|
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 cpp_regex_traits.hpp |
|
15 * VERSION see <boost/version.hpp> |
|
16 * DESCRIPTION: Declares regular expression traits class cpp_regex_traits. |
|
17 */ |
|
18 |
|
19 #ifndef BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED |
|
20 #define BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED |
|
21 |
|
22 #include <boost/config.hpp> |
|
23 |
|
24 #ifndef BOOST_NO_STD_LOCALE |
|
25 |
|
26 #ifndef BOOST_RE_PAT_EXCEPT_HPP |
|
27 #include <boost/regex/pattern_except.hpp> |
|
28 #endif |
|
29 #ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED |
|
30 #include <boost/regex/v4/regex_traits_defaults.hpp> |
|
31 #endif |
|
32 #ifdef BOOST_HAS_THREADS |
|
33 #include <boost/regex/pending/static_mutex.hpp> |
|
34 #endif |
|
35 #ifndef BOOST_REGEX_PRIMARY_TRANSFORM |
|
36 #include <boost/regex/v4/primary_transform.hpp> |
|
37 #endif |
|
38 #ifndef BOOST_REGEX_OBJECT_CACHE_HPP |
|
39 #include <boost/regex/pending/object_cache.hpp> |
|
40 #endif |
|
41 |
|
42 #include <istream> |
|
43 #include <ios> |
|
44 #include <climits> |
|
45 |
|
46 #ifdef BOOST_MSVC |
|
47 #pragma warning(push) |
|
48 #pragma warning(disable: 4103) |
|
49 #endif |
|
50 #ifdef BOOST_HAS_ABI_HEADERS |
|
51 # include BOOST_ABI_PREFIX |
|
52 #endif |
|
53 #ifdef BOOST_MSVC |
|
54 #pragma warning(pop) |
|
55 #endif |
|
56 |
|
57 #ifdef BOOST_MSVC |
|
58 #pragma warning(push) |
|
59 #pragma warning(disable:4786) |
|
60 #endif |
|
61 |
|
62 namespace boost{ |
|
63 |
|
64 // |
|
65 // forward declaration is needed by some compilers: |
|
66 // |
|
67 template <class charT> |
|
68 class cpp_regex_traits; |
|
69 |
|
70 namespace re_detail{ |
|
71 |
|
72 // |
|
73 // class parser_buf: |
|
74 // acts as a stream buffer which wraps around a pair of pointers: |
|
75 // |
|
76 template <class charT, |
|
77 class traits = ::std::char_traits<charT> > |
|
78 class parser_buf : public ::std::basic_streambuf<charT, traits> |
|
79 { |
|
80 typedef ::std::basic_streambuf<charT, traits> base_type; |
|
81 typedef typename base_type::int_type int_type; |
|
82 typedef typename base_type::char_type char_type; |
|
83 typedef typename base_type::pos_type pos_type; |
|
84 typedef ::std::streamsize streamsize; |
|
85 typedef typename base_type::off_type off_type; |
|
86 public: |
|
87 parser_buf() : base_type() { setbuf(0, 0); } |
|
88 const charT* getnext() { return this->gptr(); } |
|
89 protected: |
|
90 std::basic_streambuf<charT, traits>* setbuf(char_type* s, streamsize n); |
|
91 typename parser_buf<charT, traits>::pos_type seekpos(pos_type sp, ::std::ios_base::openmode which); |
|
92 typename parser_buf<charT, traits>::pos_type seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which); |
|
93 private: |
|
94 parser_buf& operator=(const parser_buf&); |
|
95 parser_buf(const parser_buf&); |
|
96 }; |
|
97 |
|
98 template<class charT, class traits> |
|
99 std::basic_streambuf<charT, traits>* |
|
100 parser_buf<charT, traits>::setbuf(char_type* s, streamsize n) |
|
101 { |
|
102 this->setg(s, s, s + n); |
|
103 return this; |
|
104 } |
|
105 |
|
106 template<class charT, class traits> |
|
107 typename parser_buf<charT, traits>::pos_type |
|
108 parser_buf<charT, traits>::seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which) |
|
109 { |
|
110 if(which & ::std::ios_base::out) |
|
111 return pos_type(off_type(-1)); |
|
112 std::ptrdiff_t size = this->egptr() - this->eback(); |
|
113 std::ptrdiff_t pos = this->gptr() - this->eback(); |
|
114 charT* g = this->eback(); |
|
115 switch(way) |
|
116 { |
|
117 case ::std::ios_base::beg: |
|
118 if((off < 0) || (off > size)) |
|
119 return pos_type(off_type(-1)); |
|
120 else |
|
121 this->setg(g, g + off, g + size); |
|
122 break; |
|
123 case ::std::ios_base::end: |
|
124 if((off < 0) || (off > size)) |
|
125 return pos_type(off_type(-1)); |
|
126 else |
|
127 this->setg(g, g + size - off, g + size); |
|
128 break; |
|
129 case ::std::ios_base::cur: |
|
130 { |
|
131 std::ptrdiff_t newpos = static_cast<std::ptrdiff_t>(pos + off); |
|
132 if((newpos < 0) || (newpos > size)) |
|
133 return pos_type(off_type(-1)); |
|
134 else |
|
135 this->setg(g, g + newpos, g + size); |
|
136 break; |
|
137 } |
|
138 default: ; |
|
139 } |
|
140 #ifdef BOOST_MSVC |
|
141 #pragma warning(push) |
|
142 #pragma warning(disable:4244) |
|
143 #endif |
|
144 return static_cast<pos_type>(this->gptr() - this->eback()); |
|
145 #ifdef BOOST_MSVC |
|
146 #pragma warning(pop) |
|
147 #endif |
|
148 } |
|
149 |
|
150 template<class charT, class traits> |
|
151 typename parser_buf<charT, traits>::pos_type |
