]>
Commit | Line | Data |
---|---|---|
7c673cae FG |
1 | /* |
2 | * | |
3 | * Copyright (c) 1998-2009 John Maddock | |
4 | * Copyright 2008 Eric Niebler. | |
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 regex_format.hpp | |
15 | * VERSION see <boost/version.hpp> | |
16 | * DESCRIPTION: Provides formatting output routines for search and replace | |
17 | * operations. Note this is an internal header file included | |
18 | * by regex.hpp, do not include on its own. | |
19 | */ | |
20 | ||
21 | #ifndef BOOST_REGEX_FORMAT_HPP | |
22 | #define BOOST_REGEX_FORMAT_HPP | |
23 | ||
24 | #include <boost/type_traits/is_pointer.hpp> | |
25 | #include <boost/type_traits/is_function.hpp> | |
26 | #include <boost/type_traits/is_class.hpp> | |
27 | #include <boost/type_traits/is_same.hpp> | |
28 | #include <boost/type_traits/is_convertible.hpp> | |
29 | #include <boost/type_traits/remove_pointer.hpp> | |
30 | #include <boost/type_traits/remove_cv.hpp> | |
31 | #include <boost/mpl/if.hpp> | |
32 | #include <boost/mpl/and.hpp> | |
33 | #include <boost/mpl/not.hpp> | |
34 | #ifndef BOOST_NO_SFINAE | |
35 | #include <boost/mpl/has_xxx.hpp> | |
36 | #endif | |
37 | #include <boost/ref.hpp> | |
38 | ||
39 | namespace boost{ | |
40 | ||
41 | #ifdef BOOST_MSVC | |
42 | #pragma warning(push) | |
43 | #pragma warning(disable: 4103) | |
44 | #endif | |
45 | #ifdef BOOST_HAS_ABI_HEADERS | |
46 | # include BOOST_ABI_PREFIX | |
47 | #endif | |
48 | #ifdef BOOST_MSVC | |
49 | #pragma warning(pop) | |
50 | #endif | |
51 | ||
52 | // | |
53 | // Forward declaration: | |
54 | // | |
55 | template <class BidiIterator, class Allocator = BOOST_DEDUCED_TYPENAME std::vector<sub_match<BidiIterator> >::allocator_type > | |
56 | class match_results; | |
57 | ||
58 | namespace BOOST_REGEX_DETAIL_NS{ | |
59 | ||
60 | // | |
61 | // struct trivial_format_traits: | |
62 | // defines minimum localisation support for formatting | |
63 | // in the case that the actual regex traits is unavailable. | |
64 | // | |
65 | template <class charT> | |
66 | struct trivial_format_traits | |
67 | { | |
68 | typedef charT char_type; | |
69 | ||
70 | static std::ptrdiff_t length(const charT* p) | |
71 | { | |
72 | return global_length(p); | |
73 | } | |
74 | static charT tolower(charT c) | |
75 | { | |
76 | return ::boost::BOOST_REGEX_DETAIL_NS::global_lower(c); | |
77 | } | |
78 | static charT toupper(charT c) | |
79 | { | |
80 | return ::boost::BOOST_REGEX_DETAIL_NS::global_upper(c); | |
81 | } | |
82 | static int value(const charT c, int radix) | |
83 | { | |
84 | int result = global_value(c); | |
85 | return result >= radix ? -1 : result; | |
86 | } | |
87 | int toi(const charT*& p1, const charT* p2, int radix)const | |
88 | { | |
89 | return (int)global_toi(p1, p2, radix, *this); | |
90 | } | |
91 | }; | |
92 | ||
93 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
94 | class basic_regex_formatter | |
95 | { | |
96 | public: | |
97 | typedef typename traits::char_type char_type; | |
98 | basic_regex_formatter(OutputIterator o, const Results& r, const traits& t) | |
99 | : m_traits(t), m_results(r), m_out(o), m_position(), m_end(), m_flags(), m_state(output_copy), m_restore_state(output_copy), m_have_conditional(false) {} | |
100 | OutputIterator format(ForwardIter p1, ForwardIter p2, match_flag_type f); | |
101 | OutputIterator format(ForwardIter p1, match_flag_type f) | |
102 | { | |
103 | return format(p1, p1 + m_traits.length(p1), f); | |
104 | } | |
105 | private: | |
106 | typedef typename Results::value_type sub_match_type; | |
107 | enum output_state | |
108 | { | |
109 | output_copy, | |
110 | output_next_lower, | |
111 | output_next_upper, | |
112 | output_lower, | |
113 | output_upper, | |
114 | output_none | |
115 | }; | |
116 | ||
117 | void put(char_type c); | |
118 | void put(const sub_match_type& sub); | |
119 | void format_all(); | |
120 | void format_perl(); | |
121 | void format_escape(); | |
122 | void format_conditional(); | |
123 | void format_until_scope_end(); | |
124 | bool handle_perl_verb(bool have_brace); | |
125 | ||
126 | inline typename Results::value_type const& get_named_sub(ForwardIter i, ForwardIter j, const mpl::false_&) | |
127 | { | |
128 | std::vector<char_type> v(i, j); | |
129 | return (i != j) ? this->m_results.named_subexpression(&v[0], &v[0] + v.size()) | |
130 | : this->m_results.named_subexpression(static_cast<const char_type*>(0), static_cast<const char_type*>(0)); | |
131 | } | |
132 | inline typename Results::value_type const& get_named_sub(ForwardIter i, ForwardIter j, const mpl::true_&) | |
133 | { | |
134 | return this->m_results.named_subexpression(i, j); | |
135 | } | |
136 | inline typename Results::value_type const& get_named_sub(ForwardIter i, ForwardIter j) | |
137 | { | |
138 | typedef typename boost::is_convertible<ForwardIter, const char_type*>::type tag_type; | |
