]>
Commit | Line | Data |
---|---|---|
7c673cae FG |
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) | |
20effc67 TL |
25 | #if BOOST_MSVC >= 1800 |
26 | #pragma warning(disable: 26812) | |
27 | #endif | |
7c673cae FG |
28 | #endif |
29 | #ifdef BOOST_HAS_ABI_HEADERS | |
30 | # include BOOST_ABI_PREFIX | |
31 | #endif | |
32 | #ifdef BOOST_MSVC | |
33 | #pragma warning(pop) | |
34 | #endif | |
35 | ||
36 | namespace boost{ | |
37 | namespace BOOST_REGEX_DETAIL_NS{ | |
38 | ||
39 | #ifdef BOOST_MSVC | |
40 | #pragma warning(push) | |
92f5a8d4 TL |
41 | #pragma warning(disable:4244) |
42 | #if BOOST_MSVC < 1910 | |
43 | #pragma warning(disable:4800) | |
44 | #endif | |
7c673cae FG |
45 | #endif |
46 | ||
47 | inline boost::intmax_t umax(mpl::false_ const&) | |
48 | { | |
49 | // Get out clause here, just in case numeric_limits is unspecialized: | |
50 | return std::numeric_limits<boost::intmax_t>::is_specialized ? (std::numeric_limits<boost::intmax_t>::max)() : INT_MAX; | |
51 | } | |
52 | inline boost::intmax_t umax(mpl::true_ const&) | |
53 | { | |
54 | return (std::numeric_limits<std::size_t>::max)(); | |
55 | } | |
56 | ||
57 | inline boost::intmax_t umax() | |
58 | { | |
59 | return umax(mpl::bool_<std::numeric_limits<boost::intmax_t>::digits >= std::numeric_limits<std::size_t>::digits>()); | |
60 | } | |
61 | ||
62 | template <class charT, class traits> | |
63 | class basic_regex_parser : public basic_regex_creator<charT, traits> | |
64 | { | |
65 | public: | |
66 | basic_regex_parser(regex_data<charT, traits>* data); | |
67 | void parse(const charT* p1, const charT* p2, unsigned flags); | |
68 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position); | |
69 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position, std::string message, std::ptrdiff_t start_pos); | |
70 | void fail(regex_constants::error_type error_code, std::ptrdiff_t position, const std::string& message) | |
71 | { | |
72 | fail(error_code, position, message, position); | |
73 | } | |
74 | ||
75 | bool parse_all(); | |
76 | bool parse_basic(); | |
77 | bool parse_extended(); | |
78 | bool parse_literal(); | |
79 | bool parse_open_paren(); | |
80 | bool parse_basic_escape(); | |
81 | bool parse_extended_escape(); | |
82 | bool parse_match_any(); | |
83 | bool parse_repeat(std::size_t low = 0, std::size_t high = (std::numeric_limits<std::size_t>::max)()); | |
84 | bool parse_repeat_range(bool isbasic); | |
85 | bool parse_alt(); | |
86 | bool parse_set(); | |
87 | bool parse_backref(); | |
88 | void parse_set_literal(basic_char_set<charT, traits>& char_set); | |
89 | bool parse_inner_set(basic_char_set<charT, traits>& char_set); | |
90 | bool parse_QE(); | |
91 | bool parse_perl_extension(); | |
92 | bool parse_perl_verb(); | |
93 | bool match_verb(const char*); | |
94 | bool add_emacs_code(bool negate); | |
95 | bool unwind_alts(std::ptrdiff_t last_paren_start); | |
96 | digraph<charT> get_next_set_literal(basic_char_set<charT, traits>& char_set); | |
97 | charT unescape_character(); | |
98 | regex_constants::syntax_option_type parse_options(); | |
99 | ||
100 | private: | |
101 | typedef bool (basic_regex_parser::*parser_proc_type)(); | |
102 | typedef typename traits::string_type string_type; | |
103 | typedef typename traits::char_class_type char_class_type; | |
104 | parser_proc_type m_parser_proc; // the main parser to use | |
105 | const charT* m_base; // the start of the string being parsed | |
106 | const charT* m_end; // the end of the string being parsed | |
107 | const charT* m_position; // our current parser position | |
108 | unsigned m_mark_count; // how many sub-expressions we have | |
109 | int m_mark_reset; // used to indicate that we're inside a (?|...) block. | |
110 | unsigned m_max_mark; // largest mark count seen inside a (?|...) block. | |
111 | std::ptrdiff_t m_paren_start; // where the last seen ')' began (where repeats are inserted). | |
112 | std::ptrdiff_t m_alt_insert_point; // where to insert the next alternative | |
113 | bool m_has_case_change; // true if somewhere in the current block the case has changed | |
b32b8144 | 114 | unsigned m_recursion_count; // How many times we've called parse_all. |
7c673cae FG |
115 | #if defined(BOOST_MSVC) && defined(_M_IX86) |
116 | // This is an ugly warning suppression workaround (for warnings *inside* std::vector | |
117 | // that can not otherwise be suppressed)... | |
118 | BOOST_STATIC_ASSERT(sizeof(long) >= sizeof(void*)); | |
119 | std::vector<long> m_alt_jumps; // list of alternative in the current scope. | |
120 | #else | |
121 | std::vector<std::ptrdiff_t> m_alt_jumps; // list of alternative in the current scope. | |
122 | #endif | |
123 | ||
124 | basic_regex_parser& operator=(const basic_regex_parser&); | |
125 | basic_regex_parser(const basic_regex_parser&); | |
126 | }; | |
127 | ||
128 | template <class charT, class traits> | |
129 | basic_regex_parser<charT, traits>::basic_regex_parser(regex_data<charT, traits>* data) | |
20effc67 TL |
130 | : basic_regex_creator<charT, traits>(data), m_parser_proc(), m_base(0), m_end(0), m_position(0), |
131 | 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), m_recursion_count(0) | |
7c673cae FG |
132 | { |
133 | } | |
134 | ||
135 | template <class charT, class traits> | |
136 | void basic_regex_parser<charT, traits>::parse(const charT* p1, const charT* p2, unsigned l_flags) | |
137 | { | |
138 | // pass l_flags on to base class: | |
139 | this->init(l_flags); | |
140 | // set up pointers: | |
141 | m_position = m_base = p1; | |
142 | m_end = p2; | |
143 | // empty strings are errors: | |
144 | if((p1 == p2) && | |
145 | ( | |
146 | ((l_flags & regbase::main_option_type) != regbase::perl_syntax_group) | |
147 | || (l_flags & regbase::no_empty_expressions) | |
148 | ) | |
149 | ) | |
150 | { | |
151 | fail(regex_constants::error_empty, 0); | |
152 | return; | |
153 | } | |
154 | // select which parser to use: | |
155 | switch(l_flags & regbase::main_option_type) | |
156 | { | |
157 | case regbase::perl_syntax_group: | |
158 | { | |
159 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_extended; | |
160 | // | |
161 | // Add a leading paren with index zero to give recursions a target: | |
162 | // | |
163 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); | |
164 | br->index = 0; | |
165 | br->icase = this->flags() & regbase::icase; | |
166 | break; | |
167 | } | |
168 | case regbase::basic_syntax_group: | |
169 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_basic; | |
170 | break; | |
171 | case regbase::literal: | |
172 | m_parser_proc = &basic_regex_parser<charT, traits>::parse_literal; | |
173 | break; | |
174 | default: | |
1e59de90 | 175 | // Oops, someone has managed to set more than one of the main option flags, |
7c673cae FG |
176 | // so this must be an error: |
177 | fail(regex_constants::error_unknown, 0, "An invalid combination of regular expression syntax flags was used."); | |
178 | return; | |
179 | } | |
180 | ||
181 | // parse all our characters: | |
182 | bool result = parse_all(); | |
183 | // | |
184 | // Unwind our alternatives: | |
185 | // | |
186 | unwind_alts(-1); | |
187 | // reset l_flags as a global scope (?imsx) may have altered them: | |
188 | this->flags(l_flags); | |
189 | // if we haven't gobbled up all the characters then we must | |
190 | // have had an unexpected ')' : | |
191 | if(!result) | |
192 | { | |
92f5a8d4 | 193 | fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_position), "Found a closing ) with no corresponding opening parenthesis."); |
7c673cae FG |
194 | return; |
195 | } | |
196 | // if an error has been set then give up now: | |
197 | if(this->m_pdata->m_status) | |
198 | return; | |
199 | // fill in our sub-expression count: | |
20effc67 | 200 | this->m_pdata->m_mark_count = 1u + (std::size_t)m_mark_count; |
7c673cae FG |
201 | this->finalize(p1, p2); |
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) | |
206 | { | |
207 | // get the error message: | |
208 | std::string message = this->m_pdata->m_ptraits->error_string(error_code); | |
209 | fail(error_code, position, message); | |
210 | } | |
211 | ||
212 | template <class charT, class traits> | |
213 | 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) | |
214 | { | |
215 | if(0 == this->m_pdata->m_status) // update the error code if not already set | |
216 | this->m_pdata->m_status = error_code; | |
217 | m_position = m_end; // don't bother parsing anything else | |
218 | ||
219 | #ifndef BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS | |
220 | // | |
221 | // Augment error message with the regular expression text: | |
222 | // | |
223 | if(start_pos == position) | |
224 | start_pos = (std::max)(static_cast<std::ptrdiff_t>(0), position - static_cast<std::ptrdiff_t>(10)); | |
225 | std::ptrdiff_t end_pos = (std::min)(position + static_cast<std::ptrdiff_t>(10), static_cast<std::ptrdiff_t>(m_end - m_base)); | |
226 | if(error_code != regex_constants::error_empty) | |
227 | { | |
228 | if((start_pos != 0) || (end_pos != (m_end - m_base))) | |
229 | message += " The error occurred while parsing the regular expression fragment: '"; | |
230 | else | |
231 | message += " The error occurred while parsing the regular expression: '"; | |
232 | if(start_pos != end_pos) | |
233 | { | |
234 | message += std::string(m_base + start_pos, m_base + position); | |
235 | message += ">>>HERE>>>"; | |
236 | message += std::string(m_base + position, m_base + end_pos); | |
237 | } | |
238 | message += "'."; | |
239 | } | |
240 | #endif | |
241 | ||
242 | #ifndef BOOST_NO_EXCEPTIONS | |
243 | if(0 == (this->flags() & regex_constants::no_except)) | |
244 | { | |
245 | boost::regex_error e(message, error_code, position); | |
246 | e.raise(); | |
247 | } | |
248 | #else | |
249 | (void)position; // suppress warnings. | |
250 | #endif | |
251 | } | |
252 | ||
253 | template <class charT, class traits> | |
254 | bool basic_regex_parser<charT, traits>::parse_all() | |
255 | { | |
b32b8144 FG |
256 | if (++m_recursion_count > 400) |
257 | { | |
258 | // exceeded internal limits | |
259 | fail(boost::regex_constants::error_complexity, m_position - m_base, "Exceeded nested brace limit."); | |
260 | } | |
7c673cae FG |
261 | bool result = true; |
262 | while(result && (m_position != m_end)) | |
263 | { | |
264 | result = (this->*m_parser_proc)(); | |
265 | } | |
b32b8144 | 266 | --m_recursion_count; |
7c673cae FG |
267 | return result; |
268 | } | |
269 | ||
270 | #ifdef BOOST_MSVC | |
271 | #pragma warning(push) | |
272 | #pragma warning(disable:4702) | |
273 | #endif | |
274 | template <class charT, class traits> | |
275 | bool basic_regex_parser<charT, traits>::parse_basic() | |
276 | { | |
277 | switch(this->m_traits.syntax_type(*m_position)) | |
278 | { | |
279 | case regex_constants::syntax_escape: | |
280 | return parse_basic_escape(); | |
281 | case regex_constants::syntax_dot: | |
282 | return parse_match_any(); | |
283 | case regex_constants::syntax_caret: | |
284 | ++m_position; | |
285 | this->append_state(syntax_element_start_line); | |
286 | break; | |
287 | case regex_constants::syntax_dollar: | |
288 | ++m_position; | |
289 | this->append_state(syntax_element_end_line); | |
290 | break; | |
291 | case regex_constants::syntax_star: | |
292 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line)) | |
293 | return parse_literal(); | |
294 | else | |
295 | { | |
296 | ++m_position; | |
297 | return parse_repeat(); | |
298 | } | |
299 | case regex_constants::syntax_plus: | |
300 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line) || !(this->flags() & regbase::emacs_ex)) | |
301 | return parse_literal(); | |
302 | else | |
303 | { | |
304 | ++m_position; | |
305 | return parse_repeat(1); | |
306 | } | |
307 | case regex_constants::syntax_question: | |
308 | if(!(this->m_last_state) || (this->m_last_state->type == syntax_element_start_line) || !(this->flags() & regbase::emacs_ex)) | |
309 | return parse_literal(); | |
310 | else | |
311 | { | |
312 | ++m_position; | |
313 | return parse_repeat(0, 1); | |
314 | } | |
315 | case regex_constants::syntax_open_set: | |
316 | return parse_set(); | |
317 | case regex_constants::syntax_newline: | |
318 | if(this->flags() & regbase::newline_alt) | |
319 | return parse_alt(); | |
320 | else | |
321 | return parse_literal(); | |
322 | default: | |
323 | return parse_literal(); | |
324 | } | |
325 | return true; | |
326 | } | |
327 | ||
20effc67 TL |
328 | #ifdef BOOST_MSVC |
329 | # pragma warning(push) | |
330 | #if BOOST_MSVC >= 1800 | |
331 | #pragma warning(disable:26812) | |
332 | #endif | |
333 | #endif | |
7c673cae FG |
334 | template <class charT, class traits> |
335 | bool basic_regex_parser<charT, traits>::parse_extended() | |
336 | { | |
337 | bool result = true; | |
338 | switch(this->m_traits.syntax_type(*m_position)) | |
339 | { | |
340 | case regex_constants::syntax_open_mark: | |
341 | return parse_open_paren(); | |
342 | case regex_constants::syntax_close_mark: | |
343 | return false; | |
344 | case regex_constants::syntax_escape: | |
345 | return parse_extended_escape(); | |
346 | case regex_constants::syntax_dot: | |
347 | return parse_match_any(); | |
348 | case regex_constants::syntax_caret: | |
349 | ++m_position; | |
350 | this->append_state( | |
351 | (this->flags() & regex_constants::no_mod_m ? syntax_element_buffer_start : syntax_element_start_line)); | |
352 | break; | |
353 | case regex_constants::syntax_dollar: | |
354 | ++m_position; | |
355 | this->append_state( | |
356 | (this->flags() & regex_constants::no_mod_m ? syntax_element_buffer_end : syntax_element_end_line)); | |
357 | break; | |
358 | case regex_constants::syntax_star: | |
359 | if(m_position == this->m_base) | |
360 | { | |
361 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"*\" cannot start a regular expression."); | |
362 | return false; | |
363 | } | |
364 | ++m_position; | |
365 | return parse_repeat(); | |
366 | case regex_constants::syntax_question: | |
367 | if(m_position == this->m_base) | |
368 | { | |
369 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"?\" cannot start a regular expression."); | |
370 | return false; | |
371 | } | |
372 | ++m_position; | |
373 | return parse_repeat(0,1); | |
374 | case regex_constants::syntax_plus: | |
375 | if(m_position == this->m_base) | |
376 | { | |
377 | fail(regex_constants::error_badrepeat, 0, "The repeat operator \"+\" cannot start a regular expression."); | |
378 | return false; | |
379 | } | |
380 | ++m_position; | |
381 | return parse_repeat(1); | |
382 | case regex_constants::syntax_open_brace: | |
383 | ++m_position; | |
