2024 | --m_position; |
2025 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2026 | fail(regex_constants::error_perl_extension, m_position - m_base, "The recursive sub-expression refers to an invalid marking group, or is unterminated." ); |
2027 | return false; |
2028 | } |
2029 | insert_recursion: |
2030 | pb->index = markid = 0; |
2031 | re_recurse* pr = static_cast<re_recurse*>(this->append_state(syntax_element_recurse, sizeof(re_recurse))); |
2032 | pr->alt.i = v; |
2033 | pr->state_id = 0; |
2034 | static_cast<re_case*>( |
2035 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) |
2036 | )->icase = this->flags() & regbase::icase; |
2037 | break; |
2038 | } |
2039 | case regex_constants::syntax_plus: |
2040 | // |
2041 | // A forward-relative recursive subexpression: |
2042 | // |
2043 | ++m_position; |
2044 | v = this->m_traits.toi(m_position, m_end, 10); |
2045 | if((v <= 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2046 | { |
2047 | // Rewind to start of (? sequence: |
2048 | --m_position; |
2049 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2050 | fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression." ); |
2051 | return false; |
2052 | } |
2053 | v += m_mark_count; |
2054 | goto insert_recursion; |
2055 | case regex_constants::syntax_dash: |
2056 | // |
2057 | // Possibly a backward-relative recursive subexpression: |
2058 | // |
2059 | ++m_position; |
2060 | v = this->m_traits.toi(m_position, m_end, 10); |
2061 | if(v <= 0) |
2062 | { |
2063 | --m_position; |
2064 | // Oops not a relative recursion at all, but a (?-imsx) group: |
2065 | goto option_group_jump; |
2066 | } |
2067 | v = m_mark_count + 1 - v; |
2068 | if(v <= 0) |
2069 | { |
2070 | // Rewind to start of (? sequence: |
2071 | --m_position; |
2072 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2073 | fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression." ); |
2074 | return false; |
2075 | } |
2076 | goto insert_recursion; |
2077 | case regex_constants::syntax_equal: |
2078 | pb->index = markid = -1; |
2079 | ++m_position; |
2080 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); |
2081 | this->m_pdata->m_data.align(); |
2082 | m_alt_insert_point = this->m_pdata->m_data.size(); |
2083 | break; |
2084 | case regex_constants::syntax_not: |
2085 | pb->index = markid = -2; |
2086 | ++m_position; |
2087 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); |
2088 | this->m_pdata->m_data.align(); |
2089 | m_alt_insert_point = this->m_pdata->m_data.size(); |
2090 | break; |
2091 | case regex_constants::escape_type_left_word: |
2092 | { |
2093 | // a lookbehind assertion: |
2094 | if(++m_position == m_end) |
2095 | { |
2096 | // Rewind to start of (? sequence: |
2097 | --m_position; |
2098 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2099 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2100 | return false; |
2101 | } |
2102 | regex_constants::syntax_type t = this->m_traits.syntax_type(*m_position); |
2103 | if(t == regex_constants::syntax_not) |
2104 | pb->index = markid = -2; |
2105 | else if(t == regex_constants::syntax_equal) |
2106 | pb->index = markid = -1; |
2107 | else |
2108 | { |
2109 | // Probably a named capture which also starts (?< : |
2110 | name_delim = '>'; |
2111 | --m_position; |
2112 | goto named_capture_jump; |
2113 | } |
2114 | ++m_position; |
2115 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); |
2116 | this->append_state(syntax_element_backstep, sizeof(re_brace)); |
2117 | this->m_pdata->m_data.align(); |
2118 | m_alt_insert_point = this->m_pdata->m_data.size(); |
2119 | break; |
2120 | } |
2121 | case regex_constants::escape_type_right_word: |
2122 | // |
2123 | // an independent sub-expression: |
2124 | // |
2125 | pb->index = markid = -3; |
2126 | ++m_position; |
2127 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); |
2128 | this->m_pdata->m_data.align(); |
2129 | m_alt_insert_point = this->m_pdata->m_data.size(); |
2130 | break; |
2131 | case regex_constants::syntax_open_mark: |
2132 | { |
2133 | // a conditional expression: |
2134 | pb->index = markid = -4; |
2135 | if(++m_position == m_end) |
2136 | { |
2137 | // Rewind to start of (? sequence: |
2138 | --m_position; |
2139 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2140 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2141 | return false; |
2142 | } |
2143 | v = this->m_traits.toi(m_position, m_end, 10); |
2144 | if(m_position == m_end) |
2145 | { |
2146 | // Rewind to start of (? sequence: |
