Chris@16
|
1 /*
|
Chris@16
|
2 *
|
Chris@16
|
3 * Copyright (c) 2004 John Maddock
|
Chris@16
|
4 * Copyright 2011 Garmin Ltd. or its subsidiaries
|
Chris@16
|
5 *
|
Chris@16
|
6 * Use, modification and distribution are subject to the
|
Chris@16
|
7 * Boost Software License, Version 1.0. (See accompanying file
|
Chris@16
|
8 * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
|
Chris@16
|
9 *
|
Chris@16
|
10 */
|
Chris@16
|
11
|
Chris@16
|
12 /*
|
Chris@16
|
13 * LOCATION: see http://www.boost.org for most recent version.
|
Chris@16
|
14 * FILE cpp_regex_traits.hpp
|
Chris@16
|
15 * VERSION see <boost/version.hpp>
|
Chris@16
|
16 * DESCRIPTION: Declares regular expression traits class cpp_regex_traits.
|
Chris@16
|
17 */
|
Chris@16
|
18
|
Chris@16
|
19 #ifndef BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED
|
Chris@16
|
20 #define BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED
|
Chris@16
|
21
|
Chris@16
|
22 #include <boost/config.hpp>
|
Chris@16
|
23 #include <boost/integer.hpp>
|
Chris@101
|
24 #include <boost/type_traits/make_unsigned.hpp>
|
Chris@16
|
25
|
Chris@16
|
26 #ifndef BOOST_NO_STD_LOCALE
|
Chris@16
|
27
|
Chris@16
|
28 #ifndef BOOST_RE_PAT_EXCEPT_HPP
|
Chris@16
|
29 #include <boost/regex/pattern_except.hpp>
|
Chris@16
|
30 #endif
|
Chris@16
|
31 #ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED
|
Chris@16
|
32 #include <boost/regex/v4/regex_traits_defaults.hpp>
|
Chris@16
|
33 #endif
|
Chris@16
|
34 #ifdef BOOST_HAS_THREADS
|
Chris@16
|
35 #include <boost/regex/pending/static_mutex.hpp>
|
Chris@16
|
36 #endif
|
Chris@16
|
37 #ifndef BOOST_REGEX_PRIMARY_TRANSFORM
|
Chris@16
|
38 #include <boost/regex/v4/primary_transform.hpp>
|
Chris@16
|
39 #endif
|
Chris@16
|
40 #ifndef BOOST_REGEX_OBJECT_CACHE_HPP
|
Chris@16
|
41 #include <boost/regex/pending/object_cache.hpp>
|
Chris@16
|
42 #endif
|
Chris@16
|
43
|
Chris@16
|
44 #include <istream>
|
Chris@16
|
45 #include <ios>
|
Chris@16
|
46 #include <climits>
|
Chris@16
|
47
|
Chris@16
|
48 #ifdef BOOST_MSVC
|
Chris@16
|
49 #pragma warning(push)
|
Chris@16
|
50 #pragma warning(disable: 4103)
|
Chris@16
|
51 #endif
|
Chris@16
|
52 #ifdef BOOST_HAS_ABI_HEADERS
|
Chris@16
|
53 # include BOOST_ABI_PREFIX
|
Chris@16
|
54 #endif
|
Chris@16
|
55 #ifdef BOOST_MSVC
|
Chris@16
|
56 #pragma warning(pop)
|
Chris@16
|
57 #endif
|
Chris@16
|
58
|
Chris@16
|
59 #ifdef BOOST_MSVC
|
Chris@16
|
60 #pragma warning(push)
|
Chris@16
|
61 #pragma warning(disable:4786 4251)
|
Chris@16
|
62 #endif
|
Chris@16
|
63
|
Chris@16
|
64 namespace boost{
|
Chris@16
|
65
|
Chris@16
|
66 //
|
Chris@16
|
67 // forward declaration is needed by some compilers:
|
Chris@16
|
68 //
|
Chris@16
|
69 template <class charT>
|
Chris@16
|
70 class cpp_regex_traits;
|
Chris@16
|
71
|
Chris@16
|
72 namespace re_detail{
|
Chris@16
|
73
|
Chris@16
|
74 //
|
Chris@16
|
75 // class parser_buf:
|
Chris@16
|
76 // acts as a stream buffer which wraps around a pair of pointers:
|
Chris@16
|
77 //
|
Chris@16
|
78 template <class charT,
|
Chris@16
|
79 class traits = ::std::char_traits<charT> >
|
Chris@16
|
80 class parser_buf : public ::std::basic_streambuf<charT, traits>
|
Chris@16
|
81 {
|
Chris@16
|
82 typedef ::std::basic_streambuf<charT, traits> base_type;
|
Chris@16
|
83 typedef typename base_type::int_type int_type;
|
Chris@16
|
84 typedef typename base_type::char_type char_type;
|
Chris@16
|
85 typedef typename base_type::pos_type pos_type;
|
Chris@16
|
86 typedef ::std::streamsize streamsize;
|
Chris@16
|
87 typedef typename base_type::off_type off_type;
|
Chris@16
|
88 public:
|
Chris@16
|
89 parser_buf() : base_type() { setbuf(0, 0); }
|
Chris@16
|
90 const charT* getnext() { return this->gptr(); }
|
Chris@16
|
91 protected:
|
Chris@16
|
92 std::basic_streambuf<charT, traits>* setbuf(char_type* s, streamsize n);
|
Chris@16
|
93 typename parser_buf<charT, traits>::pos_type seekpos(pos_type sp, ::std::ios_base::openmode which);
|
Chris@16
|
94 typename parser_buf<charT, traits>::pos_type seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which);
|
Chris@16
|
95 private:
|
Chris@16
|
96 parser_buf& operator=(const parser_buf&);
|
Chris@16
|
97 parser_buf(const parser_buf&);
|
Chris@16
|
98 };
|
Chris@16
|
99
|
Chris@16
|
100 template<class charT, class traits>
|
Chris@16
|
101 std::basic_streambuf<charT, traits>*
|
Chris@16
|
102 parser_buf<charT, traits>::setbuf(char_type* s, streamsize n)
|
Chris@16
|
103 {
|
Chris@16
|
104 this->setg(s, s, s + n);
|
Chris@16
|
105 return this;
|
Chris@16
|
106 }
|
Chris@16
|
107
|
Chris@16
|
108 template<class charT, class traits>
|
Chris@16
|
109 typename parser_buf<charT, traits>::pos_type
|
Chris@16
|
110 parser_buf<charT, traits>::seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which)
|
Chris@16
|
111 {
|
Chris@16
|
112 typedef typename boost::int_t<sizeof(way) * CHAR_BIT>::least cast_type;
|
Chris@16
|
113
|
Chris@16
|
114 if(which & ::std::ios_base::out)
|
Chris@16
|
115 return pos_type(off_type(-1));
|
Chris@16
|
116 std::ptrdiff_t size = this->egptr() - this->eback();
|
Chris@16
|
117 std::ptrdiff_t pos = this->gptr() - this->eback();
|
Chris@16
|
118 charT* g = this->eback();
|
Chris@16
|
119 switch(static_cast<cast_type>(way))
|
Chris@16
|
120 {
|
Chris@16
|
121 case ::std::ios_base::beg:
|
Chris@16
|
122 if((off < 0) || (off > size))
|
Chris@16
|
123 return pos_type(off_type(-1));
|
Chris@16
|
124 else
|
Chris@16
|
125 this->setg(g, g + off, g + size);
|
Chris@16
|
126 break;
|
Chris@16
|
127 case ::std::ios_base::end:
|
Chris@16
|
128 if((off < 0) || (off > size))
|
Chris@16
|
129 return pos_type(off_type(-1));
|
Chris@16
|
130 else
|
Chris@16
|
131 this->setg(g, g + size - off, g + size);
|
Chris@16
|
132 break;
|
Chris@16
|
133 case ::std::ios_base::cur:
|
Chris@16
|
134 {
|
Chris@16
|
135 std::ptrdiff_t newpos = static_cast<std::ptrdiff_t>(pos + off);
|
Chris@16
|
136 if((newpos < 0) || (newpos > size))
|
Chris@16
|
137 return pos_type(off_type(-1));
|
Chris@16
|
138 else
|
Chris@16
|
139 this->setg(g, g + newpos, g + size);
|
Chris@16
|
140 break;
|
Chris@16
|
141 }
|
Chris@16
|
142 default: ;
|
Chris@16
|
143 }
|
Chris@16
|
144 #ifdef BOOST_MSVC
|
Chris@16
|
145 #pragma warning(push)
|
Chris@16
|
146 #pragma warning(disable:4244)
|
Chris@16
|
147 #endif
|
Chris@16
|
148 return static_cast<pos_type>(this->gptr() - this->eback());
|
Chris@16
|
149 #ifdef BOOST_MSVC
|
Chris@16
|
150 #pragma warning(pop)
|
Chris@16
|
151 #endif
|
Chris@16
|
152 }
|
Chris@16
|
153
|
Chris@16
|
154 template<class charT, class traits>
|
Chris@16
|
155 typename parser_buf<charT, traits>::pos_type
|
Chris@16
|
156 parser_buf<charT, traits>::seekpos(pos_type sp, ::std::ios_base::openmode which)
|
Chris@16
|
157 {
|
Chris@16
|
158 if(which & ::std::ios_base::out)
|
Chris@16
|
159 return pos_type(off_type(-1));
|
Chris@16
|
160 off_type size = static_cast<off_type>(this->egptr() - this->eback());
|
Chris@16
|
161 charT* g = this->eback();
|
Chris@16
|
162 if(off_type(sp) <= size)
|
Chris@16
|
163 {
|
Chris@16
|
164 this->setg(g, g + off_type(sp), g + size);
|
Chris@16
|
165 }
|
Chris@16
|
166 return pos_type(off_type(-1));
|
Chris@16
|
167 }
|
Chris@16
|
168
|
Chris@16
|
169 //
|
Chris@16
|
170 // class cpp_regex_traits_base:
|
Chris@16
|
171 // acts as a container for locale and the facets we are using.