384 | return parse_repeat_range(false); | |
385 | case regex_constants::syntax_close_brace: | |
386 | if((this->flags() & regbase::no_perl_ex) == regbase::no_perl_ex) | |
387 | { | |
388 | fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {."); | |
389 | return false; | |
390 | } | |
391 | result = parse_literal(); | |
392 | break; | |
393 | case regex_constants::syntax_or: | |
394 | return parse_alt(); | |
395 | case regex_constants::syntax_open_set: | |
396 | return parse_set(); | |
397 | case regex_constants::syntax_newline: | |
398 | if(this->flags() & regbase::newline_alt) | |
399 | return parse_alt(); | |
400 | else | |
401 | return parse_literal(); | |
402 | case regex_constants::syntax_hash: | |
403 | // | |
404 | // If we have a mod_x flag set, then skip until | |
405 | // we get to a newline character: | |
406 | // | |
407 | if((this->flags() | |
408 | & (regbase::no_perl_ex|regbase::mod_x)) | |
409 | == regbase::mod_x) | |
410 | { | |
411 | while((m_position != m_end) && !is_separator(*m_position++)){} | |
412 | return true; | |
413 | } | |
414 | BOOST_FALLTHROUGH; | |
415 | default: | |
416 | result = parse_literal(); | |
417 | break; | |
418 | } | |
419 | return result; | |
420 | } | |
421 | #ifdef BOOST_MSVC | |
20effc67 TL |
422 | # pragma warning(pop) |
423 | #endif | |
424 | #ifdef BOOST_MSVC | |
7c673cae FG |
425 | #pragma warning(pop) |
426 | #endif | |
427 | ||
428 | template <class charT, class traits> | |
429 | bool basic_regex_parser<charT, traits>::parse_literal() | |
430 | { | |
431 | // append this as a literal provided it's not a space character | |
432 | // or the perl option regbase::mod_x is not set: | |
433 | if( | |
434 | ((this->flags() | |
435 | & (regbase::main_option_type|regbase::mod_x|regbase::no_perl_ex)) | |
436 | != regbase::mod_x) | |
437 | || !this->m_traits.isctype(*m_position, this->m_mask_space)) | |
438 | this->append_literal(*m_position); | |
439 | ++m_position; | |
440 | return true; | |
441 | } | |
442 | ||
443 | template <class charT, class traits> | |
444 | bool basic_regex_parser<charT, traits>::parse_open_paren() | |
445 | { | |
446 | // | |
447 | // skip the '(' and error check: | |
448 | // | |
449 | if(++m_position == m_end) | |
450 | { | |
451 | fail(regex_constants::error_paren, m_position - m_base); | |
452 | return false; | |
453 | } | |
454 | // | |
455 | // begin by checking for a perl-style (?...) extension: | |
456 | // | |
457 | if( | |
458 | ((this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) == 0) | |
459 | || ((this->flags() & (regbase::main_option_type | regbase::emacs_ex)) == (regbase::basic_syntax_group|regbase::emacs_ex)) | |
460 | ) | |
461 | { | |
462 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question) | |
463 | return parse_perl_extension(); | |
464 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_star) | |
465 | return parse_perl_verb(); | |
466 | } | |
467 | // | |
468 | // update our mark count, and append the required state: | |
469 | // | |
470 | unsigned markid = 0; | |
471 | if(0 == (this->flags() & regbase::nosubs)) | |
472 | { | |
473 | markid = ++m_mark_count; | |
474 | #ifndef BOOST_NO_STD_DISTANCE | |
475 | if(this->flags() & regbase::save_subexpression_location) | |
476 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>(std::distance(m_base, m_position) - 1, 0)); | |
477 | #else | |
478 | if(this->flags() & regbase::save_subexpression_location) | |
479 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>((m_position - m_base) - 1, 0)); | |
480 | #endif | |
481 | } | |
482 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); | |
483 | pb->index = markid; | |
484 | pb->icase = this->flags() & regbase::icase; | |
485 | std::ptrdiff_t last_paren_start = this->getoffset(pb); | |
486 | // back up insertion point for alternations, and set new point: | |
487 | std::ptrdiff_t last_alt_point = m_alt_insert_point; | |
488 | this->m_pdata->m_data.align(); | |
489 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
490 | // | |
491 | // back up the current flags in case we have a nested (?imsx) group: | |
492 | // | |
493 | regex_constants::syntax_option_type opts = this->flags(); | |
494 | bool old_case_change = m_has_case_change; | |
495 | m_has_case_change = false; // no changes to this scope as yet... | |
496 | // | |
497 | // Back up branch reset data in case we have a nested (?|...) | |
498 | // | |
499 | int mark_reset = m_mark_reset; | |
500 | m_mark_reset = -1; | |
501 | // | |
502 | // now recursively add more states, this will terminate when we get to a | |
503 | // matching ')' : | |
504 | // | |
505 | parse_all(); | |
506 | // | |
507 | // Unwind pushed alternatives: | |
508 | // | |
509 | if(0 == unwind_alts(last_paren_start)) | |
510 | return false; | |
511 | // | |
512 | // restore flags: | |
513 | // | |
514 | if(m_has_case_change) | |
515 | { | |
516 | // the case has changed in one or more of the alternatives | |
517 | // within the scoped (...) block: we have to add a state | |
518 | // to reset the case sensitivity: | |
519 | static_cast<re_case*>( | |
520 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
521 | )->icase = opts & regbase::icase; | |
522 | } | |
523 | this->flags(opts); | |
524 | m_has_case_change = old_case_change; | |
525 | // | |
526 | // restore branch reset: | |
527 | // | |
528 | m_mark_reset = mark_reset; | |
529 | // | |
530 | // we either have a ')' or we have run out of characters prematurely: | |
531 | // | |
532 | if(m_position == m_end) | |
533 | { | |
534 | this->fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_end)); | |
535 | return false; | |
536 | } | |
b32b8144 FG |
537 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) |
538 | return false; | |
7c673cae FG |
539 | #ifndef BOOST_NO_STD_DISTANCE |
540 | if(markid && (this->flags() & regbase::save_subexpression_location)) | |
541 | this->m_pdata->m_subs.at(markid - 1).second = std::distance(m_base, m_position); | |
542 | #else | |
543 | if(markid && (this->flags() & regbase::save_subexpression_location)) | |
544 | this->m_pdata->m_subs.at(markid - 1).second = (m_position - m_base); | |
545 | #endif | |
546 | ++m_position; | |
547 | // | |
548 | // append closing parenthesis state: | |
549 | // | |
550 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); | |
551 | pb->index = markid; | |
552 | pb->icase = this->flags() & regbase::icase; | |
553 | this->m_paren_start = last_paren_start; | |
554 | // | |
555 | // restore the alternate insertion point: | |
556 | // | |
557 | this->m_alt_insert_point = last_alt_point; | |
558 | // | |
559 | // allow backrefs to this mark: | |
560 | // | |
20effc67 TL |
561 | if(markid > 0) |
562 | this->m_backrefs.set(markid); | |
7c673cae FG |
563 | |
564 | return true; | |
565 | } | |
566 | ||
567 | template <class charT, class traits> | |
568 | bool basic_regex_parser<charT, traits>::parse_basic_escape() | |
569 | { | |
570 | if(++m_position == m_end) | |
571 | { | |
572 | fail(regex_constants::error_paren, m_position - m_base); | |
573 | return false; | |
574 | } | |
575 | bool result = true; | |
576 | switch(this->m_traits.escape_syntax_type(*m_position)) | |
577 | { | |
578 | case regex_constants::syntax_open_mark: | |
579 | return parse_open_paren(); | |
580 | case regex_constants::syntax_close_mark: | |
581 | return false; | |
582 | case regex_constants::syntax_plus: | |
583 | if(this->flags() & regex_constants::bk_plus_qm) | |
584 | { | |
585 | ++m_position; | |
586 | return parse_repeat(1); | |
587 | } | |
588 | else | |
589 | return parse_literal(); | |
590 | case regex_constants::syntax_question: | |
591 | if(this->flags() & regex_constants::bk_plus_qm) | |
592 | { | |
593 | ++m_position; | |
594 | return parse_repeat(0, 1); | |
595 | } | |
596 | else | |
597 | return parse_literal(); | |
598 | case regex_constants::syntax_open_brace: | |
599 | if(this->flags() & regbase::no_intervals) | |
600 | return parse_literal(); | |
601 | ++m_position; | |
602 | return parse_repeat_range(true); | |
603 | case regex_constants::syntax_close_brace: | |
604 | if(this->flags() & regbase::no_intervals) | |
605 | return parse_literal(); | |
606 | fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {."); | |
607 | return false; | |
608 | case regex_constants::syntax_or: | |
609 | if(this->flags() & regbase::bk_vbar) | |
610 | return parse_alt(); | |
611 | else | |
612 | result = parse_literal(); | |
613 | break; | |
614 | case regex_constants::syntax_digit: | |
615 | return parse_backref(); | |
616 | case regex_constants::escape_type_start_buffer: | |
617 | if(this->flags() & regbase::emacs_ex) | |
618 | { | |
619 | ++m_position; | |
620 | this->append_state(syntax_element_buffer_start); | |
621 | } | |
622 | else | |
623 | result = parse_literal(); | |
624 | break; | |
625 | case regex_constants::escape_type_end_buffer: | |
626 | if(this->flags() & regbase::emacs_ex) | |
627 | { | |
628 | ++m_position; | |
629 | this->append_state(syntax_element_buffer_end); | |
630 | } | |
631 | else | |
632 | result = parse_literal(); | |
633 | break; | |
634 | case regex_constants::escape_type_word_assert: | |
635 | if(this->flags() & regbase::emacs_ex) | |
636 | { | |
637 | ++m_position; | |
638 | this->append_state(syntax_element_word_boundary); | |
639 | } | |
640 | else | |
641 | result = parse_literal(); | |
642 | break; | |
643 | case regex_constants::escape_type_not_word_assert: | |
644 | if(this->flags() & regbase::emacs_ex) | |
645 | { | |
646 | ++m_position; | |
647 | this->append_state(syntax_element_within_word); | |
648 | } | |
649 | else | |
650 | result = parse_literal(); | |
651 | break; | |
652 | case regex_constants::escape_type_left_word: | |
653 | if(this->flags() & regbase::emacs_ex) | |
654 | { | |
655 | ++m_position; | |
656 | this->append_state(syntax_element_word_start); | |
657 | } | |
658 | else | |
659 | result = parse_literal(); | |
660 | break; | |
661 | case regex_constants::escape_type_right_word: | |
662 | if(this->flags() & regbase::emacs_ex) | |
663 | { | |
664 | ++m_position; | |
665 | this->append_state(syntax_element_word_end); | |
666 | } | |
667 | else | |
668 | result = parse_literal(); | |
669 | break; | |
670 | default: | |
671 | if(this->flags() & regbase::emacs_ex) | |
672 | { | |
673 | bool negate = true; | |
674 | switch(*m_position) | |
675 | { | |
676 | case 'w': | |
677 | negate = false; | |
678 | BOOST_FALLTHROUGH; | |
679 | case 'W': | |
680 | { | |
681 | basic_char_set<charT, traits> char_set; | |
682 | if(negate) | |
683 | char_set.negate(); | |
684 | char_set.add_class(this->m_word_mask); | |
685 | if(0 == this->append_set(char_set)) | |
686 | { | |
687 | fail(regex_constants::error_ctype, m_position - m_base); | |
688 | return false; | |
689 | } | |
690 | ++m_position; | |
691 | return true; | |
692 | } | |
693 | case 's': | |
694 | negate = false; | |
695 | BOOST_FALLTHROUGH; | |
696 | case 'S': | |
697 | return add_emacs_code(negate); | |
698 | case 'c': | |
699 | case 'C': | |
700 | // not supported yet: | |
701 | 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."); | |
702 | return false; | |
703 | default: | |
704 | break; | |
705 | } | |
706 | } | |
707 | result = parse_literal(); | |
708 | break; | |
709 | } | |
710 | return result; | |
711 | } | |
712 | ||
713 | template <class charT, class traits> | |
714 | bool basic_regex_parser<charT, traits>::parse_extended_escape() | |
715 | { | |
716 | ++m_position; | |
717 | if(m_position == m_end) | |
718 | { | |
719 | fail(regex_constants::error_escape, m_position - m_base, "Incomplete escape sequence found."); | |
720 | return false; | |
721 | } | |
722 | bool negate = false; // in case this is a character class escape: \w \d etc | |
723 | switch(this->m_traits.escape_syntax_type(*m_position)) | |
724 | { | |
725 | case regex_constants::escape_type_not_class: | |
726 | negate = true; | |
727 | BOOST_FALLTHROUGH; | |
728 | case regex_constants::escape_type_class: | |
729 | { | |
730 | escape_type_class_jump: | |
731 | typedef typename traits::char_class_type m_type; | |
732 | m_type m = this->m_traits.lookup_classname(m_position, m_position+1); | |
733 | if(m != 0) | |
734 | { | |
735 | basic_char_set<charT, traits> char_set; | |
736 | if(negate) | |
737 | char_set.negate(); | |
738 | char_set.add_class(m); | |
739 | if(0 == this->append_set(char_set)) | |
740 | { | |
741 | fail(regex_constants::error_ctype, m_position - m_base); | |
742 | return false; | |
743 | } | |
744 | ++m_position; | |
745 | return true; | |
746 | } | |
747 | // | |
748 | // not a class, just a regular unknown escape: | |
749 | // | |
750 | this->append_literal(unescape_character()); | |
751 | break; | |
752 | } | |
753 | case regex_constants::syntax_digit: | |
754 | return parse_backref(); | |
755 | case regex_constants::escape_type_left_word: | |
756 | ++m_position; | |
757 | this->append_state(syntax_element_word_start); | |
758 | break; | |
759 | case regex_constants::escape_type_right_word: | |
760 | ++m_position; | |
761 | this->append_state(syntax_element_word_end); | |
762 | break; | |
763 | case regex_constants::escape_type_start_buffer: | |
764 | ++m_position; | |
765 | this->append_state(syntax_element_buffer_start); | |
766 | break; | |
767 | case regex_constants::escape_type_end_buffer: | |
768 | ++m_position; | |
769 | this->append_state(syntax_element_buffer_end); | |
770 | break; | |
771 | case regex_constants::escape_type_word_assert: | |
772 | ++m_position; | |
773 | this->append_state(syntax_element_word_boundary); | |
774 | break; | |
775 | case regex_constants::escape_type_not_word_assert: | |
776 | ++m_position; | |
777 | this->append_state(syntax_element_within_word); | |
778 | break; | |
779 | case regex_constants::escape_type_Z: | |
780 | ++m_position; | |
781 | this->append_state(syntax_element_soft_buffer_end); | |
782 | break; | |
783 | case regex_constants::escape_type_Q: | |
784 | return parse_QE(); | |
785 | case regex_constants::escape_type_C: | |
786 | return parse_match_any(); | |
787 | case regex_constants::escape_type_X: | |
788 | ++m_position; | |
789 | this->append_state(syntax_element_combining); | |
790 | break; | |
791 | case regex_constants::escape_type_G: | |
792 | ++m_position; | |
793 | this->append_state(syntax_element_restart_continue); | |
794 | break; | |
795 | case regex_constants::escape_type_not_property: | |
796 | negate = true; | |
797 | BOOST_FALLTHROUGH; | |
798 | case regex_constants::escape_type_property: | |
799 | { | |
800 | ++m_position; | |
801 | char_class_type m; | |
802 | if(m_position == m_end) | |
803 | { | |
