]>
Commit | Line | Data |
---|---|---|
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 | inline boost::intmax_t umax(mpl::false_ const&) | |
42 | { | |
43 | // Get out clause here, just in case numeric_limits is unspecialized: | |
44 | return std::numeric_limits<boost::intmax_t>::is_specialized ? (std::numeric_limits<boost::intmax_t>::max)() : INT_MAX; | |
45 | } | |
46 | inline boost::intmax_t umax(mpl::true_ const&) | |
47 | { | |
48 | return (std::numeric_limits<std::size_t>::max)(); | |
49 | } | |
50 | ||
51 | inline boost::intmax_t umax() | |
52 | { | |
53 | return umax(mpl::bool_<std::numeric_limits<boost::intmax_t>::digits >= std::numeric_limits<std::size_t>::digits>()); | |
54 | } | |
55 | ||
56 | template <class charT, class traits> | |
57 | class basic_regex_parser : public basic_regex_creator<charT, traits> | |
58 | { | |
59 | public: | |
60 | basic_regex_parser(regex_data<charT, traits>* data); | |
61 | void parse(const charT* p1, const charT* p2, unsigned flags); | |
62 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position); | |
63 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position, std::string message, std::ptrdiff_t start_pos); | |
64 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position, const std::string& message) | |
65 | { | |
66 | fail(error_code, position, message, position); | |
67 | } | |
68 | ||
69 | bool parse_all(); | |
70 | bool parse_basic(); | |
71 | bool parse_extended(); | |
72 | bool parse_literal(); | |
73 | bool parse_open_paren(); | |
74 | bool parse_basic_escape(); | |
75 | bool parse_extended_escape(); | |
76 | bool parse_match_any(); | |
77 | bool parse_repeat(std::size_t low = 0, std::size_t high = (std::numeric_limits<std::size_t>::max)()); | |
78 | bool parse_repeat_range(bool isbasic); | |
79 | bool parse_alt(); | |
80 | bool parse_set(); | |
81 | bool parse_backref(); | |
82 | void parse_set_literal(basic_char_set<charT, traits>& char_set); | |
83 | bool parse_inner_set(basic_char_set<charT, traits>& char_set); | |
84 | bool parse_QE(); | |
85 | bool parse_perl_extension(); | |
86 | bool parse_perl_verb(); | |
87 | bool match_verb(const char*); | |
88 | bool add_emacs_code(bool negate); | |
89 | bool unwind_alts(std::ptrdiff_t last_paren_start); | |
90 | digraph<charT> get_next_set_literal(basic_char_set<charT, traits>& char_set); | |
91 | charT unescape_character(); | |
92 | regex_constants::syntax_option_type parse_options(); | |
93 | ||
94 | private: | |
95 | typedef bool (basic_regex_parser::*parser_proc_type)(); | |
96 | typedef typename traits::string_type string_type; | |
97 | typedef typename traits::char_class_type char_class_type; | |
98 | parser_proc_type m_parser_proc; // the main parser to use | |
99 | const charT* m_base; // the start of the string being parsed | |
100 | const charT* m_end; // the end of the string being parsed | |
101 | const charT* m_position; // our current parser position | |
102 | unsigned m_mark_count; // how many sub-expressions we have | |
103 | int m_mark_reset; // used to indicate that we're inside a (?|...) block. | |
104 | unsigned m_max_mark; // largest mark count seen inside a (?|...) block. | |
105 | std::ptrdiff_t m_paren_start; // where the last seen ')' began (where repeats are inserted). | |
106 | std::ptrdiff_t m_alt_insert_point; // where to insert the next alternative | |
107 | bool m_has_case_change; // true if somewhere in the current block the case has changed | |
108 | #if defined(BOOST_MSVC) && defined(_M_IX86) | |
109 | // This is an ugly warning suppression workaround (for warnings *inside* std::vector | |
110 | // that can not otherwise be suppressed)... | |
111 | BOOST_STATIC_ASSERT(sizeof(long) >= sizeof(void*)); | |
112 | std::vector<long> m_alt_jumps; // list of alternative in the current scope. | |
113 | #else | |
114 | std::vector<std::ptrdiff_t> m_alt_jumps; // list of alternative in the current scope. | |
115 | #endif | |
116 | ||
117 | basic_regex_parser& operator=(const basic_regex_parser&); | |
118 | basic_regex_parser(const basic_regex_parser&); | |
119 | }; | |
120 | ||
121 | template <class charT, class traits> | |
122 | basic_regex_parser<charT, traits>::basic_regex_parser(regex_data<charT, traits>* data) | |
123 | : 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) | |
124 | { | |
125 | } | |
126 | ||
127 | template <class charT, class traits> | |
128 | void basic_regex_parser<charT, traits>::parse(const charT* p1, const charT* p2, unsigned l_flags) | |
129 | { | |
130 | // pass l_flags on to base class: | |
131 | this->init(l_flags); | |
132 | // set up pointers: | |
133 | m_position = m_base = p1; | |
134 | m_end = p2; | |
135 | // empty strings are errors: | |
136 | if((p1 == p2) && | |
137 | ( | |
138 | ((l_flags & regbase::main_option_type) != regbase::perl_syntax_group) | |
139 | || (l_flags & regbase::no_empty_expressions) | |
140 | ) | |
141 | ) | |
142 | { | |
143 | fail(regex_constants::error_empty, 0); | |
144 | return; | |
145 | } | |
146 | // select which parser to use: | |
147 | switch(l_flags & regbase::main_option_type) | |
148 | { | |
149 | case regbase::perl_syntax_group: | |
150 | { | |
151 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_extended; | |
152 | // | |
153 | // Add a leading paren with index zero to give recursions a target: | |
154 | // | |
155 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); | |
156 | br->index = 0; | |
157 | br->icase = this->flags() & regbase::icase; | |
158 | break; | |
159 | } | |
160 | case regbase::basic_syntax_group: | |
161 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_basic; | |
162 | break; | |
163 | case regbase::literal: | |
164 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_literal; | |
165 | break; | |
166 | default: | |
167 | // Ooops, someone has managed to set more than one of the main option flags, | |
168 | // so this must be an error: | |
169 | fail(regex_constants::error_unknown, 0, "An invalid combination of regular expression syntax flags was used."); | |
170 | return; | |
171 | } | |
172 | ||
173 | // parse all our characters: | |
174 | bool result = parse_all(); | |
175 | // | |
176 | // Unwind our alternatives: | |
177 | // | |
178 | unwind_alts(-1); | |
179 | // reset l_flags as a global scope (?imsx) may have altered them: | |
180 | this->flags(l_flags); | |
181 | // if we haven't gobbled up all the characters then we must | |
182 | // have had an unexpected ')' : | |
183 | if(!result) | |
184 | { | |
185 | fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_position), "Found a closing ) with no corresponding openening parenthesis."); | |
186 | return; | |
187 | } | |
188 | // if an error has been set then give up now: | |
189 | if(this->m_pdata->m_status) | |
190 | return; | |
191 | // fill in our sub-expression count: | |
192 | this->m_pdata->m_mark_count = 1 + m_mark_count; | |
193 | this->finalize(p1, p2); | |
194 | } | |
195 | ||
196 | template <class charT, class traits> | |
197 | void basic_regex_parser<charT, traits>::fail(regex_constants::error_type error_code, std::ptrdiff_t position) | |
198 | { | |
199 | // get the error message: | |
200 | std::string message = this->m_pdata->m_ptraits->error_string(error_code); | |
201 | fail(error_code, position, message); | |
202 | } | |
203 | ||
204 | template <class charT, class traits> | |
205 | 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) | |
206 | { | |
207 | if(0 == this->m_pdata->m_status) // update the error code if not already set | |
208 | this->m_pdata->m_status = error_code; | |
209 | m_position = m_end; // don't bother parsing anything else | |
210 | ||
211 | #ifndef BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS | |
212 | // | |
213 | // Augment error message with the regular expression text: | |
214 | // | |
215 | if(start_pos == position) | |
216 | start_pos = (std::max)(static_cast<std::ptrdiff_t>(0), position - static_cast<std::ptrdiff_t>(10)); | |
217 | std::ptrdiff_t end_pos = (std::min)(position + static_cast<std::ptrdiff_t>(10), static_cast<std::ptrdiff_t>(m_end - m_base)); | |
218 | if(error_code != regex_constants::error_empty) | |
219 | { | |
220 | if((start_pos != 0) || (end_pos != (m_end - m_base))) | |
221 | message += " The error occurred while parsing the regular expression fragment: '"; | |
222 | else | |
223 | message += " The error occurred while parsing the regular expression: '"; | |
224 | if(start_pos != end_pos) | |
225 | { | |
226 | message += std::string(m_base + start_pos, m_base + position); | |
227 | message += ">>>HERE>>>"; | |
228 | message += std::string(m_base + position, m_base + end_pos); | |
229 | } | |
230 | message += "'."; | |
231 | } | |
232 | #endif | |
233 | ||
234 | #ifndef BOOST_NO_EXCEPTIONS | |
235 | if(0 == (this->flags() & regex_constants::no_except)) | |
236 | { | |
237 | boost::regex_error e(message, error_code, position); | |
238 | e.raise(); | |
239 | } | |
240 | #else | |
241 | (void)position; // suppress warnings. | |
242 | #endif | |
243 | } | |
244 | ||
245 | template <class charT, class traits> | |
246 | bool basic_regex_parser<charT, traits>::parse_all() | |
247 | { | |
248 | bool result = true; | |
249 | while(result && (m_position != m_end)) | |
250 | { | |
251 | result = (this->*m_parser_proc)(); | |
252 | } | |
253 | return result; | |
254 | } | |
255 | ||
256 | #ifdef BOOST_MSVC | |
257 | #pragma warning(push) | |
258 | #pragma warning(disable:4702) | |
259 | #endif | |
260 | template <class charT, class traits> | |
261 | bool basic_regex_parser<charT, traits>::parse_basic() | |
262 | { | |
263 | switch(this->m_traits.syntax_type(*m_position)) | |
264 | { | |
265 | case regex_constants::syntax_escape: | |
266 | return parse_basic_escape(); | |
267 | case regex_constants::syntax_dot: | |
268 | return parse_match_any(); | |
269 | case regex_constants::syntax_caret: | |
270 | ++m_position; | |
271 | this->append_state(syntax_element_start_line); | |
272 | break; | |
273 | case regex_constants::syntax_dollar: | |
274 | ++m_position; | |
275 | this->append_state(syntax_element_end_line); | |
276 | break; | |
277 | case regex_constants::syntax_star: | |
278 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line)) | |
279 | return parse_literal(); | |
280 | else | |
281 | { | |
282 | ++m_position; | |
283 | return parse_repeat(); | |
284 | } | |
285 | case regex_constants::syntax_plus: | |
286 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line) || !(this->flags() & regbase::emacs_ex)) | |
287 | return parse_literal(); | |
288 | else | |
289 | { | |
290 | ++m_position; | |
291 | return parse_repeat(1); | |
292 | } | |
293 | case regex_constants::syntax_question: | |
294 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line) || !(this->flags() & regbase::emacs_ex)) | |
295 | return parse_literal(); | |
296 | else | |
297 | { | |
298 | ++m_position; | |
299 | return parse_repeat(0, 1); | |
300 | } | |
301 | case regex_constants::syntax_open_set: | |
302 | return parse_set(); | |
303 | case regex_constants::syntax_newline: | |
304 | if(this->flags() & regbase::newline_alt) | |
305 | return parse_alt(); | |
306 | else | |
307 | return parse_literal(); | |
308 | default: | |
309 | return parse_literal(); | |
310 | } | |
311 | return true; | |
312 | } | |
313 | ||
314 | template <class charT, class traits> | |
315 | bool basic_regex_parser<charT, traits>::parse_extended() | |
316 | { | |
317 | bool result = true; | |
318 | switch(this->m_traits.syntax_type(*m_position)) | |
319 | { | |
320 | case regex_constants::syntax_open_mark: | |
321 | return parse_open_paren(); | |
322 | case regex_constants::syntax_close_mark: | |
323 | return false; | |
324 | case regex_constants::syntax_escape: | |
325 | return parse_extended_escape(); | |
326 | case regex_constants::syntax_dot: | |
327 | return parse_match_any(); | |
328 | case regex_constants::syntax_caret: | |
329 | ++m_position; | |
330 | this->append_state( | |
331 | (this->flags() & regex_constants::no_mod_m ? syntax_element_buffer_start : syntax_element_start_line)); | |
332 | break; | |
333 | case regex_constants::syntax_dollar: | |
334 | ++m_position; | |
335 | this->append_state( | |
336 | (this->flags() & regex_constants::no_mod_m ? syntax_element_buffer_end : syntax_element_end_line)); | |
337 | break; | |
338 | case regex_constants::syntax_star: | |
339 | if(m_position == this->m_base) | |
340 | { | |
341 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"*\" cannot start a regular expression."); | |
342 | return false; | |
343 | } | |
344 | ++m_position; | |
345 | return parse_repeat(); | |
346 | case regex_constants::syntax_question: | |
347 | if(m_position == this->m_base) | |
348 | { | |
349 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"?\" cannot start a regular expression."); | |
350 | return false; | |
351 | } | |
352 | ++m_position; | |
353 | return parse_repeat(0,1); | |
354 | case regex_constants::syntax_plus: | |
355 | if(m_position == this->m_base) | |
356 | { | |
357 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"+\" cannot start a regular expression."); | |
358 | return false; | |
359 | } | |
360 | ++m_position; | |
361 | return parse_repeat(1); | |
362 | case regex_constants::syntax_open_brace: | |
363 | ++m_position; | |
364 | return parse_repeat_range(false); | |
365 | case regex_constants::syntax_close_brace: | |
366 | if((this->flags() & regbase::no_perl_ex) == regbase::no_perl_ex) | |
367 | { | |
368 | fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {."); | |
369 | return false; | |
370 | } | |
371 | result = parse_literal(); | |
372 | break; | |
373 | case regex_constants::syntax_or: | |
374 | return parse_alt(); | |
375 | case regex_constants::syntax_open_set: | |
376 | return parse_set(); | |
377 | case regex_constants::syntax_newline: | |