|
Chris@16
|
172 //
|
Chris@16
|
173 template <class charT>
|
Chris@16
|
174 struct cpp_regex_traits_base
|
Chris@16
|
175 {
|
Chris@16
|
176 cpp_regex_traits_base(const std::locale& l)
|
Chris@16
|
177 { imbue(l); }
|
Chris@16
|
178 std::locale imbue(const std::locale& l);
|
Chris@16
|
179
|
Chris@16
|
180 std::locale m_locale;
|
Chris@16
|
181 std::ctype<charT> const* m_pctype;
|
Chris@16
|
182 #ifndef BOOST_NO_STD_MESSAGES
|
Chris@16
|
183 std::messages<charT> const* m_pmessages;
|
Chris@16
|
184 #endif
|
Chris@16
|
185 std::collate<charT> const* m_pcollate;
|
Chris@16
|
186
|
Chris@16
|
187 bool operator<(const cpp_regex_traits_base& b)const
|
Chris@16
|
188 {
|
Chris@16
|
189 if(m_pctype == b.m_pctype)
|
Chris@16
|
190 {
|
Chris@16
|
191 #ifndef BOOST_NO_STD_MESSAGES
|
Chris@16
|
192 if(m_pmessages == b.m_pmessages)
|
Chris@16
|
193 {
|
Chris@16
|
194 return m_pcollate < b.m_pcollate;
|
Chris@16
|
195 }
|
Chris@16
|
196 return m_pmessages < b.m_pmessages;
|
Chris@16
|
197 #else
|
Chris@16
|
198 return m_pcollate < b.m_pcollate;
|
Chris@16
|
199 #endif
|
Chris@16
|
200 }
|
Chris@16
|
201 return m_pctype < b.m_pctype;
|
Chris@16
|
202 }
|
Chris@16
|
203 bool operator==(const cpp_regex_traits_base& b)const
|
Chris@16
|
204 {
|
Chris@16
|
205 return (m_pctype == b.m_pctype)
|
Chris@16
|
206 #ifndef BOOST_NO_STD_MESSAGES
|
Chris@16
|
207 && (m_pmessages == b.m_pmessages)
|
Chris@16
|
208 #endif
|
Chris@16
|
209 && (m_pcollate == b.m_pcollate);
|
Chris@16
|
210 }
|
Chris@16
|
211 };
|
Chris@16
|
212
|
Chris@16
|
213 template <class charT>
|
Chris@16
|
214 std::locale cpp_regex_traits_base<charT>::imbue(const std::locale& l)
|
Chris@16
|
215 {
|
Chris@16
|
216 std::locale result(m_locale);
|
Chris@16
|
217 m_locale = l;
|
Chris@16
|
218 m_pctype = &BOOST_USE_FACET(std::ctype<charT>, l);
|
Chris@16
|
219 #ifndef BOOST_NO_STD_MESSAGES
|
Chris@16
|
220 m_pmessages = BOOST_HAS_FACET(std::messages<charT>, l) ? &BOOST_USE_FACET(std::messages<charT>, l) : 0;
|
Chris@16
|
221 #endif
|
Chris@16
|
222 m_pcollate = &BOOST_USE_FACET(std::collate<charT>, l);
|
Chris@16
|
223 return result;
|
Chris@16
|
224 }
|
Chris@16
|
225
|
Chris@16
|
226 //
|
Chris@16
|
227 // class cpp_regex_traits_char_layer:
|
Chris@16
|
228 // implements methods that require specialisation for narrow characters:
|
Chris@16
|
229 //
|
Chris@16
|
230 template <class charT>
|
Chris@16
|
231 class cpp_regex_traits_char_layer : public cpp_regex_traits_base<charT>
|
Chris@16
|
232 {
|
Chris@16
|
233 typedef std::basic_string<charT> string_type;
|
Chris@16
|
234 typedef std::map<charT, regex_constants::syntax_type> map_type;
|
Chris@16
|
235 typedef typename map_type::const_iterator map_iterator_type;
|
Chris@16
|
236 public:
|
Chris@16
|
237 cpp_regex_traits_char_layer(const std::locale& l)
|
Chris@16
|
238 : cpp_regex_traits_base<charT>(l)
|
Chris@16
|
239 {
|
Chris@16
|
240 init();
|
Chris@16
|
241 }
|
Chris@16
|
242 cpp_regex_traits_char_layer(const cpp_regex_traits_base<charT>& b)
|
Chris@16
|
243 : cpp_regex_traits_base<charT>(b)
|
Chris@16
|
244 {
|
Chris@16
|
245 init();
|
Chris@16
|
246 }
|
Chris@16
|
247 void init();
|
Chris@16
|
248
|
Chris@16
|
249 regex_constants::syntax_type syntax_type(charT c)const
|
Chris@16
|
250 {
|
Chris@16
|
251 map_iterator_type i = m_char_map.find(c);
|
Chris@16
|
252 return ((i == m_char_map.end()) ? 0 : i->second);
|
Chris@16
|
253 }
|
Chris@16
|
254 regex_constants::escape_syntax_type escape_syntax_type(charT c) const
|
Chris@16
|
255 {
|
Chris@16
|
256 map_iterator_type i = m_char_map.find(c);
|
Chris@16
|
257 if(i == m_char_map.end())
|
Chris@16
|
258 {
|
Chris@16
|
259 if(this->m_pctype->is(std::ctype_base::lower, c)) return regex_constants::escape_type_class;
|
Chris@16
|
260 if(this->m_pctype->is(std::ctype_base::upper, c)) return regex_constants::escape_type_not_class;
|
Chris@16
|
261 return 0;
|
Chris@16
|
262 }
|
Chris@16
|
263 return i->second;
|
Chris@16
|
264 }
|
Chris@16
|
265
|
Chris@16
|
266 private:
|
Chris@16
|
267 string_type get_default_message(regex_constants::syntax_type);
|
Chris@16
|
268 // TODO: use a hash table when available!
|
Chris@16
|
269 map_type m_char_map;
|
Chris@16
|
270 };
|
Chris@16
|
271
|
Chris@16
|
272 template <class charT>
|
Chris@16
|
273 void cpp_regex_traits_char_layer<charT>::init()
|
Chris@16
|
274 {
|
Chris@16
|
275 // we need to start by initialising our syntax map so we know which
|
Chris@16
|
276 // character is used for which purpose:
|
Chris@16
|
277 #ifndef BOOST_NO_STD_MESSAGES
|
Chris@16
|
278 #ifndef __IBMCPP__
|
Chris@16
|
279 typename std::messages<charT>::catalog cat = static_cast<std::messages<char>::catalog>(-1);
|
Chris@16
|
280 #else
|
Chris@16
|
281 typename std::messages<charT>::catalog cat = reinterpret_cast<std::messages<char>::catalog>(-1);
|
Chris@16
|
282 #endif
|
Chris@16
|
283 std::string cat_name(cpp_regex_traits<charT>::get_catalog_name());
|
Chris@16
|
284 if(cat_name.size() && (this->m_pmessages != 0))
|
Chris@16
|
285 {
|
Chris@16
|
286 cat = this->m_pmessages->open(
|
Chris@16
|
287 cat_name,
|
Chris@16
|
288 this->m_locale);
|
Chris@16
|
289 if((int)cat < 0)
|
Chris@16
|
290 {
|
Chris@16
|
291 std::string m("Unable to open message catalog: ");
|
Chris@16
|
292 std::runtime_error err(m + cat_name);
|
Chris@16
|
293 boost::re_detail::raise_runtime_error(err);
|
Chris@16
|
294 }
|
Chris@16
|
295 }
|
Chris@16
|
296 //
|
Chris@16
|
297 // if we have a valid catalog then load our messages:
|
Chris@16
|
298 //
|
Chris@16
|
299 if((int)cat >= 0)
|
Chris@16
|
300 {
|
Chris@16
|
301 #ifndef BOOST_NO_EXCEPTIONS
|
Chris@16
|
302 try{
|
Chris@16
|
303 #endif
|
Chris@16
|
304 for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
|
Chris@16
|
305 {
|
Chris@16
|
306 string_type mss = this->m_pmessages->get(cat, 0, i, get_default_message(i));
|
Chris@16
|
307 for(typename string_type::size_type j = 0; j < mss.size(); ++j)
|
Chris@16
|
308 {
|
Chris@16
|
309 m_char_map[mss[j]] = i;
|
Chris@16
|
310 }
|
Chris@16
|
311 }
|
Chris@16
|
312 this->m_pmessages->close(cat);
|
Chris@16
|
313 #ifndef BOOST_NO_EXCEPTIONS
|
Chris@16
|
314 }
|
Chris@16
|
315 catch(...)