2147 | --m_position; |
2148 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2149 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2150 | return false; |
2151 | } |
2152 | if(*m_position == charT('R')) |
2153 | { |
2154 | if(++m_position == m_end) |
2155 | { |
2156 | // Rewind to start of (? sequence: |
2157 | --m_position; |
2158 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2159 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2160 | return false; |
2161 | } |
2162 | if(*m_position == charT('&')) |
2163 | { |
2164 | const charT* base = ++m_position; |
2165 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2166 | ++m_position; |
2167 | if(m_position == m_end) |
2168 | { |
2169 | // Rewind to start of (? sequence: |
2170 | --m_position; |
2171 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2172 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2173 | return false; |
2174 | } |
2175 | v = -static_cast<int>(hash_value_from_capture_name(base, m_position)); |
2176 | } |
2177 | else |
2178 | { |
2179 | v = -this->m_traits.toi(m_position, m_end, 10); |
2180 | } |
2181 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); |
2182 | br->index = v < 0 ? (v - 1) : 0; |
2183 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) |
2184 | { |
2185 | // Rewind to start of (? sequence: |
2186 | --m_position; |
2187 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2188 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2189 | return false; |
2190 | } |
2191 | if(++m_position == m_end) |
2192 | { |
2193 | // Rewind to start of (? sequence: |
2194 | --m_position; |
2195 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2196 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2197 | return false; |
2198 | } |
2199 | } |
2200 | else if((*m_position == charT('\'')) || (*m_position == charT('<'))) |
2201 | { |
2202 | const charT* base = ++m_position; |
2203 | while((m_position != m_end) && (*m_position != charT('>')) && (*m_position != charT('\''))) |
2204 | ++m_position; |
2205 | if(m_position == m_end) |
2206 | { |
2207 | // Rewind to start of (? sequence: |
2208 | --m_position; |
2209 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2210 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2211 | return false; |
2212 | } |
2213 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); |
2214 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); |
2215 | br->index = v; |
2216 | if(((*m_position != charT('>')) && (*m_position != charT('\''))) || (++m_position == m_end)) |
2217 | { |
2218 | // Rewind to start of (? sequence: |
2219 | --m_position; |
2220 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2221 | fail(regex_constants::error_perl_extension, m_position - m_base, "Unterminated named capture." ); |
2222 | return false; |
2223 | } |
2224 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) |
2225 | { |
2226 | // Rewind to start of (? sequence: |
2227 | --m_position; |
2228 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2229 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2230 | return false; |
2231 | } |
2232 | if(++m_position == m_end) |
2233 | { |
2234 | // Rewind to start of (? sequence: |
2235 | --m_position; |
2236 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2237 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2238 | return false; |
2239 | } |
2240 | } |
2241 | else if(*m_position == charT('D')) |
2242 | { |
2243 | const char* def = "DEFINE" ; |
2244 | while(*def && (m_position != m_end) && (*m_position == charT(*def))) |
2245 | ++m_position, ++def; |
2246 | if((m_position == m_end) || *def) |
2247 | { |
2248 | // Rewind to start of (? sequence: |
2249 | --m_position; |
2250 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2251 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2252 | return false; |
2253 | } |
2254 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); |
2255 | br->index = 9999; // special magic value! |
2256 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) |
2257 | { |
2258 | // Rewind to start of (? sequence: |
2259 | --m_position; |
2260 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2261 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2262 | return false; |
2263 | } |
2264 | if(++m_position == m_end) |
2265 | { |
2266 | // Rewind to start of (? sequence: |
2267 | --m_position; |
2268 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2269 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2270 | return false; |
