[857] | 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 w32_regex_traits.hpp
|
---|
| 15 | * VERSION see <boost/version.hpp>
|
---|
| 16 | * DESCRIPTION: Declares regular expression traits class w32_regex_traits.
|
---|
| 17 | */
|
---|
| 18 |
|
---|
| 19 | #ifndef BOOST_W32_REGEX_TRAITS_HPP_INCLUDED
|
---|
| 20 | #define BOOST_W32_REGEX_TRAITS_HPP_INCLUDED
|
---|
| 21 |
|
---|
| 22 | #ifndef BOOST_RE_PAT_EXCEPT_HPP
|
---|
| 23 | #include <boost/regex/pattern_except.hpp>
|
---|
| 24 | #endif
|
---|
| 25 | #ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED
|
---|
| 26 | #include <boost/regex/v4/regex_traits_defaults.hpp>
|
---|
| 27 | #endif
|
---|
| 28 | #ifdef BOOST_HAS_THREADS
|
---|
| 29 | #include <boost/regex/pending/static_mutex.hpp>
|
---|
| 30 | #endif
|
---|
| 31 | #ifndef BOOST_REGEX_PRIMARY_TRANSFORM
|
---|
| 32 | #include <boost/regex/v4/primary_transform.hpp>
|
---|
| 33 | #endif
|
---|
| 34 | #ifndef BOOST_REGEX_OBJECT_CACHE_HPP
|
---|
| 35 | #include <boost/regex/pending/object_cache.hpp>
|
---|
| 36 | #endif
|
---|
| 37 |
|
---|
| 38 | #ifdef BOOST_HAS_ABI_HEADERS
|
---|
| 39 | # include BOOST_ABI_PREFIX
|
---|
| 40 | #endif
|
---|
| 41 |
|
---|
| 42 | #ifdef BOOST_MSVC
|
---|
| 43 | #pragma warning(push)
|
---|
| 44 | #pragma warning(disable:4786)
|
---|
| 45 | #endif
|
---|
| 46 |
|
---|
| 47 | namespace boost{
|
---|
| 48 |
|
---|
| 49 | //
|
---|
| 50 | // forward declaration is needed by some compilers:
|
---|
| 51 | //
|
---|
| 52 | template <class charT>
|
---|
| 53 | class w32_regex_traits;
|
---|
| 54 |
|
---|
| 55 | namespace re_detail{
|
---|
| 56 |
|
---|
| 57 | //
|
---|
| 58 | // start by typedeffing the types we'll need:
|
---|
| 59 | //
|
---|
| 60 | typedef ::boost::uint32_t lcid_type; // placeholder for LCID.
|
---|
| 61 | typedef ::boost::shared_ptr<void> cat_type; // placeholder for dll HANDLE.
|
---|
| 62 |
|
---|
| 63 | //
|
---|
| 64 | // then add wrappers around the actual Win32 API's (ie implementation hiding):
|
---|
| 65 | //
|
---|
| 66 | BOOST_REGEX_DECL lcid_type BOOST_REGEX_CALL w32_get_default_locale();
|
---|
| 67 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(char, lcid_type);
|
---|
| 68 | #ifndef BOOST_NO_WREGEX
|
---|
| 69 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(wchar_t, lcid_type);
|
---|
| 70 | #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
|
---|
| 71 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(unsigned short ca, lcid_type id);
|
---|
| 72 | #endif
|
---|
| 73 | #endif
|
---|
| 74 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(char, lcid_type);
|
---|
| 75 | #ifndef BOOST_NO_WREGEX
|
---|
| 76 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(wchar_t, lcid_type);
|
---|
| 77 | #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
|
---|
| 78 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(unsigned short ca, lcid_type id);
|
---|
| 79 | #endif
|
---|
| 80 | #endif
|
---|
| 81 | BOOST_REGEX_DECL cat_type BOOST_REGEX_CALL w32_cat_open(const std::string& name);