|
Chris@16
|
316 {
|
Chris@16
|
317 if(this->m_pmessages)
|
Chris@16
|
318 this->m_pmessages->close(cat);
|
Chris@16
|
319 throw;
|
Chris@16
|
320 }
|
Chris@16
|
321 #endif
|
Chris@16
|
322 }
|
Chris@16
|
323 else
|
Chris@16
|
324 {
|
Chris@16
|
325 #endif
|
Chris@16
|
326 for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
|
Chris@16
|
327 {
|
Chris@16
|
328 const char* ptr = get_default_syntax(i);
|
Chris@16
|
329 while(ptr && *ptr)
|
Chris@16
|
330 {
|
Chris@16
|
331 m_char_map[this->m_pctype->widen(*ptr)] = i;
|
Chris@16
|
332 ++ptr;
|
Chris@16
|
333 }
|
Chris@16
|
334 }
|
Chris@16
|
335 #ifndef BOOST_NO_STD_MESSAGES
|
Chris@16
|
336 }
|
Chris@16
|
337 #endif
|
Chris@16
|
338 }
|
Chris@16
|
339
|
Chris@16
|
340 template <class charT>
|
Chris@16
|
341 typename cpp_regex_traits_char_layer<charT>::string_type
|
Chris@16
|
342 cpp_regex_traits_char_layer<charT>::get_default_message(regex_constants::syntax_type i)
|
Chris@16
|
343 {
|
Chris@16
|
344 const char* ptr = get_default_syntax(i);
|
Chris@16
|
345 string_type result;
|
Chris@16
|
346 while(ptr && *ptr)
|
Chris@16
|
347 {
|
Chris@16
|
348 result.append(1, this->m_pctype->widen(*ptr));
|
Chris@16
|
349 ++ptr;
|
Chris@16
|
350 }
|
Chris@16
|
351 return result;
|
Chris@16
|
352 }
|
Chris@16
|
353
|
Chris@16
|
354 //
|
Chris@16
|
355 // specialised version for narrow characters:
|
Chris@16
|
356 //
|
Chris@16
|
357 template <>
|
Chris@16
|
358 class BOOST_REGEX_DECL cpp_regex_traits_char_layer<char> : public cpp_regex_traits_base<char>
|
Chris@16
|
359 {
|
Chris@16
|
360 typedef std::string string_type;
|
Chris@16
|
361 public:
|
Chris@16
|
362 cpp_regex_traits_char_layer(const std::locale& l)
|
Chris@16
|
363 : cpp_regex_traits_base<char>(l)
|
Chris@16
|
364 {
|
Chris@16
|
365 init();
|
Chris@16
|
366 }
|
Chris@16
|
367 cpp_regex_traits_char_layer(const cpp_regex_traits_base<char>& l)
|
Chris@16
|
368 : cpp_regex_traits_base<char>(l)
|
Chris@16
|
369 {
|
Chris@16
|
370 init();
|
Chris@16
|
371 }
|
Chris@16
|
372
|
Chris@16
|
373 regex_constants::syntax_type syntax_type(char c)const
|
Chris@16
|
374 {
|
Chris@16
|
375 return m_char_map[static_cast<unsigned char>(c)];
|
Chris@16
|
376 }
|
Chris@16
|
377 regex_constants::escape_syntax_type escape_syntax_type(char c) const
|
Chris@16
|
378 {
|
Chris@16
|
379 return m_char_map[static_cast<unsigned char>(c)];
|
Chris@16
|
380 }
|
Chris@16
|
381
|
Chris@16
|
382 private:
|
Chris@16
|
383 regex_constants::syntax_type m_char_map[1u << CHAR_BIT];
|
Chris@16
|
384 void init();
|
Chris@16
|
385 };
|
Chris@16
|
386
|
Chris@16
|
387 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET
|
Chris@16
|
388 enum
|
Chris@16
|
389 {
|
Chris@16
|
390 char_class_space=1<<0,
|
Chris@16
|
391 char_class_print=1<<1,
|
Chris@16
|
392 char_class_cntrl=1<<2,
|
Chris@16
|
393 char_class_upper=1<<3,
|
Chris@16
|
394 char_class_lower=1<<4,
|
Chris@16
|
395 char_class_alpha=1<<5,
|
Chris@16
|
396 char_class_digit=1<<6,
|
Chris@16
|
397 char_class_punct=1<<7,
|
Chris@16
|
398 char_class_xdigit=1<<8,
|
Chris@16
|
399 char_class_alnum=char_class_alpha|char_class_digit,
|
Chris@16
|
400 char_class_graph=char_class_alnum|char_class_punct,
|
Chris@16
|
401 char_class_blank=1<<9,
|
Chris@16
|
402 char_class_word=1<<10,
|
Chris@16
|
403 char_class_unicode=1<<11,
|
Chris@16
|
404 char_class_horizontal_space=1<<12,
|
Chris@16
|
405 char_class_vertical_space=1<<13
|
Chris@16
|
406 };
|
Chris@16
|
407
|
Chris@16
|
408 #endif
|
Chris@16
|
409
|
Chris@16
|
410 //
|
Chris@16
|
411 // class cpp_regex_traits_implementation:
|
Chris@16
|
412 // provides pimpl implementation for cpp_regex_traits.