|
152 parser_buf<charT, traits>::seekpos(pos_type sp, ::std::ios_base::openmode which) |
|
153 { |
|
154 if(which & ::std::ios_base::out) |
|
155 return pos_type(off_type(-1)); |
|
156 off_type size = static_cast<off_type>(this->egptr() - this->eback()); |
|
157 charT* g = this->eback(); |
|
158 if(off_type(sp) <= size) |
|
159 { |
|
160 this->setg(g, g + off_type(sp), g + size); |
|
161 } |
|
162 return pos_type(off_type(-1)); |
|
163 } |
|
164 |
|
165 // |
|
166 // class cpp_regex_traits_base: |
|
167 // acts as a container for locale and the facets we are using. |
|
168 // |
|
169 template <class charT> |
|
170 struct cpp_regex_traits_base |
|
171 { |
|
172 cpp_regex_traits_base(const std::locale& l) |
|
173 { imbue(l); } |
|
174 std::locale imbue(const std::locale& l); |
|
175 |
|
176 std::locale m_locale; |
|
177 std::ctype<charT> const* m_pctype; |
|
178 #ifndef BOOST_NO_STD_MESSAGES |
|
179 std::messages<charT> const* m_pmessages; |
|
180 #endif |
|
181 std::collate<charT> const* m_pcollate; |
|
182 |
|
183 bool operator<(const cpp_regex_traits_base& b)const |
|
184 { |
|
185 if(m_pctype == b.m_pctype) |
|
186 { |
|
187 #ifndef BOOST_NO_STD_MESSAGES |
|
188 if(m_pmessages == b.m_pmessages) |
|
189 { |
|
190 } |
|
191 return m_pmessages < b.m_pmessages; |
|
192 #else |
|
193 return m_pcollate < b.m_pcollate; |
|
194 #endif |
|
195 } |
|
196 return m_pctype < b.m_pctype; |
|
197 } |
|
198 bool operator==(const cpp_regex_traits_base& b)const |
|
199 { |
|
200 return (m_pctype == b.m_pctype) |
|
201 #ifndef BOOST_NO_STD_MESSAGES |
|
202 && (m_pmessages == b.m_pmessages) |
|
203 #endif |
|
204 && (m_pcollate == b.m_pcollate); |
|
205 } |
|
206 }; |
|
207 |
|
208 template <class charT> |
|
209 std::locale cpp_regex_traits_base<charT>::imbue(const std::locale& l) |
|
210 { |
|
211 std::locale result(m_locale); |
|
212 m_locale = l; |
|
213 m_pctype = &BOOST_USE_FACET(std::ctype<charT>, l); |
|
214 #ifndef BOOST_NO_STD_MESSAGES |
|
215 m_pmessages = &BOOST_USE_FACET(std::messages<charT>, l); |
|
216 #endif |
|
217 m_pcollate = &BOOST_USE_FACET(std::collate<charT>, l); |
|
218 return result; |
|
219 } |
|
220 |
|
221 // |
|
222 // class cpp_regex_traits_char_layer: |
|
223 // implements methods that require specialisation for narrow characters: |
|
224 // |
|
225 template <class charT> |
|
226 class cpp_regex_traits_char_layer : public cpp_regex_traits_base<charT> |
|
227 { |
|
228 typedef std::basic_string<charT> string_type; |
|
229 typedef std::map<charT, regex_constants::syntax_type> map_type; |
|
230 typedef typename map_type::const_iterator map_iterator_type; |
|
231 public: |
|
232 cpp_regex_traits_char_layer(const std::locale& l) |
|
233 : cpp_regex_traits_base<charT>(l) |
|
234 { |
|
235 init(); |
|
236 } |
|
237 cpp_regex_traits_char_layer(const cpp_regex_traits_base<charT>& b) |
|
238 : cpp_regex_traits_base<charT>(b) |
|
239 { |
|
240 init(); |
|
241 } |
|
242 void init(); |
|
243 |
|
244 regex_constants::syntax_type syntax_type(charT c)const |
|
245 { |
|
246 map_iterator_type i = m_char_map.find(c); |
|
247 return ((i == m_char_map.end()) ? 0 : i->second); |
|
248 } |
|
249 regex_constants::escape_syntax_type escape_syntax_type(charT c) const |
|
250 { |
|
251 map_iterator_type i = m_char_map.find(c); |
|
252 if(i == m_char_map.end()) |
|
253 { |
|
254 if(this->m_pctype->is(std::ctype_base::lower, c)) return regex_constants::escape_type_class; |
|
255 if(this->m_pctype->is(std::ctype_base::upper, c)) return regex_constants::escape_type_not_class; |
|
256 return 0; |
|
257 } |
|
258 return i->second; |
|
259 } |
|
260 |
|
261 private: |
|
262 string_type get_default_message(regex_constants::syntax_type); |
|
263 // TODO: use a hash table when available! |
|
264 map_type m_char_map; |
|
265 }; |
|
266 |
|
267 template <class charT> |
|
268 void cpp_regex_traits_char_layer<charT>::init() |
|
269 { |
|
270 // we need to start by initialising our syntax map so we know which |
|
271 // character is used for which purpose: |
|
272 #ifndef BOOST_NO_STD_MESSAGES |
|
273 #ifndef __IBMCPP__ |
|
274 typename std::messages<charT>::catalog cat = static_cast<std::messages<char>::catalog>(-1); |
|
275 #else |
|
276 typename std::messages<charT>::catalog cat = reinterpret_cast<std::messages<char>::catalog>(-1); |
|
277 #endif |
|
278 std::string cat_name(cpp_regex_traits<charT>::get_catalog_name()); |
|
279 if(cat_name.size()) |
|
280 { |
|
281 cat = this->m_pmessages->open( |
|
282 cat_name, |
|
283 this->m_locale); |
|
284 if((int)cat < 0) |
|
285 { |
|
286 std::string m("Unable to open message catalog: "); |
|