|
---|
| 82 | BOOST_REGEX_DECL std::string BOOST_REGEX_CALL w32_cat_get(const cat_type& cat, lcid_type id, int i, const std::string& def);
|
---|
| 83 | #ifndef BOOST_NO_WREGEX
|
---|
| 84 | BOOST_REGEX_DECL std::wstring BOOST_REGEX_CALL w32_cat_get(const cat_type& cat, lcid_type id, int i, const std::wstring& def);
|
---|
| 85 | #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
|
---|
| 86 | BOOST_REGEX_DECL std::basic_string<unsigned short> BOOST_REGEX_CALL w32_cat_get(const cat_type& cat, lcid_type, int i, const std::basic_string<unsigned short>& def);
|
---|
| 87 | #endif
|
---|
| 88 | #endif
|
---|
| 89 | BOOST_REGEX_DECL std::string BOOST_REGEX_CALL w32_transform(lcid_type id, const char* p1, const char* p2);
|
---|
| 90 | #ifndef BOOST_NO_WREGEX
|
---|
| 91 | BOOST_REGEX_DECL std::wstring BOOST_REGEX_CALL w32_transform(lcid_type id, const wchar_t* p1, const wchar_t* p2);
|
---|
| 92 | #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
|
---|
| 93 | BOOST_REGEX_DECL std::basic_string<unsigned short> BOOST_REGEX_CALL w32_transform(lcid_type id, const unsigned short* p1, const unsigned short* p2);
|
---|
| 94 | #endif
|
---|
| 95 | #endif
|
---|
| 96 | BOOST_REGEX_DECL char BOOST_REGEX_CALL w32_tolower(char c, lcid_type);
|
---|
| 97 | #ifndef BOOST_NO_WREGEX
|
---|
| 98 | BOOST_REGEX_DECL wchar_t BOOST_REGEX_CALL w32_tolower(wchar_t c, lcid_type);
|
---|
| 99 | #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
|
---|
| 100 | BOOST_REGEX_DECL unsigned short BOOST_REGEX_CALL w32_tolower(unsigned short c, lcid_type id);
|
---|
| 101 | #endif
|
---|
| 102 | #endif
|
---|
| 103 | BOOST_REGEX_DECL char BOOST_REGEX_CALL w32_toupper(char c, lcid_type);
|
---|
| 104 | #ifndef BOOST_NO_WREGEX
|
---|
| 105 | BOOST_REGEX_DECL wchar_t BOOST_REGEX_CALL w32_toupper(wchar_t c, lcid_type);
|
---|
| 106 | #endif
|
---|
| 107 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type, boost::uint32_t mask, char c);
|
---|
| 108 | #ifndef BOOST_NO_WREGEX
|
---|
| 109 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type, boost::uint32_t mask, wchar_t c);
|
---|
| 110 | #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
|
---|
| 111 | BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type id, boost::uint32_t m, unsigned short c);
|
---|
| 112 | #endif
|
---|
| 113 | #endif
|
---|
| 114 | //
|
---|
| 115 | // class w32_regex_traits_base:
|
---|
| 116 | // acts as a container for locale and the facets we are using.
|
---|
| 117 | //
|
---|
| 118 | template <class charT>
|
---|
| 119 | struct w32_regex_traits_base
|
---|
| 120 | {
|
---|
| 121 | w32_regex_traits_base(lcid_type l)
|
---|
| 122 | { imbue(l); }
|
---|
| 123 | lcid_type imbue(lcid_type l);
|
---|
| 124 |
|
---|
| 125 | lcid_type m_locale;
|
---|
| 126 | };
|
---|
| 127 |
|
---|
| 128 | template <class charT>
|
---|
| 129 | inline lcid_type w32_regex_traits_base<charT>::imbue(lcid_type l)
|
---|
| 130 | {
|
---|