2271 | } |
2272 | } |
2273 | else if(v > 0) |
2274 | { |
2275 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); |
2276 | br->index = v; |
2277 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) |
2278 | { |
2279 | // Rewind to start of (? sequence: |
2280 | --m_position; |
2281 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2282 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2283 | return false; |
2284 | } |
2285 | if(++m_position == m_end) |
2286 | { |
2287 | // Rewind to start of (? sequence: |
2288 | --m_position; |
2289 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2290 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2291 | return false; |
2292 | } |
2293 | } |
2294 | else |
2295 | { |
2296 | // verify that we have a lookahead or lookbehind assert: |
2297 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_question) |
2298 | { |
2299 | // Rewind to start of (? sequence: |
2300 | --m_position; |
2301 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2302 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2303 | return false; |
2304 | } |
2305 | if(++m_position == m_end) |
2306 | { |
2307 | // Rewind to start of (? sequence: |
2308 | --m_position; |
2309 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2310 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2311 | return false; |
2312 | } |
2313 | if(this->m_traits.syntax_type(*m_position) == regex_constants::escape_type_left_word) |
2314 | { |
2315 | if(++m_position == m_end) |
2316 | { |
2317 | // Rewind to start of (? sequence: |
2318 | --m_position; |
2319 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2320 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2321 | return false; |
2322 | } |
2323 | if((this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal) |
2324 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_not)) |
2325 | { |
2326 | // Rewind to start of (? sequence: |
2327 | --m_position; |
2328 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2329 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2330 | return false; |
2331 | } |
2332 | m_position -= 3; |
2333 | } |
2334 | else |
2335 | { |
2336 | if((this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal) |
2337 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_not)) |
2338 | { |
2339 | // Rewind to start of (? sequence: |
2340 | --m_position; |
2341 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2342 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2343 | return false; |
2344 | } |
2345 | m_position -= 2; |
2346 | } |
2347 | } |
2348 | break; |
2349 | } |
2350 | case regex_constants::syntax_close_mark: |
2351 | // Rewind to start of (? sequence: |
2352 | --m_position; |
2353 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2354 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2355 | return false; |
2356 | case regex_constants::escape_type_end_buffer: |
2357 | { |
2358 | name_delim = *m_position; |
2359 | named_capture_jump: |
2360 | markid = 0; |
2361 | if(0 == (this->flags() & regbase::nosubs)) |
2362 | { |
2363 | markid = ++m_mark_count; |
2364 | #ifndef BOOST_NO_STD_DISTANCE |
2365 | if(this->flags() & regbase::save_subexpression_location) |
2366 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>(std::distance(m_base, m_position) - 2, 0)); |
2367 | #else |
2368 | if(this->flags() & regbase::save_subexpression_location) |
2369 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>((m_position - m_base) - 2, 0)); |
2370 | #endif |
2371 | } |
2372 | pb->index = markid; |
2373 | const charT* base = ++m_position; |
2374 | if(m_position == m_end) |
2375 | { |
2376 | // Rewind to start of (? sequence: |
2377 | --m_position; |
2378 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2379 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2380 | return false; |
2381 | } |
2382 | while((m_position != m_end) && (*m_position != name_delim)) |
2383 | ++m_position; |
2384 | if(m_position == m_end) |
2385 | { |
2386 | // Rewind to start of (? sequence: |
2387 | --m_position; |
2388 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2389 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2390 | return false; |
2391 | } |
2392 | this->m_pdata->set_name(base, m_position, markid); |
2393 | ++m_position; |
2394 | break; |
2395 | } |
2396 | default: |
2397 | if(*m_position == charT('R')) |
2398 | { |
2399 | ++m_position; |
2400 | v = 0; |
2401 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) |