804 | fail(regex_constants::error_escape, m_position - m_base, "Incomplete property escape found."); | |
805 | return false; | |
806 | } | |
807 | // maybe have \p{ddd} | |
808 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) | |
809 | { | |
810 | const charT* base = m_position; | |
811 | // skip forward until we find enclosing brace: | |
812 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) | |
813 | ++m_position; | |
814 | if(m_position == m_end) | |
815 | { | |
816 | fail(regex_constants::error_escape, m_position - m_base, "Closing } missing from property escape sequence."); | |
817 | return false; | |
818 | } | |
819 | m = this->m_traits.lookup_classname(++base, m_position++); | |
820 | } | |
821 | else | |
822 | { | |
823 | m = this->m_traits.lookup_classname(m_position, m_position+1); | |
824 | ++m_position; | |
825 | } | |
826 | if(m != 0) | |
827 | { | |
828 | basic_char_set<charT, traits> char_set; | |
829 | if(negate) | |
830 | char_set.negate(); | |
831 | char_set.add_class(m); | |
832 | if(0 == this->append_set(char_set)) | |
833 | { | |
834 | fail(regex_constants::error_ctype, m_position - m_base); | |
835 | return false; | |
836 | } | |
837 | return true; | |
838 | } | |
839 | fail(regex_constants::error_ctype, m_position - m_base, "Escape sequence was neither a valid property nor a valid character class name."); | |
840 | return false; | |
841 | } | |
842 | case regex_constants::escape_type_reset_start_mark: | |
843 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
844 | { | |
845 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); | |
846 | pb->index = -5; | |
847 | pb->icase = this->flags() & regbase::icase; | |
848 | this->m_pdata->m_data.align(); | |
849 | ++m_position; | |
850 | return true; | |
851 | } | |
852 | goto escape_type_class_jump; | |
853 | case regex_constants::escape_type_line_ending: | |
854 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
855 | { | |
856 | const charT* e = get_escape_R_string<charT>(); | |
857 | const charT* old_position = m_position; | |
858 | const charT* old_end = m_end; | |
859 | const charT* old_base = m_base; | |
860 | m_position = e; | |
861 | m_base = e; | |
862 | m_end = e + traits::length(e); | |
863 | bool r = parse_all(); | |
864 | m_position = ++old_position; | |
865 | m_end = old_end; | |
866 | m_base = old_base; | |
867 | return r; | |
868 | } | |
869 | goto escape_type_class_jump; | |
870 | case regex_constants::escape_type_extended_backref: | |
871 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
872 | { | |
873 | bool have_brace = false; | |
874 | bool negative = false; | |
20effc67 | 875 | static const char incomplete_message[] = "Incomplete \\g escape found."; |
7c673cae FG |
876 | if(++m_position == m_end) |
877 | { | |
878 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); | |
879 | return false; | |
880 | } | |
881 | // maybe have \g{ddd} | |
882 | regex_constants::syntax_type syn = this->m_traits.syntax_type(*m_position); | |
883 | regex_constants::syntax_type syn_end = 0; | |
884 | if((syn == regex_constants::syntax_open_brace) | |
885 | || (syn == regex_constants::escape_type_left_word) | |
886 | || (syn == regex_constants::escape_type_end_buffer)) | |
887 | { | |
888 | if(++m_position == m_end) | |
889 | { | |
890 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); | |
891 | return false; | |
892 | } | |
893 | have_brace = true; | |
894 | switch(syn) | |
895 | { | |
896 | case regex_constants::syntax_open_brace: | |
897 | syn_end = regex_constants::syntax_close_brace; | |
898 | break; | |
899 | case regex_constants::escape_type_left_word: | |
900 | syn_end = regex_constants::escape_type_right_word; | |
901 | break; | |
902 | default: | |
903 | syn_end = regex_constants::escape_type_end_buffer; | |
904 | break; | |
905 | } | |
906 | } | |
907 | negative = (*m_position == static_cast<charT>('-')); | |
908 | if((negative) && (++m_position == m_end)) | |
909 | { | |
910 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); | |
911 | return false; | |
912 | } | |
913 | const charT* pc = m_position; | |
914 | boost::intmax_t i = this->m_traits.toi(pc, m_end, 10); | |
915 | if((i < 0) && syn_end) | |
916 | { | |
917 | // Check for a named capture, get the leftmost one if there is more than one: | |
918 | const charT* base = m_position; | |
919 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != syn_end)) | |
920 | { | |
921 | ++m_position; | |
922 | } | |
923 | i = hash_value_from_capture_name(base, m_position); | |
924 | pc = m_position; | |
925 | } | |
926 | if(negative) | |
20effc67 TL |
927 | i = 1 + (static_cast<boost::intmax_t>(m_mark_count) - i); |
928 | if(((i < hash_value_mask) && (i > 0) && (this->m_backrefs.test(i))) || ((i >= hash_value_mask) && (this->m_pdata->get_id(i) > 0) && (this->m_backrefs.test(this->m_pdata->get_id(i))))) | |
7c673cae FG |
929 | { |
930 | m_position = pc; | |
931 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace))); | |
932 | pb->index = i; | |
933 | pb->icase = this->flags() & regbase::icase; | |
934 | } | |
935 | else | |
936 | { | |
937 | fail(regex_constants::error_backref, m_position - m_base); | |
938 | return false; | |
939 | } | |
940 | m_position = pc; | |
941 | if(have_brace) | |
942 | { | |
943 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != syn_end)) | |
944 | { | |
945 | fail(regex_constants::error_escape, m_position - m_base, incomplete_message); | |
946 | return false; | |
947 | } | |
948 | ++m_position; | |
949 | } | |
950 | return true; | |
951 | } | |
952 | goto escape_type_class_jump; | |
953 | case regex_constants::escape_type_control_v: | |
954 | if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
955 | goto escape_type_class_jump; | |
956 | BOOST_FALLTHROUGH; | |
957 | default: | |
958 | this->append_literal(unescape_character()); | |
959 | break; | |
960 | } | |
961 | return true; | |
962 | } | |
963 | ||
964 | template <class charT, class traits> | |
965 | bool basic_regex_parser<charT, traits>::parse_match_any() | |
966 | { | |
967 | // | |
968 | // we have a '.' that can match any character: | |
969 | // | |
970 | ++m_position; | |
971 | static_cast<re_dot*>( | |
972 | this->append_state(syntax_element_wild, sizeof(re_dot)) | |
973 | )->mask = static_cast<unsigned char>(this->flags() & regbase::no_mod_s | |
974 | ? BOOST_REGEX_DETAIL_NS::force_not_newline | |
975 | : this->flags() & regbase::mod_s ? | |
976 | BOOST_REGEX_DETAIL_NS::force_newline : BOOST_REGEX_DETAIL_NS::dont_care); | |
977 | return true; | |
978 | } | |
979 | ||
980 | template <class charT, class traits> | |
981 | bool basic_regex_parser<charT, traits>::parse_repeat(std::size_t low, std::size_t high) | |
982 | { | |
983 | bool greedy = true; | |
1e59de90 | 984 | bool possessive = false; |
7c673cae FG |
985 | std::size_t insert_point; |
986 | // | |
987 | // when we get to here we may have a non-greedy ? mark still to come: | |
988 | // | |
989 | if((m_position != m_end) | |
990 | && ( | |
991 | (0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex))) | |
992 | || ((regbase::basic_syntax_group|regbase::emacs_ex) == (this->flags() & (regbase::main_option_type | regbase::emacs_ex))) | |
993 | ) | |
994 | ) | |
995 | { | |
996 | // OK we have a perl or emacs regex, check for a '?': | |
b32b8144 FG |
997 | if ((this->flags() & (regbase::main_option_type | regbase::mod_x | regbase::no_perl_ex)) == regbase::mod_x) |
998 | { | |
999 | // whitespace skip: | |
1000 | while ((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1001 | ++m_position; | |
1002 | } | |
1003 | if((m_position != m_end) && (this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question)) | |
7c673cae FG |
1004 | { |
1005 | greedy = false; | |
1006 | ++m_position; | |
1007 | } | |
1e59de90 | 1008 | // for perl regexes only check for possessive ++ repeats. |
7c673cae FG |
1009 | if((m_position != m_end) |
1010 | && (0 == (this->flags() & regbase::main_option_type)) | |
1011 | && (this->m_traits.syntax_type(*m_position) == regex_constants::syntax_plus)) | |
1012 | { | |
1e59de90 | 1013 | possessive = true; |
7c673cae FG |
1014 | ++m_position; |
1015 | } | |
1016 | } | |
1017 | if(0 == this->m_last_state) | |
1018 | { | |
1019 | fail(regex_constants::error_badrepeat, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_position), "Nothing to repeat."); | |
1020 | return false; | |
1021 | } | |
1022 | if(this->m_last_state->type == syntax_element_endmark) | |
1023 | { | |
1024 | // insert a repeat before the '(' matching the last ')': | |
1025 | insert_point = this->m_paren_start; | |
1026 | } | |
1027 | else if((this->m_last_state->type == syntax_element_literal) && (static_cast<re_literal*>(this->m_last_state)->length > 1)) | |
1028 | { | |
1029 | // the last state was a literal with more than one character, split it in two: | |
1030 | re_literal* lit = static_cast<re_literal*>(this->m_last_state); | |
1031 | charT c = (static_cast<charT*>(static_cast<void*>(lit+1)))[lit->length - 1]; | |
1032 | lit->length -= 1; | |
1033 | // now append new state: | |
1034 | lit = static_cast<re_literal*>(this->append_state(syntax_element_literal, sizeof(re_literal) + sizeof(charT))); | |
1035 | lit->length = 1; | |
1036 | (static_cast<charT*>(static_cast<void*>(lit+1)))[0] = c; | |
1037 | insert_point = this->getoffset(this->m_last_state); | |
1038 | } | |
1039 | else | |
1040 | { | |
1041 | // repeat the last state whatever it was, need to add some error checking here: | |
1042 | switch(this->m_last_state->type) | |
1043 | { | |
1044 | case syntax_element_start_line: | |
1045 | case syntax_element_end_line: | |
1046 | case syntax_element_word_boundary: | |
1047 | case syntax_element_within_word: | |
1048 | case syntax_element_word_start: | |
1049 | case syntax_element_word_end: | |
1050 | case syntax_element_buffer_start: | |
1051 | case syntax_element_buffer_end: | |
1052 | case syntax_element_alt: | |
1053 | case syntax_element_soft_buffer_end: | |
1054 | case syntax_element_restart_continue: | |
1055 | case syntax_element_jump: | |
1056 | case syntax_element_startmark: | |
1057 | case syntax_element_backstep: | |
1e59de90 | 1058 | case syntax_element_toggle_case: |
7c673cae FG |
1059 | // can't legally repeat any of the above: |
1060 | fail(regex_constants::error_badrepeat, m_position - m_base); | |
1061 | return false; | |
1062 | default: | |
1063 | // do nothing... | |
1064 | break; | |
1065 | } | |
1066 | insert_point = this->getoffset(this->m_last_state); | |
1067 | } | |
1068 | // | |
1069 | // OK we now know what to repeat, so insert the repeat around it: | |
1070 | // | |
1071 | re_repeat* rep = static_cast<re_repeat*>(this->insert_state(insert_point, syntax_element_rep, re_repeater_size)); | |
1072 | rep->min = low; | |
1073 | rep->max = high; | |
1074 | rep->greedy = greedy; | |
1075 | rep->leading = false; | |
1076 | // store our repeater position for later: | |
1077 | std::ptrdiff_t rep_off = this->getoffset(rep); | |
1078 | // and append a back jump to the repeat: | |
1079 | re_jump* jmp = static_cast<re_jump*>(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
1080 | jmp->alt.i = rep_off - this->getoffset(jmp); | |
1081 | this->m_pdata->m_data.align(); | |
1082 | // now fill in the alt jump for the repeat: | |
1083 | rep = static_cast<re_repeat*>(this->getaddress(rep_off)); | |
1084 | rep->alt.i = this->m_pdata->m_data.size() - rep_off; | |
1085 | // | |
1e59de90 | 1086 | // If the repeat is possessive then bracket the repeat with a (?>...) |
7c673cae FG |
1087 | // independent sub-expression construct: |
1088 | // | |
1e59de90 | 1089 | if(possessive) |
7c673cae FG |
1090 | { |
1091 | if(m_position != m_end) | |
1092 | { | |
1093 | // | |
1094 | // Check for illegal following quantifier, we have to do this here, because | |
1095 | // the extra states we insert below circumvents our usual error checking :-( | |
1096 | // | |
b32b8144 FG |
1097 | bool contin = false; |
1098 | do | |
7c673cae | 1099 | { |
b32b8144 FG |
1100 | if ((this->flags() & (regbase::main_option_type | regbase::mod_x | regbase::no_perl_ex)) == regbase::mod_x) |
1101 | { | |
1102 | // whitespace skip: | |
1103 | while ((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1104 | ++m_position; | |
1105 | } | |
1106 | if (m_position != m_end) | |
1107 | { | |
1108 | switch (this->m_traits.syntax_type(*m_position)) | |
1109 | { | |
1110 | case regex_constants::syntax_star: | |
1111 | case regex_constants::syntax_plus: | |
1112 | case regex_constants::syntax_question: | |
1113 | case regex_constants::syntax_open_brace: | |
1114 | fail(regex_constants::error_badrepeat, m_position - m_base); | |
1115 | return false; | |
1116 | case regex_constants::syntax_open_mark: | |
1117 | // Do we have a comment? If so we need to skip it here... | |
1118 | if ((m_position + 2 < m_end) && this->m_traits.syntax_type(*(m_position + 1)) == regex_constants::syntax_question | |
1119 | && this->m_traits.syntax_type(*(m_position + 2)) == regex_constants::syntax_hash) | |
1120 | { | |
1121 | while ((m_position != m_end) | |
1122 | && (this->m_traits.syntax_type(*m_position++) != regex_constants::syntax_close_mark)) { | |
1123 | } | |
1124 | contin = true; | |
1125 | } | |
1126 | else | |
1127 | contin = false; | |
1e59de90 TL |
1128 | break; |
1129 | default: | |
1130 | contin = false; | |
b32b8144 FG |
1131 | } |
1132 | } | |
1133 | else | |
1134 | contin = false; | |
1135 | } while (contin); | |
7c673cae FG |
1136 | } |
1137 | re_brace* pb = static_cast<re_brace*>(this->insert_state(insert_point, syntax_element_startmark, sizeof(re_brace))); | |
1138 | pb->index = -3; | |
1139 | pb->icase = this->flags() & regbase::icase; | |
1140 | jmp = static_cast<re_jump*>(this->insert_state(insert_point + sizeof(re_brace), syntax_element_jump, sizeof(re_jump))); | |
1141 | this->m_pdata->m_data.align(); | |
1142 | jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp); | |
1143 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); | |
1144 | pb->index = -3; | |
1145 | pb->icase = this->flags() & regbase::icase; | |
1146 | } | |
1147 | return true; | |
1148 | } | |
1149 | ||
1150 | template <class charT, class traits> | |
1151 | bool basic_regex_parser<charT, traits>::parse_repeat_range(bool isbasic) | |
1152 | { | |
20effc67 | 1153 | static const char incomplete_message[] = "Missing } in quantified repetition."; |
7c673cae FG |
1154 | // |
1155 | // parse a repeat-range: | |
1156 | // | |
1157 | std::size_t min, max; | |
1158 | boost::intmax_t v; | |
1159 | // skip whitespace: | |
1160 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1161 | ++m_position; | |
1162 | if(this->m_position == this->m_end) | |
1163 | { | |