378 | if(this->flags() & regbase::newline_alt) | |
379 | return parse_alt(); | |
380 | else | |
381 | return parse_literal(); | |
382 | case regex_constants::syntax_hash: | |
383 | // | |
384 | // If we have a mod_x flag set, then skip until | |
385 | // we get to a newline character: | |
386 | // | |
387 | if((this->flags() | |
388 | & (regbase::no_perl_ex|regbase::mod_x)) | |
389 | == regbase::mod_x) | |
390 | { | |
391 | while((m_position != m_end) && !is_separator(*m_position++)){} | |
392 | return true; | |
393 | } | |
394 | BOOST_FALLTHROUGH; | |
395 | default: | |
396 | result = parse_literal(); | |
397 | break; | |
398 | } | |
399 | return result; | |
400 | } | |
401 | #ifdef BOOST_MSVC | |
402 | #pragma warning(pop) | |
403 | #endif | |
404 | ||
405 | template <class charT, class traits> | |
406 | bool basic_regex_parser<charT, traits>::parse_literal() | |
407 | { | |
408 | // append this as a literal provided it's not a space character | |
409 | // or the perl option regbase::mod_x is not set: | |
410 | if( | |
411 | ((this->flags() | |
412 | & (regbase::main_option_type|regbase::mod_x|regbase::no_perl_ex)) | |
413 | != regbase::mod_x) | |
414 | || !this->m_traits.isctype(*m_position, this->m_mask_space)) | |
415 | this->append_literal(*m_position); | |
416 | ++m_position; | |
417 | return true; | |
418 | } | |
419 | ||
420 | template <class charT, class traits> | |
421 | bool basic_regex_parser<charT, traits>::parse_open_paren() | |
422 | { | |
423 | // | |
424 | // skip the '(' and error check: | |
425 | // | |
426 | if(++m_position == m_end) | |
427 | { | |
428 | fail(regex_constants::error_paren, m_position - m_base); | |
429 | return false; | |
430 | } | |
431 | // | |
432 | // begin by checking for a perl-style (?...) extension: | |
433 | // | |
434 | if( | |
435 | ((this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) == 0) | |
436 | || ((this->flags() & (regbase::main_option_type | regbase::emacs_ex)) == (regbase::basic_syntax_group|regbase::emacs_ex)) | |
437 | ) | |
438 | { | |
439 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question) | |
440 | return parse_perl_extension(); | |
441 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_star) | |
442 | return parse_perl_verb(); | |
443 | } | |
444 | // | |
445 | // update our mark count, and append the required state: | |
446 | // | |
447 | unsigned markid = 0; | |
448 | if(0 == (this->flags() & regbase::nosubs)) | |
449 | { | |
450 | markid = ++m_mark_count; | |
451 | #ifndef BOOST_NO_STD_DISTANCE | |
452 | if(this->flags() & regbase::save_subexpression_location) | |
453 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>(std::distance(m_base, m_position) - 1, 0)); | |
454 | #else | |
455 | if(this->flags() & regbase::save_subexpression_location) | |
456 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>((m_position - m_base) - 1, 0)); | |
457 | #endif | |
458 | } | |
459 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); | |
460 | pb->index = markid; | |
461 | pb->icase = this->flags() & regbase::icase; | |
462 | std::ptrdiff_t last_paren_start = this->getoffset(pb); | |
463 | // back up insertion point for alternations, and set new point: | |
464 | std::ptrdiff_t last_alt_point = m_alt_insert_point; | |
465 | this->m_pdata->m_data.align(); | |
466 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
467 | // | |
468 | // back up the current flags in case we have a nested (?imsx) group: | |
469 | // | |
470 | regex_constants::syntax_option_type opts = this->flags(); | |
471 | bool old_case_change = m_has_case_change; | |
472 | m_has_case_change = false; // no changes to this scope as yet... | |
473 | // | |
474 | // Back up branch reset data in case we have a nested (?|...) | |
475 | // | |
476 | int mark_reset = m_mark_reset; | |
477 | m_mark_reset = -1; | |
478 | // | |
479 | // now recursively add more states, this will terminate when we get to a | |
480 | // matching ')' : | |
481 | // | |
482 | parse_all(); | |
483 | // | |
484 | // Unwind pushed alternatives: | |
485 | // | |
486 | if(0 == unwind_alts(last_paren_start)) | |
487 | return false; | |
488 | // | |
489 | // restore flags: | |
490 | // | |
491 | if(m_has_case_change) | |
492 | { | |
493 | // the case has changed in one or more of the alternatives | |
494 | // within the scoped (...) block: we have to add a state | |
495 | // to reset the case sensitivity: | |
496 | static_cast<re_case*>( | |
497 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
498 | )->icase = opts & regbase::icase; | |
499 | } | |
500 | this->flags(opts); | |
501 | m_has_case_change = old_case_change; | |
502 | // | |
503 | // restore branch reset: | |
504 | // | |
505 | m_mark_reset = mark_reset; | |
506 | // | |
507 | // we either have a ')' or we have run out of characters prematurely: | |
508 | // | |
509 | if(m_position == m_end) | |
510 | { | |
511 | this->fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_end)); | |
512 | return false; | |
513 | } | |
514 | BOOST_ASSERT(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark); | |
515 | #ifndef BOOST_NO_STD_DISTANCE | |
516 | if(markid && (this->flags() & regbase::save_subexpression_location)) | |
517 | this->m_pdata->m_subs.at(markid - 1).second = std::distance(m_base, m_position); | |
518 | #else | |
519 | if(markid && (this->flags() & regbase::save_subexpression_location)) | |
520 | this->m_pdata->m_subs.at(markid - 1).second = (m_position - m_base); | |
521 | #endif | |
522 | ++m_position; | |
523 | // | |
524 | // append closing parenthesis state: | |
525 | // | |
526 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); | |
527 | pb->index = markid; | |
528 | pb->icase = this->flags() & regbase::icase; | |
529 | this->m_paren_start = last_paren_start; | |
530 | // | |
531 | // restore the alternate insertion point: | |
532 | // | |
533 | this->m_alt_insert_point = last_alt_point; | |
534 | // | |
535 | // allow backrefs to this mark: | |
536 | // | |
537 | if((markid > 0) && (markid < sizeof(unsigned) * CHAR_BIT)) | |
538 | this->m_backrefs |= 1u << (markid - 1); | |
539 | ||
540 | return true; | |
541 | } | |
542 | ||
543 | template <class charT, class traits> | |
544 | bool basic_regex_parser<charT, traits>::parse_basic_escape() | |
545 | { | |
546 | if(++m_position == m_end) | |
547 | { | |
548 | fail(regex_constants::error_paren, m_position - m_base); | |
549 | return false; | |
550 | } | |
551 | bool result = true; | |
552 | switch(this->m_traits.escape_syntax_type(*m_position)) | |
553 | { | |
554 | case regex_constants::syntax_open_mark: | |
555 | return parse_open_paren(); | |
556 | case regex_constants::syntax_close_mark: | |
557 | return false; | |
558 | case regex_constants::syntax_plus: | |
559 | if(this->flags() & regex_constants::bk_plus_qm) | |
560 | { | |
561 | ++m_position; | |
562 | return parse_repeat(1); | |
563 | } | |
564 | else | |
565 | return parse_literal(); | |
566 | case regex_constants::syntax_question: | |
567 | if(this->flags() & regex_constants::bk_plus_qm) | |
568 | { | |
569 | ++m_position; | |
570 | return parse_repeat(0, 1); | |
571 | } | |
572 | else | |
573 | return parse_literal(); | |
574 | case regex_constants::syntax_open_brace: | |
575 | if(this->flags() & regbase::no_intervals) | |
576 | return parse_literal(); | |
577 | ++m_position; | |
578 | return parse_repeat_range(true); | |
579 | case regex_constants::syntax_close_brace: | |
580 | if(this->flags() & regbase::no_intervals) | |
581 | return parse_literal(); | |
582 | fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {."); | |
583 | return false; | |
584 | case regex_constants::syntax_or: | |
585 | if(this->flags() & regbase::bk_vbar) | |
586 | return parse_alt(); | |
587 | else | |
588 | result = parse_literal(); | |
589 | break; | |
590 | case regex_constants::syntax_digit: | |
591 | return parse_backref(); | |
592 | case regex_constants::escape_type_start_buffer: | |
593 | if(this->flags() & regbase::emacs_ex) | |
594 | { | |
595 | ++m_position; | |
596 | this->append_state(syntax_element_buffer_start); | |
597 | } | |
598 | else | |
599 | result = parse_literal(); | |
600 | break; | |
601 | case regex_constants::escape_type_end_buffer: | |
602 | if(this->flags() & regbase::emacs_ex) | |
603 | { | |
604 | ++m_position; | |
605 | this->append_state(syntax_element_buffer_end); | |
606 | } | |
607 | else | |
608 | result = parse_literal(); | |
609 | break; | |
610 | case regex_constants::escape_type_word_assert: | |
611 | if(this->flags() & regbase::emacs_ex) | |
612 | { | |
613 | ++m_position; | |
614 | this->append_state(syntax_element_word_boundary); | |
615 | } | |
616 | else | |
617 | result = parse_literal(); | |
618 | break; | |
619 | case regex_constants::escape_type_not_word_assert: | |
620 | if(this->flags() & regbase::emacs_ex) | |
621 | { | |
622 | ++m_position; | |
623 | this->append_state(syntax_element_within_word); | |
624 | } | |
625 | else | |
626 | result = parse_literal(); | |
627 | break; | |
628 | case regex_constants::escape_type_left_word: | |
629 | if(this->flags() & regbase::emacs_ex) | |
630 | { | |
631 | ++m_position; | |
632 | this->append_state(syntax_element_word_start); | |
633 | } | |
634 | else | |
635 | result = parse_literal(); | |
636 | break; | |
637 | case regex_constants::escape_type_right_word: | |
638 | if(this->flags() & regbase::emacs_ex) | |
639 | { | |
640 | ++m_position; | |
641 | this->append_state(syntax_element_word_end); | |
642 | } | |
643 | else | |
644 | result = parse_literal(); | |
645 | break; | |
646 | default: | |
647 | if(this->flags() & regbase::emacs_ex) | |
648 | { | |
649 | bool negate = true; | |
650 | switch(*m_position) | |
651 | { | |
652 | case 'w': | |
653 | negate = false; | |
654 | BOOST_FALLTHROUGH; | |
655 | case 'W': | |
656 | { | |
657 | basic_char_set<charT, traits> char_set; | |
658 | if(negate) | |
659 | char_set.negate(); | |
660 | char_set.add_class(this->m_word_mask); | |
661 | if(0 == this->append_set(char_set)) | |
662 | { | |
663 | fail(regex_constants::error_ctype, m_position - m_base); | |
664 | return false; | |
665 | } | |
666 | ++m_position; | |
667 | return true; | |
668 | } | |
669 | case 's': | |
670 | negate = false; | |
671 | BOOST_FALLTHROUGH; | |
672 | case 'S': | |
673 | return add_emacs_code(negate); | |
674 | case 'c': | |
675 | case 'C': | |
676 | // not supported yet: | |
677 | 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."); | |
678 | return false; | |
679 | default: | |
680 | break; | |
681 | } | |
682 | } | |
683 | result = parse_literal(); | |
684 | break; | |
685 | } | |
686 | return result; | |
687 | } | |
688 | ||
689 | template <class charT, class traits> | |
690 | bool basic_regex_parser<charT, traits>::parse_extended_escape() | |
691 | { | |
692 | ++m_position; | |
693 | if(m_position == m_end) | |
694 | { | |
695 | fail(regex_constants::error_escape, m_position - m_base, "Incomplete escape sequence found."); | |
696 | return false; | |
697 | } | |
698 | bool negate = false; // in case this is a character class escape: \w \d etc | |
699 | switch(this->m_traits.escape_syntax_type(*m_position)) | |
700 | { | |
701 | case regex_constants::escape_type_not_class: | |
702 | negate = true; | |
703 | BOOST_FALLTHROUGH; | |
704 | case regex_constants::escape_type_class: | |
705 | { | |
706 | escape_type_class_jump: | |
707 | typedef typename traits::char_class_type m_type; | |
708 | m_type m = this->m_traits.lookup_classname(m_position, m_position+1); | |
709 | if(m != 0) | |
710 | { | |
711 | basic_char_set<charT, traits> char_set; | |
712 | if(negate) | |
713 | char_set.negate(); | |
714 | char_set.add_class(m); | |
715 | if(0 == this->append_set(char_set)) | |
716 | { | |
717 | fail(regex_constants::error_ctype, m_position - m_base); | |
718 | return false; | |
719 | } | |
720 | ++m_position; | |
721 | return true; | |
722 | } | |
723 | // | |
724 | // not a class, just a regular unknown escape: | |
725 | // | |
726 | this->append_literal(unescape_character()); | |
727 | break; | |
728 | } | |
729 | case regex_constants::syntax_digit: | |
730 | return parse_backref(); | |
731 | case regex_constants::escape_type_left_word: | |
732 | ++m_position; | |
733 | this->append_state(syntax_element_word_start); | |
734 | break; | |
735 | case regex_constants::escape_type_right_word: | |
736 | ++m_position; | |
737 | this->append_state(syntax_element_word_end); | |
738 | break; | |
739 | case regex_constants::escape_type_start_buffer: | |
740 | ++m_position; | |
741 | this->append_state(syntax_element_buffer_start); | |
742 | break; | |
743 | case regex_constants::escape_type_end_buffer: | |
744 | ++m_position; | |
745 | this->append_state(syntax_element_buffer_end); | |
746 | break; | |
747 | case regex_constants::escape_type_word_assert: | |
748 | ++m_position; | |
749 | this->append_state(syntax_element_word_boundary); | |
750 | break; | |
751 | case regex_constants::escape_type_not_word_assert: | |
752 | ++m_position; | |
753 | this->append_state(syntax_element_within_word); | |
754 | break; | |
755 | case regex_constants::escape_type_Z: | |
756 | ++m_position; | |
757 | this->append_state(syntax_element_soft_buffer_end); | |
758 | break; | |
759 | case regex_constants::escape_type_Q: | |
760 | return parse_QE(); | |
761 | case regex_constants::escape_type_C: | |
762 | return parse_match_any(); | |
763 | case regex_constants::escape_type_X: | |
764 | ++m_position; | |
765 | this->append_state(syntax_element_combining); | |
766 | break; | |
767 | case regex_constants::escape_type_G: | |
768 | ++m_position; | |
769 | this->append_state(syntax_element_restart_continue); | |
770 | break; | |
771 | case regex_constants::escape_type_not_property: | |
772 | negate = true; | |
773 | BOOST_FALLTHROUGH; | |
774 | case regex_constants::escape_type_property: | |
775 | { | |