2402 | { |
2403 | // Rewind to start of (? sequence: |
2404 | --m_position; |
2405 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2406 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2407 | return false; |
2408 | } |
2409 | goto insert_recursion; |
2410 | } |
2411 | if(*m_position == charT('&')) |
2412 | { |
2413 | ++m_position; |
2414 | const charT* base = m_position; |
2415 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2416 | ++m_position; |
2417 | if(m_position == m_end) |
2418 | { |
2419 | // Rewind to start of (? sequence: |
2420 | --m_position; |
2421 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2422 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2423 | return false; |
2424 | } |
2425 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); |
2426 | goto insert_recursion; |
2427 | } |
2428 | if(*m_position == charT('P')) |
2429 | { |
2430 | ++m_position; |
2431 | if(m_position == m_end) |
2432 | { |
2433 | // Rewind to start of (? sequence: |
2434 | --m_position; |
2435 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2436 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2437 | return false; |
2438 | } |
2439 | if(*m_position == charT('>')) |
2440 | { |
2441 | ++m_position; |
2442 | const charT* base = m_position; |
2443 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2444 | ++m_position; |
2445 | if(m_position == m_end) |
2446 | { |
2447 | // Rewind to start of (? sequence: |
2448 | --m_position; |
2449 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2450 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2451 | return false; |
2452 | } |
2453 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); |
2454 | goto insert_recursion; |
2455 | } |
2456 | } |
2457 | // |
2458 | // lets assume that we have a (?imsx) group and try and parse it: |
2459 | // |
2460 | option_group_jump: |
2461 | regex_constants::syntax_option_type opts = parse_options(); |
2462 | if(m_position == m_end) |
2463 | { |
2464 | // Rewind to start of (? sequence: |
2465 | --m_position; |
2466 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2467 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2468 | return false; |
2469 | } |
2470 | // make a note of whether we have a case change: |
2471 | m_has_case_change = ((opts & regbase::icase) != (this->flags() & regbase::icase)); |
2472 | pb->index = markid = 0; |
2473 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark) |
2474 | { |
2475 | // update flags and carry on as normal: |
2476 | this->flags(opts); |
2477 | restore_flags = false; |
2478 | old_case_change |= m_has_case_change; // defer end of scope by one ')' |
2479 | } |
2480 | else if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_colon) |
2481 | { |
2482 | // update flags and carry on until the matching ')' is found: |
2483 | this->flags(opts); |
2484 | ++m_position; |
2485 | } |
2486 | else |
2487 | { |
2488 | // Rewind to start of (? sequence: |
2489 | --m_position; |
2490 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2491 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2492 | return false; |
2493 | } |
2494 | |
2495 | // finally append a case change state if we need it: |
2496 | if(m_has_case_change) |
2497 | { |
2498 | static_cast<re_case*>( |
2499 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) |
2500 | )->icase = opts & regbase::icase; |
2501 | } |
2502 | |
2503 | } |
2504 | // |
2505 | // now recursively add more states, this will terminate when we get to a |
2506 | // matching ')' : |
2507 | // |
2508 | parse_all(); |
2509 | // |
2510 | // Unwind alternatives: |
2511 | // |
2512 | if(0 == unwind_alts(last_paren_start)) |
2513 | { |
2514 | // Rewind to start of (? sequence: |
2515 | --m_position; |
2516 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2517 | fail(regex_constants::error_perl_extension, m_position - m_base, "Invalid alternation operators within (?...) block." ); |
2518 | return false; |
2519 | } |
2520 | // |
2521 | // we either have a ')' or we have run out of characters prematurely: |
2522 | // |
2523 | if(m_position == m_end) |
2524 | { |
2525 | // Rewind to start of (? sequence: |
2526 | --m_position; |
2527 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2528 | this->fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_end)); |
2529 | return false; |
2530 | } |
2531 | BOOST_ASSERT(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark); |
2532 | ++m_position; |
2533 | // |
2534 | // restore the flags: |
