]>
Commit | Line | Data |
---|---|---|
1 | /* | |
2 | * | |
3 | * Copyright (c) 2003 | |
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 regex_token_iterator.hpp | |
15 | * VERSION see <boost/version.hpp> | |
16 | * DESCRIPTION: Provides regex_token_iterator implementation. | |
17 | */ | |
18 | ||
19 | #ifndef BOOST_REGEX_V4_REGEX_TOKEN_ITERATOR_HPP | |
20 | #define BOOST_REGEX_V4_REGEX_TOKEN_ITERATOR_HPP | |
21 | ||
22 | #include <boost/shared_ptr.hpp> | |
23 | #include <boost/detail/workaround.hpp> | |
24 | #if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\ | |
25 | || BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) | |
26 | // | |
27 | // Borland C++ Builder 6, and Visual C++ 6, | |
28 | // can't cope with the array template constructor | |
29 | // so we have a template member that will accept any type as | |
30 | // argument, and then assert that is really is an array: | |
31 | // | |
32 | #include <boost/static_assert.hpp> | |
33 | #include <boost/type_traits/is_array.hpp> | |
34 | #endif | |
35 | ||
36 | namespace boost{ | |
37 | ||
38 | #ifdef BOOST_MSVC | |
39 | #pragma warning(push) | |
40 | #pragma warning(disable: 4103) | |
41 | #endif | |
42 | #ifdef BOOST_HAS_ABI_HEADERS | |
43 | # include BOOST_ABI_PREFIX | |
44 | #endif | |
45 | #ifdef BOOST_MSVC | |
46 | #pragma warning(pop) | |
47 | #pragma warning(push) | |
48 | #pragma warning(disable:4700) | |
49 | #endif | |
50 | ||
51 | template <class BidirectionalIterator, | |
52 | class charT, | |
53 | class traits> | |
54 | class regex_token_iterator_implementation | |
55 | { | |
56 | typedef basic_regex<charT, traits> regex_type; | |
57 | typedef sub_match<BidirectionalIterator> value_type; | |
58 | ||
59 | match_results<BidirectionalIterator> what; // current match | |
60 | BidirectionalIterator base; // start of search area | |
61 | BidirectionalIterator end; // end of search area | |
62 | const regex_type re; // the expression | |
63 | match_flag_type flags; // match flags | |
64 | value_type result; // the current string result | |
65 | int N; // the current sub-expression being enumerated | |
66 | std::vector<int> subs; // the sub-expressions to enumerate | |
67 | ||
68 | public: | |
69 | regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, int sub, match_flag_type f) | |
70 | : end(last), re(*p), flags(f){ subs.push_back(sub); } | |
71 | regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const std::vector<int>& v, match_flag_type f) | |
72 | : end(last), re(*p), flags(f), subs(v){} | |
73 | #if !BOOST_WORKAROUND(__HP_aCC, < 60700) | |
74 | #if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\ | |
75 | || BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) \ | |
76 | || BOOST_WORKAROUND(__HP_aCC, < 60700) | |
77 | template <class T> | |
78 | regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const T& submatches, match_flag_type f) | |
79 | : end(last), re(*p), flags(f) | |
80 | { | |
81 | // assert that T really is an array: | |
82 | BOOST_STATIC_ASSERT(::boost::is_array<T>::value); | |
83 | const std::size_t array_size = sizeof(T) / sizeof(submatches[0]); | |
84 | for(std::size_t i = 0; i < array_size; ++i) | |
85 | { | |
86 | subs.push_back(submatches[i]); | |
87 | } | |
88 | } | |
89 | #else | |
90 | template <std::size_t CN> | |
91 | regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const int (&submatches)[CN], match_flag_type f) | |
92 | : end(last), re(*p), flags(f) | |
93 | { | |
94 | for(std::size_t i = 0; i < CN; ++i) | |
95 | { | |
96 | subs.push_back(submatches[i]); | |
97 | } | |
98 | } | |
99 | #endif | |
100 | #endif | |
101 | bool init(BidirectionalIterator first) | |
102 | { | |
103 | N = 0; | |
104 | base = first; | |
105 | if(regex_search(first, end, what, re, flags, base) == true) | |
106 | { | |
107 | N = 0; | |
108 | result = ((subs[N] == -1) ? what.prefix() : what[(int)subs[N]]); | |
109 | return true; | |
110 | } | |
111 | else if((subs[N] == -1) && (first != end)) | |
112 | { | |
113 | result.first = first; | |
114 | result.second = end; | |
115 | result.matched = (first != end); | |
116 | N = -1; | |
117 | return true; | |
118 | } | |
119 | return false; | |
120 | } | |
121 | bool compare(const regex_token_iterator_implementation& that) | |
122 | { | |
123 | if(this == &that) return true; | |