| 131 | lcid_type result(m_locale);
|
---|
| 132 | m_locale = l;
|
---|
| 133 | return result;
|
---|
| 134 | }
|
---|
| 135 |
|
---|
| 136 | //
|
---|
| 137 | // class w32_regex_traits_char_layer:
|
---|
| 138 | // implements methods that require specialisation for narrow characters:
|
---|
| 139 | //
|
---|
| 140 | template <class charT>
|
---|
| 141 | class w32_regex_traits_char_layer : public w32_regex_traits_base<charT>
|
---|
| 142 | {
|
---|
| 143 | typedef std::basic_string<charT> string_type;
|
---|
| 144 | typedef std::map<charT, regex_constants::syntax_type> map_type;
|
---|
| 145 | typedef typename map_type::const_iterator map_iterator_type;
|
---|
| 146 | public:
|
---|
| 147 | w32_regex_traits_char_layer(const lcid_type l);
|
---|
| 148 |
|
---|
| 149 | regex_constants::syntax_type syntax_type(charT c)const
|
---|
| 150 | {
|
---|
| 151 | map_iterator_type i = m_char_map.find(c);
|
---|
| 152 | return ((i == m_char_map.end()) ? 0 : i->second);
|
---|
| 153 | }
|
---|
| 154 | regex_constants::escape_syntax_type escape_syntax_type(charT c) const
|
---|
| 155 | {
|
---|
| 156 | map_iterator_type i = m_char_map.find(c);
|
---|
| 157 | if(i == m_char_map.end())
|
---|
| 158 | {
|
---|
| 159 | if(::boost::re_detail::w32_is_lower(c, this->m_locale)) return regex_constants::escape_type_class;
|
---|
| 160 | if(::boost::re_detail::w32_is_upper(c, this->m_locale)) return regex_constants::escape_type_not_class;
|
---|
| 161 | return 0;
|
---|
| 162 | }
|
---|
| 163 | return i->second;
|
---|
| 164 | }
|
---|
| 165 | charT tolower(charT c)const
|
---|
| 166 | {
|
---|
| 167 | return ::boost::re_detail::w32_tolower(c, this->m_locale);
|
---|
| 168 | }
|
---|
| 169 | bool isctype(boost::uint32_t mask, charT c)
|
---|
| 170 | {
|
---|
| 171 | return ::boost::re_detail::w32_is(this->m_locale, mask, c);
|
---|
| 172 | }
|
---|
| 173 |
|
---|
| 174 | private:
|
---|
| 175 | string_type get_default_message(regex_constants::syntax_type);
|
---|
| 176 | // TODO: use a hash table when available!
|
---|
| 177 | map_type m_char_map;
|
---|
| 178 | };
|
---|
| 179 |
|
---|
| 180 | template <class charT>
|
---|
| 181 | w32_regex_traits_char_layer<charT>::w32_regex_traits_char_layer(::boost::re_detail::lcid_type l)
|
---|
| 182 | : w32_regex_traits_base<charT>(l)
|
---|
| 183 | {
|
---|
| 184 | // we need to start by initialising our syntax map so we know which
|
---|
| 185 | // character is used for which purpose:
|
---|
| 186 | cat_type cat;
|
---|
| 187 | std::string cat_name(w32_regex_traits<charT>::get_catalog_name());
|
---|
| 188 | if(cat_name.size())
|
---|
| 189 | {
|
---|
| 190 | cat = ::boost::re_detail::w32_cat_open(cat_name);
|
---|
| 191 | if(!cat)
|
---|
| 192 | {
|
---|
| 193 | std::string m("Unable to open message catalog: ");
|
---|
| 194 | std::runtime_error err(m + cat_name);
|
---|
| 195 | boost::re_detail::raise_runtime_error(err);
|
---|
| 196 | }
|
---|
| 197 | }
|
---|
| 198 | //
|
---|
| 199 | // if we have a valid catalog then load our messages:
|
---|
| 200 | //
|
---|
| 201 | if(cat)
|
---|
| 202 | {
|
---|
| 203 | for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
|
---|
| 204 | {
|
---|
| 205 | string_type mss = ::boost::re_detail::w32_cat_get(cat, this->m_locale, i, get_default_message(i));
|
---|
| 206 | for(typename string_type::size_type j = 0; j < mss.size(); ++j)
|
---|
| 207 | {
|
---|
| 208 | this->m_char_map[mss[j]] = i;
|
---|
| 209 | }
|
---|
| 210 | }
|
---|
| 211 | }
|
---|
| 212 | else
|
---|
| 213 | {
|
---|
| 214 | for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
|
---|
| 215 | {
|
---|
| 216 | const char* ptr = get_default_syntax(i);
|
---|
| 217 | while(ptr && *ptr)
|
---|
| 218 | {
|
---|
| 219 | this->m_char_map[static_cast<charT>(*ptr)] = i;
|
---|
| 220 | ++ptr;
|
---|
| 221 | }
|
---|
| 222 | }
|
---|
| 223 | }
|
---|
| 224 | }
|
---|
| 225 |
|
---|
| 226 | template <class charT>
|
---|
| 227 | typename w32_regex_traits_char_layer<charT>::string_type
|
---|
| 228 | w32_regex_traits_char_layer<charT>::get_default_message(regex_constants::syntax_type i)
|
---|
| 229 | {
|
---|
| 230 | const char* ptr = get_default_syntax(i);
|
---|
| 231 | string_type result;
|
---|
| 232 | while(ptr && *ptr)
|
---|
| 233 | {
|
---|
| 234 | result.append(1, static_cast<charT>(*ptr));
|
---|
| 235 | ++ptr;
|
---|
| 236 | }
|
---|
| 237 | return result;
|
---|
| 238 | }
|
---|
| 239 |
|
---|
| 240 | //
|
---|
| 241 | // specialised version for narrow characters:
|
---|
| 242 | //
|
---|
| 243 | template <>
|
---|
| 244 | class BOOST_REGEX_DECL w32_regex_traits_char_layer<char> : public w32_regex_traits_base<char>
|
---|
| 245 | {
|
---|
| 246 | typedef std::string string_type;
|
---|
| 247 | public:
|
---|
| 248 | w32_regex_traits_char_layer(::boost::re_detail::lcid_type l)
|
---|
| 249 | : w32_regex_traits_base<char>(l)
|
---|
| 250 | {
|
---|
| 251 | init();
|
---|
| 252 | }
|
---|
| 253 |
|
---|
| 254 | regex_constants::syntax_type syntax_type(char c)const
|
---|
| 255 | {
|
---|
| 256 | return m_char_map[static_cast<unsigned char>(c)];
|
---|
| 257 | }
|
---|
| 258 | regex_constants::escape_syntax_type escape_syntax_type(char c) const
|
---|
| 259 | {
|
---|
| 260 | return m_char_map[static_cast<unsigned char>(c)];
|
---|
| 261 | }
|
---|
| 262 | char tolower(char c)const
|
---|
| 263 | {
|
---|
| 264 | return m_lower_map[static_cast<unsigned char>(c)];
|
---|
| 265 | }
|
---|
| 266 | bool isctype(boost::uint32_t mask, char c)
|
---|
| 267 | {
|
---|
| 268 | return m_type_map[static_cast<unsigned char>(c)] & mask;
|
---|
| 269 | }
|
---|
| 270 |
|
---|
| 271 | private:
|
---|
| 272 | regex_constants::syntax_type m_char_map[1u << CHAR_BIT];
|
---|
| 273 | char m_lower_map[1u << CHAR_BIT];
|
---|
| 274 | boost::uint16_t m_type_map[1u << CHAR_BIT];
|
---|
| 275 | void init();
|
---|
| 276 | };
|
---|
| 277 |
|
---|
| 278 | //
|
---|
| 279 | // class w32_regex_traits_implementation:
|
---|
| 280 | // provides pimpl implementation for w32_regex_traits.