2535 | // |
2536 | if(restore_flags) |
2537 | { |
2538 | // append a case change state if we need it: |
2539 | if(m_has_case_change) |
2540 | { |
2541 | static_cast<re_case*>( |
2542 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) |
2543 | )->icase = old_flags & regbase::icase; |
2544 | } |
2545 | this->flags(old_flags); |
2546 | } |
2547 | // |
2548 | // set up the jump pointer if we have one: |
2549 | // |
2550 | if(jump_offset) |
2551 | { |
2552 | this->m_pdata->m_data.align(); |
2553 | re_jump* jmp = static_cast<re_jump*>(this->getaddress(jump_offset)); |
2554 | jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp); |
2555 | if((this->m_last_state == jmp) && (markid != -2)) |
2556 | { |
2557 | // Oops... we didn't have anything inside the assertion. |
2558 | // Note we don't get here for negated forward lookahead as (?!) |
2559 | // does have some uses. |
2560 | // Rewind to start of (? sequence: |
2561 | --m_position; |
2562 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2563 | fail(regex_constants::error_perl_extension, m_position - m_base, "Invalid or empty zero width assertion." ); |
2564 | return false; |
2565 | } |
2566 | } |
2567 | // |
2568 | // verify that if this is conditional expression, that we do have |
2569 | // an alternative, if not add one: |
2570 | // |
2571 | if(markid == -4) |
2572 | { |
2573 | re_syntax_base* b = this->getaddress(expected_alt_point); |
2574 | // Make sure we have exactly one alternative following this state: |
2575 | if(b->type != syntax_element_alt) |
2576 | { |
2577 | re_alt* alt = static_cast<re_alt*>(this->insert_state(expected_alt_point, syntax_element_alt, sizeof(re_alt))); |
2578 | alt->alt.i = this->m_pdata->m_data.size() - this->getoffset(alt); |
2579 | } |
2580 | else if(this->getaddress(static_cast<re_alt*>(b)->alt.i, b)->type == syntax_element_alt) |
2581 | { |
2582 | // Can't have seen more than one alternative: |
2583 | // Rewind to start of (? sequence: |
2584 | --m_position; |
2585 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2586 | fail(regex_constants::error_bad_pattern, m_position - m_base, "More than one alternation operator | was encountered inside a conditional expression." ); |
2587 | return false; |
2588 | } |
2589 | else |
2590 | { |
2591 | // We must *not* have seen an alternative inside a (DEFINE) block: |
2592 | b = this->getaddress(b->next.i, b); |
2593 | if((b->type == syntax_element_assert_backref) && (static_cast<re_brace*>(b)->index == 9999)) |
2594 | { |
2595 | // Rewind to start of (? sequence: |
2596 | --m_position; |
2597 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2598 | fail(regex_constants::error_bad_pattern, m_position - m_base, "Alternation operators are not allowed inside a DEFINE block." ); |
2599 | return false; |
2600 | } |
2601 | } |
2602 | // check for invalid repetition of next state: |
2603 | b = this->getaddress(expected_alt_point); |
2604 | b = this->getaddress(static_cast<re_alt*>(b)->next.i, b); |
2605 | if((b->type != syntax_element_assert_backref) |
2606 | && (b->type != syntax_element_startmark)) |
2607 | { |
2608 | // Rewind to start of (? sequence: |
2609 | --m_position; |
2610 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2611 | fail(regex_constants::error_badrepeat, m_position - m_base, "A repetition operator cannot be applied to a zero-width assertion." ); |
2612 | return false; |
2613 | } |
2614 | } |
2615 | // |
2616 | // append closing parenthesis state: |
2617 | // |
2618 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); |
2619 | pb->index = markid; |
2620 | pb->icase = this->flags() & regbase::icase; |
2621 | this->m_paren_start = last_paren_start; |
2622 | // |
2623 | // restore the alternate insertion point: |
2624 | // |
2625 | this->m_alt_insert_point = last_alt_point; |
2626 | // |
2627 | // and the case change data: |
2628 | // |
2629 | m_has_case_change = old_case_change; |
2630 | // |
2631 | // And the mark_reset data: |
2632 | // |
2633 | if(m_max_mark > m_mark_count) |
2634 | { |
2635 | m_mark_count = m_max_mark; |
2636 | } |
2637 | m_mark_reset = mark_reset; |
2638 | m_max_mark = max_mark; |
2639 | |
2640 | |
2641 | if(markid > 0) |
2642 | { |
2643 | #ifndef BOOST_NO_STD_DISTANCE |
2644 | if(this->flags() & regbase::save_subexpression_location) |
2645 | this->m_pdata->m_subs.at(markid - 1).second = std::distance(m_base, m_position) - 1; |
2646 | #else |
2647 | if(this->flags() & regbase::save_subexpression_location) |