|
Chris@16
|
413 //
|
Chris@16
|
414 template <class charT>
|
Chris@16
|
415 class cpp_regex_traits_implementation : public cpp_regex_traits_char_layer<charT>
|
Chris@16
|
416 {
|
Chris@16
|
417 public:
|
Chris@16
|
418 typedef typename cpp_regex_traits<charT>::char_class_type char_class_type;
|
Chris@16
|
419 typedef typename std::ctype<charT>::mask native_mask_type;
|
Chris@16
|
420 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
|
Chris@16
|
421 BOOST_STATIC_CONSTANT(char_class_type, mask_blank = 1u << 24);
|
Chris@16
|
422 BOOST_STATIC_CONSTANT(char_class_type, mask_word = 1u << 25);
|
Chris@16
|
423 BOOST_STATIC_CONSTANT(char_class_type, mask_unicode = 1u << 26);
|
Chris@16
|
424 BOOST_STATIC_CONSTANT(char_class_type, mask_horizontal = 1u << 27);
|
Chris@16
|
425 BOOST_STATIC_CONSTANT(char_class_type, mask_vertical = 1u << 28);
|
Chris@16
|
426 #endif
|
Chris@16
|
427
|
Chris@16
|
428 typedef std::basic_string<charT> string_type;
|
Chris@16
|
429 typedef charT char_type;
|
Chris@16
|
430 //cpp_regex_traits_implementation();
|
Chris@16
|
431 cpp_regex_traits_implementation(const std::locale& l)
|
Chris@16
|
432 : cpp_regex_traits_char_layer<charT>(l)
|
Chris@16
|
433 {
|
Chris@16
|
434 init();
|
Chris@16
|
435 }
|
Chris@16
|
436 cpp_regex_traits_implementation(const cpp_regex_traits_base<charT>& l)
|
Chris@16
|
437 : cpp_regex_traits_char_layer<charT>(l)
|
Chris@16
|
438 {
|
Chris@16
|
439 init();
|
Chris@16
|
440 }
|
Chris@16
|
441 std::string error_string(regex_constants::error_type n) const
|
Chris@16
|
442 {
|
Chris@16
|
443 if(!m_error_strings.empty())
|
Chris@16
|
444 {
|
Chris@16
|
445 std::map<int, std::string>::const_iterator p = m_error_strings.find(n);
|
Chris@16
|
446 return (p == m_error_strings.end()) ? std::string(get_default_error_string(n)) : p->second;
|
Chris@16
|
447 }
|
Chris@16
|
448 return get_default_error_string(n);
|
Chris@16
|
449 }
|
Chris@16
|
450 char_class_type lookup_classname(const charT* p1, const charT* p2) const
|
Chris@16
|
451 {
|
Chris@16
|
452 char_class_type result = lookup_classname_imp(p1, p2);
|
Chris@16
|
453 if(result == 0)
|
Chris@16
|
454 {
|
Chris@16
|
455 string_type temp(p1, p2);
|
Chris@16
|
456 this->m_pctype->tolower(&*temp.begin(), &*temp.begin() + temp.size());
|
Chris@16
|
457 result = lookup_classname_imp(&*temp.begin(), &*temp.begin() + temp.size());
|
Chris@16
|
458 }
|
Chris@16
|
459 return result;
|
Chris@16
|
460 }
|
Chris@16
|
461 string_type lookup_collatename(const charT* p1, const charT* p2) const;
|
Chris@16
|
462 string_type transform_primary(const charT* p1, const charT* p2) const;
|
Chris@16
|
463 string_type transform(const charT* p1, const charT* p2) const;
|
Chris@16
|
464 private:
|
Chris@16
|
465 std::map<int, std::string> m_error_strings; // error messages indexed by numberic ID
|
Chris@16
|
466 std::map<string_type, char_class_type> m_custom_class_names; // character class names
|
Chris@16
|
467 std::map<string_type, string_type> m_custom_collate_names; // collating element names
|
Chris@16
|
468 unsigned m_collate_type; // the form of the collation string
|
Chris@16
|
469 charT m_collate_delim; // the collation group delimiter
|
Chris@16
|
470 //
|
Chris@16
|
471 // helpers:
|
Chris@16
|
472 //
|
Chris@16
|
473 char_class_type lookup_classname_imp(const charT* p1, const charT* p2) const;
|
Chris@16
|
474 void init();
|
Chris@16
|
475 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET
|
Chris@16
|
476 public:
|
Chris@16
|
477 bool isctype(charT c, char_class_type m)const;
|
Chris@16
|
478 #endif
|
Chris@16
|
479 };
|
Chris@16
|
480
|
Chris@16
|
481 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
|
Chris@16
|
482 #if !defined(BOOST_NO_INCLASS_MEMBER_INITIALIZATION)
|
Chris@16
|
483
|
Chris@16
|
484 template <class charT>
|
Chris@16
|
485 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_blank;
|
Chris@16
|
486 template <class charT>
|
Chris@16
|
487 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_word;
|
Chris@16
|
488 template <class charT>
|
Chris@16
|
489 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_unicode;
|
Chris@16
|
490 template <class charT>
|
Chris@16
|
491 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_vertical;
|
Chris@16
|
492 template <class charT>
|
Chris@16
|
493 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_horizontal;
|
Chris@16
|
494
|
Chris@16
|
495 #endif
|
Chris@16
|
496 #endif
|
Chris@16
|
497
|
Chris@16
|
498 template <class charT>
|
Chris@16
|
499 typename cpp_regex_traits_implementation<charT>::string_type
|
Chris@16
|
500 cpp_regex_traits_implementation<charT>::transform_primary(const charT* p1, const charT* p2) const
|
Chris@16
|
501 {
|
Chris@16
|
502 //
|
Chris@16
|
503 // PRECONDITIONS:
|
Chris@16
|
504 //
|
Chris@16
|
505 // A bug in gcc 3.2 (and maybe other versions as well) treats
|
Chris@16
|
506 // p1 as a null terminated string, for efficiency reasons
|
Chris@16
|
507 // we work around this elsewhere, but just assert here that
|
Chris@16
|
508 // we adhere to gcc's (buggy) preconditions...
|
Chris@16
|
509 //
|
Chris@16
|
510 BOOST_ASSERT(*p2 == 0);
|
Chris@16
|
511 string_type result;
|
Chris@101
|
512 #if defined(_CPPLIB_VER)
|
Chris@101
|
513 //
|
Chris@101
|
514 // A bug in VC11 and 12 causes the program to hang if we pass a null-string
|
Chris@101
|
515 // to std::collate::transform, but only for certain locales :-(
|
Chris@101
|
516 // Probably effects Intel and Clang or any compiler using the VC std library (Dinkumware).
|
Chris@101
|
517 //
|
Chris@101
|
518 if(*p1 == 0)
|
Chris@101
|
519 {
|
Chris@101
|
520 return string_type(1, charT(0));
|
Chris@101
|
521 }
|
Chris@101
|
522 #endif
|
Chris@16
|
523 //
|
Chris@16
|
524 // swallowing all exceptions here is a bad idea
|
Chris@16
|
525 // however at least one std lib will always throw
|
Chris@16
|
526 // std::bad_alloc for certain arguments...
|
Chris@16
|
527 //
|
Chris@16
|
528 #ifndef BOOST_NO_EXCEPTIONS
|
Chris@16
|
529 try{
|
Chris@16
|
530 #endif
|
Chris@16
|
531 //
|
Chris@16
|
532 // What we do here depends upon the format of the sort key returned by
|
Chris@16
|
533 // sort key returned by this->transform:
|
Chris@16
|
534 //
|
Chris@16
|
535 switch(m_collate_type)
|
Chris@16
|
536 {
|
Chris@16
|
537 case sort_C:
|
Chris@16
|
538 case sort_unknown:
|
Chris@16
|
539 // the best we can do is translate to lower case, then get a regular sort key:
|
Chris@16
|
540 {
|
Chris@16
|
541 result.assign(p1, p2);
|
Chris@16
|
542 this->m_pctype->tolower(&*result.begin(), &*result.begin() + result.size());
|
Chris@16
|
543 result = this->m_pcollate->transform(&*result.begin(), &*result.begin() + result.size());
|
Chris@16
|
544 break;
|
Chris@16
|
545 }
|
Chris@16
|
546 case sort_fixed:
|
Chris@16
|
547 {
|
Chris@16
|
548 // get a regular sort key, and then truncate it:
|
Chris@16
|
549 result.assign(this->m_pcollate->transform(p1, p2));
|
Chris@16
|
550 result.erase(this->m_collate_delim);
|
Chris@16
|
551 break;
|
Chris@16
|
552 }
|
Chris@16
|
553 case sort_delim:
|
Chris@16
|
554 // get a regular sort key, and then truncate everything after the delim:
|
Chris@16
|
555 result.assign(this->m_pcollate->transform(p1, p2));
|
Chris@16
|
556 std::size_t i;
|
Chris@16
|
557 for(i = 0; i < result.size(); ++i)
|
Chris@16
|
558 {
|
Chris@16
|
559 if(result[i] == m_collate_delim)
|
Chris@16
|
560 break;
|
Chris@16
|
561 }
|
Chris@16
|
562 result.erase(i);
|
Chris@16
|
563 break;
|
Chris@16
|
564 }
|
Chris@16
|
565 #ifndef BOOST_NO_EXCEPTIONS
|
Chris@16
|
566 }catch(...){}
|
Chris@16
|
567 #endif
|
Chris@16
|
568 while(result.size() && (charT(0) == *result.rbegin()))
|
Chris@16
|
569 result.erase(result.size() - 1);
|
Chris@16
|
570 if(result.empty())
|
Chris@16
|
571 {
|
Chris@16
|
572 // character is ignorable at the primary level:
|
Chris@16
|
573 result = string_type(1, charT(0));
|
Chris@16
|
574 }
|
Chris@16
|
575 return result;
|
Chris@16
|
576 }
|
Chris@16
|
577
|
Chris@16
|
578 template <class charT>
|
Chris@16
|
579 typename cpp_regex_traits_implementation<charT>::string_type
|
Chris@16
|
580 cpp_regex_traits_implementation<charT>::transform(const charT* p1, const charT* p2) const
|
Chris@16
|
581 {
|
Chris@16
|
582 //
|
Chris@16
|
583 // PRECONDITIONS:
|
Chris@16
|
584 //
|
Chris@16
|
585 // A bug in gcc 3.2 (and maybe other versions as well) treats
|
Chris@16
|
586 // p1 as a null terminated string, for efficiency reasons
|
Chris@16
|
587 // we work around this elsewhere, but just assert here that
|
Chris@16
|
588 // we adhere to gcc's (buggy) preconditions...