1164 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1165 | { | |
1166 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1167 | return false; | |
1168 | } | |
1169 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1170 | --m_position; | |
1171 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1172 | return parse_literal(); | |
1173 | } | |
1174 | // get min: | |
1175 | v = this->m_traits.toi(m_position, m_end, 10); | |
1176 | // skip whitespace: | |
1177 | if((v < 0) || (v > umax())) | |
1178 | { | |
1179 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1180 | { | |
1181 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1182 | return false; | |
1183 | } | |
1184 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1185 | --m_position; | |
1186 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1187 | return parse_literal(); | |
1188 | } | |
1189 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1190 | ++m_position; | |
1191 | if(this->m_position == this->m_end) | |
1192 | { | |
1193 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1194 | { | |
1195 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1196 | return false; | |
1197 | } | |
1198 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1199 | --m_position; | |
1200 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1201 | return parse_literal(); | |
1202 | } | |
1203 | min = static_cast<std::size_t>(v); | |
1204 | // see if we have a comma: | |
1205 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_comma) | |
1206 | { | |
1207 | // move on and error check: | |
1208 | ++m_position; | |
1209 | // skip whitespace: | |
1210 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1211 | ++m_position; | |
1212 | if(this->m_position == this->m_end) | |
1213 | { | |
1214 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1215 | { | |
1216 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1217 | return false; | |
1218 | } | |
1219 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1220 | --m_position; | |
1221 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1222 | return parse_literal(); | |
1223 | } | |
1224 | // get the value if any: | |
1225 | v = this->m_traits.toi(m_position, m_end, 10); | |
1226 | max = ((v >= 0) && (v < umax())) ? (std::size_t)v : (std::numeric_limits<std::size_t>::max)(); | |
1227 | } | |
1228 | else | |
1229 | { | |
1230 | // no comma, max = min: | |
1231 | max = min; | |
1232 | } | |
1233 | // skip whitespace: | |
1234 | while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space)) | |
1235 | ++m_position; | |
1236 | // OK now check trailing }: | |
1237 | if(this->m_position == this->m_end) | |
1238 | { | |
1239 | if(this->flags() & (regbase::main_option_type | regbase::no_perl_ex)) | |
1240 | { | |
1241 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1242 | return false; | |
1243 | } | |
1244 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1245 | --m_position; | |
1246 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1247 | return parse_literal(); | |
1248 | } | |
1249 | if(isbasic) | |
1250 | { | |
1251 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_escape) | |
1252 | { | |
1253 | ++m_position; | |
1254 | if(this->m_position == this->m_end) | |
1255 | { | |
1256 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1257 | return false; | |
1258 | } | |
1259 | } | |
1260 | else | |
1261 | { | |
1262 | fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message); | |
1263 | return false; | |
1264 | } | |
1265 | } | |
1266 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_brace) | |
1267 | ++m_position; | |
1268 | else | |
1269 | { | |
1270 | // Treat the opening '{' as a literal character, rewind to start of error: | |
1271 | --m_position; | |
1272 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_brace) --m_position; | |
1273 | return parse_literal(); | |
1274 | } | |
1275 | // | |
1276 | // finally go and add the repeat, unless error: | |
1277 | // | |
1278 | if(min > max) | |
1279 | { | |
1280 | // Backtrack to error location: | |
1281 | m_position -= 2; | |
1282 | while(this->m_traits.isctype(*m_position, this->m_word_mask)) --m_position; | |
1283 | ++m_position; | |
1284 | fail(regex_constants::error_badbrace, m_position - m_base); | |
1285 | return false; | |
1286 | } | |
1287 | return parse_repeat(min, max); | |
1288 | } | |
1289 | ||
1290 | template <class charT, class traits> | |
1291 | bool basic_regex_parser<charT, traits>::parse_alt() | |
1292 | { | |
1293 | // | |
1294 | // error check: if there have been no previous states, | |
1295 | // or if the last state was a '(' then error: | |
1296 | // | |
1297 | if( | |
1298 | ((this->m_last_state == 0) || (this->m_last_state->type == syntax_element_startmark)) | |
1299 | && | |
1300 | !( | |
1301 | ((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) | |
1302 | && | |
1303 | ((this->flags() & regbase::no_empty_expressions) == 0) | |
1304 | ) | |
1305 | ) | |
1306 | { | |
1307 | fail(regex_constants::error_empty, this->m_position - this->m_base, "A regular expression cannot start with the alternation operator |."); | |
1308 | return false; | |
1309 | } | |
1310 | // | |
1311 | // Reset mark count if required: | |
1312 | // | |
1313 | if(m_max_mark < m_mark_count) | |
1314 | m_max_mark = m_mark_count; | |
1315 | if(m_mark_reset >= 0) | |
1316 | m_mark_count = m_mark_reset; | |
1317 | ||
1318 | ++m_position; | |
1319 | // | |
1320 | // we need to append a trailing jump: | |
1321 | // | |
1322 | re_syntax_base* pj = this->append_state(BOOST_REGEX_DETAIL_NS::syntax_element_jump, sizeof(re_jump)); | |
1323 | std::ptrdiff_t jump_offset = this->getoffset(pj); | |
1324 | // | |
1325 | // now insert the alternative: | |
1326 | // | |
1327 | re_alt* palt = static_cast<re_alt*>(this->insert_state(this->m_alt_insert_point, syntax_element_alt, re_alt_size)); | |
1328 | jump_offset += re_alt_size; | |
1329 | this->m_pdata->m_data.align(); | |
1330 | palt->alt.i = this->m_pdata->m_data.size() - this->getoffset(palt); | |
1331 | // | |
1332 | // update m_alt_insert_point so that the next alternate gets | |
1333 | // inserted at the start of the second of the two we've just created: | |
1334 | // | |
1335 | this->m_alt_insert_point = this->m_pdata->m_data.size(); | |
1336 | // | |
1337 | // the start of this alternative must have a case changes state | |
1338 | // if the current block has messed around with case changes: | |
1339 | // | |
1340 | if(m_has_case_change) | |
1341 | { | |
1342 | static_cast<re_case*>( | |
1343 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
1344 | )->icase = this->m_icase; | |
1345 | } | |
1346 | // | |
1347 | // push the alternative onto our stack, a recursive | |
1348 | // implementation here is easier to understand (and faster | |
1349 | // as it happens), but causes all kinds of stack overflow problems | |
1350 | // on programs with small stacks (COM+). | |
1351 | // | |
1352 | m_alt_jumps.push_back(jump_offset); | |
1353 | return true; | |
1354 | } | |
1355 | ||
1356 | template <class charT, class traits> | |
1357 | bool basic_regex_parser<charT, traits>::parse_set() | |
1358 | { | |
20effc67 | 1359 | static const char incomplete_message[] = "Character set declaration starting with [ terminated prematurely - either no ] was found or the set had no content."; |
7c673cae FG |
1360 | ++m_position; |
1361 | if(m_position == m_end) | |
1362 | { | |
1363 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1364 | return false; | |
1365 | } | |
1366 | basic_char_set<charT, traits> char_set; | |
1367 | ||
1368 | const charT* base = m_position; // where the '[' was | |
1369 | const charT* item_base = m_position; // where the '[' or '^' was | |
1370 | ||
1371 | while(m_position != m_end) | |
1372 | { | |
1373 | switch(this->m_traits.syntax_type(*m_position)) | |
1374 | { | |
1375 | case regex_constants::syntax_caret: | |
1376 | if(m_position == base) | |
1377 | { | |
1378 | char_set.negate(); | |
1379 | ++m_position; | |
1380 | item_base = m_position; | |
1381 | } | |
1382 | else | |
1383 | parse_set_literal(char_set); | |
1384 | break; | |
1385 | case regex_constants::syntax_close_set: | |
1386 | if(m_position == item_base) | |
1387 | { | |
1388 | parse_set_literal(char_set); | |
1389 | break; | |
1390 | } | |
1391 | else | |
1392 | { | |
1393 | ++m_position; | |
1394 | if(0 == this->append_set(char_set)) | |
1395 | { | |
1396 | fail(regex_constants::error_ctype, m_position - m_base); | |
1397 | return false; | |
1398 | } | |
1399 | } | |
1400 | return true; | |
1401 | case regex_constants::syntax_open_set: | |
1402 | if(parse_inner_set(char_set)) | |
1403 | break; | |
1404 | return true; | |
1405 | case regex_constants::syntax_escape: | |
1406 | { | |
1407 | // | |
1408 | // look ahead and see if this is a character class shortcut | |
1409 | // \d \w \s etc... | |
1410 | // | |
1411 | ++m_position; | |
1412 | if(this->m_traits.escape_syntax_type(*m_position) | |
1413 | == regex_constants::escape_type_class) | |
1414 | { | |
1415 | char_class_type m = this->m_traits.lookup_classname(m_position, m_position+1); | |
1416 | if(m != 0) | |
1417 | { | |
1418 | char_set.add_class(m); | |
1419 | ++m_position; | |
1420 | break; | |
1421 | } | |
1422 | } | |
1423 | else if(this->m_traits.escape_syntax_type(*m_position) | |
1424 | == regex_constants::escape_type_not_class) | |
1425 | { | |
1426 | // negated character class: | |
1427 | char_class_type m = this->m_traits.lookup_classname(m_position, m_position+1); | |
1428 | if(m != 0) | |
1429 | { | |
1430 | char_set.add_negated_class(m); | |
1431 | ++m_position; | |
1432 | break; | |
1433 | } | |
1434 | } | |
1435 | // not a character class, just a regular escape: | |
1436 | --m_position; | |
1437 | parse_set_literal(char_set); | |
1438 | break; | |
1439 | } | |
1440 | default: | |
1441 | parse_set_literal(char_set); | |
1442 | break; | |
1443 | } | |
1444 | } | |
1445 | return m_position != m_end; | |
1446 | } | |
1447 | ||
1448 | template <class charT, class traits> | |
1449 | bool basic_regex_parser<charT, traits>::parse_inner_set(basic_char_set<charT, traits>& char_set) | |
1450 | { | |
20effc67 | 1451 | static const char incomplete_message[] = "Character class declaration starting with [ terminated prematurely - either no ] was found or the set had no content."; |
7c673cae FG |
1452 | // |
1453 | // we have either a character class [:name:] | |
1454 | // a collating element [.name.] | |
1455 | // or an equivalence class [=name=] | |
1456 | // | |
1457 | if(m_end == ++m_position) | |
1458 | { | |
1459 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1460 | return false; | |
1461 | } | |
1462 | switch(this->m_traits.syntax_type(*m_position)) | |
1463 | { | |
1464 | case regex_constants::syntax_dot: | |
1465 | // | |
1466 | // a collating element is treated as a literal: | |
1467 | // | |
1468 | --m_position; | |
1469 | parse_set_literal(char_set); | |
1470 | return true; | |
1471 | case regex_constants::syntax_colon: | |
1472 | { | |
1473 | // check that character classes are actually enabled: | |
1474 | if((this->flags() & (regbase::main_option_type | regbase::no_char_classes)) | |
1475 | == (regbase::basic_syntax_group | regbase::no_char_classes)) | |
1476 | { | |
1477 | --m_position; | |
1478 | parse_set_literal(char_set); | |
1479 | return true; | |
1480 | } | |
1481 | // skip the ':' | |
1482 | if(m_end == ++m_position) | |
1483 | { | |
1484 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1485 | return false; | |
1486 | } | |
1487 | const charT* name_first = m_position; | |
1488 | // skip at least one character, then find the matching ':]' | |
1489 | if(m_end == ++m_position) | |
1490 | { | |
1491 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1492 | return false; | |
1493 | } | |
1494 | while((m_position != m_end) | |
1495 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_colon)) | |
1496 | ++m_position; | |
1497 | const charT* name_last = m_position; | |
1498 | if(m_end == m_position) | |
1499 | { | |
1500 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1501 | return false; | |
1502 | } | |
1503 | if((m_end == ++m_position) | |
1504 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) | |
1505 | { | |
1506 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1507 | return false; | |
1508 | } | |
1509 | // | |
1510 | // check for negated class: | |
1511 | // | |
1512 | bool negated = false; | |
1513 | if(this->m_traits.syntax_type(*name_first) == regex_constants::syntax_caret) | |
1514 | { | |
1515 | ++name_first; | |
1516 | negated = true; | |
1517 | } | |
1518 | typedef typename traits::char_class_type m_type; | |
1519 | m_type m = this->m_traits.lookup_classname(name_first, name_last); | |
1520 | if(m == 0) | |
1521 | { | |
1522 | if(char_set.empty() && (name_last - name_first == 1)) | |
1523 | { | |
1524 | // maybe a special case: | |
1525 | ++m_position; | |
1526 | if( (m_position != m_end) | |
1527 | && (this->m_traits.syntax_type(*m_position) | |
1528 | == regex_constants::syntax_close_set)) | |
1529 | { | |
1530 | if(this->m_traits.escape_syntax_type(*name_first) | |
1531 | == regex_constants::escape_type_left_word) | |
1532 | { | |
1533 | ++m_position; | |
1534 | this->append_state(syntax_element_word_start); | |
1535 | return false; | |
1536 | } | |
1537 | if(this->m_traits.escape_syntax_type(*name_first) | |
1538 | == regex_constants::escape_type_right_word) | |
1539 | { | |
1540 | ++m_position; | |
1541 | this->append_state(syntax_element_word_end); | |
1542 | return false; | |
1543 | } | |
1544 | } | |
1545 | } | |
1546 | fail(regex_constants::error_ctype, name_first - m_base); | |
1547 | return false; | |
1548 | } | |
1e59de90 | 1549 | if(!negated) |
7c673cae FG |
1550 | char_set.add_class(m); |
1551 | else | |
1552 | char_set.add_negated_class(m); | |
1553 | ++m_position; | |
1554 | break; | |
1555 | } | |
1556 | case regex_constants::syntax_equal: | |
1557 | { | |
1558 | // skip the '=' | |
1559 | if(m_end == ++m_position) | |
1560 | { | |
1561 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1562 | return false; | |
1563 | } | |
1564 | const charT* name_first = m_position; | |
1565 | // skip at least one character, then find the matching '=]' | |
1566 | if(m_end == ++m_position) | |
1567 | { | |
1568 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1569 | return false; | |
1570 | } | |
1571 | while((m_position != m_end) | |
1572 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal)) | |
1573 | ++m_position; | |
1574 | const charT* name_last = m_position; | |
1575 | if(m_end == m_position) | |
1576 | { | |