2648 | this->m_pdata->m_subs.at(markid - 1).second = (m_position - m_base) - 1; |
2649 | #endif |
2650 | // |
2651 | // allow backrefs to this mark: |
2652 | // |
2653 | if((markid > 0) && (markid < (int)(sizeof(unsigned) * CHAR_BIT))) |
2654 | this->m_backrefs |= 1u << (markid - 1); |
2655 | } |
2656 | return true; |
2657 | } |
2658 | |
2659 | template <class charT, class traits> |
2660 | bool basic_regex_parser<charT, traits>::match_verb(const char* verb) |
2661 | { |
2662 | while(*verb) |
2663 | { |
2664 | if(static_cast<charT>(*verb) != *m_position) |
2665 | { |
2666 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2667 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2668 | return false; |
2669 | } |
2670 | if(++m_position == m_end) |
2671 | { |
2672 | --m_position; |
2673 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2674 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2675 | return false; |
2676 | } |
2677 | ++verb; |
2678 | } |
2679 | return true; |
2680 | } |
2681 | |
2682 | template <class charT, class traits> |
2683 | bool basic_regex_parser<charT, traits>::parse_perl_verb() |
2684 | { |
2685 | if(++m_position == m_end) |
2686 | { |
2687 | // Rewind to start of (* sequence: |
2688 | --m_position; |
2689 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2690 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2691 | return false; |
2692 | } |
2693 | switch(*m_position) |
2694 | { |
2695 | case 'F': |
2696 | if(++m_position == m_end) |
2697 | { |
2698 | // Rewind to start of (* sequence: |
2699 | --m_position; |
2700 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2701 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2702 | return false; |
2703 | } |
2704 | if((this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark) || match_verb(verb: "AIL" )) |
2705 | { |
2706 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2707 | { |
2708 | // Rewind to start of (* sequence: |
2709 | --m_position; |
2710 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2711 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2712 | return false; |
2713 | } |
2714 | ++m_position; |
2715 | this->append_state(syntax_element_fail); |
2716 | return true; |
2717 | } |
2718 | break; |
2719 | case 'A': |
2720 | if(++m_position == m_end) |
2721 | { |
2722 | // Rewind to start of (* sequence: |
2723 | --m_position; |
2724 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2725 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2726 | return false; |
2727 | } |
2728 | if(match_verb(verb: "CCEPT" )) |
2729 | { |
2730 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2731 | { |
2732 | // Rewind to start of (* sequence: |
2733 | --m_position; |
2734 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2735 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2736 | return false; |
2737 | } |
2738 | ++m_position; |
2739 | this->append_state(syntax_element_accept); |
2740 | return true; |
2741 | } |
2742 | break; |
2743 | case 'C': |
2744 | if(++m_position == m_end) |
2745 | { |
2746 | // Rewind to start of (* sequence: |
2747 | --m_position; |
2748 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2749 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2750 | return false; |
2751 | } |
2752 | if(match_verb(verb: "OMMIT" )) |
2753 | { |
2754 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2755 | { |
2756 | // Rewind to start of (* sequence: |
2757 | --m_position; |
2758 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2759 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2760 | return false; |
2761 | } |
2762 | ++m_position; |
2763 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_commit; |
2764 | this->m_pdata->m_disable_match_any = true; |
2765 | return true; |
2766 | } |
2767 | break; |
2768 | case 'P': |
2769 | if(++m_position == m_end) |
2770 | { |
2771 | // Rewind to start of (* sequence: |
2772 | --m_position; |
2773 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2774 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2775 | return false; |
2776 | } |
2777 | if(match_verb(verb: "RUNE" )) |
2778 | { |
2779 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2780 | { |
2781 | // Rewind to start of (* sequence: |
2782 | --m_position; |
2783 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2784 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2785 | return false; |
2786 | } |
2787 | ++m_position; |
2788 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_prune; |