|
Chris@16
|
589 //
|
Chris@16
|
590 BOOST_ASSERT(*p2 == 0);
|
Chris@16
|
591 //
|
Chris@16
|
592 // swallowing all exceptions here is a bad idea
|
Chris@16
|
593 // however at least one std lib will always throw
|
Chris@16
|
594 // std::bad_alloc for certain arguments...
|
Chris@16
|
595 //
|
Chris@101
|
596 string_type result, result2;
|
Chris@101
|
597 #if defined(_CPPLIB_VER)
|
Chris@101
|
598 //
|
Chris@101
|
599 // A bug in VC11 and 12 causes the program to hang if we pass a null-string
|
Chris@101
|
600 // to std::collate::transform, but only for certain locales :-(
|
Chris@101
|
601 // Probably effects Intel and Clang or any compiler using the VC std library (Dinkumware).
|
Chris@101
|
602 //
|
Chris@101
|
603 if(*p1 == 0)
|
Chris@101
|
604 {
|
Chris@101
|
605 return result;
|
Chris@101
|
606 }
|
Chris@101
|
607 #endif
|
Chris@16
|
608 #ifndef BOOST_NO_EXCEPTIONS
|
Chris@16
|
609 try{
|
Chris@16
|
610 #endif
|
Chris@16
|
611 result = this->m_pcollate->transform(p1, p2);
|
Chris@16
|
612 //
|
Chris@16
|
613 // Borland's STLPort version returns a NULL-terminated
|
Chris@16
|
614 // string that has garbage at the end - each call to
|
Chris@16
|
615 // std::collate<wchar_t>::transform returns a different string!
|
Chris@16
|
616 // So as a workaround, we'll truncate the string at the first NULL
|
Chris@16
|
617 // which _seems_ to work....
|
Chris@16
|
618 #if BOOST_WORKAROUND(__BORLANDC__, < 0x580)
|
Chris@16
|
619 result.erase(result.find(charT(0)));
|
Chris@16
|
620 #else
|
Chris@16
|
621 //
|
Chris@16
|
622 // some implementations (Dinkumware) append unnecessary trailing \0's:
|
Chris@16
|
623 while(result.size() && (charT(0) == *result.rbegin()))
|
Chris@16
|
624 result.erase(result.size() - 1);
|
Chris@16
|
625 #endif
|
Chris@101
|
626 //
|
Chris@101
|
627 // We may have NULL's used as separators between sections of the collate string,
|
Chris@101
|
628 // an example would be Boost.Locale. We have no way to detect this case via
|
Chris@101
|
629 // #defines since this can be used with any compiler/platform combination.
|
Chris@101
|
630 // Unfortunately our state machine (which was devised when all implementations
|
Chris@101
|
631 // used underlying C language API's) can't cope with that case. One workaround
|
Chris@101
|
632 // is to replace each character with 2, fortunately this code isn't used that
|
Chris@101
|
633 // much as this is now slower than before :-(
|
Chris@101
|
634 //
|
Chris@101
|
635 typedef typename make_unsigned<charT>::type uchar_type;
|
Chris@101
|
636 result2.reserve(result.size() * 2 + 2);
|
Chris@101
|
637 for(unsigned i = 0; i < result.size(); ++i)
|
Chris@101
|
638 {
|
Chris@101
|
639 if(static_cast<uchar_type>(result[i]) == (std::numeric_limits<uchar_type>::max)())
|
Chris@101
|
640 {
|
Chris@101
|
641 result2.append(1, charT((std::numeric_limits<uchar_type>::max)())).append(1, charT('b'));
|
Chris@101
|
642 }
|
Chris@101
|
643 else
|
Chris@101
|
644 {
|
Chris@101
|
645 result2.append(1, static_cast<charT>(1 + static_cast<uchar_type>(result[i]))).append(1, charT('b') - 1);
|
Chris@101
|
646 }
|
Chris@101
|
647 }
|
Chris@101
|
648 BOOST_ASSERT(std::find(result2.begin(), result2.end(), charT(0)) == result2.end());
|
Chris@16
|
649 #ifndef BOOST_NO_EXCEPTIONS
|
Chris@16
|
650 }
|
Chris@16
|
651 catch(...)
|
Chris@16
|
652 {
|
Chris@16
|
653 }
|
Chris@16
|
654 #endif
|
Chris@101
|
655 return result2;
|
Chris@16
|
656 }
|
Chris@16
|
657
|
Chris@16
|
658
|
Chris@16
|
659 template <class charT>
|
Chris@16
|
660 typename cpp_regex_traits_implementation<charT>::string_type
|
Chris@16
|
661 cpp_regex_traits_implementation<charT>::lookup_collatename(const charT* p1, const charT* p2) const
|
Chris@16
|
662 {
|
Chris@16
|
663 typedef typename std::map<string_type, string_type>::const_iterator iter_type;
|
Chris@16
|
664 if(m_custom_collate_names.size())
|
Chris@16
|
665 {
|
Chris@16
|
666 iter_type pos = m_custom_collate_names.find(string_type(p1, p2));
|
Chris@16
|
667 if(pos != m_custom_collate_names.end())
|
Chris@16
|
668 return pos->second;
|
Chris@16
|
669 }
|
Chris@16
|
670 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
|
Chris@16
|
671 && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
|
Chris@16
|
672 std::string name(p1, p2);
|
Chris@16
|
673 #else
|
Chris@16
|
674 std::string name;
|
Chris@16
|
675 const charT* p0 = p1;
|
Chris@16
|
676 while(p0 != p2)
|
Chris@16
|
677 name.append(1, char(*p0++));
|
Chris@16
|
678 #endif
|
Chris@16
|
679 name = lookup_default_collate_name(name);
|
Chris@16
|
680 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
|
Chris@16
|
681 && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
|
Chris@16
|
682 if(name.size())
|
Chris@16
|
683 return string_type(name.begin(), name.end());
|
Chris@16
|
684 #else
|
Chris@16
|
685 if(name.size())
|
Chris@16
|
686 {
|
Chris@16
|
687 string_type result;
|
Chris@16
|
688 typedef std::string::const_iterator iter;
|
Chris@16
|
689 iter b = name.begin();
|
Chris@16
|
690 iter e = name.end();
|
Chris@16
|
691 while(b != e)
|
Chris@16
|
692 result.append(1, charT(*b++));
|
Chris@16
|
693 return result;
|
Chris@16
|
694 }
|
Chris@16
|
695 #endif
|
Chris@16
|
696 if(p2 - p1 == 1)
|
Chris@16
|
697 return string_type(1, *p1);
|
Chris@16
|
698 return string_type();
|
Chris@16
|
699 }
|
Chris@16
|
700
|
Chris@16
|
701 template <class charT>
|
Chris@16
|
702 void cpp_regex_traits_implementation<charT>::init()
|
Chris@16
|
703 {
|
Chris@16
|
704 #ifndef BOOST_NO_STD_MESSAGES
|
Chris@16
|
705 #ifndef __IBMCPP__
|
Chris@16
|
706 typename std::messages<charT>::catalog cat = static_cast<std::messages<char>::catalog>(-1);
|
Chris@16
|
707 #else
|
Chris@16
|
708 typename std::messages<charT>::catalog cat = reinterpret_cast<std::messages<char>::catalog>(-1);
|
Chris@16
|
709 #endif
|
Chris@16
|
710 std::string cat_name(cpp_regex_traits<charT>::get_catalog_name());
|
Chris@16
|
711 if(cat_name.size() && (this->m_pmessages != 0))
|
Chris@16
|
712 {
|
Chris@16
|
713 cat = this->m_pmessages->open(
|
Chris@16
|
714 cat_name,
|
Chris@16
|
715 this->m_locale);
|
Chris@16
|
716 if((int)cat < 0)
|
Chris@16
|
717 {
|
Chris@16
|
718 std::string m("Unable to open message catalog: ");
|
Chris@16
|
719 std::runtime_error err(m + cat_name);
|
Chris@16
|
720 boost::re_detail::raise_runtime_error(err);
|
Chris@16
|
721 }
|
Chris@16
|
722 }
|
Chris@16
|
723 //
|
Chris@16
|
724 // if we have a valid catalog then load our messages:
|
Chris@16
|
725 //
|
Chris@16
|
726 if((int)cat >= 0)
|
Chris@16
|
727 {
