[857] | 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(BOOST_MSVC, < 1300) \
|
---|
| 26 | || BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003))
|
---|
| 27 | //
|
---|
| 28 | // Borland C++ Builder 6, and Visual C++ 6,
|
---|
| 29 | // can't cope with the array template constructor
|
---|
| 30 | // so we have a template member that will accept any type as
|
---|
| 31 | // argument, and then assert that is really is an array:
|
---|
| 32 | //
|
---|
| 33 | #include <boost/static_assert.hpp>
|
---|
| 34 | #include <boost/type_traits/is_array.hpp>
|
---|
| 35 | #endif
|
---|
| 36 |
|
---|
| 37 | namespace boost{
|
---|
| 38 |
|
---|
| 39 | #ifdef BOOST_HAS_ABI_HEADERS
|
---|
| 40 | # include BOOST_ABI_PREFIX
|
---|
| 41 | #endif
|
---|
| 42 | #if BOOST_WORKAROUND(BOOST_MSVC, > 1300)
|
---|
| 43 | # pragma warning(push)
|
---|
| 44 | # pragma warning(disable:4700)
|
---|
| 45 | #endif
|
---|
| 46 |
|
---|
| 47 | template <class BidirectionalIterator,
|
---|
| 48 | class charT,
|
---|
| 49 | class traits>
|
---|
| 50 | class regex_token_iterator_implementation
|
---|
| 51 | {
|
---|
| 52 | typedef basic_regex<charT, traits> regex_type;
|
---|
| 53 | typedef sub_match<BidirectionalIterator> value_type;
|
---|
| 54 |
|
---|
| 55 | match_results<BidirectionalIterator> what; // current match
|
---|
| 56 | BidirectionalIterator base; // start of search area
|
---|
| 57 | BidirectionalIterator end; // end of search area
|
---|
| 58 | const regex_type re; // the expression
|
---|
| 59 | match_flag_type flags; // match flags
|
---|
| 60 | value_type result; // the current string result
|
---|
| 61 | int N; // the current sub-expression being enumerated
|
---|
| 62 | std::vector<int> subs; // the sub-expressions to enumerate
|
---|
| 63 |
|
---|
| 64 | public:
|
---|
| 65 | regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, int sub, match_flag_type f)
|
---|
| 66 | : end(last), re(*p), flags(f){ subs.push_back(sub); }
|
---|
| 67 | regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const std::vector<int>& v, match_flag_type f)
|
---|
| 68 | : end(last), re(*p), flags(f), subs(v){}
|
---|
| 69 | #if !BOOST_WORKAROUND(__HP_aCC, BOOST_TESTED_AT(55500))
|
---|
| 70 | #if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\
|
---|
| 71 | || BOOST_WORKAROUND(BOOST_MSVC, < 1300) \
|
---|
| 72 | || BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) \
|
---|
| 73 | || BOOST_WORKAROUND(__HP_aCC, BOOST_TESTED_AT(55500))
|
---|
| 74 | template <class T>
|
---|
| 75 | regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const T& submatches, match_flag_type f)
|
---|
| 76 | : end(last), re(*p), flags(f)
|
---|
| 77 | {
|
---|
| 78 | // assert that T really is an array:
|
---|
| 79 | BOOST_STATIC_ASSERT(::boost::is_array<T>::value);
|
---|
| 80 | const std::size_t array_size = sizeof(T) / sizeof(submatches[0]);
|
---|
| 81 | for(std::size_t i = 0; i < array_size; ++i)
|
---|
| 82 | {
|
---|
| 83 | subs.push_back(submatches[i]);
|
---|
| 84 | }
|
---|
| 85 | }
|
---|
| 86 | #else
|
---|
| 87 | template <std::size_t CN>
|
---|
| 88 | regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const int (&submatches)[CN], match_flag_type f)
|
---|
| 89 | : end(last), re(*p), flags(f)
|
---|
| 90 | {
|
---|