776 | ++m_position; | |
777 | char_class_type m; | |
778 | if(m_position == m_end) | |
779 | { | |
780 | fail(regex_constants::error_escape, m_position - m_base, "Incomplete property escape found."); | |
781 | return false; | |
782 | } | |
783 | // maybe have \p{ddd} | |
784 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) | |
785 | { | |
786 | const charT* base = m_position; | |
787 | // skip forward until we find enclosing brace: | |
788 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) | |
789 | ++m_position; | |
790 | if(m_position == m_end) | |
791 | { | |
792 | fail(regex_constants::error_escape, m_position - m_base, "Closing } missing from property escape sequence."); | |
793 | return false; | |
794 | } | |
795 | m = this->m_traits.lookup_classname(++base, m_position++); | |
796 | } | |
797 | else | |
798 | { | |
799 | m = this->m_traits.lookup_classname(m_position, m_position+1); | |
800 | ++m_position; | |
801 | } | |
802 | if(m != 0) | |
803 | { | |
804 | basic_char_set<charT, traits> char_set; | |
805 | if(negate) | |
806 | char_set.negate(); | |
807 | char_set.add_class(m); | |
808 | if(0 == this->append_set(char_set)) | |
809 | { | |
810 | fail(regex_constants::error_ctype, m_position - m_base); | |
811 | return false; | |
812 | } | |
813 | return true; | |
814 | } | |
815 | fail(regex_constants::error_ctype, m_position - m_base, "Escape sequence was neither a valid property nor a valid character class name."); | |
816 | return false; | |
817 | } | |
818 | case regex_constants::escape_type_reset_start_mark: | |
819 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
820 | { | |
821 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); | |
822 | pb->index = -5; | |
823 | pb->icase = this->flags() & regbase::icase; | |
824 | this->m_pdata->m_data.align(); | |
825 | ++m_position; | |
826 | return true; | |
827 | } | |
828 | goto escape_type_class_jump; | |
829 | case regex_constants::escape_type_line_ending: | |
830 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
831 | { | |
832 | const charT* e = get_escape_R_string<charT>(); | |
833 | const charT* old_position = m_position; | |
834 | const charT* old_end = m_end; | |
835 | const charT* old_base = m_base; | |
836 | m_position = e; | |
837 | m_base = e; | |
838 | m_end = e + traits::length(e); | |
839 | bool r = parse_all(); | |
840 | m_position = ++old_position; | |
841 | m_end = old_end; | |
842 | m_base = old_base; | |
843 | return r; | |
844 | } | |
845 | goto escape_type_class_jump; | |
846 | case regex_constants::escape_type_extended_backref: | |
847 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
848 | { | |
849 | bool have_brace = false; | |
850 | bool negative = false; | |
851 | static const char* incomplete_message = "Incomplete \\g escape found."; | |
852 | if(++m_position == m_end) | |
853 | { | |
854 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); | |
855 | return false; | |
856 | } | |
857 | // maybe have \g{ddd} | |
858 | regex_constants::syntax_type syn = this->m_traits.syntax_type(*m_position); | |
859 | regex_constants::syntax_type syn_end = 0; | |
860 | if((syn == regex_constants::syntax_open_brace) | |
861 | || (syn == regex_constants::escape_type_left_word) | |
862 | || (syn == regex_constants::escape_type_end_buffer)) | |
863 | { | |
864 | if(++m_position == m_end) | |
865 | { | |
866 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); | |
867 | return false; | |
868 | } | |
869 | have_brace = true; | |
870 | switch(syn) | |
871 | { | |
872 | case regex_constants::syntax_open_brace: | |
873 | syn_end = regex_constants::syntax_close_brace; | |
874 | break; | |
875 | case regex_constants::escape_type_left_word: | |
876 | syn_end = regex_constants::escape_type_right_word; | |
877 | break; | |
878 | default: | |
879 | syn_end = regex_constants::escape_type_end_buffer; | |
880 | break; | |
881 | } | |
882 | } | |
883 | negative = (*m_position == static_cast<charT>('-')); | |
884 | if((negative) && (++m_position == m_end)) | |
885 | { | |
886 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); | |
887 | return false; | |
888 | } | |
889 | const charT* pc = m_position; | |
890 | boost::intmax_t i = this->m_traits.toi(pc, m_end, 10); | |
891 | if((i < 0) && syn_end) | |
892 | { | |
893 | // Check for a named capture, get the leftmost one if there is more than one: | |
894 | const charT* base = m_position; | |
895 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != syn_end)) | |
896 | { | |
897 | ++m_position; | |
898 | } | |
899 | i = hash_value_from_capture_name(base, m_position); | |
900 | pc = m_position; | |
901 | } | |
902 | if(negative) | |
903 | i = 1 + m_mark_count - i; | |
904 | 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))))) | |
905 | { | |
906 | m_position = pc; | |
907 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace))); | |
908 | pb->index = i; | |
909 | pb->icase = this->flags() & regbase::icase; | |
910 | } | |
911 | else | |
912 | { | |
913 | fail(regex_constants::error_backref, m_position - m_base); | |
914 | return false; | |
915 | } | |
916 | m_position = pc; | |
917 | if(have_brace) | |
918 | { | |
919 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != syn_end)) | |
920 | { | |
921 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); | |
922 | return false; | |
923 | } | |
924 | ++m_position; | |
925 | } | |
926 | return true; | |
927 | } | |
928 | goto escape_type_class_jump; | |
929 | case regex_constants::escape_type_control_v: | |
930 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
931 | goto escape_type_class_jump; | |
932 | BOOST_FALLTHROUGH; | |
933 | default: | |
934 | this->append_literal(unescape_character()); | |
935 | break; | |
936 | } | |
937 | return true; | |
938 | } | |
939 | ||
940 | template <class charT, class traits> | |
941 | bool basic_regex_parser<charT, traits>::parse_match_any() | |
942 | { | |
943 | // | |
944 | // we have a '.' that can match any character: | |
945 | // | |
946 | ++m_position; | |
947 | static_cast<re_dot*>( | |
948 | this->append_state(syntax_element_wild, sizeof(re_dot)) | |
949 | )->mask = static_cast<unsigned char>(this->flags() & regbase::no_mod_s | |
950 | ? BOOST_REGEX_DETAIL_NS::force_not_newline | |
951 | : this->flags() & regbase::mod_s ? | |
952 | BOOST_REGEX_DETAIL_NS::force_newline : BOOST_REGEX_DETAIL_NS::dont_care); | |
953 | return true; | |
954 | } | |
955 | ||
956 | template <class charT, class traits> | |
957 | bool basic_regex_parser<charT, traits>::parse_repeat(std::size_t low, std::size_t high) | |
958 | { | |
959 | bool greedy = true; | |
960 | bool pocessive = false; | |
961 | std::size_t insert_point; | |
962 | // | |
963 | // when we get to here we may have a non-greedy ? mark still to come: | |
964 | // | |
965 | if((m_position != m_end) | |
966 | && ( | |
967 | (0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
968 | || ((regbase::basic_syntax_group|regbase::emacs_ex) == (this->flags() & (regbase::main_option_type | regbase::emacs_ex))) | |
969 | ) | |
970 | ) | |
971 | { | |
972 | // OK we have a perl or emacs regex, check for a '?': | |
973 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question) | |
974 | { | |
975 | greedy = false; | |
976 | ++m_position; | |
977 | } | |
978 | // for perl regexes only check for pocessive ++ repeats. | |
979 | if((m_position != m_end) | |
980 | && (0 == (this->flags() & regbase::main_option_type)) | |
981 | && (this->m_traits.syntax_type(*m_position) == regex_constants::syntax_plus)) | |
982 | { | |
983 | pocessive = true; | |
984 | ++m_position; | |
985 | } | |
986 | } | |
987 | if(0 == this->m_last_state) | |
988 | { | |
989 | fail(regex_constants::error_badrepeat, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_position), "Nothing to repeat."); | |
990 | return false; | |
991 | } | |
992 | if(this->m_last_state->type == syntax_element_endmark) | |
993 | { | |
994 | // insert a repeat before the '(' matching the last ')': | |
995 | insert_point = this->m_paren_start; | |
996 | } | |
997 | else if((this->m_last_state->type == syntax_element_literal) && (static_cast<re_literal*>(this->m_last_state)->length > 1)) | |
998 | { | |
999 | // the last state was a literal with more than one character, split it in two: | |
1000 | re_literal* lit = static_cast<re_literal*>(this->m_last_state); | |
1001 | charT c = (static_cast<charT*>(static_cast<void*>(lit+1)))[lit->length - 1]; | |
1002 | lit->length -= 1; | |
1003 | // now append new state: | |
1004 | lit = static_cast<re_literal*>(this->append_state(syntax_element_literal, sizeof(re_literal) + sizeof(charT))); | |
1005 | lit->length = 1; | |
1006 | (static_cast<charT*>(static_cast<void*>(lit+1)))[0] = c; | |
1007 | insert_point = this->getoffset(this->m_last_state); | |
1008 | } | |
1009 | else | |
1010 | { | |
1011 | // repeat the last state whatever it was, need to add some error checking here: | |
1012 | switch(this->m_last_state->type) | |
1013 | { | |
1014 | case syntax_element_start_line: | |
1015 | case syntax_element_end_line: | |
1016 | case syntax_element_word_boundary: | |
1017 | case syntax_element_within_word: | |
1018 | case syntax_element_word_start: | |
1019 | case syntax_element_word_end: | |
1020 | case syntax_element_buffer_start: | |
1021 | case syntax_element_buffer_end: | |
1022 | case syntax_element_alt: | |
1023 | case syntax_element_soft_buffer_end: | |
1024 | case syntax_element_restart_continue: | |
1025 | case syntax_element_jump: | |
1026 | case syntax_element_startmark: | |
1027 | case syntax_element_backstep: | |
1028 | // can't legally repeat any of the above: | |
1029 | fail(regex_constants::error_badrepeat, m_position - m_base); | |
1030 | return false; | |
1031 | default: | |
1032 | // do nothing... | |
1033 | break; | |
1034 | } | |
1035 | insert_point = this->getoffset(this->m_last_state); | |
1036 | } | |
1037 | // | |
1038 | // OK we now know what to repeat, so insert the repeat around it: | |
1039 | // | |
1040 | re_repeat* rep = static_cast<re_repeat*>(this->insert_state(insert_point, syntax_element_rep, re_repeater_size)); | |
1041 | rep->min = low; | |
1042 | rep->max = high; | |
1043 | rep->greedy = greedy; | |
1044 | rep->leading = false; | |
1045 | // store our repeater position for later: | |
1046 | std::ptrdiff_t rep_off = this->getoffset(rep); | |
1047 | // and append a back jump to the repeat: | |
1048 | re_jump* jmp = static_cast<re_jump*>(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
1049 | jmp->alt.i = rep_off - this->getoffset(jmp); | |
1050 | this->m_pdata->m_data.align(); | |
1051 | // now fill in the alt jump for the repeat: | |
1052 | rep = static_cast<re_repeat*>(this->getaddress(rep_off)); | |
1053 | rep->alt.i = this->m_pdata->m_data.size() - rep_off; | |
1054 | // | |
1055 | // If the repeat is pocessive then bracket the repeat with a (?>...) | |
1056 | // independent sub-expression construct: | |
1057 | // | |
1058 | if(pocessive) | |
1059 | { | |
1060 | if(m_position != m_end) | |
1061 | { | |
1062 | // | |
1063 | // Check for illegal following quantifier, we have to do this here, because | |
1064 | // the extra states we insert below circumvents our usual error checking :-( | |
1065 | // | |
1066 | switch(this->m_traits.syntax_type(*m_position)) | |
1067 | { | |
1068 | case regex_constants::syntax_star: | |
1069 | case regex_constants::syntax_plus: | |
1070 | case regex_constants::syntax_question: | |
1071 | case regex_constants::syntax_open_brace: | |
1072 | fail(regex_constants::error_badrepeat, m_position - m_base); | |
1073 | return false; | |
1074 | } | |
1075 | } | |
1076 | re_brace* pb = static_cast<re_brace*>(this->insert_state(insert_point, syntax_element_startmark, sizeof(re_brace))); | |
1077 | pb->index = -3; | |
1078 | pb->icase = this->flags() & regbase::icase; | |
1079 | jmp = static_cast<re_jump*>(this->insert_state(insert_point + sizeof(re_brace), syntax_element_jump, sizeof(re_jump))); | |
1080 | this->m_pdata->m_data.align(); | |
1081 | jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp); | |
1082 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); | |
1083 | pb->index = -3; | |
1084 | pb->icase = this->flags() & regbase::icase; | |
1085 | } | |
1086 | return true; | |
1087 | } | |
1088 | ||
1089 | template <class charT, class traits> | |
1090 | bool basic_regex_parser<charT, traits>::parse_repeat_range(bool isbasic) | |
1091 | { | |
1092 | static const char* incomplete_message = "Missing } in quantified repetition."; | |
1093 | // | |
1094 | // parse a repeat-range: | |
1095 | // | |
1096 | std::size_t min, max; | |
1097 | boost::intmax_t v; | |
1098 | // skip whitespace: | |
1099 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1100 | ++m_position; | |
1101 | if(this->m_position == this->m_end) | |
1102 | { | |
1103 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1104 | { | |
1105 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1106 | return false; | |
1107 | } | |
1108 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1109 | --m_position; | |
1110 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1111 | return parse_literal(); | |
1112 | } | |
1113 | // get min: | |
1114 | v = this->m_traits.toi(m_position, m_end, 10); | |
1115 | // skip whitespace: | |
1116 | if((v < 0) || (v > umax())) | |
1117 | { | |
1118 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1119 | { | |
1120 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1121 | return false; | |
1122 | } | |
1123 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1124 | --m_position; | |
1125 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1126 | return parse_literal(); | |
1127 | } | |
1128 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1129 | ++m_position; | |
1130 | if(this->m_position == this->m_end) | |
1131 | { | |