139 | return get_named_sub(i, j, tag_type()); | |
140 | } | |
141 | inline int get_named_sub_index(ForwardIter i, ForwardIter j, const mpl::false_&) | |
142 | { | |
143 | std::vector<char_type> v(i, j); | |
144 | return (i != j) ? this->m_results.named_subexpression_index(&v[0], &v[0] + v.size()) | |
145 | : this->m_results.named_subexpression_index(static_cast<const char_type*>(0), static_cast<const char_type*>(0)); | |
146 | } | |
147 | inline int get_named_sub_index(ForwardIter i, ForwardIter j, const mpl::true_&) | |
148 | { | |
149 | return this->m_results.named_subexpression_index(i, j); | |
150 | } | |
151 | inline int get_named_sub_index(ForwardIter i, ForwardIter j) | |
152 | { | |
153 | typedef typename boost::is_convertible<ForwardIter, const char_type*>::type tag_type; | |
154 | return get_named_sub_index(i, j, tag_type()); | |
155 | } | |
156 | #ifdef BOOST_MSVC | |
157 | // msvc-8.0 issues a spurious warning on the call to std::advance here: | |
158 | #pragma warning(push) | |
159 | #pragma warning(disable:4244) | |
160 | #endif | |
161 | inline int toi(ForwardIter& i, ForwardIter j, int base, const boost::mpl::false_&) | |
162 | { | |
163 | if(i != j) | |
164 | { | |
165 | std::vector<char_type> v(i, j); | |
166 | const char_type* start = &v[0]; | |
167 | const char_type* pos = start; | |
168 | int r = (int)m_traits.toi(pos, &v[0] + v.size(), base); | |
169 | std::advance(i, pos - start); | |
170 | return r; | |
171 | } | |
172 | return -1; | |
173 | } | |
174 | #ifdef BOOST_MSVC | |
175 | #pragma warning(pop) | |
176 | #endif | |
177 | inline int toi(ForwardIter& i, ForwardIter j, int base, const boost::mpl::true_&) | |
178 | { | |
179 | return m_traits.toi(i, j, base); | |
180 | } | |
181 | inline int toi(ForwardIter& i, ForwardIter j, int base) | |
182 | { | |
183 | #if defined(_MSC_VER) && defined(__INTEL_COMPILER) && ((__INTEL_COMPILER == 9999) || (__INTEL_COMPILER == 1210)) | |
184 | // Workaround for Intel support issue #656654. | |
185 | // See also https://svn.boost.org/trac/boost/ticket/6359 | |
186 | return toi(i, j, base, mpl::false_()); | |
187 | #else | |
188 | typedef typename boost::is_convertible<ForwardIter, const char_type*&>::type tag_type; | |
189 | return toi(i, j, base, tag_type()); | |
190 | #endif | |
191 | } | |
192 | ||
193 | const traits& m_traits; // the traits class for localised formatting operations | |
194 | const Results& m_results; // the match_results being used. | |
195 | OutputIterator m_out; // where to send output. | |
196 | ForwardIter m_position; // format string, current position | |
197 | ForwardIter m_end; // format string end | |
198 | match_flag_type m_flags; // format flags to use | |
199 | output_state m_state; // what to do with the next character | |
200 | output_state m_restore_state; // what state to restore to. | |
201 | bool m_have_conditional; // we are parsing a conditional | |
202 | private: | |
203 | basic_regex_formatter(const basic_regex_formatter&); | |
204 | basic_regex_formatter& operator=(const basic_regex_formatter&); | |
205 | }; | |
206 | ||
207 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
208 | OutputIterator basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::format(ForwardIter p1, ForwardIter p2, match_flag_type f) | |
209 | { | |
210 | m_position = p1; | |
211 | m_end = p2; | |
212 | m_flags = f; | |
213 | format_all(); | |
214 | return m_out; | |
215 | } | |
216 | ||
217 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
218 | void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::format_all() | |
219 | { | |
220 | // over and over: | |
221 | while(m_position != m_end) | |
222 | { | |
223 | switch(*m_position) | |
224 | { | |
225 | case '&': | |
226 | if(m_flags & ::boost::regex_constants::format_sed) | |
227 | { | |
228 | ++m_position; | |
229 | put(m_results[0]); | |
230 | break; | |
231 | } | |
232 | put(*m_position++); | |
233 | break; | |
234 | case '\\': | |
235 | format_escape(); | |
236 | break; | |
237 | case '(': | |
238 | if(m_flags & boost::regex_constants::format_all) | |
239 | { | |
240 | ++m_position; | |
241 | bool have_conditional = m_have_conditional; | |
242 | m_have_conditional = false; | |
243 | format_until_scope_end(); | |
244 | m_have_conditional = have_conditional; | |
245 | if(m_position == m_end) | |
246 | return; | |
247 | BOOST_ASSERT(*m_position == static_cast<char_type>(')')); | |
248 | ++m_position; // skip the closing ')' | |
249 | break; | |
250 | } | |
251 | put(*m_position); | |
252 | ++m_position; | |
253 | break; | |
254 | case ')': | |
255 | if(m_flags & boost::regex_constants::format_all) | |
256 | { | |
257 | return; | |
258 | } | |
259 | put(*m_position); | |
260 | ++m_position; | |
261 | break; | |
262 | case ':': | |
263 | if((m_flags & boost::regex_constants::format_all) && m_have_conditional) | |
264 | { | |
265 | return; | |
266 | } | |
267 | put(*m_position); | |