2789 | this->m_pdata->m_disable_match_any = true; |
2790 | return true; |
2791 | } |
2792 | break; |
2793 | case 'S': |
2794 | if(++m_position == m_end) |
2795 | { |
2796 | // Rewind to start of (* sequence: |
2797 | --m_position; |
2798 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2799 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2800 | return false; |
2801 | } |
2802 | if(match_verb(verb: "KIP" )) |
2803 | { |
2804 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2805 | { |
2806 | // Rewind to start of (* sequence: |
2807 | --m_position; |
2808 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2809 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2810 | return false; |
2811 | } |
2812 | ++m_position; |
2813 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_skip; |
2814 | this->m_pdata->m_disable_match_any = true; |
2815 | return true; |
2816 | } |
2817 | break; |
2818 | case 'T': |
2819 | if(++m_position == m_end) |
2820 | { |
2821 | // Rewind to start of (* sequence: |
2822 | --m_position; |
2823 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2824 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2825 | return false; |
2826 | } |
2827 | if(match_verb(verb: "HEN" )) |
2828 | { |
2829 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) |
2830 | { |
2831 | // Rewind to start of (* sequence: |
2832 | --m_position; |
2833 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2834 | fail(regex_constants::error_perl_extension, m_position - m_base); |
2835 | return false; |
2836 | } |
2837 | ++m_position; |
2838 | this->append_state(syntax_element_then); |
2839 | this->m_pdata->m_disable_match_any = true; |
2840 | return true; |
2841 | } |
2842 | break; |
2843 | } |
2844 | return false; |
2845 | } |
2846 | |
2847 | template <class charT, class traits> |
2848 | bool basic_regex_parser<charT, traits>::add_emacs_code(bool negate) |
2849 | { |
2850 | // |
2851 | // parses an emacs style \sx or \Sx construct. |
2852 | // |
2853 | if(++m_position == m_end) |
2854 | { |
2855 | // Rewind to start of sequence: |
2856 | --m_position; |
2857 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; |
2858 | fail(regex_constants::error_escape, m_position - m_base); |
2859 | return false; |
2860 | } |
2861 | basic_char_set<charT, traits> char_set; |
2862 | if(negate) |
2863 | char_set.negate(); |
2864 | |
2865 | static const charT s_punct[5] = { 'p', 'u', 'n', 'c', 't', }; |
2866 | |
2867 | switch(*m_position) |
2868 | { |
2869 | case 's': |
2870 | case ' ': |
2871 | char_set.add_class(this->m_mask_space); |
2872 | break; |
2873 | case 'w': |
2874 | char_set.add_class(this->m_word_mask); |
2875 | break; |
2876 | case '_': |
2877 | char_set.add_single(digraph<charT>(charT('$'))); |
2878 | char_set.add_single(digraph<charT>(charT('&'))); |
2879 | char_set.add_single(digraph<charT>(charT('*'))); |
2880 | char_set.add_single(digraph<charT>(charT('+'))); |
2881 | char_set.add_single(digraph<charT>(charT('-'))); |
2882 | char_set.add_single(digraph<charT>(charT('_'))); |
2883 | char_set.add_single(digraph<charT>(charT('<'))); |
2884 | char_set.add_single(digraph<charT>(charT('>'))); |
2885 | break; |
2886 | case '.': |
2887 | char_set.add_class(this->m_traits.lookup_classname(s_punct, s_punct+5)); |
2888 | break; |
2889 | case '(': |
2890 | char_set.add_single(digraph<charT>(charT('('))); |
2891 | char_set.add_single(digraph<charT>(charT('['))); |
2892 | char_set.add_single(digraph<charT>(charT('{'))); |
2893 | break; |
2894 | case ')': |
2895 | char_set.add_single(digraph<charT>(charT(')'))); |
2896 | char_set.add_single(digraph<charT>(charT(']'))); |
2897 | char_set.add_single(digraph<charT>(charT('}'))); |
2898 | break; |
2899 | case '"': |
2900 | char_set.add_single(digraph<charT>(charT('"'))); |
2901 | char_set.add_single(digraph<charT>(charT('\''))); |
2902 | char_set.add_single(digraph<charT>(charT('`'))); |
2903 | break; |
2904 | case '\'': |
2905 | char_set.add_single(digraph<charT>(charT('\''))); |
2906 | char_set.add_single(digraph<charT>(charT(','))); |
2907 | char_set.add_single(digraph<charT>(charT('#'))); |
2908 | break; |
2909 | case '<': |
2910 | char_set.add_single(digraph<charT>(charT(';'))); |
2911 | break; |
2912 | case '>': |
2913 | char_set.add_single(digraph<charT>(charT('\n'))); |
2914 | char_set.add_single(digraph<charT>(charT('\f'))); |
2915 | break; |
2916 | default: |
2917 | fail(regex_constants::error_ctype, m_position - m_base); |
2918 | return false; |
2919 | } |