287 std::runtime_error err(m + cat_name); |
|
288 boost::re_detail::raise_runtime_error(err); |
|
289 } |
|
290 } |
|
291 // |
|
292 // if we have a valid catalog then load our messages: |
|
293 // |
|
294 if((int)cat >= 0) |
|
295 { |
|
296 #ifndef BOOST_NO_EXCEPTIONS |
|
297 try{ |
|
298 #endif |
|
299 for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i) |
|
300 { |
|
301 string_type mss = this->m_pmessages->get(cat, 0, i, get_default_message(i)); |
|
302 for(typename string_type::size_type j = 0; j < mss.size(); ++j) |
|
303 { |
|
304 m_char_map[mss[j]] = i; |
|
305 } |
|
306 } |
|
307 this->m_pmessages->close(cat); |
|
308 #ifndef BOOST_NO_EXCEPTIONS |
|
309 } |
|
310 catch(...) |
|
311 { |
|
312 this->m_pmessages->close(cat); |
|
313 throw; |
|
314 } |
|
315 #endif |
|
316 } |
|
317 else |
|
318 { |
|
319 #endif |
|
320 for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i) |
|
321 { |
|
322 const char* ptr = get_default_syntax(i); |
|
323 while(ptr && *ptr) |
|
324 { |
|
325 m_char_map[this->m_pctype->widen(*ptr)] = i; |
|
326 ++ptr; |
|
327 } |
|
328 } |
|
329 #ifndef BOOST_NO_STD_MESSAGES |
|
330 } |
|
331 #endif |
|
332 } |
|
333 |
|
334 template <class charT> |
|
335 typename cpp_regex_traits_char_layer<charT>::string_type |
|
336 cpp_regex_traits_char_layer<charT>::get_default_message(regex_constants::syntax_type i) |
|
337 { |
|
338 const char* ptr = get_default_syntax(i); |
|
339 string_type result; |
|
340 while(ptr && *ptr) |
|
341 { |
|
342 result.append(1, this->m_pctype->widen(*ptr)); |
|
343 ++ptr; |
|
344 } |
|
345 return result; |
|
346 } |
|
347 |
|
348 // |
|
349 // specialised version for narrow characters: |
|
350 // |
|
351 template <> |
|
352 class BOOST_REGEX_DECL cpp_regex_traits_char_layer<char> : public cpp_regex_traits_base<char> |
|
353 { |
|
354 typedef std::string string_type; |
|
355 public: |
|
356 cpp_regex_traits_char_layer(const std::locale& l) |
|
357 : cpp_regex_traits_base<char>(l) |
|
358 { |
|
359 init(); |
|
360 } |
|
361 cpp_regex_traits_char_layer(const cpp_regex_traits_base<char>& l) |
|
362 : cpp_regex_traits_base<char>(l) |
|
363 { |
|
364 init(); |
|
365 } |
|
366 |
|
367 regex_constants::syntax_type syntax_type(char c)const |
|
368 { |
|
369 return m_char_map[static_cast<unsigned char>(c)]; |
|
370 } |
|
371 regex_constants::escape_syntax_type escape_syntax_type(char c) const |
|
372 { |
|
373 return m_char_map[static_cast<unsigned char>(c)]; |
|
374 } |
|
375 |
|
376 private: |
|
377 regex_constants::syntax_type m_char_map[1u << CHAR_BIT]; |
|
378 void init(); |
|
379 }; |
|
380 |
|
381 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET |
|
382 enum |
|
383 { |
|
384 char_class_space=1<<0, |
|
385 char_class_print=1<<1, |
|
386 char_class_cntrl=1<<2, |
|
387 char_class_upper=1<<3, |
|
388 char_class_lower=1<<4, |
|
389 char_class_alpha=1<<5, |
|
390 char_class_digit=1<<6, |
|
391 char_class_punct=1<<7, |
|
392 char_class_xdigit=1<<8, |
|
393 char_class_alnum=char_class_alpha|char_class_digit, |
|
394 char_class_graph=char_class_alnum|char_class_punct, |
|
395 char_class_blank=1<<9, |
|
396 char_class_word=1<<10, |
|
397 char_class_unicode=1<<11 |
|
398 }; |
|
399 |
|
400 #endif |
|
401 |
|
402 // |
|
403 // class cpp_regex_traits_implementation: |
|
404 // provides pimpl implementation for cpp_regex_traits. |
|
405 // |
|
406 template <class charT> |
|
407 class cpp_regex_traits_implementation : public cpp_regex_traits_char_layer<charT> |
|
408 { |
|
409 public: |
|
410 typedef typename cpp_regex_traits<charT>::char_class_type char_class_type; |
|
411 typedef typename std::ctype<charT>::mask native_mask_type; |
|
412 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET |
|
413 BOOST_STATIC_CONSTANT(char_class_type, mask_blank = 1u << 24); |
|
414 BOOST_STATIC_CONSTANT(char_class_type, mask_word = 1u << 25); |
|
415 BOOST_STATIC_CONSTANT(char_class_type, mask_unicode = 1u << 26); |
|
416 #endif |
|
417 |
|
418 typedef std::basic_string<charT> string_type; |
|
419 typedef charT char_type; |
|
420 //cpp_regex_traits_implementation(); |
|
421 cpp_regex_traits_implementation(const std::locale& l) |
|
422 : cpp_regex_traits_char_layer<charT>(l) |
|
423 { |
|
424 init(); |
|
425 } |
|
426 cpp_regex_traits_implementation(const cpp_regex_traits_base<charT>& l) |
|
427 : cpp_regex_traits_char_layer<charT>(l) |
|
428 { |
|
429 init(); |
|
430 } |
|
431 std::string error_string(regex_constants::error_type n) const |
|
432 { |
|
433 if(!m_error_strings.empty()) |
|
434 { |
|
435 std::map<int, std::string>::const_iterator p = m_error_strings.find(n); |
|