1577 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1578 | return false; | |
1579 | } | |
1580 | if((m_end == ++m_position) | |
1581 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) | |
1582 | { | |
1583 | fail(regex_constants::error_brack, m_position - m_base, incomplete_message); | |
1584 | return false; | |
1585 | } | |
1586 | string_type m = this->m_traits.lookup_collatename(name_first, name_last); | |
1e59de90 | 1587 | if(m.empty() || (m.size() > 2)) |
7c673cae FG |
1588 | { |
1589 | fail(regex_constants::error_collate, name_first - m_base); | |
1590 | return false; | |
1591 | } | |
1592 | digraph<charT> d; | |
1593 | d.first = m[0]; | |
1594 | if(m.size() > 1) | |
1595 | d.second = m[1]; | |
1596 | else | |
1597 | d.second = 0; | |
1598 | char_set.add_equivalent(d); | |
1599 | ++m_position; | |
1600 | break; | |
1601 | } | |
1602 | default: | |
1603 | --m_position; | |
1604 | parse_set_literal(char_set); | |
1605 | break; | |
1606 | } | |
1607 | return true; | |
1608 | } | |
1609 | ||
1610 | template <class charT, class traits> | |
1611 | void basic_regex_parser<charT, traits>::parse_set_literal(basic_char_set<charT, traits>& char_set) | |
1612 | { | |
1613 | digraph<charT> start_range(get_next_set_literal(char_set)); | |
1614 | if(m_end == m_position) | |
1615 | { | |
1616 | fail(regex_constants::error_brack, m_position - m_base); | |
1617 | return; | |
1618 | } | |
1619 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_dash) | |
1620 | { | |
1621 | // we have a range: | |
1622 | if(m_end == ++m_position) | |
1623 | { | |
1624 | fail(regex_constants::error_brack, m_position - m_base); | |
1625 | return; | |
1626 | } | |
1627 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set) | |
1628 | { | |
1629 | digraph<charT> end_range = get_next_set_literal(char_set); | |
1630 | char_set.add_range(start_range, end_range); | |
1631 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_dash) | |
1632 | { | |
1633 | if(m_end == ++m_position) | |
1634 | { | |
1635 | fail(regex_constants::error_brack, m_position - m_base); | |
1636 | return; | |
1637 | } | |
1638 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_set) | |
1639 | { | |
1640 | // trailing - : | |
1641 | --m_position; | |
1642 | return; | |
1643 | } | |
1644 | fail(regex_constants::error_range, m_position - m_base); | |
1645 | return; | |
1646 | } | |
1647 | return; | |
1648 | } | |
1649 | --m_position; | |
1650 | } | |
1651 | char_set.add_single(start_range); | |
1652 | } | |
1653 | ||
1654 | template <class charT, class traits> | |
1655 | digraph<charT> basic_regex_parser<charT, traits>::get_next_set_literal(basic_char_set<charT, traits>& char_set) | |
1656 | { | |
1657 | digraph<charT> result; | |
1658 | switch(this->m_traits.syntax_type(*m_position)) | |
1659 | { | |
1660 | case regex_constants::syntax_dash: | |
1661 | if(!char_set.empty()) | |
1662 | { | |
1663 | // see if we are at the end of the set: | |
1664 | if((++m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) | |
1665 | { | |
1666 | fail(regex_constants::error_range, m_position - m_base); | |
1667 | return result; | |
1668 | } | |
1669 | --m_position; | |
1670 | } | |
1671 | result.first = *m_position++; | |
1672 | return result; | |
1673 | case regex_constants::syntax_escape: | |
1674 | // check to see if escapes are supported first: | |
1675 | if(this->flags() & regex_constants::no_escape_in_lists) | |
1676 | { | |
1677 | result = *m_position++; | |
1678 | break; | |
1679 | } | |
1680 | ++m_position; | |
1681 | result = unescape_character(); | |
1682 | break; | |
1683 | case regex_constants::syntax_open_set: | |
1684 | { | |
1685 | if(m_end == ++m_position) | |
1686 | { | |
1687 | fail(regex_constants::error_collate, m_position - m_base); | |
1688 | return result; | |
1689 | } | |
1690 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_dot) | |
1691 | { | |
1692 | --m_position; | |
1693 | result.first = *m_position; | |
1694 | ++m_position; | |
1695 | return result; | |
1696 | } | |
1697 | if(m_end == ++m_position) | |
1698 | { | |
1699 | fail(regex_constants::error_collate, m_position - m_base); | |
1700 | return result; | |
1701 | } | |
1702 | const charT* name_first = m_position; | |
1703 | // skip at least one character, then find the matching ':]' | |
1704 | if(m_end == ++m_position) | |
1705 | { | |
1706 | fail(regex_constants::error_collate, name_first - m_base); | |
1707 | return result; | |
1708 | } | |
1709 | while((m_position != m_end) | |
1710 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_dot)) | |
1711 | ++m_position; | |
1712 | const charT* name_last = m_position; | |
1713 | if(m_end == m_position) | |
1714 | { | |
1715 | fail(regex_constants::error_collate, name_first - m_base); | |
1716 | return result; | |
1717 | } | |
1718 | if((m_end == ++m_position) | |
1719 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_set)) | |
1720 | { | |
1721 | fail(regex_constants::error_collate, name_first - m_base); | |
1722 | return result; | |
1723 | } | |
1724 | ++m_position; | |
1725 | string_type s = this->m_traits.lookup_collatename(name_first, name_last); | |
1726 | if(s.empty() || (s.size() > 2)) | |
1727 | { | |
1728 | fail(regex_constants::error_collate, name_first - m_base); | |
1729 | return result; | |
1730 | } | |
1731 | result.first = s[0]; | |
1732 | if(s.size() > 1) | |
1733 | result.second = s[1]; | |
1734 | else | |
1735 | result.second = 0; | |
1736 | return result; | |
1737 | } | |
1738 | default: | |
1739 | result = *m_position++; | |
1740 | } | |
1741 | return result; | |
1742 | } | |
1743 | ||
1744 | // | |
1745 | // does a value fit in the specified charT type? | |
1746 | // | |
1747 | template <class charT> | |
1748 | bool valid_value(charT, boost::intmax_t v, const mpl::true_&) | |
1749 | { | |
1750 | return (v >> (sizeof(charT) * CHAR_BIT)) == 0; | |
1751 | } | |
1752 | template <class charT> | |
1753 | bool valid_value(charT, boost::intmax_t, const mpl::false_&) | |
1754 | { | |
1755 | return true; // v will alsways fit in a charT | |
1756 | } | |
1757 | template <class charT> | |
1758 | bool valid_value(charT c, boost::intmax_t v) | |
1759 | { | |
1760 | return valid_value(c, v, mpl::bool_<(sizeof(charT) < sizeof(boost::intmax_t))>()); | |
1761 | } | |
1762 | ||
1763 | template <class charT, class traits> | |
1764 | charT basic_regex_parser<charT, traits>::unescape_character() | |
1765 | { | |
1766 | #ifdef BOOST_MSVC | |
1767 | #pragma warning(push) | |
1768 | #pragma warning(disable:4127) | |
1769 | #endif | |
1770 | charT result(0); | |
1771 | if(m_position == m_end) | |
1772 | { | |
1773 | fail(regex_constants::error_escape, m_position - m_base, "Escape sequence terminated prematurely."); | |
1774 | return false; | |
1775 | } | |
1776 | switch(this->m_traits.escape_syntax_type(*m_position)) | |
1777 | { | |
1778 | case regex_constants::escape_type_control_a: | |
1779 | result = charT('\a'); | |
1780 | break; | |
1781 | case regex_constants::escape_type_e: | |
1782 | result = charT(27); | |
1783 | break; | |
1784 | case regex_constants::escape_type_control_f: | |
1785 | result = charT('\f'); | |
1786 | break; | |
1787 | case regex_constants::escape_type_control_n: | |
1788 | result = charT('\n'); | |
1789 | break; | |
1790 | case regex_constants::escape_type_control_r: | |
1791 | result = charT('\r'); | |
1792 | break; | |
1793 | case regex_constants::escape_type_control_t: | |
1794 | result = charT('\t'); | |
1795 | break; | |
1796 | case regex_constants::escape_type_control_v: | |
1797 | result = charT('\v'); | |
1798 | break; | |
1799 | case regex_constants::escape_type_word_assert: | |
1800 | result = charT('\b'); | |
1801 | break; | |
1802 | case regex_constants::escape_type_ascii_control: | |
1803 | ++m_position; | |
1804 | if(m_position == m_end) | |
1805 | { | |
1806 | // Rewind to start of escape: | |
1807 | --m_position; | |
1808 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1809 | fail(regex_constants::error_escape, m_position - m_base, "ASCII escape sequence terminated prematurely."); | |
1810 | return result; | |
1811 | } | |
1812 | result = static_cast<charT>(*m_position % 32); | |
1813 | break; | |
1814 | case regex_constants::escape_type_hex: | |
1815 | ++m_position; | |
1816 | if(m_position == m_end) | |
1817 | { | |
1818 | // Rewind to start of escape: | |
1819 | --m_position; | |
1820 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1821 | fail(regex_constants::error_escape, m_position - m_base, "Hexadecimal escape sequence terminated prematurely."); | |
1822 | return result; | |
1823 | } | |
1824 | // maybe have \x{ddd} | |
1825 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) | |
1826 | { | |
1827 | ++m_position; | |
1828 | if(m_position == m_end) | |
1829 | { | |
1830 | // Rewind to start of escape: | |
1831 | --m_position; | |
1832 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1833 | fail(regex_constants::error_escape, m_position - m_base, "Missing } in hexadecimal escape sequence."); | |
1834 | return result; | |
1835 | } | |
1836 | boost::intmax_t i = this->m_traits.toi(m_position, m_end, 16); | |
1837 | if((m_position == m_end) | |
1838 | || (i < 0) | |
1839 | || ((std::numeric_limits<charT>::is_specialized) && (i > (boost::intmax_t)(std::numeric_limits<charT>::max)())) | |
1840 | || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) | |
1841 | { | |
1842 | // Rewind to start of escape: | |
1843 | --m_position; | |
1844 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1845 | fail(regex_constants::error_badbrace, m_position - m_base, "Hexadecimal escape sequence was invalid."); | |
1846 | return result; | |
1847 | } | |
1848 | ++m_position; | |
1849 | result = charT(i); | |
1850 | } | |
1851 | else | |
1852 | { | |
1853 | std::ptrdiff_t len = (std::min)(static_cast<std::ptrdiff_t>(2), static_cast<std::ptrdiff_t>(m_end - m_position)); | |
1854 | boost::intmax_t i = this->m_traits.toi(m_position, m_position + len, 16); | |
1855 | if((i < 0) | |
1856 | || !valid_value(charT(0), i)) | |
1857 | { | |
1858 | // Rewind to start of escape: | |
1859 | --m_position; | |
1860 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1861 | fail(regex_constants::error_escape, m_position - m_base, "Escape sequence did not encode a valid character."); | |
1862 | return result; | |
1863 | } | |
1864 | result = charT(i); | |
1865 | } | |
1866 | return result; | |
1867 | case regex_constants::syntax_digit: | |
1868 | { | |
1869 | // an octal escape sequence, the first character must be a zero | |
1870 | // followed by up to 3 octal digits: | |
1871 | std::ptrdiff_t len = (std::min)(::boost::BOOST_REGEX_DETAIL_NS::distance(m_position, m_end), static_cast<std::ptrdiff_t>(4)); | |
1872 | const charT* bp = m_position; | |
1873 | boost::intmax_t val = this->m_traits.toi(bp, bp + 1, 8); | |
1874 | if(val != 0) | |
1875 | { | |
1876 | // Rewind to start of escape: | |
1877 | --m_position; | |
1878 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1879 | // Oops not an octal escape after all: | |
1880 | fail(regex_constants::error_escape, m_position - m_base, "Invalid octal escape sequence."); | |
1881 | return result; | |
1882 | } | |
1883 | val = this->m_traits.toi(m_position, m_position + len, 8); | |
1884 | if((val < 0) || (val > (boost::intmax_t)(std::numeric_limits<charT>::max)())) | |
1885 | { | |
1886 | // Rewind to start of escape: | |
1887 | --m_position; | |
1888 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1889 | fail(regex_constants::error_escape, m_position - m_base, "Octal escape sequence is invalid."); | |
1890 | return result; | |
1891 | } | |
1892 | return static_cast<charT>(val); | |
1893 | } | |
1894 | case regex_constants::escape_type_named_char: | |
1895 | { | |
1896 | ++m_position; | |
1897 | if(m_position == m_end) | |
1898 | { | |
1899 | // Rewind to start of escape: | |
1900 | --m_position; | |
1901 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1902 | fail(regex_constants::error_escape, m_position - m_base); | |
1903 | return false; | |
1904 | } | |
1905 | // maybe have \N{name} | |
1906 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace) | |
1907 | { | |
1908 | const charT* base = m_position; | |
1909 | // skip forward until we find enclosing brace: | |
1910 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace)) | |
1911 | ++m_position; | |
1912 | if(m_position == m_end) | |
1913 | { | |
1914 | // Rewind to start of escape: | |
1915 | --m_position; | |
1916 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1917 | fail(regex_constants::error_escape, m_position - m_base); | |
1918 | return false; | |
1919 | } | |
1920 | string_type s = this->m_traits.lookup_collatename(++base, m_position++); | |
1921 | if(s.empty()) | |
1922 | { | |
1923 | // Rewind to start of escape: | |
1924 | --m_position; | |
1925 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1926 | fail(regex_constants::error_collate, m_position - m_base); | |
1927 | return false; | |
1928 | } | |
1929 | if(s.size() == 1) | |
1930 | { | |
1931 | return s[0]; | |
1932 | } | |
1933 | } | |
1934 | // fall through is a failure: | |
1935 | // Rewind to start of escape: | |
1936 | --m_position; | |
1937 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1938 | fail(regex_constants::error_escape, m_position - m_base); | |
1939 | return false; | |
1940 | } | |
1941 | default: | |
1942 | result = *m_position; | |
1943 | break; | |
1944 | } | |
1945 | ++m_position; | |
1946 | return result; | |
1947 | #ifdef BOOST_MSVC | |
1948 | #pragma warning(pop) | |
1949 | #endif | |
1950 | } | |
1951 | ||
1952 | template <class charT, class traits> | |
1953 | bool basic_regex_parser<charT, traits>::parse_backref() | |
1954 | { | |
1e59de90 | 1955 | BOOST_REGEX_ASSERT(m_position != m_end); |
7c673cae FG |
1956 | const charT* pc = m_position; |
1957 | boost::intmax_t i = this->m_traits.toi(pc, pc + 1, 10); | |
1958 | if((i == 0) || (((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) && (this->flags() & regbase::no_bk_refs))) | |
1959 | { | |
1960 | // not a backref at all but an octal escape sequence: | |
1961 | charT c = unescape_character(); | |
1962 | this->append_literal(c); | |
1963 | } | |
20effc67 | 1964 | else if((i > 0) && (this->m_backrefs.test(i))) |
7c673cae FG |
1965 | { |
1966 | m_position = pc; | |
1967 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace))); | |
1968 | pb->index = i; | |
1969 | pb->icase = this->flags() & regbase::icase; | |
1970 | } | |
1971 | else | |
1972 | { | |
1973 | // Rewind to start of escape: | |
1974 | --m_position; | |
1975 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
1976 | fail(regex_constants::error_backref, m_position - m_base); | |