268 | ++m_position; | |
269 | break; | |
270 | case '?': | |
271 | if(m_flags & boost::regex_constants::format_all) | |
272 | { | |
273 | ++m_position; | |
274 | format_conditional(); | |
275 | break; | |
276 | } | |
277 | put(*m_position); | |
278 | ++m_position; | |
279 | break; | |
280 | case '$': | |
281 | if((m_flags & format_sed) == 0) | |
282 | { | |
283 | format_perl(); | |
284 | break; | |
285 | } | |
286 | // not a special character: | |
287 | BOOST_FALLTHROUGH; | |
288 | default: | |
289 | put(*m_position); | |
290 | ++m_position; | |
291 | break; | |
292 | } | |
293 | } | |
294 | } | |
295 | ||
296 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
297 | void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::format_perl() | |
298 | { | |
299 | // | |
300 | // On entry *m_position points to a '$' character | |
301 | // output the information that goes with it: | |
302 | // | |
303 | BOOST_ASSERT(*m_position == '$'); | |
304 | // | |
305 | // see if this is a trailing '$': | |
306 | // | |
307 | if(++m_position == m_end) | |
308 | { | |
309 | --m_position; | |
310 | put(*m_position); | |
311 | ++m_position; | |
312 | return; | |
313 | } | |
314 | // | |
315 | // OK find out what kind it is: | |
316 | // | |
317 | bool have_brace = false; | |
318 | ForwardIter save_position = m_position; | |
319 | switch(*m_position) | |
320 | { | |
321 | case '&': | |
322 | ++m_position; | |
323 | put(this->m_results[0]); | |
324 | break; | |
325 | case '`': | |
326 | ++m_position; | |
327 | put(this->m_results.prefix()); | |
328 | break; | |
329 | case '\'': | |
330 | ++m_position; | |
331 | put(this->m_results.suffix()); | |
332 | break; | |
333 | case '$': | |
334 | put(*m_position++); | |
335 | break; | |
336 | case '+': | |
337 | if((++m_position != m_end) && (*m_position == '{')) | |
338 | { | |
339 | ForwardIter base = ++m_position; | |
340 | while((m_position != m_end) && (*m_position != '}')) ++m_position; | |
341 | if(m_position != m_end) | |
342 | { | |
343 | // Named sub-expression: | |
344 | put(get_named_sub(base, m_position)); | |
345 | ++m_position; | |
346 | break; | |
347 | } | |
348 | else | |
349 | { | |
350 | m_position = --base; | |
351 | } | |
352 | } | |
353 | put((this->m_results)[this->m_results.size() > 1 ? static_cast<int>(this->m_results.size() - 1) : 1]); | |
354 | break; | |
355 | case '{': | |
356 | have_brace = true; | |
357 | ++m_position; | |
358 | BOOST_FALLTHROUGH; | |
359 | default: | |
360 | // see if we have a number: | |
361 | { | |
362 | std::ptrdiff_t len = ::boost::BOOST_REGEX_DETAIL_NS::distance(m_position, m_end); | |
363 | //len = (std::min)(static_cast<std::ptrdiff_t>(2), len); | |
364 | int v = this->toi(m_position, m_position + len, 10); | |
365 | if((v < 0) || (have_brace && ((m_position == m_end) || (*m_position != '}')))) | |
366 | { | |
367 | // Look for a Perl-5.10 verb: | |
368 | if(!handle_perl_verb(have_brace)) | |
369 | { | |
370 | // leave the $ as is, and carry on: | |
371 | m_position = --save_position; | |
372 | put(*m_position); | |
373 | ++m_position; | |
374 | } | |
375 | break; | |
376 | } | |
377 | // otherwise output sub v: | |
378 | put(this->m_results[v]); | |
379 | if(have_brace) | |
380 | ++m_position; | |
381 | } | |
382 | } | |
383 | } | |
384 | ||
385 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
386 | bool basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::handle_perl_verb(bool have_brace) | |
387 | { | |
388 | // | |
389 | // We may have a capitalised string containing a Perl action: | |
390 | // | |
391 | static const char_type MATCH[] = { 'M', 'A', 'T', 'C', 'H' }; | |
392 | static const char_type PREMATCH[] = { 'P', 'R', 'E', 'M', 'A', 'T', 'C', 'H' }; | |
393 | static const char_type POSTMATCH[] = { 'P', 'O', 'S', 'T', 'M', 'A', 'T', 'C', 'H' }; | |
394 | static const char_type LAST_PAREN_MATCH[] = { 'L', 'A', 'S', 'T', '_', 'P', 'A', 'R', 'E', 'N', '_', 'M', 'A', 'T', 'C', 'H' }; | |
395 | static const char_type LAST_SUBMATCH_RESULT[] = { 'L', 'A', 'S', 'T', '_', 'S', 'U', 'B', 'M', 'A', 'T', 'C', 'H', '_', 'R', 'E', 'S', 'U', 'L', 'T' }; | |
396 | static const char_type LAST_SUBMATCH_RESULT_ALT[] = { '^', 'N' }; | |
397 | ||
398 | if(m_position == m_end) | |
399 | return false; | |
400 | if(have_brace && (*m_position == '^')) | |
401 | ++m_position; | |
402 | ||
403 | std::ptrdiff_t max_len = m_end - m_position; | |
404 | ||
405 | if((max_len >= 5) && std::equal(m_position, m_position + 5, MATCH)) | |
406 | { | |
407 | m_position += 5; | |
408 | if(have_brace) | |
409 | { | |
410 | if((m_position != m_end) && (*m_position == '}')) | |
411 | ++m_position; | |
412 | else | |
413 | { | |
414 | m_position -= 5; | |
415 | return false; | |
416 | } | |
417 | } | |
418 | put(this->m_results[0]); | |
419 | return true; | |
420 | } | |
421 | if((max_len >= 8) && std::equal(m_position, m_position + 8, PREMATCH)) | |