2920 | if(0 == this->append_set(char_set)) |
2921 | { |
2922 | fail(regex_constants::error_ctype, m_position - m_base); |
2923 | return false; |
2924 | } |
2925 | ++m_position; |
2926 | return true; |
2927 | } |
2928 | |
2929 | template <class charT, class traits> |
2930 | regex_constants::syntax_option_type basic_regex_parser<charT, traits>::parse_options() |
2931 | { |
2932 | // we have a (?imsx-imsx) group, convert it into a set of flags: |
2933 | regex_constants::syntax_option_type f = this->flags(); |
2934 | bool breakout = false; |
2935 | do |
2936 | { |
2937 | switch(*m_position) |
2938 | { |
2939 | case 's': |
2940 | f |= regex_constants::mod_s; |
2941 | f &= ~regex_constants::no_mod_s; |
2942 | break; |
2943 | case 'm': |
2944 | f &= ~regex_constants::no_mod_m; |
2945 | break; |
2946 | case 'i': |
2947 | f |= regex_constants::icase; |
2948 | break; |
2949 | case 'x': |
2950 | f |= regex_constants::mod_x; |
2951 | break; |
2952 | default: |
2953 | breakout = true; |
2954 | continue; |
2955 | } |
2956 | if(++m_position == m_end) |
2957 | { |
2958 | // Rewind to start of (? sequence: |
2959 | --m_position; |
2960 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2961 | fail(regex_constants::error_paren, m_position - m_base); |
2962 | return false; |
2963 | } |
2964 | } |
2965 | while(!breakout); |
2966 | |
2967 | breakout = false; |
2968 | |
2969 | if(*m_position == static_cast<charT>('-')) |
2970 | { |
2971 | if(++m_position == m_end) |
2972 | { |
2973 | // Rewind to start of (? sequence: |
2974 | --m_position; |
2975 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
2976 | fail(regex_constants::error_paren, m_position - m_base); |
2977 | return false; |
2978 | } |
2979 | do |
2980 | { |
2981 | switch(*m_position) |
2982 | { |
2983 | case 's': |
2984 | f &= ~regex_constants::mod_s; |
2985 | f |= regex_constants::no_mod_s; |
2986 | break; |
2987 | case 'm': |
2988 | f |= regex_constants::no_mod_m; |
2989 | break; |
2990 | case 'i': |
2991 | f &= ~regex_constants::icase; |
2992 | break; |
2993 | case 'x': |
2994 | f &= ~regex_constants::mod_x; |
2995 | break; |
2996 | default: |
2997 | breakout = true; |
2998 | continue; |
2999 | } |
3000 | if(++m_position == m_end) |
3001 | { |
3002 | // Rewind to start of (? sequence: |
3003 | --m_position; |
3004 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; |
3005 | fail(regex_constants::error_paren, m_position - m_base); |
3006 | return false; |
3007 | } |
3008 | } |
3009 | while(!breakout); |
3010 | } |
3011 | return f; |
3012 | } |
3013 | |
3014 | template <class charT, class traits> |
3015 | bool basic_regex_parser<charT, traits>::unwind_alts(std::ptrdiff_t last_paren_start) |
3016 | { |
3017 | // |
3018 | // If we didn't actually add any states after the last |
3019 | // alternative then that's an error: |
3020 | // |
3021 | if((this->m_alt_insert_point == static_cast<std::ptrdiff_t>(this->m_pdata->m_data.size())) |
3022 | && m_alt_jumps.size() && (m_alt_jumps.back() > last_paren_start) |
3023 | && |
3024 | !( |
3025 | ((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) |
3026 | && |
3027 | ((this->flags() & regbase::no_empty_expressions) == 0) |
3028 | ) |
3029 | ) |
3030 | { |
3031 | fail(regex_constants::error_empty, this->m_position - this->m_base, "Can't terminate a sub-expression with an alternation operator |." ); |
3032 | return false; |
3033 | } |
3034 | // |
3035 | // Fix up our alternatives: |
3036 | // |
3037 | while(m_alt_jumps.size() && (m_alt_jumps.back() > last_paren_start)) |
3038 | { |
3039 | // |
3040 | // fix up the jump to point to the end of the states |
3041 | // that we've just added: |
3042 | // |
3043 | std::ptrdiff_t jump_offset = m_alt_jumps.back(); |
3044 | m_alt_jumps.pop_back(); |
3045 | this->m_pdata->m_data.align(); |
3046 | re_jump* jmp = static_cast<re_jump*>(this->getaddress(jump_offset)); |
3047 | BOOST_ASSERT(jmp->type == syntax_element_jump); |
3048 | jmp->alt.i = this->m_pdata->m_data.size() - jump_offset; |
3049 | } |
3050 | return true; |
3051 | } |
3052 | |
3053 | #ifdef BOOST_MSVC |
3054 | #pragma warning(pop) |
3055 | #endif |
3056 | |
3057 | } // namespace BOOST_REGEX_DETAIL_NS |
3058 | } // namespace boost |
3059 | |
3060 | #ifdef BOOST_MSVC |
3061 | #pragma warning(push) |
3062 | #pragma warning(disable: 4103) |
3063 | #endif |
3064 | #ifdef BOOST_HAS_ABI_HEADERS |
3065 | # include BOOST_ABI_SUFFIX |
3066 | #endif |
3067 | #ifdef BOOST_MSVC |
3068 | #pragma warning(pop) |
3069 | #endif |
3070 | |
3071 | #endif |
3072 | |