1132 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1133 | { | |
1134 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1135 | return false; | |
1136 | } | |
1137 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1138 | --m_position; | |
1139 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1140 | return parse_literal(); | |
1141 | } | |
1142 | min = static_cast<std::size_t>(v); | |
1143 | // see if we have a comma: | |
1144 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_comma) | |
1145 | { | |
1146 | // move on and error check: | |
1147 | ++m_position; | |
1148 | // skip whitespace: | |
1149 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1150 | ++m_position; | |
1151 | if(this->m_position == this->m_end) | |
1152 | { | |
1153 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1154 | { | |
1155 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1156 | return false; | |
1157 | } | |
1158 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1159 | --m_position; | |
1160 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1161 | return parse_literal(); | |
1162 | } | |
1163 | // get the value if any: | |
1164 | v = this->m_traits.toi(m_position, m_end, 10); | |
1165 | max = ((v >= 0) && (v < umax())) ? (std::size_t)v : (std::numeric_limits<std::size_t>::max)(); | |
1166 | } | |
1167 | else | |
1168 | { | |
1169 | // no comma, max = min: | |
1170 | max = min; | |
1171 | } | |
1172 | // skip whitespace: | |
1173 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1174 | ++m_position; | |
1175 | // OK now check trailing }: | |
1176 | if(this->m_position == this->m_end) | |
1177 | { | |
1178 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1179 | { | |
1180 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1181 | return false; | |
1182 | } | |
1183 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1184 | --m_position; | |
1185 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1186 | return parse_literal(); | |
1187 | } | |
1188 | if(isbasic) | |
1189 | { | |
1190 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_escape) | |
1191 | { | |
1192 | ++m_position; | |
1193 | if(this->m_position == this->m_end) | |
1194 | { | |
1195 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1196 | return false; | |
1197 | } | |
1198 | } | |
1199 | else | |
1200 | { | |
1201 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1202 | return false; | |
1203 | } | |
1204 | } | |
1205 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_brace) | |
1206 | ++m_position; | |
1207 | else | |
1208 | { | |
1209 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1210 | --m_position; | |
1211 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1212 | return parse_literal(); | |
1213 | } | |
1214 | // | |
1215 | // finally go and add the repeat, unless error: | |
1216 | // | |
1217 | if(min > max) | |
1218 | { | |
1219 | // Backtrack to error location: | |
1220 | m_position -= 2; | |
1221 | while(this->m_traits.isctype(*m_position, this->m_word_mask)) --m_position; | |
1222 | ++m_position; | |
1223 | fail(regex_constants::error_badbrace, m_position - m_base); | |
1224 | return false; | |
1225 | } | |
1226 | return parse_repeat(min, max); | |
1227 | } | |
1228 | ||
1229 | template <class charT, class traits> | |
1230 | bool basic_regex_parser<charT, traits>::parse_alt() | |
1231 | { | |
1232 | // | |
1233 | // error check: if there have been no previous states, | |
1234 | // or if the last state was a '(' then error: | |
1235 | // | |
1236 | if( | |
1237 | ((this->m_last_state == 0) || (this->m_last_state->type == syntax_element_startmark)) | |
1238 | && | |
1239 | !( | |
1240 | ((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) | |
1241 | && | |
1242 | ((this->flags() & regbase::no_empty_expressions) == 0) | |
1243 | ) | |
1244 | ) | |
1245 | { | |
1246 | fail(regex_constants::error_empty, this->m_position - this->m_base, "A regular expression cannot start with the alternation operator |."); | |
1247 | return false; | |
1248 | } | |
1249 | // | |
1250 | // Reset mark count if required: | |
1251 | // | |
1252 | if(m_max_mark < m_mark_count) | |
1253 | m_max_mark = m_mark_count; | |
1254 | if(m_mark_reset >= 0) | |
1255 | m_mark_count = m_mark_reset; | |
1256 | ||
1257 | ++m_position; | |
1258 | // | |
1259 | // we need to append a trailing jump: | |
1260 | // | |
1261 | re_syntax_base* pj = this->append_state(BOOST_REGEX_DETAIL_NS::syntax_element_jump, sizeof(re_jump)); | |
1262 | std::ptrdiff_t jump_offset = this->getoffset(pj); | |
1263 | // | |
1264 | // now insert the alternative: | |
1265 | // | |
1266 | re_alt* palt = static_cast<re_alt*>(this->insert_state(this->m_alt_insert_point, syntax_element_alt, re_alt_size)); | |
1267 | jump_offset += re_alt_size; | |
1268 | this->m_pdata->m_data.align(); | |
1269 | palt->alt.i = this->m_pdata->m_data.size() - this->getoffset(palt); | |
1270 | // | |
1271 | // update m_alt_insert_point so that the next alternate gets | |
1272 | // inserted at the start of the second of the two we've just created: | |
1273 | // | |
1274 | this->m_alt_insert_point = this->m_pdata->m_data.size(); | |
1275 | // | |
1276 | // the start of this alternative must have a case changes state | |
1277 | // if the current block has messed around with case changes: | |
1278 | // | |
1279 | if(m_has_case_change) | |
1280 | { | |
1281 | static_cast<re_case*>( | |
1282 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
1283 | )->icase = this->m_icase; | |
1284 | } | |
1285 | // | |
1286 | // push the alternative onto our stack, a recursive | |
1287 | // implementation here is easier to understand (and faster | |
1288 | // as it happens), but causes all kinds of stack overflow problems | |
1289 | // on programs with small stacks (COM+). | |
1290 | // | |
1291 | m_alt_jumps.push_back(jump_offset); | |
1292 | return true; | |
1293 | } | |
1294 | ||
1295 | template <class charT, class traits> | |
1296 | bool basic_regex_parser<charT, traits>::parse_set() | |
1297 | { | |
1298 | static const char* incomplete_message = "Character set declaration starting with [ terminated prematurely - either no ] was found or the set had no content."; | |
1299 | ++m_position; | |
1300 | if(m_position == m_end) | |
1301 | { | |
1302 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1303 | return false; | |
1304 | } | |
1305 | basic_char_set<charT, traits> char_set; | |
1306 | ||
1307 | const charT* base = m_position; // where the '[' was | |
1308 | const charT* item_base = m_position; // where the '[' or '^' was | |
1309 | ||
1310 | while(m_position != m_end) | |
1311 | { | |
1312 | switch(this->m_traits.syntax_type(*m_position)) | |
1313 | { | |
1314 | case regex_constants::syntax_caret: | |
1315 | if(m_position == base) | |
1316 | { | |
1317 | char_set.negate(); | |
1318 | ++m_position; | |
1319 | item_base = m_position; | |
1320 | } | |
1321 | else | |
1322 | parse_set_literal(char_set); | |
1323 | break; | |
1324 | case regex_constants::syntax_close_set: | |
1325 | if(m_position == item_base) | |
1326 | { | |
1327 | parse_set_literal(char_set); | |
1328 | break; | |
1329 | } | |
1330 | else | |
1331 | { | |
1332 | ++m_position; | |
1333 | if(0 == this->append_set(char_set)) | |
1334 | { | |
1335 | fail(regex_constants::error_ctype, m_position - m_base); | |
1336 | return false; | |
1337 | } | |
1338 | } | |
1339 | return true; | |
1340 | case regex_constants::syntax_open_set: | |
1341 | if(parse_inner_set(char_set)) | |
1342 | break; | |
1343 | return true; | |
1344 | case regex_constants::syntax_escape: | |
1345 | { | |
1346 | // | |
1347 | // look ahead and see if this is a character class shortcut | |
1348 | // \d \w \s etc... | |
1349 | // | |
1350 | ++m_position; | |
1351 | if(this->m_traits.escape_syntax_type(*m_position) | |
1352 | == regex_constants::escape_type_class) | |
1353 | { | |
1354 | char_class_type m = this->m_traits.lookup_classname(m_position, m_position+1); | |
1355 | if(m != 0) | |
1356 | { | |
1357 | char_set.add_class(m); | |
1358 | ++m_position; | |
1359 | break; | |
1360 | } | |
1361 | } | |
1362 | else if(this->m_traits.escape_syntax_type(*m_position) | |
1363 | == regex_constants::escape_type_not_class) | |
1364 | { | |
1365 | // negated character class: | |
1366 | char_class_type m = this->m_traits.lookup_classname(m_position, m_position+1); | |
1367 | if(m != 0) | |
1368 | { | |
1369 | char_set.add_negated_class(m); | |
1370 | ++m_position; | |
1371 | break; | |
1372 | } | |
1373 | } | |
1374 | // not a character class, just a regular escape: | |
1375 | --m_position; | |
1376 | parse_set_literal(char_set); | |
1377 | break; | |
1378 | } | |
1379 | default: | |
1380 | parse_set_literal(char_set); | |
1381 | break; | |
1382 | } | |
1383 | } | |
1384 | return m_position != m_end; | |
1385 | } | |
1386 | ||
1387 | template <class charT, class traits> | |
1388 | bool basic_regex_parser<charT, traits>::parse_inner_set(basic_char_set<charT, traits>& char_set) | |
1389 | { | |
1390 | static const char* incomplete_message = "Character class declaration starting with [ terminated prematurely - either no ] was found or the set had no content."; | |
1391 | // | |
1392 | // we have either a character class [:name:] | |
1393 | // a collating element [.name.] | |
1394 | // or an equivalence class [=name=] | |
1395 | // | |
1396 | if(m_end == ++m_position) | |
1397 | { | |
1398 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1399 | return false; | |
1400 | } | |
1401 | switch(this->m_traits.syntax_type(*m_position)) | |
1402 | { | |
1403 | case regex_constants::syntax_dot: | |
1404 | // | |
1405 | // a collating element is treated as a literal: | |
1406 | // | |
1407 | --m_position; | |
1408 | parse_set_literal(char_set); | |
1409 | return true; | |
1410 | case regex_constants::syntax_colon: | |
1411 | { | |
1412 | // check that character classes are actually enabled: | |
1413 | if((this->flags() & (regbase::main_option_type | regbase::no_char_classes)) | |
1414 | == (regbase::basic_syntax_group | regbase::no_char_classes)) | |
1415 | { | |
1416 | --m_position; | |
1417 | parse_set_literal(char_set); | |
1418 | return true; | |
1419 | } | |
1420 | // skip the ':' | |
1421 | if(m_end == ++m_position) | |
1422 | { | |
1423 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1424 | return false; | |
1425 | } | |
1426 | const charT* name_first = m_position; | |
1427 | // skip at least one character, then find the matching ':]' | |
1428 | if(m_end == ++m_position) | |
1429 | { | |
1430 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1431 | return false; | |
1432 | } | |
1433 | while((m_position != m_end) | |
1434 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_colon)) | |
1435 | ++m_position; | |
1436 | const charT* name_last = m_position; | |
1437 | if(m_end == m_position) | |
1438 | { | |
1439 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1440 | return false; | |
1441 | } | |
1442 | if((m_end == ++m_position) | |
1443 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) | |
1444 | { | |
1445 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1446 | return false; | |
1447 | } | |
1448 | // | |
1449 | // check for negated class: | |
1450 | // | |
1451 | bool negated = false; | |
1452 | if(this->m_traits.syntax_type(*name_first) == regex_constants::syntax_caret) | |
1453 | { | |
1454 | ++name_first; | |
1455 | negated = true; | |
1456 | } | |
1457 | typedef typename traits::char_class_type m_type; | |
1458 | m_type m = this->m_traits.lookup_classname(name_first, name_last); | |
1459 | if(m == 0) | |
1460 | { | |
1461 | if(char_set.empty() && (name_last - name_first == 1)) | |
1462 | { | |
1463 | // maybe a special case: | |
1464 | ++m_position; | |
1465 | if( (m_position != m_end) | |
1466 | && (this->m_traits.syntax_type(*m_position) | |
1467 | == regex_constants::syntax_close_set)) | |
1468 | { | |
1469 | if(this->m_traits.escape_syntax_type(*name_first) | |
1470 | == regex_constants::escape_type_left_word) | |
1471 | { | |
1472 | ++m_position; | |
1473 | this->append_state(syntax_element_word_start); | |
1474 | return false; | |
1475 | } | |
1476 | if(this->m_traits.escape_syntax_type(*name_first) | |
1477 | == regex_constants::escape_type_right_word) | |
1478 | { | |
1479 | ++m_position; | |
1480 | this->append_state(syntax_element_word_end); | |
1481 | return false; | |
1482 | } | |
1483 | } | |
1484 | } | |
1485 | fail(regex_constants::error_ctype, name_first - m_base); | |
1486 | return false; | |
1487 | } | |
1488 | if(negated == false) | |
1489 | char_set.add_class(m); | |
1490 | else | |
1491 | char_set.add_negated_class(m); | |
1492 | ++m_position; | |
1493 | break; | |
1494 | } | |
1495 | case regex_constants::syntax_equal: | |
1496 | { | |
1497 | // skip the '=' | |
1498 | if(m_end == ++m_position) | |
1499 | { | |
1500 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1501 | return false; | |
1502 | } | |
1503 | const charT* name_first = m_position; | |
1504 | // skip at least one character, then find the matching '=]' | |
1505 | if(m_end == ++m_position) | |
1506 | { | |
1507 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1508 | return false; | |
1509 | } | |
1510 | while((m_position != m_end) | |
1511 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal)) | |
1512 | ++m_position; | |
1513 | const charT* name_last = m_position; | |
1514 | if(m_end == m_position) | |
1515 | { | |
1516 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1517 | return false; | |
1518 | } | |
1519 | if((m_end == ++m_position) | |
1520 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) | |
1521 | { | |
1522 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1523 | return false; | |
1524 | } | |
1525 | string_type m = this->m_traits.lookup_collatename(name_first, name_last); | |
1526 | if((0 == m.size()) || (m.size() > 2)) | |