422 | { | |
423 | m_position += 8; | |
424 | if(have_brace) | |
425 | { | |
426 | if((m_position != m_end) && (*m_position == '}')) | |
427 | ++m_position; | |
428 | else | |
429 | { | |
430 | m_position -= 8; | |
431 | return false; | |
432 | } | |
433 | } | |
434 | put(this->m_results.prefix()); | |
435 | return true; | |
436 | } | |
437 | if((max_len >= 9) && std::equal(m_position, m_position + 9, POSTMATCH)) | |
438 | { | |
439 | m_position += 9; | |
440 | if(have_brace) | |
441 | { | |
442 | if((m_position != m_end) && (*m_position == '}')) | |
443 | ++m_position; | |
444 | else | |
445 | { | |
446 | m_position -= 9; | |
447 | return false; | |
448 | } | |
449 | } | |
450 | put(this->m_results.suffix()); | |
451 | return true; | |
452 | } | |
453 | if((max_len >= 16) && std::equal(m_position, m_position + 16, LAST_PAREN_MATCH)) | |
454 | { | |
455 | m_position += 16; | |
456 | if(have_brace) | |
457 | { | |
458 | if((m_position != m_end) && (*m_position == '}')) | |
459 | ++m_position; | |
460 | else | |
461 | { | |
462 | m_position -= 16; | |
463 | return false; | |
464 | } | |
465 | } | |
466 | put((this->m_results)[this->m_results.size() > 1 ? static_cast<int>(this->m_results.size() - 1) : 1]); | |
467 | return true; | |
468 | } | |
469 | if((max_len >= 20) && std::equal(m_position, m_position + 20, LAST_SUBMATCH_RESULT)) | |
470 | { | |
471 | m_position += 20; | |
472 | if(have_brace) | |
473 | { | |
474 | if((m_position != m_end) && (*m_position == '}')) | |
475 | ++m_position; | |
476 | else | |
477 | { | |
478 | m_position -= 20; | |
479 | return false; | |
480 | } | |
481 | } | |
482 | put(this->m_results.get_last_closed_paren()); | |
483 | return true; | |
484 | } | |
485 | if((max_len >= 2) && std::equal(m_position, m_position + 2, LAST_SUBMATCH_RESULT_ALT)) | |
486 | { | |
487 | m_position += 2; | |
488 | if(have_brace) | |
489 | { | |
490 | if((m_position != m_end) && (*m_position == '}')) | |
491 | ++m_position; | |
492 | else | |
493 | { | |
494 | m_position -= 2; | |
495 | return false; | |
496 | } | |
497 | } | |
498 | put(this->m_results.get_last_closed_paren()); | |
499 | return true; | |
500 | } | |
501 | return false; | |
502 | } | |
503 | ||
504 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
505 | void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::format_escape() | |
506 | { | |
507 | // skip the escape and check for trailing escape: | |
508 | if(++m_position == m_end) | |
509 | { | |
510 | put(static_cast<char_type>('\\')); | |
511 | return; | |
512 | } | |
513 | // now switch on the escape type: | |
514 | switch(*m_position) | |
515 | { | |
516 | case 'a': | |
517 | put(static_cast<char_type>('\a')); | |
518 | ++m_position; | |
519 | break; | |
520 | case 'f': | |
521 | put(static_cast<char_type>('\f')); | |
522 | ++m_position; | |
523 | break; | |
524 | case 'n': | |
525 | put(static_cast<char_type>('\n')); | |
526 | ++m_position; | |
527 | break; | |
528 | case 'r': | |
529 | put(static_cast<char_type>('\r')); | |
530 | ++m_position; | |
531 | break; | |
532 | case 't': | |
533 | put(static_cast<char_type>('\t')); | |
534 | ++m_position; | |
535 | break; | |
536 | case 'v': | |
537 | put(static_cast<char_type>('\v')); | |
538 | ++m_position; | |
539 | break; | |
540 | case 'x': | |
541 | if(++m_position == m_end) | |
542 | { | |
543 | put(static_cast<char_type>('x')); | |
544 | return; | |
545 | } | |
546 | // maybe have \x{ddd} | |
547 | if(*m_position == static_cast<char_type>('{')) | |
548 | { | |
549 | ++m_position; | |
550 | int val = this->toi(m_position, m_end, 16); | |
551 | if(val < 0) | |
552 | { | |
553 | // invalid value treat everything as literals: | |
554 | put(static_cast<char_type>('x')); | |
555 | put(static_cast<char_type>('{')); | |
556 | return; | |
557 | } | |
558 | if((m_position == m_end) || (*m_position != static_cast<char_type>('}'))) | |
559 | { | |
560 | --m_position; | |
561 | while(*m_position != static_cast<char_type>('\\')) | |
562 | --m_position; | |
563 | ++m_position; | |
564 | put(*m_position++); | |
565 | return; | |
566 | } | |
567 | ++m_position; | |
568 | put(static_cast<char_type>(val)); | |
569 | return; | |
570 | } | |
571 | else | |
572 | { | |
573 | std::ptrdiff_t len = ::boost::BOOST_REGEX_DETAIL_NS::distance(m_position, m_end); | |
574 | len = (std::min)(static_cast<std::ptrdiff_t>(2), len); | |
575 | int val = this->toi(m_position, m_position + len, 16); | |
576 | if(val < 0) | |
577 | { | |
578 | --m_position; | |
579 | put(*m_position++); | |
580 | return; | |
581 | } | |
582 | put(static_cast<char_type>(val)); | |
583 | } | |
584 | break; | |
585 | case 'c': | |
586 | if(++m_position == m_end) | |
587 | { | |
588 | --m_position; | |
589 | put(*m_position++); | |
590 | return; | |
591 | } | |
592 | put(static_cast<char_type>(*m_position++ % 32)); | |
593 | break; | |
594 | case 'e': | |
595 | put(static_cast<char_type>(27)); | |