|
---|
| 281 | //
|
---|
| 282 | template <class charT>
|
---|
| 283 | class w32_regex_traits_implementation : public w32_regex_traits_char_layer<charT>
|
---|
| 284 | {
|
---|
| 285 | public:
|
---|
| 286 | typedef typename w32_regex_traits<charT>::char_class_type char_class_type;
|
---|
| 287 | BOOST_STATIC_CONSTANT(char_class_type, mask_word = 0x0400); // must be C1_DEFINED << 1
|
---|
| 288 | BOOST_STATIC_CONSTANT(char_class_type, mask_unicode = 0x0800); // must be C1_DEFINED << 2
|
---|
| 289 | BOOST_STATIC_CONSTANT(char_class_type, mask_base = 0x3ff); // all the masks used by the CT_CTYPE1 group
|
---|
| 290 |
|
---|
| 291 | typedef std::basic_string<charT> string_type;
|
---|
| 292 | typedef charT char_type;
|
---|
| 293 | w32_regex_traits_implementation(::boost::re_detail::lcid_type l);
|
---|
| 294 | std::string error_string(regex_constants::error_type n) const
|
---|
| 295 | {
|
---|
| 296 | if(!m_error_strings.empty())
|
---|
| 297 | {
|
---|
| 298 | std::map<int, std::string>::const_iterator p = m_error_strings.find(n);
|
---|
| 299 | return (p == m_error_strings.end()) ? std::string(get_default_error_string(n)) : p->second;
|
---|
| 300 | }
|
---|
| 301 | return get_default_error_string(n);
|
---|
| 302 | }
|
---|
| 303 | char_class_type lookup_classname(const charT* p1, const charT* p2) const
|
---|
| 304 | {
|
---|
| 305 | char_class_type result = lookup_classname_imp(p1, p2);
|
---|
| 306 | if(result == 0)
|
---|
| 307 | {
|
---|
| 308 | typedef typename string_type::size_type size_type;
|
---|
| 309 | string_type temp(p1, p2);
|
---|
| 310 | for(size_type i = 0; i < temp.size(); ++i)
|
---|
| 311 | temp[i] = this->tolower(temp[i]);
|
---|
| 312 | result = lookup_classname_imp(&*temp.begin(), &*temp.begin() + temp.size());
|
---|
| 313 | }
|
---|
| 314 | return result;
|
---|
| 315 | }
|
---|
| 316 | string_type lookup_collatename(const charT* p1, const charT* p2) const;
|
---|
| 317 | string_type transform_primary(const charT* p1, const charT* p2) const;
|
---|
| 318 | string_type transform(const charT* p1, const charT* p2) const
|
---|
| 319 | {
|
---|
| 320 | return ::boost::re_detail::w32_transform(this->m_locale, p1, p2);
|
---|
| 321 | }
|
---|
| 322 | private:
|
---|
| 323 | std::map<int, std::string> m_error_strings; // error messages indexed by numberic ID
|
---|
| 324 | std::map<string_type, char_class_type> m_custom_class_names; // character class names
|
---|
| 325 | std::map<string_type, string_type> m_custom_collate_names; // collating element names
|
---|
| 326 | unsigned m_collate_type; // the form of the collation string
|
---|
| 327 | charT m_collate_delim; // the collation group delimiter
|
---|
| 328 | //
|
---|
| 329 | // helpers:
|
---|
| 330 | //
|
---|
| 331 | char_class_type lookup_classname_imp(const charT* p1, const charT* p2) const;
|
---|
| 332 | };
|
---|
| 333 |
|
---|
| 334 | template <class charT>
|
---|
| 335 | typename w32_regex_traits_implementation<charT>::string_type
|
---|
| 336 | w32_regex_traits_implementation<charT>::transform_primary(const charT* p1, const charT* p2) const
|
---|
| 337 | {
|
---|
| 338 | string_type result;
|
---|
| 339 | //
|
---|
| 340 | // What we do here depends upon the format of the sort key returned by
|
---|
| 341 | // sort key returned by this->transform:
|
---|
| 342 | //
|
---|
| 343 | switch(m_collate_type)
|
---|
| 344 | {
|
---|
| 345 | case sort_C:
|
---|
| 346 | case sort_unknown:
|
---|
| 347 | // the best we can do is translate to lower case, then get a regular sort key:
|
---|
| 348 | {
|
---|
| 349 | result.assign(p1, p2);
|
---|
| 350 | typedef typename string_type::size_type size_type;
|
---|
| 351 | for(size_type i = 0; i < result.size(); ++i)
|
---|
| 352 | result[i] = this->tolower(result[i]);
|
---|
| 353 | result = this->transform(&*result.begin(), &*result.begin() + result.size());
|
---|
| 354 | break;