|
Chris@16
|
728 //
|
Chris@16
|
729 // Error messages:
|
Chris@16
|
730 //
|
Chris@16
|
731 for(boost::regex_constants::error_type i = static_cast<boost::regex_constants::error_type>(0);
|
Chris@16
|
732 i <= boost::regex_constants::error_unknown;
|
Chris@16
|
733 i = static_cast<boost::regex_constants::error_type>(i + 1))
|
Chris@16
|
734 {
|
Chris@16
|
735 const char* p = get_default_error_string(i);
|
Chris@16
|
736 string_type default_message;
|
Chris@16
|
737 while(*p)
|
Chris@16
|
738 {
|
Chris@16
|
739 default_message.append(1, this->m_pctype->widen(*p));
|
Chris@16
|
740 ++p;
|
Chris@16
|
741 }
|
Chris@16
|
742 string_type s = this->m_pmessages->get(cat, 0, i+200, default_message);
|
Chris@16
|
743 std::string result;
|
Chris@16
|
744 for(std::string::size_type j = 0; j < s.size(); ++j)
|
Chris@16
|
745 {
|
Chris@16
|
746 result.append(1, this->m_pctype->narrow(s[j], 0));
|
Chris@16
|
747 }
|
Chris@16
|
748 m_error_strings[i] = result;
|
Chris@16
|
749 }
|
Chris@16
|
750 //
|
Chris@16
|
751 // Custom class names:
|
Chris@16
|
752 //
|
Chris@16
|
753 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
|
Chris@16
|
754 static const char_class_type masks[16] =
|
Chris@16
|
755 {
|
Chris@16
|
756 std::ctype<charT>::alnum,
|
Chris@16
|
757 std::ctype<charT>::alpha,
|
Chris@16
|
758 std::ctype<charT>::cntrl,
|
Chris@16
|
759 std::ctype<charT>::digit,
|
Chris@16
|
760 std::ctype<charT>::graph,
|
Chris@16
|
761 cpp_regex_traits_implementation<charT>::mask_horizontal,
|
Chris@16
|
762 std::ctype<charT>::lower,
|
Chris@16
|
763 std::ctype<charT>::print,
|
Chris@16
|
764 std::ctype<charT>::punct,
|
Chris@16
|
765 std::ctype<charT>::space,
|
Chris@16
|
766 std::ctype<charT>::upper,
|
Chris@16
|
767 cpp_regex_traits_implementation<charT>::mask_vertical,
|
Chris@16
|
768 std::ctype<charT>::xdigit,
|
Chris@16
|
769 cpp_regex_traits_implementation<charT>::mask_blank,
|
Chris@16
|
770 cpp_regex_traits_implementation<charT>::mask_word,
|
Chris@16
|
771 cpp_regex_traits_implementation<charT>::mask_unicode,
|
Chris@16
|
772 };
|
Chris@16
|
773 #else
|
Chris@16
|
774 static const char_class_type masks[16] =
|
Chris@16
|
775 {
|
Chris@16
|
776 ::boost::re_detail::char_class_alnum,
|
Chris@16
|
777 ::boost::re_detail::char_class_alpha,
|
Chris@16
|
778 ::boost::re_detail::char_class_cntrl,
|
Chris@16
|
779 ::boost::re_detail::char_class_digit,
|
Chris@16
|
780 ::boost::re_detail::char_class_graph,
|
Chris@16
|
781 ::boost::re_detail::char_class_horizontal_space,
|
Chris@16
|
782 ::boost::re_detail::char_class_lower,
|
Chris@16
|
783 ::boost::re_detail::char_class_print,
|
Chris@16
|
784 ::boost::re_detail::char_class_punct,
|
Chris@16
|
785 ::boost::re_detail::char_class_space,
|
Chris@16
|
786 ::boost::re_detail::char_class_upper,
|
Chris@16
|
787 ::boost::re_detail::char_class_vertical_space,
|
Chris@16
|
788 ::boost::re_detail::char_class_xdigit,
|
Chris@16
|
789 ::boost::re_detail::char_class_blank,
|
Chris@16
|
790 ::boost::re_detail::char_class_word,
|
Chris@16
|
791 ::boost::re_detail::char_class_unicode,
|
Chris@16
|
792 };
|
Chris@16
|
793 #endif
|
Chris@16
|
794 static const string_type null_string;
|
Chris@16
|
795 for(unsigned int j = 0; j <= 13; ++j)
|
Chris@16
|
796 {
|
Chris@16
|
797 string_type s(this->m_pmessages->get(cat, 0, j+300, null_string));
|
Chris@16
|
798 if(s.size())
|
Chris@16
|
799 this->m_custom_class_names[s] = masks[j];
|
Chris@16
|
800 }
|
Chris@16
|
801 }
|
Chris@16
|
802 #endif
|
Chris@16
|
803 //
|
Chris@16
|
804 // get the collation format used by m_pcollate:
|
Chris@16
|
805 //
|
Chris@16
|
806 m_collate_type = re_detail::find_sort_syntax(this, &m_collate_delim);
|
Chris@16
|
807 }
|
Chris@16
|
808
|
Chris@16
|
809 template <class charT>
|
Chris@16
|
810 typename cpp_regex_traits_implementation<charT>::char_class_type
|
Chris@16
|
811 cpp_regex_traits_implementation<charT>::lookup_classname_imp(const charT* p1, const charT* p2) const
|
Chris@16
|
812 {
|
Chris@16
|
813 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
|
Chris@16
|
814 static const char_class_type masks[22] =
|
Chris@16
|
815 {
|
Chris@16
|
816 0,
|
Chris@16
|
817 std::ctype<char>::alnum,
|
Chris@16
|
818 std::ctype<char>::alpha,
|
Chris@16
|
819 cpp_regex_traits_implementation<charT>::mask_blank,
|
Chris@16
|
820 std::ctype<char>::cntrl,
|
Chris@16
|
821 std::ctype<char>::digit,
|
Chris@16
|
822 std::ctype<char>::digit,
|
Chris@16
|
823 std::ctype<char>::graph,
|
Chris@16
|
824 cpp_regex_traits_implementation<charT>::mask_horizontal,
|
Chris@16
|
825 std::ctype<char>::lower,
|
Chris@16
|
826 std::ctype<char>::lower,
|
Chris@16
|
827 std::ctype<char>::print,
|
Chris@16
|
828 std::ctype<char>::punct,
|
Chris@16
|
829 std::ctype<char>::space,
|
Chris@16
|
830 std::ctype<char>::space,
|
Chris@16
|
831 std::ctype<char>::upper,
|
Chris@16
|
832 cpp_regex_traits_implementation<charT>::mask_unicode,
|
Chris@16
|
833 std::ctype<char>::upper,
|
Chris@16
|
834 cpp_regex_traits_implementation<charT>::mask_vertical,
|
Chris@16
|
835 std::ctype<char>::alnum | cpp_regex_traits_implementation<charT>::mask_word,
|
Chris@16
|
836 std::ctype<char>::alnum | cpp_regex_traits_implementation<charT>::mask_word,
|
Chris@16
|
837 std::ctype<char>::xdigit,
|
Chris@16
|
838 };
|
Chris@16
|
839 #else
|
Chris@16
|
840 static const char_class_type masks[22] =
|
Chris@16
|
841 {
|
Chris@16
|
842 0,
|
Chris@16
|
843 ::boost::re_detail::char_class_alnum,
|
Chris@16
|
844 ::boost::re_detail::char_class_alpha,
|
Chris@16
|
845 ::boost::re_detail::char_class_blank,
|
Chris@16
|
846 ::boost::re_detail::char_class_cntrl,
|
Chris@16
|
847 ::boost::re_detail::char_class_digit,
|
Chris@16
|
848 ::boost::re_detail::char_class_digit,
|
Chris@16
|
849 ::boost::re_detail::char_class_graph,
|
Chris@16
|
850 ::boost::re_detail::char_class_horizontal_space,
|
Chris@16
|
851 ::boost::re_detail::char_class_lower,
|
Chris@16
|
852 ::boost::re_detail::char_class_lower,
|
Chris@16
|
853 ::boost::re_detail::char_class_print,
|
Chris@16
|
854 ::boost::re_detail::char_class_punct,
|
Chris@16
|
855 ::boost::re_detail::char_class_space,
|
Chris@16
|
856 ::boost::re_detail::char_class_space,
|
Chris@16
|
857 ::boost::re_detail::char_class_upper,
|
Chris@16
|
858 ::boost::re_detail::char_class_unicode,
|
Chris@16
|
859 ::boost::re_detail::char_class_upper,
|
Chris@16
|
860 ::boost::re_detail::char_class_vertical_space,
|
Chris@16
|
861 ::boost::re_detail::char_class_alnum | ::boost::re_detail::char_class_word,
|
Chris@16
|