596 | ++m_position; | |
597 | break; | |
598 | default: | |
599 | // see if we have a perl specific escape: | |
600 | if((m_flags & boost::regex_constants::format_sed) == 0) | |
601 | { | |
602 | bool breakout = false; | |
603 | switch(*m_position) | |
604 | { | |
605 | case 'l': | |
606 | ++m_position; | |
607 | m_restore_state = m_state; | |
608 | m_state = output_next_lower; | |
609 | breakout = true; | |
610 | break; | |
611 | case 'L': | |
612 | ++m_position; | |
613 | m_state = output_lower; | |
614 | breakout = true; | |
615 | break; | |
616 | case 'u': | |
617 | ++m_position; | |
618 | m_restore_state = m_state; | |
619 | m_state = output_next_upper; | |
620 | breakout = true; | |
621 | break; | |
622 | case 'U': | |
623 | ++m_position; | |
624 | m_state = output_upper; | |
625 | breakout = true; | |
626 | break; | |
627 | case 'E': | |
628 | ++m_position; | |
629 | m_state = output_copy; | |
630 | breakout = true; | |
631 | break; | |
632 | } | |
633 | if(breakout) | |
634 | break; | |
635 | } | |
636 | // see if we have a \n sed style backreference: | |
637 | std::ptrdiff_t len = ::boost::BOOST_REGEX_DETAIL_NS::distance(m_position, m_end); | |
638 | len = (std::min)(static_cast<std::ptrdiff_t>(1), len); | |
639 | int v = this->toi(m_position, m_position+len, 10); | |
640 | if((v > 0) || ((v == 0) && (m_flags & ::boost::regex_constants::format_sed))) | |
641 | { | |
642 | put(m_results[v]); | |
643 | break; | |
644 | } | |
645 | else if(v == 0) | |
646 | { | |
647 | // octal ecape sequence: | |
648 | --m_position; | |
649 | len = ::boost::BOOST_REGEX_DETAIL_NS::distance(m_position, m_end); | |
650 | len = (std::min)(static_cast<std::ptrdiff_t>(4), len); | |
651 | v = this->toi(m_position, m_position + len, 8); | |
652 | BOOST_ASSERT(v >= 0); | |
653 | put(static_cast<char_type>(v)); | |
654 | break; | |
655 | } | |
656 | // Otherwise output the character "as is": | |
657 | put(*m_position++); | |
658 | break; | |
659 | } | |
660 | } | |
661 | ||
662 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
663 | void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::format_conditional() | |
664 | { | |
665 | if(m_position == m_end) | |
666 | { | |
667 | // oops trailing '?': | |
668 | put(static_cast<char_type>('?')); | |
669 | return; | |
670 | } | |
671 | int v; | |
672 | if(*m_position == '{') | |
673 | { | |
674 | ForwardIter base = m_position; | |
675 | ++m_position; | |
676 | v = this->toi(m_position, m_end, 10); | |
677 | if(v < 0) | |
678 | { | |
679 | // Try a named subexpression: | |
680 | while((m_position != m_end) && (*m_position != '}')) | |
681 | ++m_position; | |
682 | v = this->get_named_sub_index(base + 1, m_position); | |
683 | } | |
684 | if((v < 0) || (*m_position != '}')) | |
685 | { | |
686 | m_position = base; | |
687 | // oops trailing '?': | |
688 | put(static_cast<char_type>('?')); | |
689 | return; | |
690 | } | |
691 | // Skip trailing '}': | |
692 | ++m_position; | |
693 | } | |
694 | else | |
695 | { | |
696 | std::ptrdiff_t len = ::boost::BOOST_REGEX_DETAIL_NS::distance(m_position, m_end); | |
697 | len = (std::min)(static_cast<std::ptrdiff_t>(2), len); | |
698 | v = this->toi(m_position, m_position + len, 10); | |
699 | } | |
700 | if(v < 0) | |
701 | { | |
702 | // oops not a number: | |
703 | put(static_cast<char_type>('?')); | |
704 | return; | |
705 | } | |
706 | ||
707 | // output varies depending upon whether sub-expression v matched or not: | |
708 | if(m_results[v].matched) | |
709 | { | |
710 | m_have_conditional = true; | |
711 | format_all(); | |
712 | m_have_conditional = false; | |
713 | if((m_position != m_end) && (*m_position == static_cast<char_type>(':'))) | |
714 | { | |
715 | // skip the ':': | |
716 | ++m_position; | |
717 | // save output state, then turn it off: | |
718 | output_state saved_state = m_state; | |
719 | m_state = output_none; | |
720 | // format the rest of this scope: | |
721 | format_until_scope_end(); | |
722 | // restore output state: | |
723 | m_state = saved_state; | |
724 | } | |
725 | } | |
726 | else | |
727 | { | |
728 | // save output state, then turn it off: | |
729 | output_state saved_state = m_state; | |
730 | m_state = output_none; | |
731 | // format until ':' or ')': | |
732 | m_have_conditional = true; | |
733 | format_all(); | |
734 | m_have_conditional = false; | |
735 | // restore state: | |
736 | m_state = saved_state; | |
737 | if((m_position != m_end) && (*m_position == static_cast<char_type>(':'))) | |
738 | { | |
739 | // skip the ':': | |
740 | ++m_position; | |
741 | // format the rest of this scope: | |
742 | format_until_scope_end(); | |
743 | } | |
744 | } | |
745 | } | |
746 | ||
747 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
748 | void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::format_until_scope_end() | |
749 | { | |
750 | do | |
751 | { | |