| 91 | for(std::size_t i = 0; i < CN; ++i)
|
---|
| 92 | {
|
---|
| 93 | subs.push_back(submatches[i]);
|
---|
| 94 | }
|
---|
| 95 | }
|
---|
| 96 | #endif
|
---|
| 97 | #endif
|
---|
| 98 | bool init(BidirectionalIterator first)
|
---|
| 99 | {
|
---|
| 100 | N = 0;
|
---|
| 101 | base = first;
|
---|
| 102 | if(regex_search(first, end, what, re, flags, base) == true)
|
---|
| 103 | {
|
---|
| 104 | N = 0;
|
---|
| 105 | result = ((subs[N] == -1) ? what.prefix() : what[(int)subs[N]]);
|
---|
| 106 | return true;
|
---|
| 107 | }
|
---|
| 108 | else if((subs[N] == -1) && (first != end))
|
---|
| 109 | {
|
---|
| 110 | result.first = first;
|
---|
| 111 | result.second = end;
|
---|
| 112 | result.matched = (first != end);
|
---|
| 113 | return true;
|
---|
| 114 | }
|
---|
| 115 | return false;
|
---|
| 116 | }
|
---|
| 117 | bool compare(const regex_token_iterator_implementation& that)
|
---|
| 118 | {
|
---|
| 119 | if(this == &that) return true;
|
---|
| 120 | return (&re.get_data() == &that.re.get_data())
|
---|
| 121 | && (end == that.end)
|
---|
| 122 | && (flags == that.flags)
|
---|
| 123 | && (N == that.N)
|
---|
| 124 | && (what[0].first == that.what[0].first)
|
---|
| 125 | && (what[0].second == that.what[0].second);
|
---|
| 126 | }
|
---|
| 127 | const value_type& get()
|
---|
| 128 | { return result; }
|
---|
| 129 | bool next()
|
---|
| 130 | {
|
---|
| 131 | if(N == -1)
|
---|
| 132 | return false;
|
---|
| 133 | if(N+1 < (int)subs.size())
|
---|
| 134 | {
|
---|
| 135 | ++N;
|
---|
| 136 | result =((subs[N] == -1) ? what.prefix() : what[subs[N]]);
|
---|
| 137 | return true;
|
---|
| 138 | }
|
---|
| 139 | //if(what.prefix().first != what[0].second)
|
---|
| 140 | // flags |= /*match_prev_avail |*/ regex_constants::match_not_bob;
|
---|
| 141 | BidirectionalIterator last_end(what[0].second);
|
---|
| 142 | if(regex_search(last_end, end, what, re, ((what[0].first == what[0].second) ? flags | regex_constants::match_not_initial_null : flags), base))
|
---|
| 143 | {
|
---|
| 144 | N =0;
|
---|
| 145 | result =((subs[N] == -1) ? what.prefix() : what[subs[N]]);
|
---|
| 146 | return true;
|
---|
| 147 | }
|
---|
| 148 | else if((last_end != end) && (subs[0] == -1))
|
---|
| 149 | {
|
---|
| 150 | N =-1;
|
---|
| 151 | result.first = last_end;
|
---|
| 152 | result.second = end;
|
---|
| 153 | result.matched = (last_end != end);
|
---|
| 154 | return true;
|
---|
| 155 | }
|
---|
| 156 | return false;
|
---|
| 157 | }
|
---|
| 158 | private:
|
---|
| 159 | regex_token_iterator_implementation& operator=(const regex_token_iterator_implementation&);
|
---|
| 160 | };
|
---|
| 161 |
|
---|
| 162 | template <class BidirectionalIterator,
|
---|
| 163 | class charT = BOOST_DEDUCED_TYPENAME re_detail::regex_iterator_traits<BidirectionalIterator>::value_type,
|
---|
| 164 | class traits = regex_traits<charT> >
|
---|
| 165 | class regex_token_iterator
|
---|
| 166 | #ifndef BOOST_NO_STD_ITERATOR
|
---|
| 167 | : public std::iterator<
|
---|
| 168 | std::forward_iterator_tag,
|
---|
| 169 | sub_match<BidirectionalIterator>,
|
---|
| 170 | typename re_detail::regex_iterator_traits<BidirectionalIterator>::difference_type,
|
---|
| 171 | const sub_match<BidirectionalIterator>*,
|
---|
| 172 | const sub_match<BidirectionalIterator>& >