1527 | { | |
1528 | fail(regex_constants::error_collate, name_first - m_base); | |
1529 | return false; | |
1530 | } | |
1531 | digraph<charT> d; | |
1532 | d.first = m[0]; | |
1533 | if(m.size() > 1) | |
1534 | d.second = m[1]; | |
1535 | else | |
1536 | d.second = 0; | |
1537 | char_set.add_equivalent(d); | |
1538 | ++m_position; | |
1539 | break; | |
1540 | } | |
1541 | default: | |
1542 | --m_position; | |
1543 | parse_set_literal(char_set); | |
1544 | break; | |
1545 | } | |
1546 | return true; | |
1547 | } | |
1548 | ||
1549 | template <class charT, class traits> | |
1550 | void basic_regex_parser<charT, traits>::parse_set_literal(basic_char_set<charT, traits>& char_set) | |
1551 | { | |
1552 | digraph<charT> start_range(get_next_set_literal(char_set)); | |
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_dash) | |
1559 | { | |
1560 | // we have a range: | |
1561 | if(m_end == ++m_position) | |
1562 | { | |
1563 | fail(regex_constants::error_brack, m_position - m_base); | |
1564 | return; | |
1565 | } | |
1566 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set) | |
1567 | { | |
1568 | digraph<charT> end_range = get_next_set_literal(char_set); | |
1569 | char_set.add_range(start_range, end_range); | |
1570 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_dash) | |
1571 | { | |
1572 | if(m_end == ++m_position) | |
1573 | { | |
1574 | fail(regex_constants::error_brack, m_position - m_base); | |
1575 | return; | |
1576 | } | |
1577 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_set) | |
1578 | { | |
1579 | // trailing - : | |
1580 | --m_position; | |
1581 | return; | |
1582 | } | |
1583 | fail(regex_constants::error_range, m_position - m_base); | |
1584 | return; | |
1585 | } | |
1586 | return; | |
1587 | } | |
1588 | --m_position; | |
1589 | } | |
1590 | char_set.add_single(start_range); | |
1591 | } | |
1592 | ||
1593 | template <class charT, class traits> | |
1594 | digraph<charT> basic_regex_parser<charT, traits>::get_next_set_literal(basic_char_set<charT, traits>& char_set) | |
1595 | { | |
1596 | digraph<charT> result; | |
1597 | switch(this->m_traits.syntax_type(*m_position)) | |
1598 | { | |
1599 | case regex_constants::syntax_dash: | |
1600 | if(!char_set.empty()) | |
1601 | { | |
1602 | // see if we are at the end of the set: | |
1603 | if((++m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) | |
1604 | { | |
1605 | fail(regex_constants::error_range, m_position - m_base); | |
1606 | return result; | |
1607 | } | |
1608 | --m_position; | |
1609 | } | |
1610 | result.first = *m_position++; | |
1611 | return result; | |
1612 | case regex_constants::syntax_escape: | |
1613 | // check to see if escapes are supported first: | |
1614 | if(this->flags() & regex_constants::no_escape_in_lists) | |
1615 | { | |
1616 | result = *m_position++; | |
1617 | break; | |
1618 | } | |
1619 | ++m_position; | |
1620 | result = unescape_character(); | |
1621 | break; | |
1622 | case regex_constants::syntax_open_set: | |
1623 | { | |
1624 | if(m_end == ++m_position) | |
1625 | { | |
1626 | fail(regex_constants::error_collate, m_position - m_base); | |
1627 | return result; | |
1628 | } | |
1629 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_dot) | |
1630 | { | |
1631 | --m_position; | |
1632 | result.first = *m_position; | |
1633 | ++m_position; | |
1634 | return result; | |
1635 | } | |
1636 | if(m_end == ++m_position) | |
1637 | { | |
1638 | fail(regex_constants::error_collate, m_position - m_base); | |
1639 | return result; | |
1640 | } | |
1641 | const charT* name_first = m_position; | |
1642 | // skip at least one character, then find the matching ':]' | |
1643 | if(m_end == ++m_position) | |
1644 | { | |
1645 | fail(regex_constants::error_collate, name_first - m_base); | |
1646 | return result; | |
1647 | } | |
1648 | while((m_position != m_end) | |
1649 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_dot)) | |
1650 | ++m_position; | |
1651 | const charT* name_last = m_position; | |
1652 | if(m_end == m_position) | |
1653 | { | |
1654 | fail(regex_constants::error_collate, name_first - m_base); | |
1655 | return result; | |
1656 | } | |
1657 | if((m_end == ++m_position) | |
1658 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) | |
1659 | { | |
1660 | fail(regex_constants::error_collate, name_first - m_base); | |
1661 | return result; | |
1662 | } | |
1663 | ++m_position; | |
1664 | string_type s = this->m_traits.lookup_collatename(name_first, name_last); | |
1665 | if(s.empty() || (s.size() > 2)) | |
1666 | { | |
1667 | fail(regex_constants::error_collate, name_first - m_base); | |
1668 | return result; | |
1669 | } | |
1670 | result.first = s[0]; | |
1671 | if(s.size() > 1) | |
1672 | result.second = s[1]; | |
1673 | else | |
1674 | result.second = 0; | |
1675 | return result; | |
1676 | } | |
1677 | default: | |
1678 | result = *m_position++; | |
1679 | } | |
1680 | return result; | |
1681 | } | |
1682 | ||
1683 | // | |
1684 | // does a value fit in the specified charT type? | |
1685 | // | |
1686 | template <class charT> | |
1687 | bool valid_value(charT, boost::intmax_t v, const mpl::true_&) | |
1688 | { | |
1689 | return (v >> (sizeof(charT) * CHAR_BIT)) == 0; | |
1690 | } | |
1691 | template <class charT> | |
1692 | bool valid_value(charT, boost::intmax_t, const mpl::false_&) | |
1693 | { | |
1694 | return true; // v will alsways fit in a charT | |
1695 | } | |
1696 | template <class charT> | |
1697 | bool valid_value(charT c, boost::intmax_t v) | |
1698 | { | |
1699 | return valid_value(c, v, mpl::bool_<(sizeof(charT) < sizeof(boost::intmax_t))>()); | |
1700 | } | |
1701 | ||
1702 | template <class charT, class traits> | |
1703 | charT basic_regex_parser<charT, traits>::unescape_character() | |
1704 | { | |
1705 | #ifdef BOOST_MSVC | |
1706 | #pragma warning(push) | |
1707 | #pragma warning(disable:4127) | |
1708 | #endif | |
1709 | charT result(0); | |
1710 | if(m_position == m_end) | |
1711 | { | |
1712 | fail(regex_constants::error_escape, m_position - m_base, "Escape sequence terminated prematurely."); | |
1713 | return false; | |
1714 | } | |
1715 | switch(this->m_traits.escape_syntax_type(*m_position)) | |
1716 | { | |
1717 | case regex_constants::escape_type_control_a: | |
1718 | result = charT('\a'); | |
1719 | break; | |
1720 | case regex_constants::escape_type_e: | |
1721 | result = charT(27); | |
1722 | break; | |
1723 | case regex_constants::escape_type_control_f: | |
1724 | result = charT('\f'); | |
1725 | break; | |
1726 | case regex_constants::escape_type_control_n: | |
1727 | result = charT('\n'); | |
1728 | break; | |
1729 | case regex_constants::escape_type_control_r: | |
1730 | result = charT('\r'); | |
1731 | break; | |
1732 | case regex_constants::escape_type_control_t: | |
1733 | result = charT('\t'); | |
1734 | break; | |
1735 | case regex_constants::escape_type_control_v: | |
1736 | result = charT('\v'); | |
1737 | break; | |
1738 | case regex_constants::escape_type_word_assert: | |
1739 | result = charT('\b'); | |
1740 | break; | |
1741 | case regex_constants::escape_type_ascii_control: | |
1742 | ++m_position; | |
1743 | if(m_position == m_end) | |
1744 | { | |
1745 | // Rewind to start of escape: | |
1746 | --m_position; | |
1747 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1748 | fail(regex_constants::error_escape, m_position - m_base, "ASCII escape sequence terminated prematurely."); | |
1749 | return result; | |
1750 | } | |
1751 | result = static_cast<charT>(*m_position % 32); | |
1752 | break; | |
1753 | case regex_constants::escape_type_hex: | |
1754 | ++m_position; | |
1755 | if(m_position == m_end) | |
1756 | { | |
1757 | // Rewind to start of escape: | |
1758 | --m_position; | |
1759 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1760 | fail(regex_constants::error_escape, m_position - m_base, "Hexadecimal escape sequence terminated prematurely."); | |
1761 | return result; | |
1762 | } | |
1763 | // maybe have \x{ddd} | |
1764 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) | |
1765 | { | |
1766 | ++m_position; | |
1767 | if(m_position == m_end) | |
1768 | { | |
1769 | // Rewind to start of escape: | |
1770 | --m_position; | |
1771 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1772 | fail(regex_constants::error_escape, m_position - m_base, "Missing } in hexadecimal escape sequence."); | |
1773 | return result; | |
1774 | } | |
1775 | boost::intmax_t i = this->m_traits.toi(m_position, m_end, 16); | |
1776 | if((m_position == m_end) | |
1777 | || (i < 0) | |
1778 | || ((std::numeric_limits<charT>::is_specialized) && (i > (boost::intmax_t)(std::numeric_limits<charT>::max)())) | |
1779 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) | |
1780 | { | |
1781 | // Rewind to start of escape: | |
1782 | --m_position; | |
1783 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1784 | fail(regex_constants::error_badbrace, m_position - m_base, "Hexadecimal escape sequence was invalid."); | |
1785 | return result; | |
1786 | } | |
1787 | ++m_position; | |
1788 | result = charT(i); | |
1789 | } | |
1790 | else | |
1791 | { | |
1792 | std::ptrdiff_t len = (std::min)(static_cast<std::ptrdiff_t>(2), static_cast<std::ptrdiff_t>(m_end - m_position)); | |
1793 | boost::intmax_t i = this->m_traits.toi(m_position, m_position + len, 16); | |
1794 | if((i < 0) | |
1795 | || !valid_value(charT(0), i)) | |
1796 | { | |
1797 | // Rewind to start of escape: | |
1798 | --m_position; | |
1799 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1800 | fail(regex_constants::error_escape, m_position - m_base, "Escape sequence did not encode a valid character."); | |
1801 | return result; | |
1802 | } | |
1803 | result = charT(i); | |
1804 | } | |
1805 | return result; | |
1806 | case regex_constants::syntax_digit: | |
1807 | { | |
1808 | // an octal escape sequence, the first character must be a zero | |
1809 | // followed by up to 3 octal digits: | |
1810 | std::ptrdiff_t len = (std::min)(::boost::BOOST_REGEX_DETAIL_NS::distance(m_position, m_end), static_cast<std::ptrdiff_t>(4)); | |
1811 | const charT* bp = m_position; | |
1812 | boost::intmax_t val = this->m_traits.toi(bp, bp + 1, 8); | |
1813 | if(val != 0) | |
1814 | { | |
1815 | // Rewind to start of escape: | |
1816 | --m_position; | |
1817 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1818 | // Oops not an octal escape after all: | |
1819 | fail(regex_constants::error_escape, m_position - m_base, "Invalid octal escape sequence."); | |
1820 | return result; | |
1821 | } | |
1822 | val = this->m_traits.toi(m_position, m_position + len, 8); | |
1823 | if((val < 0) || (val > (boost::intmax_t)(std::numeric_limits<charT>::max)())) | |
1824 | { | |
1825 | // Rewind to start of escape: | |
1826 | --m_position; | |
1827 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1828 | fail(regex_constants::error_escape, m_position - m_base, "Octal escape sequence is invalid."); | |
1829 | return result; | |
1830 | } | |
1831 | return static_cast<charT>(val); | |
1832 | } | |
1833 | case regex_constants::escape_type_named_char: | |
1834 | { | |
1835 | ++m_position; | |
1836 | if(m_position == m_end) | |
1837 | { | |
1838 | // Rewind to start of escape: | |
1839 | --m_position; | |
1840 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1841 | fail(regex_constants::error_escape, m_position - m_base); | |
1842 | return false; | |
1843 | } | |
1844 | // maybe have \N{name} | |
1845 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) | |
1846 | { | |
1847 | const charT* base = m_position; | |
1848 | // skip forward until we find enclosing brace: | |
1849 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) | |
1850 | ++m_position; | |
1851 | if(m_position == m_end) | |
1852 | { | |
1853 | // Rewind to start of escape: | |
1854 | --m_position; | |
1855 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1856 | fail(regex_constants::error_escape, m_position - m_base); | |
1857 | return false; | |
1858 | } | |
1859 | string_type s = this->m_traits.lookup_collatename(++base, m_position++); | |
1860 | if(s.empty()) | |
1861 | { | |
1862 | // Rewind to start of escape: | |
1863 | --m_position; | |
1864 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1865 | fail(regex_constants::error_collate, m_position - m_base); | |
1866 | return false; | |
1867 | } | |
1868 | if(s.size() == 1) | |
1869 | { | |
1870 | return s[0]; | |
1871 | } | |
1872 | } | |
1873 | // fall through is a failure: | |
1874 | // Rewind to start of escape: | |
1875 | --m_position; | |
1876 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1877 | fail(regex_constants::error_escape, m_position - m_base); | |
1878 | return false; | |
1879 | } | |
1880 | default: | |
1881 | result = *m_position; | |
1882 | break; | |
1883 | } | |
1884 | ++m_position; | |
1885 | return result; | |
1886 | #ifdef BOOST_MSVC | |
1887 | #pragma warning(pop) | |
1888 | #endif | |
1889 | } | |
1890 | ||
1891 | template <class charT, class traits> | |
1892 | bool basic_regex_parser<charT, traits>::parse_backref() | |
1893 | { | |
1894 | BOOST_ASSERT(m_position != m_end); | |
1895 | const charT* pc = m_position; | |
1896 | boost::intmax_t i = this->m_traits.toi(pc, pc + 1, 10); | |
1897 | if((i == 0) || (((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) && (this->flags() & regbase::no_bk_refs))) | |
1898 | { | |
1899 | // not a backref at all but an octal escape sequence: | |
1900 | charT c = unescape_character(); | |
1901 | this->append_literal(c); | |
1902 | } | |
1903 | else if((i > 0) && (this->m_backrefs & (1u << (i-1)))) | |
1904 | { | |
1905 | m_position = pc; | |
1906 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace))); | |
1907 | pb->index = i; | |
1908 | pb->icase = this->flags() & regbase::icase; | |
1909 | } | |
1910 | else | |
1911 | { | |