436 return (p == m_error_strings.end()) ? std::string(get_default_error_string(n)) : p->second; |
|
437 } |
|
438 return get_default_error_string(n); |
|
439 } |
|
440 char_class_type lookup_classname(const charT* p1, const charT* p2) const |
|
441 { |
|
442 char_class_type result = lookup_classname_imp(p1, p2); |
|
443 if(result == 0) |
|
444 { |
|
445 string_type temp(p1, p2); |
|
446 this->m_pctype->tolower(&*temp.begin(), &*temp.begin() + temp.size()); |
|
447 result = lookup_classname_imp(&*temp.begin(), &*temp.begin() + temp.size()); |
|
448 } |
|
449 return result; |
|
450 } |
|
451 string_type lookup_collatename(const charT* p1, const charT* p2) const; |
|
452 string_type transform_primary(const charT* p1, const charT* p2) const; |
|
453 string_type transform(const charT* p1, const charT* p2) const; |
|
454 private: |
|
455 std::map<int, std::string> m_error_strings; // error messages indexed by numberic ID |
|
456 std::map<string_type, char_class_type> m_custom_class_names; // character class names |
|
457 std::map<string_type, string_type> m_custom_collate_names; // collating element names |
|
458 unsigned m_collate_type; // the form of the collation string |
|
459 charT m_collate_delim; // the collation group delimiter |
|
460 // |
|
461 // helpers: |
|
462 // |
|
463 char_class_type lookup_classname_imp(const charT* p1, const charT* p2) const; |
|
464 void init(); |
|
465 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET |
|
466 public: |
|
467 bool isctype(charT c, char_class_type m)const; |
|
468 #endif |
|
469 }; |
|
470 |
|
471 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET |
|
472 #if !defined(BOOST_NO_INCLASS_MEMBER_INITIALIZATION) |
|
473 |
|
474 template <class charT> |
|
475 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_blank; |
|
476 template <class charT> |
|
477 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_word; |
|
478 template <class charT> |
|
479 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_unicode; |
|
480 |
|
481 #endif |
|
482 #endif |
|
483 |
|
484 template <class charT> |
|
485 typename cpp_regex_traits_implementation<charT>::string_type |
|
486 cpp_regex_traits_implementation<charT>::transform_primary(const charT* p1, const charT* p2) const |
|
487 { |
|
488 // |
|
489 // PRECONDITIONS: |
|
490 // |
|
491 // A bug in gcc 3.2 (and maybe other versions as well) treats |
|
492 // p1 as a null terminated string, for efficiency reasons |
|
493 // we work around this elsewhere, but just assert here that |
|
494 // we adhere to gcc's (buggy) preconditions... |
|
495 // |
|
496 BOOST_ASSERT(*p2 == 0); |
|
497 |
|
498 string_type result; |
|
499 // |
|
500 // swallowing all exceptions here is a bad idea |
|
501 // however at least one std lib will always throw |
|
502 // std::bad_alloc for certain arguments... |
|
503 // |
|
504 try{ |
|
505 // |
|
506 // What we do here depends upon the format of the sort key returned by |
|
507 // sort key returned by this->transform: |
|
508 // |
|
509 switch(m_collate_type) |
|
510 { |
|
511 case sort_C: |
|
512 case sort_unknown: |
|
513 // the best we can do is translate to lower case, then get a regular sort key: |
|
514 { |
|
515 result.assign(p1, p2); |
|
516 this->m_pctype->tolower(&*result.begin(), &*result.begin() + result.size()); |
|
517 result = this->m_pcollate->transform(&*result.begin(), &*result.begin() + result.size()); |
|
518 break; |
|
519 } |
|
520 case sort_fixed: |
|
521 { |
|
522 // get a regular sort key, and then truncate it: |
|
523 result.assign(this->m_pcollate->transform(p1, p2)); |
|
524 result.erase(this->m_collate_delim); |
|
525 break; |
|
526 } |
|
527 case sort_delim: |
|
528 // get a regular sort key, and then truncate everything after the delim: |
|
529 result.assign(this->m_pcollate->transform(p1, p2)); |
|
530 std::size_t i; |
|
531 for(i = 0; i < result.size(); ++i) |
|
532 { |
|
533 if(result[i] == m_collate_delim) |
|
534 break; |
|
535 } |
|
536 result.erase(i); |
|
537 break; |
|
538 } |
|
539 }catch(...){} |
|
540 while(result.size() && (charT(0) == *result.rbegin())) |
|
541 result.erase(result.size() - 1); |
|
542 if(result.empty()) |
|
543 { |
|
544 // character is ignorable at the primary level: |
|
545 result = string_type(1, charT(0)); |
|
546 } |
|
547 return result; |
|
548 } |
|
549 |
|
550 template <class charT> |
|
551 typename cpp_regex_traits_implementation<charT>::string_type |
|
552 cpp_regex_traits_implementation<charT>::transform(const charT* p1, const charT* p2) const |
|
553 { |
|
554 // |
|
555 // PRECONDITIONS: |
|
556 // |
|
557 // A bug in gcc 3.2 (and maybe other versions as well) treats |