|
---|
| 173 | #endif
|
---|
| 174 | {
|
---|
| 175 | private:
|
---|
| 176 | typedef regex_token_iterator_implementation<BidirectionalIterator, charT, traits> impl;
|
---|
| 177 | typedef shared_ptr<impl> pimpl;
|
---|
| 178 | public:
|
---|
| 179 | typedef basic_regex<charT, traits> regex_type;
|
---|
| 180 | typedef sub_match<BidirectionalIterator> value_type;
|
---|
| 181 | typedef typename re_detail::regex_iterator_traits<BidirectionalIterator>::difference_type
|
---|
| 182 | difference_type;
|
---|
| 183 | typedef const value_type* pointer;
|
---|
| 184 | typedef const value_type& reference;
|
---|
| 185 | typedef std::forward_iterator_tag iterator_category;
|
---|
| 186 |
|
---|
| 187 | regex_token_iterator(){}
|
---|
| 188 | regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,
|
---|
| 189 | int submatch = 0, match_flag_type m = match_default)
|
---|
| 190 | : pdata(new impl(&re, b, submatch, m))
|
---|
| 191 | {
|
---|
| 192 | if(!pdata->init(a))
|
---|
| 193 | pdata.reset();
|
---|
| 194 | }
|
---|
| 195 | regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,
|
---|
| 196 | const std::vector<int>& submatches, match_flag_type m = match_default)
|
---|
| 197 | : pdata(new impl(&re, b, submatches, m))
|
---|
| 198 | {
|
---|
| 199 | if(!pdata->init(a))
|
---|
| 200 | pdata.reset();
|
---|
| 201 | }
|
---|
| 202 | #if !BOOST_WORKAROUND(__HP_aCC, BOOST_TESTED_AT(55500))
|
---|
| 203 | #if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\
|
---|
| 204 | || BOOST_WORKAROUND(BOOST_MSVC, < 1300) \
|
---|
| 205 | || BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) \
|
---|
| 206 | || BOOST_WORKAROUND(__HP_aCC, BOOST_TESTED_AT(55500))
|
---|
| 207 | template <class T>
|
---|
| 208 | regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,
|
---|
| 209 | const T& submatches, match_flag_type m = match_default)
|
---|
| 210 | : pdata(new impl(&re, b, submatches, m))
|
---|
| 211 | {
|
---|
| 212 | if(!pdata->init(a))
|
---|
| 213 | pdata.reset();
|
---|
| 214 | }
|
---|
| 215 | #else
|
---|
| 216 | template <std::size_t N>
|
---|
| 217 | regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,
|
---|
| 218 | const int (&submatches)[N], match_flag_type m = match_default)
|
---|
| 219 | : pdata(new impl(&re, b, submatches, m))
|
---|
| 220 | {
|
---|
| 221 | if(!pdata->init(a))
|
---|
| 222 | pdata.reset();
|
---|
| 223 | }
|
---|
| 224 | #endif
|
---|
| 225 | #endif
|
---|
| 226 | regex_token_iterator(const regex_token_iterator& that)
|
---|
| 227 | : pdata(that.pdata) {}
|
---|
| 228 | regex_token_iterator& operator=(const regex_token_iterator& that)
|
---|
| 229 | {
|
---|
| 230 | pdata = that.pdata;
|
---|
| 231 | return *this;
|
---|
| 232 | }
|
---|
| 233 | bool operator==(const regex_token_iterator& that)const
|
---|
| 234 | {
|
---|
| 235 | if((pdata.get() == 0) || (that.pdata.get() == 0))
|
---|
| 236 | return pdata.get() == that.pdata.get();
|
---|
| 237 | return pdata->compare(*(that.pdata.get()));
|
---|
| 238 | }
|
---|
| 239 | bool operator!=(const regex_token_iterator& that)const
|
---|
| 240 | { return !(*this == that); }
|
---|
| 241 | const value_type& operator*()const
|
---|
| 242 | { return pdata->get(); }
|
---|
| 243 | const value_type* operator->()const