1912 | // Rewind to start of escape: | |
1913 | --m_position; | |
1914 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1915 | fail(regex_constants::error_backref, m_position - m_base); | |
1916 | return false; | |
1917 | } | |
1918 | return true; | |
1919 | } | |
1920 | ||
1921 | template <class charT, class traits> | |
1922 | bool basic_regex_parser<charT, traits>::parse_QE() | |
1923 | { | |
1924 | #ifdef BOOST_MSVC | |
1925 | #pragma warning(push) | |
1926 | #pragma warning(disable:4127) | |
1927 | #endif | |
1928 | // | |
1929 | // parse a \Q...\E sequence: | |
1930 | // | |
1931 | ++m_position; // skip the Q | |
1932 | const charT* start = m_position; | |
1933 | const charT* end; | |
1934 | do | |
1935 | { | |
1936 | while((m_position != m_end) | |
1937 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape)) | |
1938 | ++m_position; | |
1939 | if(m_position == m_end) | |
1940 | { | |
1941 | // a \Q...\E sequence may terminate with the end of the expression: | |
1942 | end = m_position; | |
1943 | break; | |
1944 | } | |
1945 | if(++m_position == m_end) // skip the escape | |
1946 | { | |
1947 | fail(regex_constants::error_escape, m_position - m_base, "Unterminated \\Q...\\E sequence."); | |
1948 | return false; | |
1949 | } | |
1950 | // check to see if it's a \E: | |
1951 | if(this->m_traits.escape_syntax_type(*m_position) == regex_constants::escape_type_E) | |
1952 | { | |
1953 | ++m_position; | |
1954 | end = m_position - 2; | |
1955 | break; | |
1956 | } | |
1957 | // otherwise go round again: | |
1958 | }while(true); | |
1959 | // | |
1960 | // now add all the character between the two escapes as literals: | |
1961 | // | |
1962 | while(start != end) | |
1963 | { | |
1964 | this->append_literal(*start); | |
1965 | ++start; | |
1966 | } | |
1967 | return true; | |
1968 | #ifdef BOOST_MSVC | |
1969 | #pragma warning(pop) | |
1970 | #endif | |
1971 | } | |
1972 | ||
1973 | template <class charT, class traits> | |
1974 | bool basic_regex_parser<charT, traits>::parse_perl_extension() | |
1975 | { | |
1976 | if(++m_position == m_end) | |
1977 | { | |
1978 | // Rewind to start of (? sequence: | |
1979 | --m_position; | |
1980 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
1981 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
1982 | return false; | |
1983 | } | |
1984 | // | |
1985 | // treat comments as a special case, as these | |
1986 | // are the only ones that don't start with a leading | |
1987 | // startmark state: | |
1988 | // | |
1989 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_hash) | |
1990 | { | |
1991 | while((m_position != m_end) | |
1992 | && (this->m_traits.syntax_type(*m_position++) != regex_constants::syntax_close_mark)) | |
1993 | {} | |
1994 | return true; | |
1995 | } | |
1996 | // | |
1997 | // backup some state, and prepare the way: | |
1998 | // | |
1999 | int markid = 0; | |
2000 | std::ptrdiff_t jump_offset = 0; | |
2001 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); | |
2002 | pb->icase = this->flags() & regbase::icase; | |
2003 | std::ptrdiff_t last_paren_start = this->getoffset(pb); | |
2004 | // back up insertion point for alternations, and set new point: | |
2005 | std::ptrdiff_t last_alt_point = m_alt_insert_point; | |
2006 | this->m_pdata->m_data.align(); | |
2007 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2008 | std::ptrdiff_t expected_alt_point = m_alt_insert_point; | |
2009 | bool restore_flags = true; | |
2010 | regex_constants::syntax_option_type old_flags = this->flags(); | |
2011 | bool old_case_change = m_has_case_change; | |
2012 | m_has_case_change = false; | |
2013 | charT name_delim; | |
2014 | int mark_reset = m_mark_reset; | |
2015 | int max_mark = m_max_mark; | |
2016 | m_mark_reset = -1; | |
2017 | m_max_mark = m_mark_count; | |
2018 | boost::intmax_t v; | |
2019 | // | |
2020 | // select the actual extension used: | |
2021 | // | |
2022 | switch(this->m_traits.syntax_type(*m_position)) | |
2023 | { | |
2024 | case regex_constants::syntax_or: | |
2025 | m_mark_reset = m_mark_count; | |
2026 | BOOST_FALLTHROUGH; | |
2027 | case regex_constants::syntax_colon: | |
2028 | // | |
2029 | // a non-capturing mark: | |
2030 | // | |
2031 | pb->index = markid = 0; | |
2032 | ++m_position; | |
2033 | break; | |
2034 | case regex_constants::syntax_digit: | |
2035 | { | |
2036 | // | |
2037 | // a recursive subexpression: | |
2038 | // | |
2039 | v = this->m_traits.toi(m_position, m_end, 10); | |
2040 | if((v < 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2041 | { | |
2042 | // Rewind to start of (? sequence: | |
2043 | --m_position; | |
2044 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2045 | fail(regex_constants::error_perl_extension, m_position - m_base, "The recursive sub-expression refers to an invalid marking group, or is unterminated."); | |
2046 | return false; | |
2047 | } | |
2048 | insert_recursion: | |
2049 | pb->index = markid = 0; | |
2050 | re_recurse* pr = static_cast<re_recurse*>(this->append_state(syntax_element_recurse, sizeof(re_recurse))); | |
2051 | pr->alt.i = v; | |
2052 | pr->state_id = 0; | |
2053 | static_cast<re_case*>( | |
2054 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
2055 | )->icase = this->flags() & regbase::icase; | |
2056 | break; | |
2057 | } | |
2058 | case regex_constants::syntax_plus: | |
2059 | // | |
2060 | // A forward-relative recursive subexpression: | |
2061 | // | |
2062 | ++m_position; | |
2063 | v = this->m_traits.toi(m_position, m_end, 10); | |
2064 | if((v <= 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2065 | { | |
2066 | // Rewind to start of (? sequence: | |
2067 | --m_position; | |
2068 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2069 | fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression."); | |
2070 | return false; | |
2071 | } | |
2072 | v += m_mark_count; | |
2073 | goto insert_recursion; | |
2074 | case regex_constants::syntax_dash: | |
2075 | // | |
2076 | // Possibly a backward-relative recursive subexpression: | |
2077 | // | |
2078 | ++m_position; | |
2079 | v = this->m_traits.toi(m_position, m_end, 10); | |
2080 | if(v <= 0) | |
2081 | { | |
2082 | --m_position; | |
2083 | // Oops not a relative recursion at all, but a (?-imsx) group: | |
2084 | goto option_group_jump; | |
2085 | } | |
2086 | v = m_mark_count + 1 - v; | |
2087 | if(v <= 0) | |
2088 | { | |
2089 | // Rewind to start of (? sequence: | |
2090 | --m_position; | |
2091 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2092 | fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression."); | |
2093 | return false; | |
2094 | } | |
2095 | goto insert_recursion; | |
2096 | case regex_constants::syntax_equal: | |
2097 | pb->index = markid = -1; | |
2098 | ++m_position; | |
2099 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
2100 | this->m_pdata->m_data.align(); | |
2101 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2102 | break; | |
2103 | case regex_constants::syntax_not: | |
2104 | pb->index = markid = -2; | |
2105 | ++m_position; | |
2106 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
2107 | this->m_pdata->m_data.align(); | |
2108 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2109 | break; | |
2110 | case regex_constants::escape_type_left_word: | |
2111 | { | |
2112 | // a lookbehind assertion: | |
2113 | if(++m_position == m_end) | |
2114 | { | |
2115 | // Rewind to start of (? sequence: | |
2116 | --m_position; | |
2117 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2118 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2119 | return false; | |
2120 | } | |
2121 | regex_constants::syntax_type t = this->m_traits.syntax_type(*m_position); | |
2122 | if(t == regex_constants::syntax_not) | |
2123 | pb->index = markid = -2; | |
2124 | else if(t == regex_constants::syntax_equal) | |
2125 | pb->index = markid = -1; | |
2126 | else | |
2127 | { | |
2128 | // Probably a named capture which also starts (?< : | |
2129 | name_delim = '>'; | |
2130 | --m_position; | |
2131 | goto named_capture_jump; | |
2132 | } | |
2133 | ++m_position; | |
2134 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
2135 | this->append_state(syntax_element_backstep, sizeof(re_brace)); | |
2136 | this->m_pdata->m_data.align(); | |
2137 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2138 | break; | |
2139 | } | |
2140 | case regex_constants::escape_type_right_word: | |
2141 | // | |
2142 | // an independent sub-expression: | |
2143 | // | |
2144 | pb->index = markid = -3; | |
2145 | ++m_position; | |
2146 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
2147 | this->m_pdata->m_data.align(); | |
2148 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2149 | break; | |
2150 | case regex_constants::syntax_open_mark: | |
2151 | { | |
2152 | // a conditional expression: | |
2153 | pb->index = markid = -4; | |
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 | v = this->m_traits.toi(m_position, m_end, 10); | |
2163 | if(m_position == m_end) | |
2164 | { | |
2165 | // Rewind to start of (? sequence: | |
2166 | --m_position; | |
2167 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2168 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2169 | return false; | |
2170 | } | |
2171 | if(*m_position == charT('R')) | |
2172 | { | |
2173 | if(++m_position == m_end) | |
2174 | { | |
2175 | // Rewind to start of (? sequence: | |
2176 | --m_position; | |
2177 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2178 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2179 | return false; | |
2180 | } | |
2181 | if(*m_position == charT('&')) | |
2182 | { | |
2183 | const charT* base = ++m_position; | |
2184 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2185 | ++m_position; | |
2186 | if(m_position == m_end) | |
2187 | { | |
2188 | // Rewind to start of (? sequence: | |
2189 | --m_position; | |
2190 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2191 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2192 | return false; | |
2193 | } | |
2194 | v = -static_cast<int>(hash_value_from_capture_name(base, m_position)); | |
2195 | } | |
2196 | else | |
2197 | { | |
2198 | v = -this->m_traits.toi(m_position, m_end, 10); | |
2199 | } | |
2200 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); | |
2201 | br->index = v < 0 ? (v - 1) : 0; | |
2202 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2203 | { | |
2204 | // Rewind to start of (? sequence: | |
2205 | --m_position; | |
2206 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2207 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2208 | return false; | |
2209 | } | |
2210 | if(++m_position == m_end) | |
2211 | { | |
2212 | // Rewind to start of (? sequence: | |
2213 | --m_position; | |
2214 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2215 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2216 | return false; | |
2217 | } | |
2218 | } | |
2219 | else if((*m_position == charT('\'')) || (*m_position == charT('<'))) | |
2220 | { | |
2221 | const charT* base = ++m_position; | |
2222 | while((m_position != m_end) && (*m_position != charT('>')) && (*m_position != charT('\''))) | |
2223 | ++m_position; | |
2224 | if(m_position == m_end) | |
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 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); | |
2233 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); | |
2234 | br->index = v; | |
2235 | if(((*m_position != charT('>')) && (*m_position != charT('\''))) || (++m_position == m_end)) | |
2236 | { | |
2237 | // Rewind to start of (? sequence: | |
2238 | --m_position; | |
2239 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2240 | fail(regex_constants::error_perl_extension, m_position - m_base, "Unterminated named capture."); | |
2241 | return false; | |
2242 | } | |
2243 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2244 | { | |
2245 | // Rewind to start of (? sequence: | |
2246 | --m_position; | |
2247 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2248 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2249 | return false; | |
2250 | } | |
2251 | if(++m_position == m_end) | |
2252 | { | |
2253 | // Rewind to start of (? sequence: | |
2254 | --m_position; | |
2255 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2256 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2257 | return false; | |
2258 | } | |
2259 | } | |
2260 | else if(*m_position == charT('D')) | |
2261 | { | |
2262 | const char* def = "DEFINE"; | |
2263 | while(*def && (m_position != m_end) && (*m_position == charT(*def))) | |
2264 | ++m_position, ++def; | |
2265 | if((m_position == m_end) || *def) | |
2266 | { | |
2267 | // Rewind to start of (? sequence: | |
2268 | --m_position; | |
2269 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2270 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2271 | return false; | |
2272 | } | |
2273 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); | |
2274 | br->index = 9999; // special magic value! | |
2275 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2276 | { | |
2277 | // Rewind to start of (? sequence: | |
2278 | --m_position; | |
2279 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2280 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2281 | return false; | |
2282 | } | |
2283 | if(++m_position == m_end) | |
2284 | { | |
2285 | // Rewind to start of (? sequence: | |
2286 | --m_position; | |
2287 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2288 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2289 | return false; | |
2290 | } | |
2291 | } | |
2292 | else if(v > 0) | |
2293 | { | |
2294 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); | |
2295 | br->index = v; | |
2296 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2297 | { | |
2298 | // Rewind to start of (? sequence: | |