|
558 // p1 as a null terminated string, for efficiency reasons |
|
559 // we work around this elsewhere, but just assert here that |
|
560 // we adhere to gcc's (buggy) preconditions... |
|
561 // |
|
562 BOOST_ASSERT(*p2 == 0); |
|
563 // |
|
564 // swallowing all exceptions here is a bad idea |
|
565 // however at least one std lib will always throw |
|
566 // std::bad_alloc for certain arguments... |
|
567 // |
|
568 string_type result; |
|
569 try{ |
|
570 result = this->m_pcollate->transform(p1, p2); |
|
571 // |
|
572 // Borland's STLPort version returns a NULL-terminated |
|
573 // string that has garbage at the end - each call to |
|
574 // std::collate<wchar_t>::transform returns a different string! |
|
575 // So as a workaround, we'll truncate the string at the first NULL |
|
576 // which _seems_ to work.... |
|
577 #if BOOST_WORKAROUND(__BORLANDC__, < 0x580) |
|
578 result.erase(result.find(charT(0))); |
|
579 #else |
|
580 // |
|
581 // some implementations (Dinkumware) append unnecessary trailing \0's: |
|
582 while(result.size() && (charT(0) == *result.rbegin())) |
|
583 result.erase(result.size() - 1); |
|
584 #endif |
|
585 BOOST_ASSERT(std::find(result.begin(), result.end(), charT(0)) == result.end()); |
|
586 } |
|
587 catch(...) |
|
588 { |
|
589 } |
|
590 return result; |
|
591 } |
|
592 |
|
593 |
|
594 template <class charT> |
|
595 typename cpp_regex_traits_implementation<charT>::string_type |
|
596 cpp_regex_traits_implementation<charT>::lookup_collatename(const charT* p1, const charT* p2) const |
|
597 { |
|
598 typedef typename std::map<string_type, string_type>::const_iterator iter_type; |
|
599 if(m_custom_collate_names.size()) |
|
600 { |
|
601 iter_type pos = m_custom_collate_names.find(string_type(p1, p2)); |
|
602 if(pos != m_custom_collate_names.end()) |
|
603 return pos->second; |
|
604 } |
|
605 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\ |
|
606 && !BOOST_WORKAROUND(BOOST_MSVC, < 1300)\ |
|
607 && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551) |
|
608 std::string name(p1, p2); |
|
609 #else |
|
610 std::string name; |
|
611 const charT* p0 = p1; |
|
612 while(p0 != p2) |
|
613 name.append(1, char(*p0++)); |
|
614 #endif |
|
615 name = lookup_default_collate_name(name); |
|
616 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\ |
|
617 && !BOOST_WORKAROUND(BOOST_MSVC, < 1300)\ |
|
618 && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551) |
|
619 if(name.size()) |
|
620 return string_type(name.begin(), name.end()); |
|
621 #else |
|
622 if(name.size()) |
|
623 { |
|
624 string_type result; |
|
625 typedef std::string::const_iterator iter; |
|
626 iter b = name.begin(); |
|
627 iter e = name.end(); |
|
628 while(b != e) |
|
629 result.append(1, charT(*b++)); |
|
630 return result; |
|
631 } |
|
632 #endif |
|
633 if(p2 - p1 == 1) |
|
634 return string_type(1, *p1); |
|
635 return string_type(); |
|
636 } |
|
637 |
|
638 template <class charT> |
|
639 void cpp_regex_traits_implementation<charT>::init() |
|
640 { |
|
641 #ifndef BOOST_NO_STD_MESSAGES |
|
642 #ifndef __IBMCPP__ |
|
643 typename std::messages<charT>::catalog cat = static_cast<std::messages<char>::catalog>(-1); |
|
644 #else |
|
645 typename std::messages<charT>::catalog cat = reinterpret_cast<std::messages<char>::catalog>(-1); |
|
646 #endif |
|
647 std::string cat_name(cpp_regex_traits<charT>::get_catalog_name()); |
|
648 if(cat_name.size()) |
|
649 { |
|
650 cat = this->m_pmessages->open( |
|
651 cat_name, |
|
652 this->m_locale); |
|
653 if((int)cat < 0) |
|
654 { |
|
655 std::string m("Unable to open message catalog: "); |
|
656 std::runtime_error err(m + cat_name); |
|
657 boost::re_detail::raise_runtime_error(err); |
|
658 } |
|
659 } |
|
660 // |
|
661 // if we have a valid catalog then load our messages: |
|
662 // |
|
663 if((int)cat >= 0) |
|
664 { |
|
665 // |
|
666 // Error messages: |
|
667 // |
|
668 for(boost::regex_constants::error_type i = static_cast<boost::regex_constants::error_type>(0); |
|
669 i <= boost::regex_constants::error_unknown; |
|
670 i = static_cast<boost::regex_constants::error_type>(i + 1)) |
|
671 { |
|
672 const char* p = get_default_error_string(i); |
|
673 string_type default_message; |
|
674 while(*p) |
|
675 { |
|
676 default_message.append(1, this->m_pctype->widen(*p)); |
|
677 ++p; |
|
678 } |
|
679 string_type s = this->m_pmessages->get(cat, 0, i+200, default_message); |
|
680 std::string result; |
|
681 for(std::string::size_type j = 0; j < s.size(); ++j) |
|
682 { |
|
683 result.append(1, this->m_pctype->narrow(s[j], 0)); |
|
684 } |
|