862 ::boost::re_detail::char_class_alnum | ::boost::re_detail::char_class_word,
|
Chris@16
|
863 ::boost::re_detail::char_class_xdigit,
|
Chris@16
|
864 };
|
Chris@16
|
865 #endif
|
Chris@16
|
866 if(m_custom_class_names.size())
|
Chris@16
|
867 {
|
Chris@16
|
868 typedef typename std::map<std::basic_string<charT>, char_class_type>::const_iterator map_iter;
|
Chris@16
|
869 map_iter pos = m_custom_class_names.find(string_type(p1, p2));
|
Chris@16
|
870 if(pos != m_custom_class_names.end())
|
Chris@16
|
871 return pos->second;
|
Chris@16
|
872 }
|
Chris@16
|
873 std::size_t state_id = 1 + re_detail::get_default_class_id(p1, p2);
|
Chris@16
|
874 BOOST_ASSERT(state_id < sizeof(masks) / sizeof(masks[0]));
|
Chris@16
|
875 return masks[state_id];
|
Chris@16
|
876 }
|
Chris@16
|
877
|
Chris@16
|
878 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET
|
Chris@16
|
879 template <class charT>
|
Chris@16
|
880 bool cpp_regex_traits_implementation<charT>::isctype(const charT c, char_class_type mask) const
|
Chris@16
|
881 {
|
Chris@16
|
882 return
|
Chris@16
|
883 ((mask & ::boost::re_detail::char_class_space) && (this->m_pctype->is(std::ctype<charT>::space, c)))
|
Chris@16
|
884 || ((mask & ::boost::re_detail::char_class_print) && (this->m_pctype->is(std::ctype<charT>::print, c)))
|
Chris@16
|
885 || ((mask & ::boost::re_detail::char_class_cntrl) && (this->m_pctype->is(std::ctype<charT>::cntrl, c)))
|
Chris@16
|
886 || ((mask & ::boost::re_detail::char_class_upper) && (this->m_pctype->is(std::ctype<charT>::upper, c)))
|
Chris@16
|
887 || ((mask & ::boost::re_detail::char_class_lower) && (this->m_pctype->is(std::ctype<charT>::lower, c)))
|
Chris@16
|
888 || ((mask & ::boost::re_detail::char_class_alpha) && (this->m_pctype->is(std::ctype<charT>::alpha, c)))
|
Chris@16
|
889 || ((mask & ::boost::re_detail::char_class_digit) && (this->m_pctype->is(std::ctype<charT>::digit, c)))
|
Chris@16
|
890 || ((mask & ::boost::re_detail::char_class_punct) && (this->m_pctype->is(std::ctype<charT>::punct, c)))
|
Chris@16
|
891 || ((mask & ::boost::re_detail::char_class_xdigit) && (this->m_pctype->is(std::ctype<charT>::xdigit, c)))
|
Chris@16
|
892 || ((mask & ::boost::re_detail::char_class_blank) && (this->m_pctype->is(std::ctype<charT>::space, c)) && !::boost::re_detail::is_separator(c))
|
Chris@16
|
893 || ((mask & ::boost::re_detail::char_class_word) && (c == '_'))
|
Chris@16
|
894 || ((mask & ::boost::re_detail::char_class_unicode) && ::boost::re_detail::is_extended(c))
|
Chris@16
|
895 || ((mask & ::boost::re_detail::char_class_vertical_space) && (is_separator(c) || (c == '\v')))
|
Chris@16
|
896 || ((mask & ::boost::re_detail::char_class_horizontal_space) && this->m_pctype->is(std::ctype<charT>::space, c) && !(is_separator(c) || (c == '\v')));
|
Chris@16
|
897 }
|
Chris@16
|
898 #endif
|
Chris@16
|
899
|
Chris@16
|
900
|
Chris@16
|
901 template <class charT>
|
Chris@101
|
902 inline boost::shared_ptr<const cpp_regex_traits_implementation<charT> > create_cpp_regex_traits(const std::locale& l)
|
Chris@16
|
903 {
|
Chris@16
|
904 cpp_regex_traits_base<charT> key(l);
|
Chris@16
|
905 return ::boost::object_cache<cpp_regex_traits_base<charT>, cpp_regex_traits_implementation<charT> >::get(key, 5);
|
Chris@16
|
906 }
|
Chris@16
|
907
|
Chris@16
|
908 } // re_detail
|
Chris@16
|
909
|
Chris@16
|
910 template <class charT>
|
Chris@16
|
911 class cpp_regex_traits
|
Chris@16
|
912 {
|
Chris@16
|
913 private:
|
Chris@16
|
914 typedef std::ctype<charT> ctype_type;
|
Chris@16
|
915 public:
|
Chris@16
|
916 typedef charT char_type;
|
Chris@16
|
917 typedef std::size_t size_type;
|
Chris@16
|
918 typedef std::basic_string<char_type> string_type;
|
Chris@16
|
919 typedef std::locale locale_type;
|
Chris@16
|
920 typedef boost::uint_least32_t char_class_type;
|
Chris@16
|
921
|
Chris@16
|
922 struct boost_extensions_tag{};
|
Chris@16
|
923
|
Chris@16
|
924 cpp_regex_traits()
|
Chris@16
|
925 : m_pimpl(re_detail::create_cpp_regex_traits<charT>(std::locale()))
|
Chris@16
|
926 { }
|
Chris@16
|
927 static size_type length(const char_type* p)
|
Chris@16
|
928 {
|
Chris@16
|
929 return std::char_traits<charT>::length(p);
|
Chris@16
|
930 }
|
Chris@16
|
931 regex_constants::syntax_type syntax_type(charT c)const
|
Chris@16
|
932 {
|
Chris@16
|
933 return m_pimpl->syntax_type(c);
|
Chris@16
|
934 }
|
Chris@16
|
935 regex_constants::escape_syntax_type escape_syntax_type(charT c) const
|
Chris@16
|
936 {
|
Chris@16
|
937 return m_pimpl->escape_syntax_type(c);
|
Chris@16
|
938 }
|
Chris@16
|
939 charT translate(charT c) const
|
Chris@16
|
940 {
|
Chris@16
|
941 return c;
|
Chris@16
|
942 }
|
Chris@16
|
943 charT translate_nocase(charT c) const
|
Chris@16
|
944 {
|
Chris@16
|
945 return m_pimpl->m_pctype->tolower(c);
|
Chris@16
|
946 }
|
Chris@16
|
947 charT translate(charT c, bool icase) const
|
Chris@16
|
948 {
|
Chris@16
|
949 return icase ? m_pimpl->m_pctype->tolower(c) : c;
|
Chris@16
|
950 }
|
Chris@16
|
951 charT tolower(charT c) const
|
Chris@16
|
952 {
|
Chris@16
|
953 return m_pimpl->m_pctype->tolower(c);
|
Chris@16
|
954 }
|
Chris@16
|
955 charT toupper(charT c) const
|
Chris@16
|
956 {
|
Chris@16
|
957 return m_pimpl->m_pctype->toupper(c);
|
Chris@16
|
958 }
|
Chris@16
|
959 string_type transform(const charT* p1, const charT* p2) const
|
Chris@16
|
960 {
|
Chris@16
|
961 return m_pimpl->transform(p1, p2);
|
Chris@16
|
962 }
|
Chris@16
|
963 string_type transform_primary(const charT* p1, const charT* p2) const
|
Chris@16
|
964 {
|
Chris@16
|
965 return m_pimpl->transform_primary(p1, p2);
|
Chris@16
|
966 }
|
Chris@16
|
967 char_class_type lookup_classname(const charT* p1, const charT* p2) const
|
Chris@16
|
968 {
|
Chris@16
|
969 return m_pimpl->lookup_classname(p1, p2);
|
Chris@16
|
970 }
|
Chris@16
|
971 string_type lookup_collatename(const charT* p1, const charT* p2) const
|
Chris@16
|
972 {
|
Chris@16
|
973 return m_pimpl->lookup_collatename(p1, p2);
|
Chris@16
|
974 }
|
Chris@16
|
975 bool isctype(charT c, char_class_type f) const
|
Chris@16
|
976 {
|
Chris@16
|
977 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
|
Chris@16
|
978 typedef typename std::ctype<charT>::mask ctype_mask;
|
Chris@16
|
979
|
Chris@16
|
980 static const ctype_mask mask_base =
|
Chris@16
|
981 static_cast<ctype_mask>(
|
Chris@16
|
982 std::ctype<charT>::alnum
|
Chris@16
|
983 | std::ctype<charT>::alpha
|
Chris@16
|
984 | std::ctype<charT>::cntrl
|
Chris@16
|
985 | std::ctype<charT>::digit
|
Chris@16
|
986 | std::ctype<charT>::graph
|
Chris@16
|
987 | std::ctype<charT>::lower