2299 | --m_position; | |
2300 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2301 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2302 | return false; | |
2303 | } | |
2304 | if(++m_position == m_end) | |
2305 | { | |
2306 | // Rewind to start of (? sequence: | |
2307 | --m_position; | |
2308 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2309 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2310 | return false; | |
2311 | } | |
2312 | } | |
2313 | else | |
2314 | { | |
2315 | // verify that we have a lookahead or lookbehind assert: | |
2316 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_question) | |
2317 | { | |
2318 | // Rewind to start of (? sequence: | |
2319 | --m_position; | |
2320 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2321 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2322 | return false; | |
2323 | } | |
2324 | if(++m_position == m_end) | |
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 | if(this->m_traits.syntax_type(*m_position) == regex_constants::escape_type_left_word) | |
2333 | { | |
2334 | if(++m_position == m_end) | |
2335 | { | |
2336 | // Rewind to start of (? sequence: | |
2337 | --m_position; | |
2338 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2339 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2340 | return false; | |
2341 | } | |
2342 | if((this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal) | |
2343 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_not)) | |
2344 | { | |
2345 | // Rewind to start of (? sequence: | |
2346 | --m_position; | |
2347 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2348 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2349 | return false; | |
2350 | } | |
2351 | m_position -= 3; | |
2352 | } | |
2353 | else | |
2354 | { | |
2355 | if((this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal) | |
2356 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_not)) | |
2357 | { | |
2358 | // Rewind to start of (? sequence: | |
2359 | --m_position; | |
2360 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2361 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2362 | return false; | |
2363 | } | |
2364 | m_position -= 2; | |
2365 | } | |
2366 | } | |
2367 | break; | |
2368 | } | |
2369 | case regex_constants::syntax_close_mark: | |
2370 | // Rewind to start of (? sequence: | |
2371 | --m_position; | |
2372 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2373 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2374 | return false; | |
2375 | case regex_constants::escape_type_end_buffer: | |
2376 | { | |
2377 | name_delim = *m_position; | |
2378 | named_capture_jump: | |
2379 | markid = 0; | |
2380 | if(0 == (this->flags() & regbase::nosubs)) | |
2381 | { | |
2382 | markid = ++m_mark_count; | |
2383 | #ifndef BOOST_NO_STD_DISTANCE | |
2384 | if(this->flags() & regbase::save_subexpression_location) | |
2385 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>(std::distance(m_base, m_position) - 2, 0)); | |
2386 | #else | |
2387 | if(this->flags() & regbase::save_subexpression_location) | |
2388 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>((m_position - m_base) - 2, 0)); | |
2389 | #endif | |
2390 | } | |
2391 | pb->index = markid; | |
2392 | const charT* base = ++m_position; | |
2393 | if(m_position == m_end) | |
2394 | { | |
2395 | // Rewind to start of (? sequence: | |
2396 | --m_position; | |
2397 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2398 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2399 | return false; | |
2400 | } | |
2401 | while((m_position != m_end) && (*m_position != name_delim)) | |
2402 | ++m_position; | |
2403 | if(m_position == m_end) | |
2404 | { | |
2405 | // Rewind to start of (? sequence: | |
2406 | --m_position; | |
2407 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2408 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2409 | return false; | |
2410 | } | |
2411 | this->m_pdata->set_name(base, m_position, markid); | |
2412 | ++m_position; | |
2413 | break; | |
2414 | } | |
2415 | default: | |
2416 | if(*m_position == charT('R')) | |
2417 | { | |
2418 | ++m_position; | |
2419 | v = 0; | |
2420 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2421 | { | |
2422 | // Rewind to start of (? sequence: | |
2423 | --m_position; | |
2424 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2425 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2426 | return false; | |
2427 | } | |
2428 | goto insert_recursion; | |
2429 | } | |
2430 | if(*m_position == charT('&')) | |
2431 | { | |
2432 | ++m_position; | |
2433 | const charT* base = m_position; | |
2434 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2435 | ++m_position; | |
2436 | if(m_position == m_end) | |
2437 | { | |
2438 | // Rewind to start of (? sequence: | |
2439 | --m_position; | |
2440 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2441 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2442 | return false; | |
2443 | } | |
2444 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); | |
2445 | goto insert_recursion; | |
2446 | } | |
2447 | if(*m_position == charT('P')) | |
2448 | { | |
2449 | ++m_position; | |
2450 | if(m_position == m_end) | |
2451 | { | |
2452 | // Rewind to start of (? sequence: | |
2453 | --m_position; | |
2454 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2455 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2456 | return false; | |
2457 | } | |
2458 | if(*m_position == charT('>')) | |
2459 | { | |
2460 | ++m_position; | |
2461 | const charT* base = m_position; | |
2462 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2463 | ++m_position; | |
2464 | if(m_position == m_end) | |
2465 | { | |
2466 | // Rewind to start of (? sequence: | |
2467 | --m_position; | |
2468 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2469 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2470 | return false; | |
2471 | } | |
2472 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); | |
2473 | goto insert_recursion; | |
2474 | } | |
2475 | } | |
2476 | // | |
2477 | // lets assume that we have a (?imsx) group and try and parse it: | |
2478 | // | |
2479 | option_group_jump: | |
2480 | regex_constants::syntax_option_type opts = parse_options(); | |
2481 | if(m_position == m_end) | |
2482 | { | |
2483 | // Rewind to start of (? sequence: | |
2484 | --m_position; | |
2485 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2486 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2487 | return false; | |
2488 | } | |
2489 | // make a note of whether we have a case change: | |
2490 | m_has_case_change = ((opts & regbase::icase) != (this->flags() & regbase::icase)); | |
2491 | pb->index = markid = 0; | |
2492 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark) | |
2493 | { | |
2494 | // update flags and carry on as normal: | |
2495 | this->flags(opts); | |
2496 | restore_flags = false; | |
2497 | old_case_change |= m_has_case_change; // defer end of scope by one ')' | |
2498 | } | |
2499 | else if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_colon) | |
2500 | { | |
2501 | // update flags and carry on until the matching ')' is found: | |
2502 | this->flags(opts); | |
2503 | ++m_position; | |
2504 | } | |
2505 | else | |
2506 | { | |
2507 | // Rewind to start of (? sequence: | |
2508 | --m_position; | |
2509 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2510 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2511 | return false; | |
2512 | } | |
2513 | ||
2514 | // finally append a case change state if we need it: | |
2515 | if(m_has_case_change) | |
2516 | { | |
2517 | static_cast<re_case*>( | |
2518 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
2519 | )->icase = opts & regbase::icase; | |
2520 | } | |
2521 | ||
2522 | } | |
2523 | // | |
2524 | // now recursively add more states, this will terminate when we get to a | |
2525 | // matching ')' : | |
2526 | // | |
2527 | parse_all(); | |
2528 | // | |
2529 | // Unwind alternatives: | |
2530 | // | |
2531 | if(0 == unwind_alts(last_paren_start)) | |
2532 | { | |
2533 | // Rewind to start of (? sequence: | |
2534 | --m_position; | |
2535 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2536 | fail(regex_constants::error_perl_extension, m_position - m_base, "Invalid alternation operators within (?...) block."); | |
2537 | return false; | |
2538 | } | |
2539 | // | |
2540 | // we either have a ')' or we have run out of characters prematurely: | |
2541 | // | |
2542 | if(m_position == m_end) | |
2543 | { | |
2544 | // Rewind to start of (? sequence: | |
2545 | --m_position; | |
2546 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2547 | this->fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_end)); | |
2548 | return false; | |
2549 | } | |
2550 | BOOST_ASSERT(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark); | |
2551 | ++m_position; | |
2552 | // | |
2553 | // restore the flags: | |
2554 | // | |
2555 | if(restore_flags) | |
2556 | { | |
2557 | // append a case change state if we need it: | |
2558 | if(m_has_case_change) | |
2559 | { | |
2560 | static_cast<re_case*>( | |
2561 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
2562 | )->icase = old_flags & regbase::icase; | |
2563 | } | |
2564 | this->flags(old_flags); | |
2565 | } | |
2566 | // | |
2567 | // set up the jump pointer if we have one: | |
2568 | // | |
2569 | if(jump_offset) | |
2570 | { | |
2571 | this->m_pdata->m_data.align(); | |
2572 | re_jump* jmp = static_cast<re_jump*>(this->getaddress(jump_offset)); | |
2573 | jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp); | |
2574 | if((this->m_last_state == jmp) && (markid != -2)) | |
2575 | { | |
2576 | // Oops... we didn't have anything inside the assertion. | |
2577 | // Note we don't get here for negated forward lookahead as (?!) | |
2578 | // does have some uses. | |
2579 | // Rewind to start of (? sequence: | |
2580 | --m_position; | |
2581 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2582 | fail(regex_constants::error_perl_extension, m_position - m_base, "Invalid or empty zero width assertion."); | |
2583 | return false; | |
2584 | } | |
2585 | } | |
2586 | // | |
2587 | // verify that if this is conditional expression, that we do have | |
2588 | // an alternative, if not add one: | |
2589 | // | |
2590 | if(markid == -4) | |
2591 | { | |
2592 | re_syntax_base* b = this->getaddress(expected_alt_point); | |
2593 | // Make sure we have exactly one alternative following this state: | |
2594 | if(b->type != syntax_element_alt) | |
2595 | { | |
2596 | re_alt* alt = static_cast<re_alt*>(this->insert_state(expected_alt_point, syntax_element_alt, sizeof(re_alt))); | |
2597 | alt->alt.i = this->m_pdata->m_data.size() - this->getoffset(alt); | |
2598 | } | |
2599 | else if(this->getaddress(static_cast<re_alt*>(b)->alt.i, b)->type == syntax_element_alt) | |
2600 | { | |
2601 | // Can't have seen more than one alternative: | |
2602 | // Rewind to start of (? sequence: | |
2603 | --m_position; | |
2604 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2605 | fail(regex_constants::error_bad_pattern, m_position - m_base, "More than one alternation operator | was encountered inside a conditional expression."); | |
2606 | return false; | |
2607 | } | |
2608 | else | |
2609 | { | |
2610 | // We must *not* have seen an alternative inside a (DEFINE) block: | |
2611 | b = this->getaddress(b->next.i, b); | |
2612 | if((b->type == syntax_element_assert_backref) && (static_cast<re_brace*>(b)->index == 9999)) | |
2613 | { | |
2614 | // Rewind to start of (? sequence: | |
2615 | --m_position; | |
2616 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2617 | fail(regex_constants::error_bad_pattern, m_position - m_base, "Alternation operators are not allowed inside a DEFINE block."); | |
2618 | return false; | |
2619 | } | |
2620 | } | |
2621 | // check for invalid repetition of next state: | |
2622 | b = this->getaddress(expected_alt_point); | |
2623 | b = this->getaddress(static_cast<re_alt*>(b)->next.i, b); | |
2624 | if((b->type != syntax_element_assert_backref) | |
2625 | && (b->type != syntax_element_startmark)) | |
2626 | { | |
2627 | // Rewind to start of (? sequence: | |
2628 | --m_position; | |
2629 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2630 | fail(regex_constants::error_badrepeat, m_position - m_base, "A repetition operator cannot be applied to a zero-width assertion."); | |
2631 | return false; | |
2632 | } | |
2633 | } | |
2634 | // | |
2635 | // append closing parenthesis state: | |
2636 | // | |
2637 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); | |
2638 | pb->index = markid; | |
2639 | pb->icase = this->flags() & regbase::icase; | |
2640 | this->m_paren_start = last_paren_start; | |
2641 | // | |
2642 | // restore the alternate insertion point: | |
2643 | // | |
2644 | this->m_alt_insert_point = last_alt_point; | |
2645 | // | |
2646 | // and the case change data: | |
2647 | // | |
2648 | m_has_case_change = old_case_change; | |
2649 | // | |
2650 | // And the mark_reset data: | |
2651 | // | |
2652 | if(m_max_mark > m_mark_count) | |
2653 | { | |
2654 | m_mark_count = m_max_mark; | |
2655 | } | |
2656 | m_mark_reset = mark_reset; | |
2657 | m_max_mark = max_mark; | |
2658 | ||
2659 | ||
2660 | if(markid > 0) | |
2661 | { | |
2662 | #ifndef BOOST_NO_STD_DISTANCE | |
2663 | if(this->flags() & regbase::save_subexpression_location) | |
2664 | this->m_pdata->m_subs.at(markid - 1).second = std::distance(m_base, m_position) - 1; | |
2665 | #else | |
2666 | if(this->flags() & regbase::save_subexpression_location) | |
2667 | this->m_pdata->m_subs.at(markid - 1).second = (m_position - m_base) - 1; | |
2668 | #endif | |
2669 | // | |
2670 | // allow backrefs to this mark: | |
2671 | // | |
2672 | if(markid < (int)(sizeof(unsigned) * CHAR_BIT)) | |