752 | format_all(); | |
753 | if((m_position == m_end) || (*m_position == static_cast<char_type>(')'))) | |
754 | return; | |
755 | put(*m_position++); | |
756 | }while(m_position != m_end); | |
757 | } | |
758 | ||
759 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
760 | void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::put(char_type c) | |
761 | { | |
762 | // write a single character to output | |
763 | // according to which case translation mode we are in: | |
764 | switch(this->m_state) | |
765 | { | |
766 | case output_none: | |
767 | return; | |
768 | case output_next_lower: | |
769 | c = m_traits.tolower(c); | |
770 | this->m_state = m_restore_state; | |
771 | break; | |
772 | case output_next_upper: | |
773 | c = m_traits.toupper(c); | |
774 | this->m_state = m_restore_state; | |
775 | break; | |
776 | case output_lower: | |
777 | c = m_traits.tolower(c); | |
778 | break; | |
779 | case output_upper: | |
780 | c = m_traits.toupper(c); | |
781 | break; | |
782 | default: | |
783 | break; | |
784 | } | |
785 | *m_out = c; | |
786 | ++m_out; | |
787 | } | |
788 | ||
789 | template <class OutputIterator, class Results, class traits, class ForwardIter> | |
790 | void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::put(const sub_match_type& sub) | |
791 | { | |
792 | typedef typename sub_match_type::iterator iterator_type; | |
793 | iterator_type i = sub.first; | |
794 | while(i != sub.second) | |
795 | { | |
796 | put(*i); | |
797 | ++i; | |
798 | } | |
799 | } | |
800 | ||
801 | template <class S> | |
802 | class string_out_iterator | |
803 | #ifndef BOOST_NO_STD_ITERATOR | |
804 | : public std::iterator<std::output_iterator_tag, typename S::value_type> | |
805 | #endif | |
806 | { | |
807 | S* out; | |
808 | public: | |
809 | string_out_iterator(S& s) : out(&s) {} | |
810 | string_out_iterator& operator++() { return *this; } | |
811 | string_out_iterator& operator++(int) { return *this; } | |
812 | string_out_iterator& operator*() { return *this; } | |
813 | string_out_iterator& operator=(typename S::value_type v) | |
814 | { | |
815 | out->append(1, v); | |
816 | return *this; | |
817 | } | |
818 | ||
819 | #ifdef BOOST_NO_STD_ITERATOR | |
820 | typedef std::ptrdiff_t difference_type; | |
821 | typedef typename S::value_type value_type; | |
822 | typedef value_type* pointer; | |
823 | typedef value_type& reference; | |
824 | typedef std::output_iterator_tag iterator_category; | |
825 | #endif | |
826 | }; | |
827 | ||
828 | template <class OutputIterator, class Iterator, class Alloc, class ForwardIter, class traits> | |
829 | OutputIterator regex_format_imp(OutputIterator out, | |
830 | const match_results<Iterator, Alloc>& m, | |
831 | ForwardIter p1, ForwardIter p2, | |
832 | match_flag_type flags, | |
833 | const traits& t | |
834 | ) | |
835 | { | |
836 | if(flags & regex_constants::format_literal) | |
837 | { | |
838 | return BOOST_REGEX_DETAIL_NS::copy(p1, p2, out); | |
839 | } | |
840 | ||
841 | BOOST_REGEX_DETAIL_NS::basic_regex_formatter< | |
842 | OutputIterator, | |
843 | match_results<Iterator, Alloc>, | |
844 | traits, ForwardIter> f(out, m, t); | |
845 | return f.format(p1, p2, flags); | |
846 | } | |
847 | ||
848 | #ifndef BOOST_NO_SFINAE | |
849 | ||
850 | BOOST_MPL_HAS_XXX_TRAIT_DEF(const_iterator) | |
851 | ||
852 | struct any_type | |
853 | { | |
854 | template <class T> | |
855 | any_type(const T&); | |
856 | template <class T, class U> | |
857 | any_type(const T&, const U&); | |
858 | template <class T, class U, class V> | |
859 | any_type(const T&, const U&, const V&); | |
860 | }; | |
861 | typedef char no_type; | |
862 | typedef char (&unary_type)[2]; | |
863 | typedef char (&binary_type)[3]; | |
864 | typedef char (&ternary_type)[4]; | |
865 | ||
866 | no_type check_is_formatter(unary_type, binary_type, ternary_type); | |
867 | template<typename T> | |
868 | unary_type check_is_formatter(T const &, binary_type, ternary_type); | |
869 | template<typename T> | |
870 | binary_type check_is_formatter(unary_type, T const &, ternary_type); | |
871 | template<typename T, typename U> | |
872 | binary_type check_is_formatter(T const &, U const &, ternary_type); | |
873 | template<typename T> | |
874 | ternary_type check_is_formatter(unary_type, binary_type, T const &); | |
875 | template<typename T, typename U> | |
876 | ternary_type check_is_formatter(T const &, binary_type, U const &); | |
877 | template<typename T, typename U> | |
878 | ternary_type check_is_formatter(unary_type, T const &, U const &); | |
879 | template<typename T, typename U, typename V> | |
880 | ternary_type check_is_formatter(T const &, U const &, V const &); | |
881 | ||
882 | struct unary_binary_ternary | |
883 | { | |
884 | typedef unary_type (*unary_fun)(any_type); | |
885 | typedef binary_type (*binary_fun)(any_type, any_type); | |
886 | typedef ternary_type (*ternary_fun)(any_type, any_type, any_type); | |