|
Chris@16
|
988 | std::ctype<charT>::print
|
Chris@16
|
989 | std::ctype<charT>::punct
|
Chris@16
|
990 | std::ctype<charT>::space
|
Chris@16
|
991 | std::ctype<charT>::upper
|
Chris@16
|
992 | std::ctype<charT>::xdigit);
|
Chris@16
|
993
|
Chris@16
|
994 if((f & mask_base)
|
Chris@16
|
995 && (m_pimpl->m_pctype->is(
|
Chris@16
|
996 static_cast<ctype_mask>(f & mask_base), c)))
|
Chris@16
|
997 return true;
|
Chris@16
|
998 else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_unicode) && re_detail::is_extended(c))
|
Chris@16
|
999 return true;
|
Chris@16
|
1000 else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_word) && (c == '_'))
|
Chris@16
|
1001 return true;
|
Chris@16
|
1002 else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_blank)
|
Chris@16
|
1003 && m_pimpl->m_pctype->is(std::ctype<charT>::space, c)
|
Chris@16
|
1004 && !re_detail::is_separator(c))
|
Chris@16
|
1005 return true;
|
Chris@16
|
1006 else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_vertical)
|
Chris@16
|
1007 && (::boost::re_detail::is_separator(c) || (c == '\v')))
|
Chris@16
|
1008 return true;
|
Chris@16
|
1009 else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_horizontal)
|
Chris@16
|
1010 && this->isctype(c, std::ctype<charT>::space) && !this->isctype(c, re_detail::cpp_regex_traits_implementation<charT>::mask_vertical))
|
Chris@16
|
1011 return true;
|
Chris@16
|
1012 return false;
|
Chris@16
|
1013 #else
|
Chris@16
|
1014 return m_pimpl->isctype(c, f);
|
Chris@16
|
1015 #endif
|
Chris@16
|
1016 }
|
Chris@16
|
1017 int toi(const charT*& p1, const charT* p2, int radix)const;
|
Chris@16
|
1018 int value(charT c, int radix)const
|
Chris@16
|
1019 {
|
Chris@16
|
1020 const charT* pc = &c;
|
Chris@16
|
1021 return toi(pc, pc + 1, radix);
|
Chris@16
|
1022 }
|
Chris@16
|
1023 locale_type imbue(locale_type l)
|
Chris@16
|
1024 {
|
Chris@16
|
1025 std::locale result(getloc());
|
Chris@16
|
1026 m_pimpl = re_detail::create_cpp_regex_traits<charT>(l);
|
Chris@16
|
1027 return result;
|
Chris@16
|
1028 }
|
Chris@16
|
1029 locale_type getloc()const
|
Chris@16
|
1030 {
|
Chris@16
|
1031 return m_pimpl->m_locale;
|
Chris@16
|
1032 }
|
Chris@16
|
1033 std::string error_string(regex_constants::error_type n) const
|
Chris@16
|
1034 {
|
Chris@16
|
1035 return m_pimpl->error_string(n);
|
Chris@16
|
1036 }
|
Chris@16
|
1037
|
Chris@16
|
1038 //
|
Chris@16
|
1039 // extension:
|
Chris@16
|
1040 // set the name of the message catalog in use (defaults to "boost_regex").
|
Chris@16
|
1041 //
|
Chris@16
|
1042 static std::string catalog_name(const std::string& name);
|
Chris@16
|
1043 static std::string get_catalog_name();
|
Chris@16
|
1044
|
Chris@16
|
1045 private:
|
Chris@16
|
1046 boost::shared_ptr<const re_detail::cpp_regex_traits_implementation<charT> > m_pimpl;
|
Chris@16
|
1047 //
|
Chris@16
|
1048 // catalog name handler:
|
Chris@16
|
1049 //
|
Chris@16
|
1050 static std::string& get_catalog_name_inst();
|
Chris@16
|
1051
|
Chris@16
|
1052 #ifdef BOOST_HAS_THREADS
|
Chris@16
|
1053 static static_mutex& get_mutex_inst();
|
Chris@16
|
1054 #endif
|
Chris@16
|
1055 };
|
Chris@16
|
1056
|
Chris@16
|
1057
|
Chris@16
|
1058 template <class charT>
|
Chris@16
|
1059 int cpp_regex_traits<charT>::toi(const charT*& first, const charT* last, int radix)const
|
Chris@16
|
1060 {
|
Chris@16
|
1061 re_detail::parser_buf<charT> sbuf; // buffer for parsing numbers.
|
Chris@16
|
1062 std::basic_istream<charT> is(&sbuf); // stream for parsing numbers.
|
Chris@16
|
1063
|
Chris@16
|
1064 // we do NOT want to parse any thousands separators inside the stream:
|
Chris@16
|
1065 last = std::find(first, last, BOOST_USE_FACET(std::numpunct<charT>, is.getloc()).thousands_sep());
|
Chris@16
|
1066
|
Chris@16
|
1067 sbuf.pubsetbuf(const_cast<charT*>(static_cast<const charT*>(first)), static_cast<std::streamsize>(last-first));
|
Chris@16
|
1068 is.clear();
|
Chris@16
|
1069 if(std::abs(radix) == 16) is >> std::hex;
|
Chris@16
|
1070 else if(std::abs(radix) == 8) is >> std::oct;
|
Chris@16
|
1071 else is >> std::dec;
|
Chris@16
|
1072 int val;
|
Chris@16
|
1073 if(is >> val)
|
Chris@16
|
1074 {
|
Chris@16
|
1075 first = first + ((last - first) - sbuf.in_avail());
|
Chris@16
|
1076 return val;
|
Chris@16
|
1077 }
|
Chris@16
|
1078 else
|
Chris@16
|
1079 return -1;
|
Chris@16
|
1080 }
|
Chris@16
|
1081
|
Chris@16
|
1082 template <class charT>
|
Chris@16
|
1083 std::string cpp_regex_traits<charT>::catalog_name(const std::string& name)
|
Chris@16
|
1084 {
|
Chris@16
|
1085 #ifdef BOOST_HAS_THREADS
|
Chris@16
|
1086 static_mutex::scoped_lock lk(get_mutex_inst());
|
Chris@16
|
1087 #endif
|
Chris@16
|
1088 std::string result(get_catalog_name_inst());
|
Chris@16
|
1089 get_catalog_name_inst() = name;
|
Chris@16
|
1090 return result;
|
Chris@16
|
1091 }
|
Chris@16
|
1092
|
Chris@16
|
1093 template <class charT>
|
Chris@16
|
1094 std::string& cpp_regex_traits<charT>::get_catalog_name_inst()
|
Chris@16
|
1095 {
|
Chris@16
|
1096 static std::string s_name;
|
Chris@16
|
1097 return s_name;
|
Chris@16
|
1098 }
|
Chris@16
|
1099
|
Chris@16
|
1100 template <class charT>
|
Chris@16
|
1101 std::string cpp_regex_traits<charT>::get_catalog_name()
|
Chris@16
|
1102 {
|
Chris@16
|
1103 #ifdef BOOST_HAS_THREADS
|
Chris@16
|
1104 static_mutex::scoped_lock lk(get_mutex_inst());
|
Chris@16
|
1105 #endif
|
Chris@16
|
1106 std::string result(get_catalog_name_inst());
|
Chris@16
|
1107 return result;
|
Chris@16
|
1108 }
|
Chris@16
|
1109
|
Chris@16
|
1110 #ifdef BOOST_HAS_THREADS
|
Chris@16
|
1111 template <class charT>
|
Chris@16
|
1112 static_mutex& cpp_regex_traits<charT>::get_mutex_inst()
|
Chris@16
|
1113 {
|
Chris@16
|
1114 static static_mutex s_mutex = BOOST_STATIC_MUTEX_INIT;
|
Chris@16
|
1115 return s_mutex;
|
Chris@16
|
1116 }
|
Chris@16
|
1117 #endif
|
Chris@16
|
1118
|
Chris@16
|
1119
|
Chris@16
|
1120 } // boost
|
Chris@16
|
1121
|
Chris@16
|
1122 #ifdef BOOST_MSVC
|
Chris@16
|
1123 #pragma warning(pop)
|
Chris@16
|
1124 #endif
|
Chris@16
|
1125
|
Chris@16
|
1126 #ifdef BOOST_MSVC
|
Chris@16
|
1127 #pragma warning(push)
|
Chris@16
|
1128 #pragma warning(disable: 4103)
|
Chris@16
|
1129 #endif
|
Chris@16
|
1130 #ifdef BOOST_HAS_ABI_HEADERS
|
Chris@16
|
1131 # include BOOST_ABI_SUFFIX
|
Chris@16
|
1132 #endif
|
Chris@16
|
1133 #ifdef BOOST_MSVC
|
Chris@16
|
1134 #pragma warning(pop)
|
Chris@16
|
1135 #endif
|
Chris@16
|
1136
|
Chris@16
|
1137 #endif
|
Chris@16
|
1138
|
Chris@16
|
1139 #endif
|
Chris@16
|
1140
|
Chris@16
|
1141
|