2673 | this->m_backrefs |= 1u << (markid - 1); | |
2674 | } | |
2675 | return true; | |
2676 | } | |
2677 | ||
2678 | template <class charT, class traits> | |
2679 | bool basic_regex_parser<charT, traits>::match_verb(const char* verb) | |
2680 | { | |
2681 | while(*verb) | |
2682 | { | |
2683 | if(static_cast<charT>(*verb) != *m_position) | |
2684 | { | |
2685 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2686 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2687 | return false; | |
2688 | } | |
2689 | if(++m_position == m_end) | |
2690 | { | |
2691 | --m_position; | |
2692 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2693 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2694 | return false; | |
2695 | } | |
2696 | ++verb; | |
2697 | } | |
2698 | return true; | |
2699 | } | |
2700 | ||
2701 | template <class charT, class traits> | |
2702 | bool basic_regex_parser<charT, traits>::parse_perl_verb() | |
2703 | { | |
2704 | if(++m_position == m_end) | |
2705 | { | |
2706 | // Rewind to start of (* sequence: | |
2707 | --m_position; | |
2708 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2709 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2710 | return false; | |
2711 | } | |
2712 | switch(*m_position) | |
2713 | { | |
2714 | case 'F': | |
2715 | if(++m_position == m_end) | |
2716 | { | |
2717 | // Rewind to start of (* sequence: | |
2718 | --m_position; | |
2719 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2720 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2721 | return false; | |
2722 | } | |
2723 | if((this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark) || match_verb("AIL")) | |
2724 | { | |
2725 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2726 | { | |
2727 | // Rewind to start of (* sequence: | |
2728 | --m_position; | |
2729 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2730 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2731 | return false; | |
2732 | } | |
2733 | ++m_position; | |
2734 | this->append_state(syntax_element_fail); | |
2735 | return true; | |
2736 | } | |
2737 | break; | |
2738 | case 'A': | |
2739 | if(++m_position == m_end) | |
2740 | { | |
2741 | // Rewind to start of (* sequence: | |
2742 | --m_position; | |
2743 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2744 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2745 | return false; | |
2746 | } | |
2747 | if(match_verb("CCEPT")) | |
2748 | { | |
2749 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2750 | { | |
2751 | // Rewind to start of (* sequence: | |
2752 | --m_position; | |
2753 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2754 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2755 | return false; | |
2756 | } | |
2757 | ++m_position; | |
2758 | this->append_state(syntax_element_accept); | |
2759 | return true; | |
2760 | } | |
2761 | break; | |
2762 | case 'C': | |
2763 | if(++m_position == m_end) | |
2764 | { | |
2765 | // Rewind to start of (* sequence: | |
2766 | --m_position; | |
2767 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2768 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2769 | return false; | |
2770 | } | |
2771 | if(match_verb("OMMIT")) | |
2772 | { | |
2773 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2774 | { | |
2775 | // Rewind to start of (* sequence: | |
2776 | --m_position; | |
2777 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2778 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2779 | return false; | |
2780 | } | |
2781 | ++m_position; | |
2782 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_commit; | |
2783 | this->m_pdata->m_disable_match_any = true; | |
2784 | return true; | |
2785 | } | |
2786 | break; | |
2787 | case 'P': | |
2788 | if(++m_position == m_end) | |
2789 | { | |
2790 | // Rewind to start of (* sequence: | |
2791 | --m_position; | |
2792 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2793 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2794 | return false; | |
2795 | } | |
2796 | if(match_verb("RUNE")) | |
2797 | { | |
2798 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2799 | { | |
2800 | // Rewind to start of (* sequence: | |
2801 | --m_position; | |
2802 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2803 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2804 | return false; | |
2805 | } | |
2806 | ++m_position; | |
2807 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_prune; | |
2808 | this->m_pdata->m_disable_match_any = true; | |
2809 | return true; | |
2810 | } | |
2811 | break; | |
2812 | case 'S': | |
2813 | if(++m_position == m_end) | |
2814 | { | |
2815 | // Rewind to start of (* sequence: | |
2816 | --m_position; | |
2817 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2818 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2819 | return false; | |
2820 | } | |
2821 | if(match_verb("KIP")) | |
2822 | { | |
2823 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2824 | { | |
2825 | // Rewind to start of (* sequence: | |
2826 | --m_position; | |
2827 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2828 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2829 | return false; | |
2830 | } | |
2831 | ++m_position; | |
2832 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_skip; | |
2833 | this->m_pdata->m_disable_match_any = true; | |
2834 | return true; | |
2835 | } | |
2836 | break; | |
2837 | case 'T': | |
2838 | if(++m_position == m_end) | |
2839 | { | |
2840 | // Rewind to start of (* sequence: | |
2841 | --m_position; | |
2842 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2843 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2844 | return false; | |
2845 | } | |
2846 | if(match_verb("HEN")) | |
2847 | { | |
2848 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2849 | { | |
2850 | // Rewind to start of (* sequence: | |
2851 | --m_position; | |
2852 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2853 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2854 | return false; | |
2855 | } | |
2856 | ++m_position; | |
2857 | this->append_state(syntax_element_then); | |
2858 | this->m_pdata->m_disable_match_any = true; | |
2859 | return true; | |
2860 | } | |
2861 | break; | |
2862 | } | |
2863 | return false; | |
2864 | } | |
2865 | ||
2866 | template <class charT, class traits> | |
2867 | bool basic_regex_parser<charT, traits>::add_emacs_code(bool negate) | |
2868 | { | |
2869 | // | |
2870 | // parses an emacs style \sx or \Sx construct. | |
2871 | // | |
2872 | if(++m_position == m_end) | |
2873 | { | |
2874 | // Rewind to start of sequence: | |
2875 | --m_position; | |
2876 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
2877 | fail(regex_constants::error_escape, m_position - m_base); | |
2878 | return false; | |
2879 | } | |
2880 | basic_char_set<charT, traits> char_set; | |
2881 | if(negate) | |
2882 | char_set.negate(); | |
2883 | ||
2884 | static const charT s_punct[5] = { 'p', 'u', 'n', 'c', 't', }; | |
2885 | ||
2886 | switch(*m_position) | |
2887 | { | |
2888 | case 's': | |
2889 | case ' ': | |
2890 | char_set.add_class(this->m_mask_space); | |
2891 | break; | |
2892 | case 'w': | |
2893 | char_set.add_class(this->m_word_mask); | |
2894 | break; | |
2895 | case '_': | |
2896 | char_set.add_single(digraph<charT>(charT('$'))); | |
2897 | char_set.add_single(digraph<charT>(charT('&'))); | |
2898 | char_set.add_single(digraph<charT>(charT('*'))); | |
2899 | char_set.add_single(digraph<charT>(charT('+'))); | |
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 | char_set.add_single(digraph<charT>(charT('>'))); | |
2904 | break; | |
2905 | case '.': | |
2906 | char_set.add_class(this->m_traits.lookup_classname(s_punct, s_punct+5)); | |
2907 | break; | |
2908 | case '(': | |
2909 | char_set.add_single(digraph<charT>(charT('('))); | |
2910 | char_set.add_single(digraph<charT>(charT('['))); | |
2911 | char_set.add_single(digraph<charT>(charT('{'))); | |
2912 | break; | |
2913 | case ')': | |
2914 | char_set.add_single(digraph<charT>(charT(')'))); | |
2915 | char_set.add_single(digraph<charT>(charT(']'))); | |
2916 | char_set.add_single(digraph<charT>(charT('}'))); | |
2917 | break; | |
2918 | case '"': | |
2919 | char_set.add_single(digraph<charT>(charT('"'))); | |
2920 | char_set.add_single(digraph<charT>(charT('\''))); | |
2921 | char_set.add_single(digraph<charT>(charT('`'))); | |
2922 | break; | |
2923 | case '\'': | |
2924 | char_set.add_single(digraph<charT>(charT('\''))); | |
2925 | char_set.add_single(digraph<charT>(charT(','))); | |
2926 | char_set.add_single(digraph<charT>(charT('#'))); | |
2927 | break; | |
2928 | case '<': | |
2929 | char_set.add_single(digraph<charT>(charT(';'))); | |
2930 | break; | |
2931 | case '>': | |
2932 | char_set.add_single(digraph<charT>(charT('\n'))); | |
2933 | char_set.add_single(digraph<charT>(charT('\f'))); | |
2934 | break; | |
2935 | default: | |
2936 | fail(regex_constants::error_ctype, m_position - m_base); | |
2937 | return false; | |
2938 | } | |
2939 | if(0 == this->append_set(char_set)) | |
2940 | { | |
2941 | fail(regex_constants::error_ctype, m_position - m_base); | |
2942 | return false; | |
2943 | } | |
2944 | ++m_position; | |
2945 | return true; | |
2946 | } | |
2947 | ||
2948 | template <class charT, class traits> | |
2949 | regex_constants::syntax_option_type basic_regex_parser<charT, traits>::parse_options() | |
2950 | { | |
2951 | // we have a (?imsx-imsx) group, convert it into a set of flags: | |
2952 | regex_constants::syntax_option_type f = this->flags(); | |
2953 | bool breakout = false; | |
2954 | do | |
2955 | { | |
2956 | switch(*m_position) | |
2957 | { | |
2958 | case 's': | |
2959 | f |= regex_constants::mod_s; | |
2960 | f &= ~regex_constants::no_mod_s; | |
2961 | break; | |
2962 | case 'm': | |
2963 | f &= ~regex_constants::no_mod_m; | |
2964 | break; | |
2965 | case 'i': | |
2966 | f |= regex_constants::icase; | |
2967 | break; | |
2968 | case 'x': | |
2969 | f |= regex_constants::mod_x; | |
2970 | break; | |
2971 | default: | |
2972 | breakout = true; | |
2973 | continue; | |
2974 | } | |
2975 | if(++m_position == m_end) | |
2976 | { | |
2977 | // Rewind to start of (? sequence: | |
2978 | --m_position; | |
2979 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2980 | fail(regex_constants::error_paren, m_position - m_base); | |
2981 | return false; | |
2982 | } | |
2983 | } | |
2984 | while(!breakout); | |
2985 | ||
2986 | breakout = false; | |
2987 | ||
2988 | if(*m_position == static_cast<charT>('-')) | |
2989 | { | |
2990 | if(++m_position == m_end) | |
2991 | { | |
2992 | // Rewind to start of (? sequence: | |
2993 | --m_position; | |
2994 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2995 | fail(regex_constants::error_paren, m_position - m_base); | |
2996 | return false; | |
2997 | } | |
2998 | do | |
2999 | { | |
3000 | switch(*m_position) | |
3001 | { | |
3002 | case 's': | |
3003 | f &= ~regex_constants::mod_s; | |
3004 | f |= regex_constants::no_mod_s; | |
3005 | break; | |
3006 | case 'm': | |
3007 | f |= regex_constants::no_mod_m; | |
3008 | break; | |
3009 | case 'i': | |
3010 | f &= ~regex_constants::icase; | |
3011 | break; | |
3012 | case 'x': | |
3013 | f &= ~regex_constants::mod_x; | |
3014 | break; | |
3015 | default: | |
3016 | breakout = true; | |
3017 | continue; | |
3018 | } | |
3019 | if(++m_position == m_end) | |
3020 | { | |
3021 | // Rewind to start of (? sequence: | |
3022 | --m_position; | |
3023 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
3024 | fail(regex_constants::error_paren, m_position - m_base); | |
3025 | return false; | |
3026 | } | |
3027 | } | |
3028 | while(!breakout); | |
3029 | } | |
3030 | return f; | |
3031 | } | |
3032 | ||
3033 | template <class charT, class traits> | |
3034 | bool basic_regex_parser<charT, traits>::unwind_alts(std::ptrdiff_t last_paren_start) | |
3035 | { | |
3036 | // | |
3037 | // If we didn't actually add any states after the last | |
3038 | // alternative then that's an error: | |
3039 | // | |
3040 | if((this->m_alt_insert_point == static_cast<std::ptrdiff_t>(this->m_pdata->m_data.size())) | |
3041 | && m_alt_jumps.size() && (m_alt_jumps.back() > last_paren_start) | |
3042 | && | |
3043 | !( | |
3044 | ((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) | |
3045 | && | |
3046 | ((this->flags() & regbase::no_empty_expressions) == 0) | |
3047 | ) | |
3048 | ) | |
3049 | { | |
3050 | fail(regex_constants::error_empty, this->m_position - this->m_base, "Can't terminate a sub-expression with an alternation operator |."); | |
3051 | return false; | |
3052 | } | |
3053 | // | |
3054 | // Fix up our alternatives: | |
3055 | // | |
3056 | while(m_alt_jumps.size() && (m_alt_jumps.back() > last_paren_start)) | |
3057 | { | |
3058 | // | |
3059 | // fix up the jump to point to the end of the states | |
3060 | // that we've just added: | |
3061 | // | |
3062 | std::ptrdiff_t jump_offset = m_alt_jumps.back(); | |
3063 | m_alt_jumps.pop_back(); | |
3064 | this->m_pdata->m_data.align(); | |
3065 | re_jump* jmp = static_cast<re_jump*>(this->getaddress(jump_offset)); | |
3066 | BOOST_ASSERT(jmp->type == syntax_element_jump); | |
3067 | jmp->alt.i = this->m_pdata->m_data.size() - jump_offset; | |
3068 | } | |
3069 | return true; | |
3070 | } | |
3071 | ||
3072 | #ifdef BOOST_MSVC | |
3073 | #pragma warning(pop) | |
3074 | #endif | |
3075 | ||
3076 | } // namespace BOOST_REGEX_DETAIL_NS | |
3077 | } // namespace boost | |
3078 | ||
3079 | #ifdef BOOST_MSVC | |
3080 | #pragma warning(push) | |
3081 | #pragma warning(disable: 4103) | |
3082 | #endif | |
3083 | #ifdef BOOST_HAS_ABI_HEADERS | |
3084 | # include BOOST_ABI_SUFFIX | |
3085 | #endif | |
3086 | #ifdef BOOST_MSVC | |
3087 | #pragma warning(pop) | |
3088 | #endif | |
3089 | ||
3090 | #endif |