887 | operator unary_fun(); | |
888 | operator binary_fun(); | |
889 | operator ternary_fun(); | |
890 | }; | |
891 | ||
892 | template<typename Formatter, bool IsFunction = boost::is_function<Formatter>::value> | |
893 | struct formatter_wrapper | |
894 | : Formatter | |
895 | , unary_binary_ternary | |
896 | { | |
897 | formatter_wrapper(){} | |
898 | }; | |
899 | ||
900 | template<typename Formatter> | |
901 | struct formatter_wrapper<Formatter, true> | |
902 | : unary_binary_ternary | |
903 | { | |
904 | operator Formatter *(); | |
905 | }; | |
906 | ||
907 | template<typename Formatter> | |
908 | struct formatter_wrapper<Formatter *, false> | |
909 | : unary_binary_ternary | |
910 | { | |
911 | operator Formatter *(); | |
912 | }; | |
913 | ||
914 | template <class F, class M, class O> | |
915 | struct format_traits_imp | |
916 | { | |
917 | private: | |
918 | // | |
919 | // F must be a pointer, a function, or a class with a function call operator: | |
920 | // | |
921 | BOOST_STATIC_ASSERT((::boost::is_pointer<F>::value || ::boost::is_function<F>::value || ::boost::is_class<F>::value)); | |
922 | static formatter_wrapper<typename unwrap_reference<F>::type> f; | |
923 | static M m; | |
924 | static O out; | |
925 | static boost::regex_constants::match_flag_type flags; | |
926 | public: | |
927 | BOOST_STATIC_CONSTANT(int, value = sizeof(check_is_formatter(f(m), f(m, out), f(m, out, flags)))); | |
928 | }; | |
929 | ||
930 | template <class F, class M, class O> | |
931 | struct format_traits | |
932 | { | |
933 | public: | |
934 | // | |
935 | // Type is mpl::int_<N> where N is one of: | |
936 | // | |
937 | // 0 : F is a pointer to a presumably null-terminated string. | |
938 | // 1 : F is a character-container such as a std::string. | |
939 | // 2 : F is a Unary Functor. | |
940 | // 3 : F is a Binary Functor. | |
941 | // 4 : F is a Ternary Functor. | |
942 | // | |
943 | typedef typename boost::mpl::if_< | |
944 | boost::mpl::and_<boost::is_pointer<F>, boost::mpl::not_<boost::is_function<typename boost::remove_pointer<F>::type> > >, | |
945 | boost::mpl::int_<0>, | |
946 | typename boost::mpl::if_< | |
947 | has_const_iterator<F>, | |
948 | boost::mpl::int_<1>, | |
949 | boost::mpl::int_<format_traits_imp<F, M, O>::value> | |
950 | >::type | |
951 | >::type type; | |
952 | // | |
953 | // This static assertion will fail if the functor passed does not accept | |
954 | // the same type of arguments passed. | |
955 | // | |
956 | BOOST_STATIC_ASSERT( boost::is_class<F>::value && !has_const_iterator<F>::value ? (type::value > 1) : true); | |
957 | }; | |
958 | ||
959 | #else // BOOST_NO_SFINAE | |
960 | ||
961 | template <class F, class M, class O> | |
962 | struct format_traits | |
963 | { | |
964 | public: | |
965 | // | |
966 | // Type is mpl::int_<N> where N is one of: | |
967 | // | |
968 | // 0 : F is a pointer to a presumably null-terminated string. | |
969 | // 1 : F is a character-container such as a std::string. | |
970 | // | |
971 | // Other options such as F being a Functor are not supported without | |
972 | // SFINAE support. | |
973 | // | |
974 | typedef typename boost::mpl::if_< | |
975 | boost::is_pointer<F>, | |
976 | boost::mpl::int_<0>, | |
977 | boost::mpl::int_<1> | |
978 | >::type type; | |
979 | }; | |
980 | ||
981 | #endif // BOOST_NO_SFINAE | |
982 | ||
983 | template <class Base, class Match> | |
984 | struct format_functor3 | |
985 | { | |
986 | format_functor3(Base b) : func(b) {} | |
987 | template <class OutputIter> | |
988 | OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f) | |
989 | { | |
990 | return boost::unwrap_ref(func)(m, i, f); | |
991 | } | |
992 | template <class OutputIter, class Traits> | |
993 | OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f, const Traits&) | |
994 | { | |
995 | return (*this)(m, i, f); | |
996 | } | |
997 | private: | |
998 | Base func; | |
999 | format_functor3(const format_functor3&); | |
1000 | format_functor3& operator=(const format_functor3&); | |
1001 | }; | |
1002 | ||
1003 | template <class Base, class Match> | |
1004 | struct format_functor2 | |
1005 | { | |
1006 | format_functor2(Base b) : func(b) {} | |
1007 | template <class OutputIter> | |
1008 | OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type /*f*/) | |
1009 | { | |
1010 | return boost::unwrap_ref(func)(m, i); | |
1011 | } | |
1012 | template <class OutputIter, class Traits> | |
1013 | OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f, const Traits&) | |
1014 | { | |
1015 | return (*this)(m, i, f); | |
1016 | } | |
1017 | private: | |
1018 | Base func; | |
1019 | format_functor2(const format_functor2&); | |
1020 | format_functor2& operator=(const format_functor2&); | |
1021 | }; | |
1022 | ||
1023 | template <class Base, class Match> | |
1024 | struct format_functor1 | |
1025 | { | |
1026 | format_functor1(Base b) : func(b) {} | |