124 | return (&re.get_data() == &that.re.get_data()) | |
125 | && (end == that.end) | |
126 | && (flags == that.flags) | |
127 | && (N == that.N) | |
128 | && (what[0].first == that.what[0].first) | |
129 | && (what[0].second == that.what[0].second); | |
130 | } | |
131 | const value_type& get() | |
132 | { return result; } | |
133 | bool next() | |
134 | { | |
135 | if(N == -1) | |
136 | return false; | |
137 | if(N+1 < (int)subs.size()) | |
138 | { | |
139 | ++N; | |
140 | result =((subs[N] == -1) ? what.prefix() : what[subs[N]]); | |
141 | return true; | |
142 | } | |
143 | //if(what.prefix().first != what[0].second) | |
144 | // flags |= /*match_prev_avail |*/ regex_constants::match_not_bob; | |
145 | BidirectionalIterator last_end(what[0].second); | |
146 | if(regex_search(last_end, end, what, re, ((what[0].first == what[0].second) ? flags | regex_constants::match_not_initial_null : flags), base)) | |
147 | { | |
148 | N =0; | |
149 | result =((subs[N] == -1) ? what.prefix() : what[subs[N]]); | |
150 | return true; | |
151 | } | |
152 | else if((last_end != end) && (subs[0] == -1)) | |
153 | { | |
154 | N =-1; | |
155 | result.first = last_end; | |
156 | result.second = end; | |
157 | result.matched = (last_end != end); | |
158 | return true; | |
159 | } | |
160 | return false; | |
161 | } | |
162 | private: | |
163 | regex_token_iterator_implementation& operator=(const regex_token_iterator_implementation&); | |
164 | }; | |
165 | ||
166 | template <class BidirectionalIterator, | |
167 | class charT = BOOST_DEDUCED_TYPENAME BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::value_type, | |
168 | class traits = regex_traits<charT> > | |
169 | class regex_token_iterator | |
170 | #ifndef BOOST_NO_STD_ITERATOR | |
171 | : public std::iterator< | |
172 | std::forward_iterator_tag, | |
173 | sub_match<BidirectionalIterator>, | |
174 | typename BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::difference_type, | |
175 | const sub_match<BidirectionalIterator>*, | |
176 | const sub_match<BidirectionalIterator>& > | |
177 | #endif | |
178 | { | |
179 | private: | |
180 | typedef regex_token_iterator_implementation<BidirectionalIterator, charT, traits> impl; | |
181 | typedef shared_ptr<impl> pimpl; | |
182 | public: | |
183 | typedef basic_regex<charT, traits> regex_type; | |
184 | typedef sub_match<BidirectionalIterator> value_type; | |
185 | typedef typename BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::difference_type | |
186 | difference_type; | |
187 | typedef const value_type* pointer; | |
188 | typedef const value_type& reference; | |
189 | typedef std::forward_iterator_tag iterator_category; | |
190 | ||
191 | regex_token_iterator(){} | |
192 | regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re, | |
193 | int submatch = 0, match_flag_type m = match_default) | |
194 | : pdata(new impl(&re, b, submatch, m)) | |
195 | { | |
196 | if(!pdata->init(a)) | |
197 | pdata.reset(); | |
198 | } | |
199 | regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re, | |
200 | const std::vector<int>& submatches, match_flag_type m = match_default) | |
201 | : pdata(new impl(&re, b, submatches, m)) | |
202 | { | |
203 | if(!pdata->init(a)) | |
204 | pdata.reset(); | |
205 | } | |
206 | #if !BOOST_WORKAROUND(__HP_aCC, < 60700) | |
207 | #if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\ | |
208 | || BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) \ | |
209 | || BOOST_WORKAROUND(__HP_aCC, < 60700) | |
210 | template <class T> | |
211 | regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re, | |
212 | const T& submatches, match_flag_type m = match_default) | |
213 | : pdata(new impl(&re, b, submatches, m)) | |
214 | { | |
215 | if(!pdata->init(a)) | |
216 | pdata.reset(); | |
217 | } | |
218 | #else | |
219 | template <std::size_t N> | |
220 | regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re, | |
221 | const int (&submatches)[N], match_flag_type m = match_default) | |
222 | : pdata(new impl(&re, b, submatches, m)) | |
223 | { | |
224 | if(!pdata->init(a)) | |
225 | pdata.reset(); | |
226 | } | |
227 | #endif | |
228 | #endif | |
229 | regex_token_iterator(const regex_token_iterator& that) | |
230 | : pdata(that.pdata) {} | |
231 | regex_token_iterator& operator=(const regex_token_iterator& that) | |
232 | { | |
233 | pdata = that.pdata; | |