685 m_error_strings[i] = result; |
|
686 } |
|
687 // |
|
688 // Custom class names: |
|
689 // |
|
690 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET |
|
691 static const char_class_type masks[14] = |
|
692 { |
|
693 std::ctype<charT>::alnum, |
|
694 std::ctype<charT>::alpha, |
|
695 std::ctype<charT>::cntrl, |
|
696 std::ctype<charT>::digit, |
|
697 std::ctype<charT>::graph, |
|
698 std::ctype<charT>::lower, |
|
699 std::ctype<charT>::print, |
|
700 std::ctype<charT>::punct, |
|
701 std::ctype<charT>::space, |
|
702 std::ctype<charT>::upper, |
|
703 std::ctype<charT>::xdigit, |
|
704 cpp_regex_traits_implementation<charT>::mask_blank, |
|
705 cpp_regex_traits_implementation<charT>::mask_word, |
|
706 cpp_regex_traits_implementation<charT>::mask_unicode, |
|
707 }; |
|
708 #else |
|
709 static const char_class_type masks[14] = |
|
710 { |
|
711 ::boost::re_detail::char_class_alnum, |
|
712 ::boost::re_detail::char_class_alpha, |
|
713 ::boost::re_detail::char_class_cntrl, |
|
714 ::boost::re_detail::char_class_digit, |
|
715 ::boost::re_detail::char_class_graph, |
|
716 ::boost::re_detail::char_class_lower, |
|
717 ::boost::re_detail::char_class_print, |
|
718 ::boost::re_detail::char_class_punct, |
|
719 ::boost::re_detail::char_class_space, |
|
720 ::boost::re_detail::char_class_upper, |
|
721 ::boost::re_detail::char_class_xdigit, |
|
722 ::boost::re_detail::char_class_blank, |
|
723 ::boost::re_detail::char_class_word, |
|
724 ::boost::re_detail::char_class_unicode, |
|
725 }; |
|
726 #endif |
|
727 static const string_type null_string; |
|
728 for(unsigned int j = 0; j <= 13; ++j) |
|
729 { |
|
730 string_type s(this->m_pmessages->get(cat, 0, j+300, null_string)); |
|
731 if(s.size()) |
|
732 this->m_custom_class_names[s] = masks[j]; |
|
733 } |
|
734 } |
|
735 #endif |
|
736 // |
|
737 // get the collation format used by m_pcollate: |
|
738 // |
|
739 m_collate_type = re_detail::find_sort_syntax(this, &m_collate_delim); |
|
740 } |
|
741 |
|
742 template <class charT> |
|
743 typename cpp_regex_traits_implementation<charT>::char_class_type |
|
744 cpp_regex_traits_implementation<charT>::lookup_classname_imp(const charT* p1, const charT* p2) const |
|
745 { |
|
746 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET |
|
747 static const char_class_type masks[20] = |
|
748 { |
|
749 0, |
|
750 std::ctype<char>::alnum, |
|
751 std::ctype<char>::alpha, |
|
752 cpp_regex_traits_implementation<charT>::mask_blank, |
|
753 std::ctype<char>::cntrl, |
|
754 std::ctype<char>::digit, |
|
755 std::ctype<char>::digit, |
|
756 std::ctype<char>::graph, |
|
757 std::ctype<char>::lower, |
|
758 std::ctype<char>::lower, |
|
759 std::ctype<char>::print, |
|
760 std::ctype<char>::punct, |
|
761 std::ctype<char>::space, |
|
762 std::ctype<char>::space, |
|
763 std::ctype<char>::upper, |
|
764 cpp_regex_traits_implementation<charT>::mask_unicode, |
|
765 std::ctype<char>::upper, |
|
766 std::ctype<char>::alnum | cpp_regex_traits_implementation<charT>::mask_word, |
|
767 std::ctype<char>::alnum | cpp_regex_traits_implementation<charT>::mask_word, |
|
768 std::ctype<char>::xdigit, |
|
769 }; |
|
770 #else |
|
771 static const char_class_type masks[20] = |
|
772 { |
|
773 0, |
|
774 ::boost::re_detail::char_class_alnum, |
|
775 ::boost::re_detail::char_class_alpha, |
|
776 ::boost::re_detail::char_class_blank, |
|
777 ::boost::re_detail::char_class_cntrl, |
|
778 ::boost::re_detail::char_class_digit, |
|
779 ::boost::re_detail::char_class_digit, |
|
780 ::boost::re_detail::char_class_graph, |
|
781 ::boost::re_detail::char_class_lower, |
|
782 ::boost::re_detail::char_class_lower, |
|
783 ::boost::re_detail::char_class_print, |
|
784 ::boost::re_detail::char_class_punct, |
|
785 ::boost::re_detail::char_class_space, |
|
786 ::boost::re_detail::char_class_space, |
|
787 ::boost::re_detail::char_class_upper, |
|
788 ::boost::re_detail::char_class_unicode, |
|
789 ::boost::re_detail::char_class_upper, |
|
790 ::boost::re_detail::char_class_alnum | ::boost::re_detail::char_class_word, |
|
791 ::boost::re_detail::char_class_alnum | ::boost::re_detail::char_class_word, |
|
792 ::boost::re_detail::char_class_xdigit, |
|
793 }; |
|
794 #endif |
|
795 if(m_custom_class_names.size()) |
|
796 { |
|
797 typedef typename std::map<std::basic_string<charT>, char_class_type>::const_iterator map_iter; |
|
798 map_iter pos = m_custom_class_names.find(string_type(p1, p2)); |
|
799 if(pos != m_custom_class_names.end()) |
|
800 return pos->second; |
|
801 } |
|
802 std::size_t state_id = 1 + re_detail::get_default_class_id(p1, p2); |
|