1027 | ||
1028 | template <class S, class OutputIter> | |
1029 | OutputIter do_format_string(const S& s, OutputIter i) | |
1030 | { | |
1031 | return BOOST_REGEX_DETAIL_NS::copy(s.begin(), s.end(), i); | |
1032 | } | |
1033 | template <class S, class OutputIter> | |
1034 | inline OutputIter do_format_string(const S* s, OutputIter i) | |
1035 | { | |
1036 | while(s && *s) | |
1037 | { | |
1038 | *i = *s; | |
1039 | ++i; | |
1040 | ++s; | |
1041 | } | |
1042 | return i; | |
1043 | } | |
1044 | template <class OutputIter> | |
1045 | OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type /*f*/) | |
1046 | { | |
1047 | return do_format_string(boost::unwrap_ref(func)(m), i); | |
1048 | } | |
1049 | template <class OutputIter, class Traits> | |
1050 | OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f, const Traits&) | |
1051 | { | |
1052 | return (*this)(m, i, f); | |
1053 | } | |
1054 | private: | |
1055 | Base func; | |
1056 | format_functor1(const format_functor1&); | |
1057 | format_functor1& operator=(const format_functor1&); | |
1058 | }; | |
1059 | ||
1060 | template <class charT, class Match, class Traits> | |
1061 | struct format_functor_c_string | |
1062 | { | |
1063 | format_functor_c_string(const charT* ps) : func(ps) {} | |
1064 | ||
1065 | template <class OutputIter> | |
1066 | OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f, const Traits& t = Traits()) | |
1067 | { | |
1068 | //typedef typename Match::char_type char_type; | |
1069 | const charT* end = func; | |
1070 | while(*end) ++end; | |
1071 | return regex_format_imp(i, m, func, end, f, t); | |
1072 | } | |
1073 | private: | |
1074 | const charT* func; | |
1075 | format_functor_c_string(const format_functor_c_string&); | |
1076 | format_functor_c_string& operator=(const format_functor_c_string&); | |
1077 | }; | |
1078 | ||
1079 | template <class Container, class Match, class Traits> | |
1080 | struct format_functor_container | |
1081 | { | |
1082 | format_functor_container(const Container& c) : func(c) {} | |
1083 | ||
1084 | template <class OutputIter> | |
1085 | OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f, const Traits& t = Traits()) | |
1086 | { | |
1087 | //typedef typename Match::char_type char_type; | |
1088 | return BOOST_REGEX_DETAIL_NS::regex_format_imp(i, m, func.begin(), func.end(), f, t); | |
1089 | } | |
1090 | private: | |
1091 | const Container& func; | |
1092 | format_functor_container(const format_functor_container&); | |
1093 | format_functor_container& operator=(const format_functor_container&); | |
1094 | }; | |
1095 | ||
1096 | template <class Func, class Match, class OutputIterator, class Traits = BOOST_REGEX_DETAIL_NS::trivial_format_traits<typename Match::char_type> > | |
1097 | struct compute_functor_type | |
1098 | { | |
1099 | typedef typename format_traits<Func, Match, OutputIterator>::type tag; | |
1100 | typedef typename boost::remove_cv< typename boost::remove_pointer<Func>::type>::type maybe_char_type; | |
1101 | ||
1102 | typedef typename mpl::if_< | |
1103 | ::boost::is_same<tag, mpl::int_<0> >, format_functor_c_string<maybe_char_type, Match, Traits>, | |
1104 | typename mpl::if_< | |
1105 | ::boost::is_same<tag, mpl::int_<1> >, format_functor_container<Func, Match, Traits>, | |
1106 | typename mpl::if_< | |
1107 | ::boost::is_same<tag, mpl::int_<2> >, format_functor1<Func, Match>, | |
1108 | typename mpl::if_< | |
1109 | ::boost::is_same<tag, mpl::int_<3> >, format_functor2<Func, Match>, | |
1110 | format_functor3<Func, Match> | |
1111 | >::type | |
1112 | >::type | |
1113 | >::type | |
1114 | >::type type; | |
1115 | }; | |
1116 | ||
1117 | } // namespace BOOST_REGEX_DETAIL_NS | |
1118 | ||
1119 | template <class OutputIterator, class Iterator, class Allocator, class Functor> | |
1120 | inline OutputIterator regex_format(OutputIterator out, | |
1121 | const match_results<Iterator, Allocator>& m, | |
1122 | Functor fmt, | |
1123 | match_flag_type flags = format_all | |
1124 | ) | |
1125 | { | |
1126 | return m.format(out, fmt, flags); | |
1127 | } | |
1128 | ||
1129 | template <class Iterator, class Allocator, class Functor> | |
1130 | inline std::basic_string<typename match_results<Iterator, Allocator>::char_type> regex_format(const match_results<Iterator, Allocator>& m, | |
1131 | Functor fmt, | |
1132 | match_flag_type flags = format_all) | |
1133 | { | |
1134 | return m.format(fmt, flags); | |
1135 | } | |
1136 | ||
1137 | #ifdef BOOST_MSVC | |
1138 | #pragma warning(push) | |
1139 | #pragma warning(disable: 4103) | |
1140 | #endif | |
1141 | #ifdef BOOST_HAS_ABI_HEADERS | |
1142 | # include BOOST_ABI_SUFFIX | |
1143 | #endif | |
1144 | #ifdef BOOST_MSVC | |
1145 | #pragma warning(pop) | |
1146 | #endif | |
1147 | ||
1148 | } // namespace boost | |
1149 | ||
1150 | #endif // BOOST_REGEX_FORMAT_HPP | |
1151 | ||
1152 | ||
1153 | ||
1154 | ||
1155 | ||
1156 |