|
---|
| 244 | { return &(pdata->get()); }
|
---|
| 245 | regex_token_iterator& operator++()
|
---|
| 246 | {
|
---|
| 247 | cow();
|
---|
| 248 | if(0 == pdata->next())
|
---|
| 249 | {
|
---|
| 250 | pdata.reset();
|
---|
| 251 | }
|
---|
| 252 | return *this;
|
---|
| 253 | }
|
---|
| 254 | regex_token_iterator operator++(int)
|
---|
| 255 | {
|
---|
| 256 | regex_token_iterator result(*this);
|
---|
| 257 | ++(*this);
|
---|
| 258 | return result;
|
---|
| 259 | }
|
---|
| 260 | private:
|
---|
| 261 |
|
---|
| 262 | pimpl pdata;
|
---|
| 263 |
|
---|
| 264 | void cow()
|
---|
| 265 | {
|
---|
| 266 | // copy-on-write
|
---|
| 267 | if(pdata.get() && !pdata.unique())
|
---|
| 268 | {
|
---|
| 269 | pdata.reset(new impl(*(pdata.get())));
|
---|
| 270 | }
|
---|
| 271 | }
|
---|
| 272 | };
|
---|
| 273 |
|
---|
| 274 | typedef regex_token_iterator<const char*> cregex_token_iterator;
|
---|
| 275 | typedef regex_token_iterator<std::string::const_iterator> sregex_token_iterator;
|
---|
| 276 | #ifndef BOOST_NO_WREGEX
|
---|
| 277 | typedef regex_token_iterator<const wchar_t*> wcregex_token_iterator;
|
---|
| 278 | typedef regex_token_iterator<std::wstring::const_iterator> wsregex_token_iterator;
|
---|
| 279 | #endif
|
---|
| 280 |
|
---|
| 281 | template <class charT, class traits>
|
---|
| 282 | 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)
|
---|
| 283 | {
|
---|
| 284 | return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m);
|
---|
| 285 | }
|
---|
| 286 | template <class charT, class traits, class ST, class SA>
|
---|
| 287 | 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)
|
---|
| 288 | {
|
---|
| 289 | return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m);
|
---|
| 290 | }
|
---|
| 291 | #if !BOOST_WORKAROUND(BOOST_MSVC, < 1300)
|
---|
| 292 | template <class charT, class traits, std::size_t N>
|
---|
| 293 | 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)
|
---|
| 294 | {
|
---|
| 295 | return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m);
|
---|
| 296 | }
|
---|
| 297 | template <class charT, class traits, class ST, class SA, std::size_t N>
|
---|
| 298 | 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)
|
---|
| 299 | {
|
---|
| 300 | return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m);
|
---|
| 301 | }
|
---|
| 302 | #endif
|
---|
| 303 | template <class charT, class traits>
|
---|
| 304 | 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)
|
---|
| 305 | {
|
---|
| 306 | return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m);
|
---|
| 307 | }
|
---|
| 308 | template <class charT, class traits, class ST, class SA>
|
---|
| 309 | 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)
|
---|
| 310 | {
|
---|
| 311 | return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m);
|
---|
| 312 | }
|
---|
| 313 |
|
---|
| 314 | #if BOOST_WORKAROUND(BOOST_MSVC, == 1310)
|
---|
| 315 | # pragma warning(pop)
|
---|
| 316 | #endif
|
---|
| 317 | #ifdef BOOST_HAS_ABI_HEADERS
|
---|
| 318 | # include BOOST_ABI_SUFFIX
|
---|
| 319 | #endif
|
---|
| 320 |
|
---|
| 321 | } // namespace boost
|
---|
| 322 |
|
---|
| 323 | #endif // BOOST_REGEX_V4_REGEX_TOKEN_ITERATOR_HPP
|
---|
| 324 |
|
---|
| 325 |
|
---|
| 326 |
|
---|
| 327 |
|
---|