1977 | return false; | |
1978 | } | |
1979 | return true; | |
1980 | } | |
1981 | ||
1982 | template <class charT, class traits> | |
1983 | bool basic_regex_parser<charT, traits>::parse_QE() | |
1984 | { | |
1985 | #ifdef BOOST_MSVC | |
1986 | #pragma warning(push) | |
1987 | #pragma warning(disable:4127) | |
1988 | #endif | |
1989 | // | |
1990 | // parse a \Q...\E sequence: | |
1991 | // | |
1992 | ++m_position; // skip the Q | |
1993 | const charT* start = m_position; | |
1994 | const charT* end; | |
1995 | do | |
1996 | { | |
1997 | while((m_position != m_end) | |
1998 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape)) | |
1999 | ++m_position; | |
2000 | if(m_position == m_end) | |
2001 | { | |
2002 | // a \Q...\E sequence may terminate with the end of the expression: | |
2003 | end = m_position; | |
2004 | break; | |
2005 | } | |
2006 | if(++m_position == m_end) // skip the escape | |
2007 | { | |
2008 | fail(regex_constants::error_escape, m_position - m_base, "Unterminated \\Q...\\E sequence."); | |
2009 | return false; | |
2010 | } | |
2011 | // check to see if it's a \E: | |
2012 | if(this->m_traits.escape_syntax_type(*m_position) == regex_constants::escape_type_E) | |
2013 | { | |
2014 | ++m_position; | |
2015 | end = m_position - 2; | |
2016 | break; | |
2017 | } | |
2018 | // otherwise go round again: | |
2019 | }while(true); | |
2020 | // | |
2021 | // now add all the character between the two escapes as literals: | |
2022 | // | |
2023 | while(start != end) | |
2024 | { | |
2025 | this->append_literal(*start); | |
2026 | ++start; | |
2027 | } | |
2028 | return true; | |
2029 | #ifdef BOOST_MSVC | |
2030 | #pragma warning(pop) | |
2031 | #endif | |
2032 | } | |
2033 | ||
2034 | template <class charT, class traits> | |
2035 | bool basic_regex_parser<charT, traits>::parse_perl_extension() | |
2036 | { | |
2037 | if(++m_position == m_end) | |
2038 | { | |
2039 | // Rewind to start of (? sequence: | |
2040 | --m_position; | |
2041 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2042 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2043 | return false; | |
2044 | } | |
2045 | // | |
2046 | // treat comments as a special case, as these | |
2047 | // are the only ones that don't start with a leading | |
2048 | // startmark state: | |
2049 | // | |
2050 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_hash) | |
2051 | { | |
2052 | while((m_position != m_end) | |
2053 | && (this->m_traits.syntax_type(*m_position++) != regex_constants::syntax_close_mark)) | |
b32b8144 | 2054 | {} |
7c673cae FG |
2055 | return true; |
2056 | } | |
2057 | // | |
2058 | // backup some state, and prepare the way: | |
2059 | // | |
2060 | int markid = 0; | |
2061 | std::ptrdiff_t jump_offset = 0; | |
2062 | re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace))); | |
2063 | pb->icase = this->flags() & regbase::icase; | |
2064 | std::ptrdiff_t last_paren_start = this->getoffset(pb); | |
2065 | // back up insertion point for alternations, and set new point: | |
2066 | std::ptrdiff_t last_alt_point = m_alt_insert_point; | |
2067 | this->m_pdata->m_data.align(); | |
2068 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2069 | std::ptrdiff_t expected_alt_point = m_alt_insert_point; | |
2070 | bool restore_flags = true; | |
2071 | regex_constants::syntax_option_type old_flags = this->flags(); | |
2072 | bool old_case_change = m_has_case_change; | |
2073 | m_has_case_change = false; | |
2074 | charT name_delim; | |
2075 | int mark_reset = m_mark_reset; | |
2076 | int max_mark = m_max_mark; | |
2077 | m_mark_reset = -1; | |
2078 | m_max_mark = m_mark_count; | |
2079 | boost::intmax_t v; | |
2080 | // | |
2081 | // select the actual extension used: | |
2082 | // | |
2083 | switch(this->m_traits.syntax_type(*m_position)) | |
2084 | { | |
2085 | case regex_constants::syntax_or: | |
2086 | m_mark_reset = m_mark_count; | |
2087 | BOOST_FALLTHROUGH; | |
2088 | case regex_constants::syntax_colon: | |
2089 | // | |
2090 | // a non-capturing mark: | |
2091 | // | |
2092 | pb->index = markid = 0; | |
2093 | ++m_position; | |
2094 | break; | |
2095 | case regex_constants::syntax_digit: | |
2096 | { | |
2097 | // | |
2098 | // a recursive subexpression: | |
2099 | // | |
2100 | v = this->m_traits.toi(m_position, m_end, 10); | |
2101 | if((v < 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2102 | { | |
2103 | // Rewind to start of (? sequence: | |
2104 | --m_position; | |
2105 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2106 | fail(regex_constants::error_perl_extension, m_position - m_base, "The recursive sub-expression refers to an invalid marking group, or is unterminated."); | |
2107 | return false; | |
2108 | } | |
2109 | insert_recursion: | |
2110 | pb->index = markid = 0; | |
2111 | re_recurse* pr = static_cast<re_recurse*>(this->append_state(syntax_element_recurse, sizeof(re_recurse))); | |
2112 | pr->alt.i = v; | |
2113 | pr->state_id = 0; | |
2114 | static_cast<re_case*>( | |
2115 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
2116 | )->icase = this->flags() & regbase::icase; | |
2117 | break; | |
2118 | } | |
2119 | case regex_constants::syntax_plus: | |
2120 | // | |
2121 | // A forward-relative recursive subexpression: | |
2122 | // | |
2123 | ++m_position; | |
2124 | v = this->m_traits.toi(m_position, m_end, 10); | |
2125 | if((v <= 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2126 | { | |
2127 | // Rewind to start of (? sequence: | |
2128 | --m_position; | |
2129 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2130 | fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression."); | |
2131 | return false; | |
2132 | } | |
b32b8144 FG |
2133 | if ((std::numeric_limits<boost::intmax_t>::max)() - m_mark_count < v) |
2134 | { | |
2135 | fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression."); | |
2136 | return false; | |
2137 | } | |
7c673cae FG |
2138 | v += m_mark_count; |
2139 | goto insert_recursion; | |
2140 | case regex_constants::syntax_dash: | |
2141 | // | |
2142 | // Possibly a backward-relative recursive subexpression: | |
2143 | // | |
2144 | ++m_position; | |
2145 | v = this->m_traits.toi(m_position, m_end, 10); | |
2146 | if(v <= 0) | |
2147 | { | |
2148 | --m_position; | |
2149 | // Oops not a relative recursion at all, but a (?-imsx) group: | |
2150 | goto option_group_jump; | |
2151 | } | |
20effc67 | 2152 | v = static_cast<boost::intmax_t>(m_mark_count) + 1 - v; |
7c673cae FG |
2153 | if(v <= 0) |
2154 | { | |
2155 | // Rewind to start of (? sequence: | |
2156 | --m_position; | |
2157 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2158 | fail(regex_constants::error_perl_extension, m_position - m_base, "An invalid or unterminated recursive sub-expression."); | |
2159 | return false; | |
2160 | } | |
2161 | goto insert_recursion; | |
2162 | case regex_constants::syntax_equal: | |
2163 | pb->index = markid = -1; | |
2164 | ++m_position; | |
2165 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
2166 | this->m_pdata->m_data.align(); | |
2167 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2168 | break; | |
2169 | case regex_constants::syntax_not: | |
2170 | pb->index = markid = -2; | |
2171 | ++m_position; | |
2172 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
2173 | this->m_pdata->m_data.align(); | |
2174 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2175 | break; | |
2176 | case regex_constants::escape_type_left_word: | |
2177 | { | |
2178 | // a lookbehind assertion: | |
2179 | if(++m_position == m_end) | |
2180 | { | |
2181 | // Rewind to start of (? sequence: | |
2182 | --m_position; | |
2183 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2184 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2185 | return false; | |
2186 | } | |
2187 | regex_constants::syntax_type t = this->m_traits.syntax_type(*m_position); | |
2188 | if(t == regex_constants::syntax_not) | |
2189 | pb->index = markid = -2; | |
2190 | else if(t == regex_constants::syntax_equal) | |
2191 | pb->index = markid = -1; | |
2192 | else | |
2193 | { | |
2194 | // Probably a named capture which also starts (?< : | |
2195 | name_delim = '>'; | |
2196 | --m_position; | |
2197 | goto named_capture_jump; | |
2198 | } | |
2199 | ++m_position; | |
2200 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
2201 | this->append_state(syntax_element_backstep, sizeof(re_brace)); | |
2202 | this->m_pdata->m_data.align(); | |
2203 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2204 | break; | |
2205 | } | |
2206 | case regex_constants::escape_type_right_word: | |
2207 | // | |
2208 | // an independent sub-expression: | |
2209 | // | |
2210 | pb->index = markid = -3; | |
2211 | ++m_position; | |
2212 | jump_offset = this->getoffset(this->append_state(syntax_element_jump, sizeof(re_jump))); | |
2213 | this->m_pdata->m_data.align(); | |
2214 | m_alt_insert_point = this->m_pdata->m_data.size(); | |
2215 | break; | |
2216 | case regex_constants::syntax_open_mark: | |
2217 | { | |
2218 | // a conditional expression: | |
2219 | pb->index = markid = -4; | |
2220 | if(++m_position == m_end) | |
2221 | { | |
2222 | // Rewind to start of (? sequence: | |
2223 | --m_position; | |
2224 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2225 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2226 | return false; | |
2227 | } | |
2228 | v = this->m_traits.toi(m_position, m_end, 10); | |
2229 | if(m_position == m_end) | |
2230 | { | |
2231 | // Rewind to start of (? sequence: | |
2232 | --m_position; | |
2233 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2234 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2235 | return false; | |
2236 | } | |
2237 | if(*m_position == charT('R')) | |
2238 | { | |
2239 | if(++m_position == m_end) | |
2240 | { | |
2241 | // Rewind to start of (? sequence: | |
2242 | --m_position; | |
2243 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2244 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2245 | return false; | |
2246 | } | |
2247 | if(*m_position == charT('&')) | |
2248 | { | |
2249 | const charT* base = ++m_position; | |
2250 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2251 | ++m_position; | |
2252 | if(m_position == m_end) | |
2253 | { | |
2254 | // Rewind to start of (? sequence: | |
2255 | --m_position; | |
2256 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2257 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2258 | return false; | |
2259 | } | |
2260 | v = -static_cast<int>(hash_value_from_capture_name(base, m_position)); | |
2261 | } | |
2262 | else | |
2263 | { | |
2264 | v = -this->m_traits.toi(m_position, m_end, 10); | |
2265 | } | |
2266 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); | |
2267 | br->index = v < 0 ? (v - 1) : 0; | |
2268 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2269 | { | |
2270 | // Rewind to start of (? sequence: | |
2271 | --m_position; | |
2272 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2273 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2274 | return false; | |
2275 | } | |
2276 | if(++m_position == m_end) | |
2277 | { | |
2278 | // Rewind to start of (? sequence: | |
2279 | --m_position; | |
2280 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2281 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2282 | return false; | |
2283 | } | |
2284 | } | |
2285 | else if((*m_position == charT('\'')) || (*m_position == charT('<'))) | |
2286 | { | |
2287 | const charT* base = ++m_position; | |
2288 | while((m_position != m_end) && (*m_position != charT('>')) && (*m_position != charT('\''))) | |
2289 | ++m_position; | |
2290 | if(m_position == m_end) | |
2291 | { | |
2292 | // Rewind to start of (? sequence: | |
2293 | --m_position; | |
2294 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2295 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2296 | return false; | |
2297 | } | |
2298 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); | |
2299 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); | |
2300 | br->index = v; | |
2301 | if(((*m_position != charT('>')) && (*m_position != charT('\''))) || (++m_position == m_end)) | |
2302 | { | |
2303 | // Rewind to start of (? sequence: | |
2304 | --m_position; | |
2305 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2306 | fail(regex_constants::error_perl_extension, m_position - m_base, "Unterminated named capture."); | |
2307 | return false; | |
2308 | } | |
2309 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2310 | { | |
2311 | // Rewind to start of (? sequence: | |
2312 | --m_position; | |
2313 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2314 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2315 | return false; | |
2316 | } | |
2317 | if(++m_position == m_end) | |
2318 | { | |
2319 | // Rewind to start of (? sequence: | |
2320 | --m_position; | |
2321 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2322 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2323 | return false; | |
2324 | } | |
2325 | } | |
2326 | else if(*m_position == charT('D')) | |
2327 | { | |
2328 | const char* def = "DEFINE"; | |
2329 | while(*def && (m_position != m_end) && (*m_position == charT(*def))) | |
2330 | ++m_position, ++def; | |
2331 | if((m_position == m_end) || *def) | |
2332 | { | |
2333 | // Rewind to start of (? sequence: | |
2334 | --m_position; | |
2335 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2336 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2337 | return false; | |
2338 | } | |
2339 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); | |
2340 | br->index = 9999; // special magic value! | |
2341 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2342 | { | |
2343 | // Rewind to start of (? sequence: | |
2344 | --m_position; | |
2345 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2346 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2347 | return false; | |
2348 | } | |
2349 | if(++m_position == m_end) | |
2350 | { | |
2351 | // Rewind to start of (? sequence: | |
2352 | --m_position; | |
2353 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2354 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2355 | return false; | |
2356 | } | |
2357 | } | |
2358 | else if(v > 0) | |
2359 | { | |
2360 | re_brace* br = static_cast<re_brace*>(this->append_state(syntax_element_assert_backref, sizeof(re_brace))); | |
2361 | br->index = v; | |
2362 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2363 | { | |
2364 | // Rewind to start of (? sequence: | |
2365 | --m_position; | |
2366 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2367 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2368 | return false; | |