234 | return *this; | |
235 | } | |
236 | bool operator==(const regex_token_iterator& that)const | |
237 | { | |
238 | if((pdata.get() == 0) || (that.pdata.get() == 0)) | |
239 | return pdata.get() == that.pdata.get(); | |
240 | return pdata->compare(*(that.pdata.get())); | |
241 | } | |
242 | bool operator!=(const regex_token_iterator& that)const | |
243 | { return !(*this == that); } | |
244 | const value_type& operator*()const | |
245 | { return pdata->get(); } | |
246 | const value_type* operator->()const | |
247 | { return &(pdata->get()); } | |
248 | regex_token_iterator& operator++() | |
249 | { | |
250 | cow(); | |
251 | if(0 == pdata->next()) | |
252 | { | |
253 | pdata.reset(); | |
254 | } | |
255 | return *this; | |
256 | } | |
257 | regex_token_iterator operator++(int) | |
258 | { | |
259 | regex_token_iterator result(*this); | |
260 | ++(*this); | |
261 | return result; | |
262 | } | |
263 | private: | |
264 | ||
265 | pimpl pdata; | |
266 | ||
267 | void cow() | |
268 | { | |
269 | // copy-on-write | |
270 | if(pdata.get() && !pdata.unique()) | |
271 | { | |
272 | pdata.reset(new impl(*(pdata.get()))); | |
273 | } | |
274 | } | |
275 | }; | |
276 | ||
277 | typedef regex_token_iterator<const char*> cregex_token_iterator; | |
278 | typedef regex_token_iterator<std::string::const_iterator> sregex_token_iterator; | |
279 | #ifndef BOOST_NO_WREGEX | |
280 | typedef regex_token_iterator<const wchar_t*> wcregex_token_iterator; | |
281 | typedef regex_token_iterator<std::wstring::const_iterator> wsregex_token_iterator; | |
282 | #endif | |
283 | ||
284 | template <class charT, class traits> | |
285 | inline regex_token_iterator<const charT*, charT, traits> make_regex_token_iterator(const charT* p, const basic_regex<charT, traits>& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default) | |
286 | { | |
287 | return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m); | |
288 | } | |
289 | template <class charT, class traits, class ST, class SA> | |
290 | inline regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits> make_regex_token_iterator(const std::basic_string<charT, ST, SA>& p, const basic_regex<charT, traits>& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default) | |
291 | { | |
292 | return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m); | |
293 | } | |
294 | template <class charT, class traits, std::size_t N> | |
295 | inline regex_token_iterator<const charT*, charT, traits> make_regex_token_iterator(const charT* p, const basic_regex<charT, traits>& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default) | |
296 | { | |
297 | return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m); | |
298 | } | |
299 | template <class charT, class traits, class ST, class SA, std::size_t N> | |
300 | inline regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits> make_regex_token_iterator(const std::basic_string<charT, ST, SA>& p, const basic_regex<charT, traits>& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default) | |
301 | { | |
302 | return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m); | |
303 | } | |
304 | template <class charT, class traits> | |
305 | inline regex_token_iterator<const charT*, charT, traits> make_regex_token_iterator(const charT* p, const basic_regex<charT, traits>& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default) | |
306 | { | |
307 | return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m); | |
308 | } | |
309 | template <class charT, class traits, class ST, class SA> | |
310 | inline regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits> make_regex_token_iterator(const std::basic_string<charT, ST, SA>& p, const basic_regex<charT, traits>& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default) | |
311 | { | |
312 | return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m); | |
313 | } | |
314 | ||
315 | #ifdef BOOST_MSVC | |
316 | #pragma warning(pop) | |
317 | #pragma warning(push) | |
318 | #pragma warning(disable: 4103) | |
319 | #endif | |
320 | #ifdef BOOST_HAS_ABI_HEADERS | |
321 | # include BOOST_ABI_SUFFIX | |
322 | #endif | |
323 | #ifdef BOOST_MSVC | |
324 | #pragma warning(pop) | |
325 | #endif | |
326 | ||
327 | } // namespace boost | |
328 | ||
329 | #endif // BOOST_REGEX_V4_REGEX_TOKEN_ITERATOR_HPP | |
330 | ||
331 | ||
332 | ||
333 |