803 BOOST_ASSERT(state_id < sizeof(masks) / sizeof(masks[0])); |
|
804 return masks[state_id]; |
|
805 } |
|
806 |
|
807 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET |
|
808 template <class charT> |
|
809 bool cpp_regex_traits_implementation<charT>::isctype(const charT c, char_class_type mask) const |
|
810 { |
|
811 return |
|
812 ((mask & ::boost::re_detail::char_class_space) && (m_pctype->is(std::ctype<charT>::space, c))) |
|
813 || ((mask & ::boost::re_detail::char_class_print) && (m_pctype->is(std::ctype<charT>::print, c))) |
|
814 || ((mask & ::boost::re_detail::char_class_cntrl) && (m_pctype->is(std::ctype<charT>::cntrl, c))) |
|
815 || ((mask & ::boost::re_detail::char_class_upper) && (m_pctype->is(std::ctype<charT>::upper, c))) |
|
816 || ((mask & ::boost::re_detail::char_class_lower) && (m_pctype->is(std::ctype<charT>::lower, c))) |
|
817 || ((mask & ::boost::re_detail::char_class_alpha) && (m_pctype->is(std::ctype<charT>::alpha, c))) |
|
818 || ((mask & ::boost::re_detail::char_class_digit) && (m_pctype->is(std::ctype<charT>::digit, c))) |
|
819 || ((mask & ::boost::re_detail::char_class_punct) && (m_pctype->is(std::ctype<charT>::punct, c))) |
|
820 || ((mask & ::boost::re_detail::char_class_xdigit) && (m_pctype->is(std::ctype<charT>::xdigit, c))) |
|
821 || ((mask & ::boost::re_detail::char_class_blank) && (m_pctype->is(std::ctype<charT>::space, c)) && !::boost::re_detail::is_separator(c)) |
|
822 || ((mask & ::boost::re_detail::char_class_word) && (c == '_')) |
|
823 || ((mask & ::boost::re_detail::char_class_unicode) && ::boost::re_detail::is_extended(c)); |
|
824 } |
|
825 #endif |
|
826 |
|
827 |
|
828 template <class charT> |
|
829 inline boost::shared_ptr<const cpp_regex_traits_implementation<charT> > create_cpp_regex_traits(const std::locale& l BOOST_APPEND_EXPLICIT_TEMPLATE_TYPE(charT)) |
|
830 { |
|
831 cpp_regex_traits_base<charT> key(l); |
|
832 return ::boost::object_cache<cpp_regex_traits_base<charT>, cpp_regex_traits_implementation<charT> >::get(key, 5); |
|
833 } |
|
834 |
|
835 } // re_detail |
|
836 |
|
837 template <class charT> |
|
838 class cpp_regex_traits |
|
839 { |
|
840 private: |
|
841 typedef std::ctype<charT> ctype_type; |
|
842 public: |
|
843 typedef charT char_type; |
|
844 typedef std::size_t size_type; |
|
845 typedef std::basic_string<char_type> string_type; |
|
846 typedef std::locale locale_type; |
|
847 typedef boost::uint_least32_t char_class_type; |
|
848 |
|
849 struct boost_extensions_tag{}; |
|
850 |
|
851 cpp_regex_traits() |
|
852 : m_pimpl(re_detail::create_cpp_regex_traits<charT>(std::locale())) |
|
853 { } |
|
854 static size_type length(const char_type* p) |
|
855 { |
|
856 return std::char_traits<charT>::length(p); |
|
857 } |
|
858 regex_constants::syntax_type syntax_type(charT c)const |
|
859 { |
|
860 return m_pimpl->syntax_type(c); |
|
861 } |
|
862 regex_constants::escape_syntax_type escape_syntax_type(charT c) const |
|
863 { |
|
864 return m_pimpl->escape_syntax_type(c); |
|
865 } |
|
866 charT translate(charT c) const |
|
867 { |
|
868 return c; |
|
869 } |
|
870 charT translate_nocase(charT c) const |
|
871 { |
|
872 return m_pimpl->m_pctype->tolower(c); |
|
873 } |
|
874 charT translate(charT c, bool icase) const |
|
875 { |
|
876 return icase ? m_pimpl->m_pctype->tolower(c) : c; |
|
877 } |
|
878 charT tolower(charT c) const |
|
879 { |
|
880 return m_pimpl->m_pctype->tolower(c); |
|
881 } |
|
882 charT toupper(charT c) const |
|
883 { |
|
884 return m_pimpl->m_pctype->toupper(c); |
|
885 } |
|
886 string_type transform(const charT* p1, const charT* p2) const |
|
887 { |
|
888 return m_pimpl->transform(p1, p2); |
|
889 } |
|
890 string_type transform_primary(const charT* p1, const charT* p2) const |
|
891 { |
|
892 return m_pimpl->transform_primary(p1, p2); |
|
893 } |
|
894 char_class_type lookup_classname(const charT* p1, const charT* p2) const |
|
895 { |
|
896 return m_pimpl->lookup_classname(p1, p2); |
|
897 } |
|
898 string_type lookup_collatename(const charT* p1, const charT* p2) const |
|
899 { |
|
900 return m_pimpl->lookup_collatename(p1, p2); |
|
901 } |
|
902 bool isctype(charT c, char_class_type f) const |
|
903 { |
|
904 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET |
|
905 typedef typename std::ctype<charT>::mask ctype_mask; |
|
906 |
|
907 static const ctype_mask mask_base = |
|
908 static_cast<ctype_mask>( |
|
909 std::ctype<charT>::alnum |
|
910 | std::ctype<charT>::alpha |
|
911 | std::ctype<charT>::cntrl |
|
912 | std::ctype<charT>::digit |
|
913 | std::ctype<charT>::graph |
|
914 | std::ctype<charT>::lower |
|