2369 | } | |
2370 | if(++m_position == m_end) | |
2371 | { | |
2372 | // Rewind to start of (? sequence: | |
2373 | --m_position; | |
2374 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2375 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2376 | return false; | |
2377 | } | |
2378 | } | |
2379 | else | |
2380 | { | |
2381 | // verify that we have a lookahead or lookbehind assert: | |
2382 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_question) | |
2383 | { | |
2384 | // Rewind to start of (? sequence: | |
2385 | --m_position; | |
2386 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2387 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2388 | return false; | |
2389 | } | |
2390 | if(++m_position == m_end) | |
2391 | { | |
2392 | // Rewind to start of (? sequence: | |
2393 | --m_position; | |
2394 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2395 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2396 | return false; | |
2397 | } | |
2398 | if(this->m_traits.syntax_type(*m_position) == regex_constants::escape_type_left_word) | |
2399 | { | |
2400 | if(++m_position == m_end) | |
2401 | { | |
2402 | // Rewind to start of (? sequence: | |
2403 | --m_position; | |
2404 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2405 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2406 | return false; | |
2407 | } | |
2408 | if((this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal) | |
2409 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_not)) | |
2410 | { | |
2411 | // Rewind to start of (? sequence: | |
2412 | --m_position; | |
2413 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2414 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2415 | return false; | |
2416 | } | |
2417 | m_position -= 3; | |
2418 | } | |
2419 | else | |
2420 | { | |
2421 | if((this->m_traits.syntax_type(*m_position) != regex_constants::syntax_equal) | |
2422 | && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_not)) | |
2423 | { | |
2424 | // Rewind to start of (? sequence: | |
2425 | --m_position; | |
2426 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2427 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2428 | return false; | |
2429 | } | |
2430 | m_position -= 2; | |
2431 | } | |
2432 | } | |
2433 | break; | |
2434 | } | |
2435 | case regex_constants::syntax_close_mark: | |
2436 | // Rewind to start of (? sequence: | |
2437 | --m_position; | |
2438 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2439 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2440 | return false; | |
2441 | case regex_constants::escape_type_end_buffer: | |
2442 | { | |
2443 | name_delim = *m_position; | |
2444 | named_capture_jump: | |
2445 | markid = 0; | |
2446 | if(0 == (this->flags() & regbase::nosubs)) | |
2447 | { | |
2448 | markid = ++m_mark_count; | |
2449 | #ifndef BOOST_NO_STD_DISTANCE | |
2450 | if(this->flags() & regbase::save_subexpression_location) | |
2451 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>(std::distance(m_base, m_position) - 2, 0)); | |
2452 | #else | |
2453 | if(this->flags() & regbase::save_subexpression_location) | |
2454 | this->m_pdata->m_subs.push_back(std::pair<std::size_t, std::size_t>((m_position - m_base) - 2, 0)); | |
2455 | #endif | |
2456 | } | |
2457 | pb->index = markid; | |
2458 | const charT* base = ++m_position; | |
2459 | if(m_position == m_end) | |
2460 | { | |
2461 | // Rewind to start of (? sequence: | |
2462 | --m_position; | |
2463 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2464 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2465 | return false; | |
2466 | } | |
2467 | while((m_position != m_end) && (*m_position != name_delim)) | |
2468 | ++m_position; | |
2469 | if(m_position == m_end) | |
2470 | { | |
2471 | // Rewind to start of (? sequence: | |
2472 | --m_position; | |
2473 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2474 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2475 | return false; | |
2476 | } | |
2477 | this->m_pdata->set_name(base, m_position, markid); | |
2478 | ++m_position; | |
2479 | break; | |
2480 | } | |
2481 | default: | |
2482 | if(*m_position == charT('R')) | |
2483 | { | |
2484 | ++m_position; | |
2485 | v = 0; | |
2486 | if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark) | |
2487 | { | |
2488 | // Rewind to start of (? sequence: | |
2489 | --m_position; | |
2490 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2491 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2492 | return false; | |
2493 | } | |
2494 | goto insert_recursion; | |
2495 | } | |
2496 | if(*m_position == charT('&')) | |
2497 | { | |
2498 | ++m_position; | |
2499 | const charT* base = m_position; | |
2500 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2501 | ++m_position; | |
2502 | if(m_position == m_end) | |
2503 | { | |
2504 | // Rewind to start of (? sequence: | |
2505 | --m_position; | |
2506 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2507 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2508 | return false; | |
2509 | } | |
2510 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); | |
2511 | goto insert_recursion; | |
2512 | } | |
2513 | if(*m_position == charT('P')) | |
2514 | { | |
2515 | ++m_position; | |
2516 | if(m_position == m_end) | |
2517 | { | |
2518 | // Rewind to start of (? sequence: | |
2519 | --m_position; | |
2520 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2521 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2522 | return false; | |
2523 | } | |
2524 | if(*m_position == charT('>')) | |
2525 | { | |
2526 | ++m_position; | |
2527 | const charT* base = m_position; | |
2528 | while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2529 | ++m_position; | |
2530 | if(m_position == m_end) | |
2531 | { | |
2532 | // Rewind to start of (? sequence: | |
2533 | --m_position; | |
2534 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2535 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2536 | return false; | |
2537 | } | |
2538 | v = static_cast<int>(hash_value_from_capture_name(base, m_position)); | |
2539 | goto insert_recursion; | |
2540 | } | |
2541 | } | |
2542 | // | |
2543 | // lets assume that we have a (?imsx) group and try and parse it: | |
2544 | // | |
2545 | option_group_jump: | |
2546 | regex_constants::syntax_option_type opts = parse_options(); | |
2547 | if(m_position == m_end) | |
2548 | { | |
2549 | // Rewind to start of (? sequence: | |
2550 | --m_position; | |
2551 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2552 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2553 | return false; | |
2554 | } | |
2555 | // make a note of whether we have a case change: | |
2556 | m_has_case_change = ((opts & regbase::icase) != (this->flags() & regbase::icase)); | |
2557 | pb->index = markid = 0; | |
2558 | if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark) | |
2559 | { | |
2560 | // update flags and carry on as normal: | |
2561 | this->flags(opts); | |
2562 | restore_flags = false; | |
2563 | old_case_change |= m_has_case_change; // defer end of scope by one ')' | |
2564 | } | |
2565 | else if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_colon) | |
2566 | { | |
2567 | // update flags and carry on until the matching ')' is found: | |
2568 | this->flags(opts); | |
2569 | ++m_position; | |
2570 | } | |
2571 | else | |
2572 | { | |
2573 | // Rewind to start of (? sequence: | |
2574 | --m_position; | |
2575 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2576 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2577 | return false; | |
2578 | } | |
2579 | ||
2580 | // finally append a case change state if we need it: | |
2581 | if(m_has_case_change) | |
2582 | { | |
2583 | static_cast<re_case*>( | |
2584 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
2585 | )->icase = opts & regbase::icase; | |
2586 | } | |
2587 | ||
2588 | } | |
2589 | // | |
2590 | // now recursively add more states, this will terminate when we get to a | |
2591 | // matching ')' : | |
2592 | // | |
2593 | parse_all(); | |
2594 | // | |
2595 | // Unwind alternatives: | |
2596 | // | |
2597 | if(0 == unwind_alts(last_paren_start)) | |
2598 | { | |
2599 | // Rewind to start of (? sequence: | |
2600 | --m_position; | |
2601 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2602 | fail(regex_constants::error_perl_extension, m_position - m_base, "Invalid alternation operators within (?...) block."); | |
2603 | return false; | |
2604 | } | |
2605 | // | |
2606 | // we either have a ')' or we have run out of characters prematurely: | |
2607 | // | |
2608 | if(m_position == m_end) | |
2609 | { | |
2610 | // Rewind to start of (? sequence: | |
2611 | --m_position; | |
2612 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2613 | this->fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_end)); | |
2614 | return false; | |
2615 | } | |
1e59de90 | 2616 | BOOST_REGEX_ASSERT(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark); |
7c673cae FG |
2617 | ++m_position; |
2618 | // | |
2619 | // restore the flags: | |
2620 | // | |
2621 | if(restore_flags) | |
2622 | { | |
2623 | // append a case change state if we need it: | |
2624 | if(m_has_case_change) | |
2625 | { | |
2626 | static_cast<re_case*>( | |
2627 | this->append_state(syntax_element_toggle_case, sizeof(re_case)) | |
2628 | )->icase = old_flags & regbase::icase; | |
2629 | } | |
2630 | this->flags(old_flags); | |
2631 | } | |
2632 | // | |
2633 | // set up the jump pointer if we have one: | |
2634 | // | |
2635 | if(jump_offset) | |
2636 | { | |
2637 | this->m_pdata->m_data.align(); | |
2638 | re_jump* jmp = static_cast<re_jump*>(this->getaddress(jump_offset)); | |
2639 | jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp); | |
2640 | if((this->m_last_state == jmp) && (markid != -2)) | |
2641 | { | |
2642 | // Oops... we didn't have anything inside the assertion. | |
2643 | // Note we don't get here for negated forward lookahead as (?!) | |
2644 | // does have some uses. | |
2645 | // Rewind to start of (? sequence: | |
2646 | --m_position; | |
2647 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2648 | fail(regex_constants::error_perl_extension, m_position - m_base, "Invalid or empty zero width assertion."); | |
2649 | return false; | |
2650 | } | |
2651 | } | |
2652 | // | |
2653 | // verify that if this is conditional expression, that we do have | |
2654 | // an alternative, if not add one: | |
2655 | // | |
2656 | if(markid == -4) | |
2657 | { | |
2658 | re_syntax_base* b = this->getaddress(expected_alt_point); | |
2659 | // Make sure we have exactly one alternative following this state: | |
2660 | if(b->type != syntax_element_alt) | |
2661 | { | |
2662 | re_alt* alt = static_cast<re_alt*>(this->insert_state(expected_alt_point, syntax_element_alt, sizeof(re_alt))); | |
2663 | alt->alt.i = this->m_pdata->m_data.size() - this->getoffset(alt); | |
2664 | } | |
b32b8144 | 2665 | else if(((std::ptrdiff_t)this->m_pdata->m_data.size() > (static_cast<re_alt*>(b)->alt.i + this->getoffset(b))) && (static_cast<re_alt*>(b)->alt.i > 0) && this->getaddress(static_cast<re_alt*>(b)->alt.i, b)->type == syntax_element_alt) |
7c673cae FG |
2666 | { |
2667 | // Can't have seen more than one alternative: | |
2668 | // Rewind to start of (? sequence: | |
2669 | --m_position; | |
2670 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2671 | fail(regex_constants::error_bad_pattern, m_position - m_base, "More than one alternation operator | was encountered inside a conditional expression."); | |
2672 | return false; | |
2673 | } | |
2674 | else | |
2675 | { | |
2676 | // We must *not* have seen an alternative inside a (DEFINE) block: | |
2677 | b = this->getaddress(b->next.i, b); | |
2678 | if((b->type == syntax_element_assert_backref) && (static_cast<re_brace*>(b)->index == 9999)) | |
2679 | { | |
2680 | // Rewind to start of (? sequence: | |
2681 | --m_position; | |
2682 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2683 | fail(regex_constants::error_bad_pattern, m_position - m_base, "Alternation operators are not allowed inside a DEFINE block."); | |
2684 | return false; | |
2685 | } | |
2686 | } | |
2687 | // check for invalid repetition of next state: | |
2688 | b = this->getaddress(expected_alt_point); | |
2689 | b = this->getaddress(static_cast<re_alt*>(b)->next.i, b); | |
2690 | if((b->type != syntax_element_assert_backref) | |
2691 | && (b->type != syntax_element_startmark)) | |
2692 | { | |
2693 | // Rewind to start of (? sequence: | |
2694 | --m_position; | |
2695 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2696 | fail(regex_constants::error_badrepeat, m_position - m_base, "A repetition operator cannot be applied to a zero-width assertion."); | |
2697 | return false; | |
2698 | } | |
2699 | } | |
2700 | // | |
2701 | // append closing parenthesis state: | |
2702 | // | |
2703 | pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace))); | |
2704 | pb->index = markid; | |
2705 | pb->icase = this->flags() & regbase::icase; | |
2706 | this->m_paren_start = last_paren_start; | |
2707 | // | |
2708 | // restore the alternate insertion point: | |
2709 | // | |
2710 | this->m_alt_insert_point = last_alt_point; | |
2711 | // | |
2712 | // and the case change data: | |
2713 | // | |
2714 | m_has_case_change = old_case_change; | |
2715 | // | |
2716 | // And the mark_reset data: | |
2717 | // | |
2718 | if(m_max_mark > m_mark_count) | |
2719 | { | |
2720 | m_mark_count = m_max_mark; | |
2721 | } | |
2722 | m_mark_reset = mark_reset; | |
2723 | m_max_mark = max_mark; | |
2724 | ||
2725 | ||
2726 | if(markid > 0) | |
2727 | { | |
2728 | #ifndef BOOST_NO_STD_DISTANCE | |
2729 | if(this->flags() & regbase::save_subexpression_location) | |
20effc67 | 2730 | this->m_pdata->m_subs.at((std::size_t)markid - 1).second = std::distance(m_base, m_position) - 1; |
7c673cae FG |
2731 | #else |
2732 | if(this->flags() & regbase::save_subexpression_location) | |
2733 | this->m_pdata->m_subs.at(markid - 1).second = (m_position - m_base) - 1; | |
2734 | #endif | |
2735 | // | |
2736 | // allow backrefs to this mark: | |
2737 | // | |
20effc67 | 2738 | this->m_backrefs.set(markid); |
7c673cae FG |
2739 | } |
2740 | return true; | |
2741 | } | |
2742 | ||
2743 | template <class charT, class traits> | |
2744 | bool basic_regex_parser<charT, traits>::match_verb(const char* verb) | |
2745 | { | |
2746 | while(*verb) | |
2747 | { | |
2748 | if(static_cast<charT>(*verb) != *m_position) | |
2749 | { | |
2750 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2751 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2752 | return false; | |
2753 | } | |
2754 | if(++m_position == m_end) | |
2755 | { | |
2756 | --m_position; | |
2757 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2758 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2759 | return false; | |
2760 | } | |
2761 | ++verb; | |
2762 | } | |
2763 | return true; | |
2764 | } | |
2765 | ||
20effc67 TL |
2766 | #ifdef BOOST_MSVC |
2767 | # pragma warning(push) | |
2768 | #if BOOST_MSVC >= 1800 | |
2769 | #pragma warning(disable:26812) | |
2770 | #endif | |
2771 | #endif | |
7c673cae FG |
2772 | template <class charT, class traits> |
2773 | bool basic_regex_parser<charT, traits>::parse_perl_verb() | |
2774 | { | |
2775 | if(++m_position == m_end) | |
2776 | { | |
2777 | // Rewind to start of (* sequence: | |
2778 | --m_position; | |
2779 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2780 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2781 | return false; | |
2782 | } | |
2783 | switch(*m_position) | |
2784 | { | |
2785 | case 'F': | |
2786 | if(++m_position == m_end) | |
2787 | { | |
2788 | // Rewind to start of (* sequence: | |
2789 | --m_position; | |
2790 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2791 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2792 | return false; | |
2793 | } | |
2794 | if((this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark) || match_verb("AIL")) | |
2795 | { | |
2796 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2797 | { | |
2798 | // Rewind to start of (* sequence: | |
2799 | --m_position; | |
2800 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2801 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2802 | return false; | |
2803 | } | |
2804 | ++m_position; | |
2805 | this->append_state(syntax_element_fail); | |
2806 | return true; | |
2807 | } | |
2808 | break; | |
2809 | case 'A': | |
2810 | if(++m_position == m_end) | |
2811 | { | |
2812 | // Rewind to start of (* sequence: | |
2813 | --m_position; | |
2814 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2815 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2816 | return false; | |
2817 | } | |
2818 | if(match_verb("CCEPT")) | |
2819 | { | |
2820 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2821 | { | |
2822 | // Rewind to start of (* sequence: | |
2823 | --m_position; | |
2824 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2825 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2826 | return false; | |
2827 | } | |
2828 | ++m_position; | |
2829 | this->append_state(syntax_element_accept); | |
2830 | return true; | |
2831 | } | |
2832 | break; | |
2833 | case 'C': | |
2834 | if(++m_position == m_end) | |
2835 | { | |
2836 | // Rewind to start of (* sequence: | |
2837 | --m_position; | |
2838 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2839 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2840 | return false; | |
2841 | } | |
2842 | if(match_verb("OMMIT")) | |
2843 | { | |
2844 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2845 | { | |
2846 | // Rewind to start of (* sequence: | |
2847 | --m_position; | |
2848 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2849 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2850 | return false; | |
2851 | } | |
2852 | ++m_position; | |
2853 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_commit; | |
2854 | this->m_pdata->m_disable_match_any = true; | |
2855 | return true; | |
2856 | } | |
2857 | break; | |
2858 | case 'P': | |
2859 | if(++m_position == m_end) | |
2860 | { | |
2861 | // Rewind to start of (* sequence: | |
2862 | --m_position; | |
2863 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2864 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2865 | return false; | |
2866 | } | |
2867 | if(match_verb("RUNE")) | |
2868 | { | |
2869 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2870 | { | |
2871 | // Rewind to start of (* sequence: | |
2872 | --m_position; | |
2873 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2874 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2875 | return false; | |
2876 | } | |
2877 | ++m_position; | |
2878 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_prune; | |
2879 | this->m_pdata->m_disable_match_any = true; | |
2880 | return true; | |
2881 | } | |
2882 | break; | |
2883 | case 'S': | |
2884 | if(++m_position == m_end) | |
2885 | { | |
2886 | // Rewind to start of (* sequence: | |
2887 | --m_position; | |
2888 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2889 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2890 | return false; | |
2891 | } | |
2892 | if(match_verb("KIP")) | |
2893 | { | |
2894 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2895 | { | |
2896 | // Rewind to start of (* sequence: | |
2897 | --m_position; | |
2898 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2899 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2900 | return false; | |
2901 | } | |
2902 | ++m_position; | |
2903 | static_cast<re_commit*>(this->append_state(syntax_element_commit, sizeof(re_commit)))->action = commit_skip; | |
2904 | this->m_pdata->m_disable_match_any = true; | |
2905 | return true; | |
2906 | } | |
2907 | break; | |
2908 | case 'T': | |
2909 | if(++m_position == m_end) | |
2910 | { | |
2911 | // Rewind to start of (* sequence: | |
2912 | --m_position; | |
2913 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2914 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2915 | return false; | |
2916 | } | |
2917 | if(match_verb("HEN")) | |
2918 | { | |
2919 | if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)) | |
2920 | { | |
2921 | // Rewind to start of (* sequence: | |
2922 | --m_position; | |
2923 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2924 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
2925 | return false; | |
2926 | } | |
2927 | ++m_position; | |
2928 | this->append_state(syntax_element_then); | |
2929 | this->m_pdata->m_disable_match_any = true; | |
2930 | return true; | |
2931 | } | |
2932 | break; | |
2933 | } | |
b32b8144 FG |
2934 | // Rewind to start of (* sequence: |
2935 | --m_position; | |
2936 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
2937 | fail(regex_constants::error_perl_extension, m_position - m_base); | |
7c673cae FG |
2938 | return false; |
2939 | } | |
20effc67 TL |
2940 | #ifdef BOOST_MSVC |
2941 | # pragma warning(pop) | |
2942 | #endif | |
7c673cae FG |
2943 | |
2944 | template <class charT, class traits> | |
2945 | bool basic_regex_parser<charT, traits>::add_emacs_code(bool negate) | |
2946 | { | |
2947 | // | |
2948 | // parses an emacs style \sx or \Sx construct. | |
2949 | // | |
2950 | if(++m_position == m_end) | |
2951 | { | |
2952 | // Rewind to start of sequence: | |
2953 | --m_position; | |
2954 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_escape) --m_position; | |
2955 | fail(regex_constants::error_escape, m_position - m_base); | |
2956 | return false; | |
2957 | } | |
2958 | basic_char_set<charT, traits> char_set; | |
2959 | if(negate) | |
2960 | char_set.negate(); | |
2961 | ||
2962 | static const charT s_punct[5] = { 'p', 'u', 'n', 'c', 't', }; | |
2963 | ||
2964 | switch(*m_position) | |
2965 | { | |
2966 | case 's': | |
2967 | case ' ': | |
2968 | char_set.add_class(this->m_mask_space); | |
2969 | break; | |
2970 | case 'w': | |
2971 | char_set.add_class(this->m_word_mask); | |
2972 | break; | |
2973 | case '_': | |
2974 | char_set.add_single(digraph<charT>(charT('$'))); | |
2975 | char_set.add_single(digraph<charT>(charT('&'))); | |
2976 | char_set.add_single(digraph<charT>(charT('*'))); | |
2977 | char_set.add_single(digraph<charT>(charT('+'))); | |
2978 | char_set.add_single(digraph<charT>(charT('-'))); | |
2979 | char_set.add_single(digraph<charT>(charT('_'))); | |
2980 | char_set.add_single(digraph<charT>(charT('<'))); | |
2981 | char_set.add_single(digraph<charT>(charT('>'))); | |
2982 | break; | |
2983 | case '.': | |
2984 | char_set.add_class(this->m_traits.lookup_classname(s_punct, s_punct+5)); | |
2985 | break; | |
2986 | case '(': | |
2987 | char_set.add_single(digraph<charT>(charT('('))); | |
2988 | char_set.add_single(digraph<charT>(charT('['))); | |
2989 | char_set.add_single(digraph<charT>(charT('{'))); | |
2990 | break; | |
2991 | case ')': | |
2992 | char_set.add_single(digraph<charT>(charT(')'))); | |
2993 | char_set.add_single(digraph<charT>(charT(']'))); | |
2994 | char_set.add_single(digraph<charT>(charT('}'))); | |
2995 | break; | |
2996 | case '"': | |
2997 | char_set.add_single(digraph<charT>(charT('"'))); | |
2998 | char_set.add_single(digraph<charT>(charT('\''))); | |
2999 | char_set.add_single(digraph<charT>(charT('`'))); | |
3000 | break; | |
3001 | case '\'': | |
3002 | char_set.add_single(digraph<charT>(charT('\''))); | |
3003 | char_set.add_single(digraph<charT>(charT(','))); | |
3004 | char_set.add_single(digraph<charT>(charT('#'))); | |
3005 | break; | |
3006 | case '<': | |
3007 | char_set.add_single(digraph<charT>(charT(';'))); | |
3008 | break; | |
3009 | case '>': | |
3010 | char_set.add_single(digraph<charT>(charT('\n'))); | |
3011 | char_set.add_single(digraph<charT>(charT('\f'))); | |
3012 | break; | |
3013 | default: | |
3014 | fail(regex_constants::error_ctype, m_position - m_base); | |
3015 | return false; | |
3016 | } | |
3017 | if(0 == this->append_set(char_set)) | |
3018 | { | |
3019 | fail(regex_constants::error_ctype, m_position - m_base); | |
3020 | return false; | |
3021 | } | |
3022 | ++m_position; | |
3023 | return true; | |
3024 | } | |
3025 | ||
3026 | template <class charT, class traits> | |
3027 | regex_constants::syntax_option_type basic_regex_parser<charT, traits>::parse_options() | |
3028 | { | |
3029 | // we have a (?imsx-imsx) group, convert it into a set of flags: | |
3030 | regex_constants::syntax_option_type f = this->flags(); | |
3031 | bool breakout = false; | |
3032 | do | |
3033 | { | |
3034 | switch(*m_position) | |
3035 | { | |
3036 | case 's': | |
3037 | f |= regex_constants::mod_s; | |
3038 | f &= ~regex_constants::no_mod_s; | |
3039 | break; | |
3040 | case 'm': | |
3041 | f &= ~regex_constants::no_mod_m; | |
3042 | break; | |
3043 | case 'i': | |
3044 | f |= regex_constants::icase; | |
3045 | break; | |
3046 | case 'x': | |
3047 | f |= regex_constants::mod_x; | |
3048 | break; | |
3049 | default: | |
3050 | breakout = true; | |
3051 | continue; | |
3052 | } | |
3053 | if(++m_position == m_end) | |
3054 | { | |
3055 | // Rewind to start of (? sequence: | |
3056 | --m_position; | |
3057 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
3058 | fail(regex_constants::error_paren, m_position - m_base); | |
3059 | return false; | |
3060 | } | |
3061 | } | |
3062 | while(!breakout); | |
3063 | ||
3064 | breakout = false; | |
3065 | ||
3066 | if(*m_position == static_cast<charT>('-')) | |
3067 | { | |
3068 | if(++m_position == m_end) | |
3069 | { | |
3070 | // Rewind to start of (? sequence: | |
3071 | --m_position; | |
3072 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
3073 | fail(regex_constants::error_paren, m_position - m_base); | |
3074 | return false; | |
3075 | } | |
3076 | do | |
3077 | { | |
3078 | switch(*m_position) | |
3079 | { | |
3080 | case 's': | |
3081 | f &= ~regex_constants::mod_s; | |
3082 | f |= regex_constants::no_mod_s; | |
3083 | break; | |
3084 | case 'm': | |
3085 | f |= regex_constants::no_mod_m; | |
3086 | break; | |
3087 | case 'i': | |
3088 | f &= ~regex_constants::icase; | |
3089 | break; | |
3090 | case 'x': | |
3091 | f &= ~regex_constants::mod_x; | |
3092 | break; | |
3093 | default: | |
3094 | breakout = true; | |
3095 | continue; | |
3096 | } | |
3097 | if(++m_position == m_end) | |
3098 | { | |
3099 | // Rewind to start of (? sequence: | |
3100 | --m_position; | |
3101 | while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position; | |
3102 | fail(regex_constants::error_paren, m_position - m_base); | |
3103 | return false; | |
3104 | } | |
3105 | } | |
3106 | while(!breakout); | |
3107 | } | |
3108 | return f; | |
3109 | } | |
3110 | ||
3111 | template <class charT, class traits> | |
3112 | bool basic_regex_parser<charT, traits>::unwind_alts(std::ptrdiff_t last_paren_start) | |
3113 | { | |
3114 | // | |
3115 | // If we didn't actually add any states after the last | |
3116 | // alternative then that's an error: | |
3117 | // | |
3118 | if((this->m_alt_insert_point == static_cast<std::ptrdiff_t>(this->m_pdata->m_data.size())) | |
1e59de90 | 3119 | && (!m_alt_jumps.empty()) && (m_alt_jumps.back() > last_paren_start) |
7c673cae FG |
3120 | && |
3121 | !( | |
3122 | ((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group) | |
3123 | && | |
3124 | ((this->flags() & regbase::no_empty_expressions) == 0) | |
3125 | ) | |
3126 | ) | |
3127 | { | |
3128 | fail(regex_constants::error_empty, this->m_position - this->m_base, "Can't terminate a sub-expression with an alternation operator |."); | |
3129 | return false; | |
3130 | } | |
3131 | // | |
3132 | // Fix up our alternatives: | |
3133 | // | |
1e59de90 | 3134 | while((!m_alt_jumps.empty()) && (m_alt_jumps.back() > last_paren_start)) |
7c673cae FG |
3135 | { |
3136 | // | |
3137 | // fix up the jump to point to the end of the states | |
3138 | // that we've just added: | |
3139 | // | |
3140 | std::ptrdiff_t jump_offset = m_alt_jumps.back(); | |
3141 | m_alt_jumps.pop_back(); | |
3142 | this->m_pdata->m_data.align(); | |
3143 | re_jump* jmp = static_cast<re_jump*>(this->getaddress(jump_offset)); | |
1e59de90 TL |
3144 | if (jmp->type != syntax_element_jump) |
3145 | { | |
3146 | // Something really bad happened, this used to be an assert, | |
3147 | // but we'll make it an error just in case we should ever get here. | |
3148 | fail(regex_constants::error_unknown, this->m_position - this->m_base, "Internal logic failed while compiling the expression, probably you added a repeat to something non-repeatable!"); | |
3149 | return false; | |
3150 | } | |
7c673cae FG |
3151 | jmp->alt.i = this->m_pdata->m_data.size() - jump_offset; |
3152 | } | |
3153 | return true; | |
3154 | } | |
3155 | ||
3156 | #ifdef BOOST_MSVC | |
3157 | #pragma warning(pop) | |
3158 | #endif | |
3159 | ||
3160 | } // namespace BOOST_REGEX_DETAIL_NS | |
3161 | } // namespace boost | |
3162 | ||
3163 | #ifdef BOOST_MSVC | |
3164 | #pragma warning(push) | |
3165 | #pragma warning(disable: 4103) | |
3166 | #endif | |
3167 | #ifdef BOOST_HAS_ABI_HEADERS | |
3168 | # include BOOST_ABI_SUFFIX | |
3169 | #endif | |
3170 | #ifdef BOOST_MSVC | |
3171 | #pragma warning(pop) | |
3172 | #endif | |
3173 | ||
3174 | #endif |