915 | std::ctype<charT>::print |
|
916 | std::ctype<charT>::punct |
|
917 | std::ctype<charT>::space |
|
918 | std::ctype<charT>::upper |
|
919 | std::ctype<charT>::xdigit); |
|
920 |
|
921 if((f & mask_base) |
|
922 && (m_pimpl->m_pctype->is( |
|
923 static_cast<ctype_mask>(f & mask_base), c))) |
|
924 return true; |
|
925 else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_unicode) && re_detail::is_extended(c)) |
|
926 return true; |
|
927 else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_word) && (c == '_')) |
|
928 return true; |
|
929 else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_blank) |
|
930 && m_pimpl->m_pctype->is(std::ctype<charT>::space, c) |
|
931 && !re_detail::is_separator(c)) |
|
932 return true; |
|
933 return false; |
|
934 #else |
|
935 return m_pimpl->isctype(c, f); |
|
936 #endif |
|
937 } |
|
938 int toi(const charT*& p1, const charT* p2, int radix)const; |
|
939 int value(charT c, int radix)const |
|
940 { |
|
941 const charT* pc = &c; |
|
942 return toi(pc, pc + 1, radix); |
|
943 } |
|
944 locale_type imbue(locale_type l) |
|
945 { |
|
946 std::locale result(getloc()); |
|
947 m_pimpl = re_detail::create_cpp_regex_traits<charT>(l); |
|
948 return result; |
|
949 } |
|
950 locale_type getloc()const |
|
951 { |
|
952 return m_pimpl->m_locale; |
|
953 } |
|
954 std::string error_string(regex_constants::error_type n) const |
|
955 { |
|
956 return m_pimpl->error_string(n); |
|
957 } |
|
958 |
|
959 // |
|
960 // extension: |
|
961 // set the name of the message catalog in use (defaults to "boost_regex"). |
|
962 // |
|
963 static std::string catalog_name(const std::string& name); |
|
964 static std::string get_catalog_name(); |
|
965 |
|
966 private: |
|
967 boost::shared_ptr<const re_detail::cpp_regex_traits_implementation<charT> > m_pimpl; |
|
968 // |
|
969 // catalog name handler: |
|
970 // |
|
971 static std::string& get_catalog_name_inst(); |
|
972 |
|
973 #ifdef BOOST_HAS_THREADS |
|
974 static static_mutex& get_mutex_inst(); |
|
975 #endif |
|
976 }; |
|
977 |
|
978 |
|
979 template <class charT> |
|
980 int cpp_regex_traits<charT>::toi(const charT*& first, const charT* last, int radix)const |
|
981 { |
|
982 re_detail::parser_buf<charT> sbuf; // buffer for parsing numbers. |
|
983 std::basic_istream<charT> is(&sbuf); // stream for parsing numbers. |
|
984 |
|
985 // we do NOT want to parse any thousands separators inside the stream: |
|
986 last = std::find(first, last, BOOST_USE_FACET(std::numpunct<charT>, is.getloc()).thousands_sep()); |
|
987 |
|
988 sbuf.pubsetbuf(const_cast<charT*>(static_cast<const charT*>(first)), static_cast<std::streamsize>(last-first)); |
|
989 is.clear(); |
|
990 if(std::abs(radix) == 16) is >> std::hex; |
|
991 else if(std::abs(radix) == 8) is >> std::oct; |
|
992 else is >> std::dec; |
|
993 int val; |
|
994 if(is >> val) |
|
995 { |
|
996 first = first + ((last - first) - sbuf.in_avail()); |
|
997 return val; |
|
998 } |
|
999 else |
|
1000 return -1; |
|
1001 } |
|
1002 |
|
1003 template <class charT> |
|
1004 std::string cpp_regex_traits<charT>::catalog_name(const std::string& name) |
|
1005 { |
|
1006 #ifdef BOOST_HAS_THREADS |
|
1007 static_mutex::scoped_lock lk(get_mutex_inst()); |
|
1008 #endif |
|
1009 std::string result(get_catalog_name_inst()); |
|
1010 get_catalog_name_inst() = name; |
|
1011 return result; |
|
1012 } |
|
1013 |
|
1014 template <class charT> |
|
1015 std::string& cpp_regex_traits<charT>::get_catalog_name_inst() |
|
1016 { |
|
1017 static std::string s_name; |
|
1018 return s_name; |
|
1019 } |
|
1020 |
|
1021 template <class charT> |
|
1022 std::string cpp_regex_traits<charT>::get_catalog_name() |
|
1023 { |
|
1024 #ifdef BOOST_HAS_THREADS |
|
1025 static_mutex::scoped_lock lk(get_mutex_inst()); |
|
1026 #endif |
|
1027 std::string result(get_catalog_name_inst()); |
|
1028 return result; |
|
1029 } |
|
1030 |
|
1031 #ifdef BOOST_HAS_THREADS |
|
1032 template <class charT> |
|
1033 static_mutex& cpp_regex_traits<charT>::get_mutex_inst() |
|
1034 { |
|
1035 static static_mutex s_mutex = BOOST_STATIC_MUTEX_INIT; |
|
1036 return s_mutex; |
|
1037 } |
|
1038 #endif |
|
1039 |
|
1040 |
|
1041 } // boost |
|
1042 |
|
1043 #ifdef BOOST_MSVC |
|
1044 #pragma warning(pop) |
|
1045 #endif |
|
1046 |
|
1047 #ifdef BOOST_MSVC |
|
1048 #pragma warning(push) |
|
1049 #pragma warning(disable: 4103) |
|
1050 #endif |
|
1051 #ifdef BOOST_HAS_ABI_HEADERS |
|
1052 # include BOOST_ABI_SUFFIX |
|
1053 #endif |
|
1054 #ifdef BOOST_MSVC |
|
1055 #pragma warning(pop) |
|
1056 #endif |
|
1057 |
|
1058 #endif |
|
1059 |
|
1060 #endif |
|
1061 |
|
1062 |