|
---|
| 355 | }
|
---|
| 356 | case sort_fixed:
|
---|
| 357 | {
|
---|
| 358 | // get a regular sort key, and then truncate it:
|
---|
| 359 | result.assign(this->transform(p1, p2));
|
---|
| 360 | result.erase(this->m_collate_delim);
|
---|
| 361 | break;
|
---|
| 362 | }
|
---|
| 363 | case sort_delim:
|
---|
| 364 | // get a regular sort key, and then truncate everything after the delim:
|
---|
| 365 | result.assign(this->transform(p1, p2));
|
---|
| 366 | std::size_t i;
|
---|
| 367 | for(i = 0; i < result.size(); ++i)
|
---|
| 368 | {
|
---|
| 369 | if(result[i] == m_collate_delim)
|
---|
| 370 | break;
|
---|
| 371 | }
|
---|
| 372 | result.erase(i);
|
---|
| 373 | break;
|
---|
| 374 | }
|
---|
| 375 | if(result.empty())
|
---|
| 376 | result = string_type(1, charT(0));
|
---|
| 377 | return result;
|
---|
| 378 | }
|
---|
| 379 |
|
---|
| 380 | template <class charT>
|
---|
| 381 | typename w32_regex_traits_implementation<charT>::string_type
|
---|
| 382 | w32_regex_traits_implementation<charT>::lookup_collatename(const charT* p1, const charT* p2) const
|
---|
| 383 | {
|
---|
| 384 | typedef typename std::map<string_type, string_type>::const_iterator iter_type;
|
---|
| 385 | if(m_custom_collate_names.size())
|
---|
| 386 | {
|
---|
| 387 | iter_type pos = m_custom_collate_names.find(string_type(p1, p2));
|
---|
| 388 | if(pos != m_custom_collate_names.end())
|
---|
| 389 | return pos->second;
|
---|
| 390 | }
|
---|
| 391 | #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
|
---|
| 392 | && !BOOST_WORKAROUND(BOOST_MSVC, < 1300)\
|
---|
| 393 | && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
|
---|
| 394 | std::string name(p1, p2);
|
---|
| 395 | #else
|
---|
| 396 | std::string name;
|
---|
| 397 | const charT* p0 = p1;
|
---|
| 398 | while(p0 != p2)
|
---|
| 399 | name.append(1, char(*p0++));
|
---|
| 400 | #endif
|
---|
| 401 | name = lookup_default_collate_name(name);
|
---|
| 402 | #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
|
---|
| 403 | && !BOOST_WORKAROUND(BOOST_MSVC, < 1300)\
|
---|
| 404 | && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
|
---|
| 405 | if(name.size())
|
---|
| 406 | return string_type(name.begin(), name.end());
|
---|
| 407 | #else
|
---|
| 408 | if(name.size())
|
---|
| 409 | {
|
---|
| 410 | string_type result;
|
---|
| 411 | typedef std::string::const_iterator iter;
|
---|
| 412 | iter b = name.begin();
|
---|
| 413 | iter e = name.end();
|
---|
| 414 | while(b != e)
|
---|
| 415 | result.append(1, charT(*b++));
|
---|
| 416 | return result;
|
---|
| 417 | }
|
---|
| 418 | #endif
|
---|
| 419 | if(p2 - p1 == 1)
|
---|
| 420 | return string_type(1, *p1);
|
---|
| 421 | return string_type();
|
---|
| 422 | }
|
---|
| 423 |
|
---|
| 424 | template <class charT>
|
---|
| 425 | w32_regex_traits_implementation<charT>::w32_regex_traits_implementation(::boost::re_detail::lcid_type l)
|
---|
| 426 | : w32_regex_traits_char_layer<charT>(l)
|
---|
| 427 | {
|
---|
| 428 | cat_type cat;
|
---|
| 429 | std::string cat_name(w32_regex_traits<charT>::get_catalog_name());
|
---|
| 430 | if(cat_name.size())
|
---|
| 431 | {
|
---|
| 432 | cat = ::boost::re_detail::w32_cat_open(cat_name);
|
---|
| 433 | if(!cat)
|
---|
| 434 | {
|
---|
| 435 | std::string m("Unable to open message catalog: ");
|
---|
| 436 | std::runtime_error err(m + cat_name);
|
---|
| 437 | boost::re_detail::raise_runtime_error(err);
|
---|
| 438 | }
|
---|
| 439 | }
|
---|
| 440 | //
|
---|
| 441 | // if we have a valid catalog then load our messages:
|
---|
| 442 | //
|
---|
| 443 | if(cat)
|
---|
| 444 | {
|
---|
| 445 | //
|
---|
| 446 | // Error messages:
|
---|
| 447 | //
|
---|
| 448 | for(boost::regex_constants::error_type i = static_cast<boost::regex_constants::error_type>(0);
|
---|
| 449 | i <= boost::regex_constants::error_unknown;
|
---|
| 450 | i = static_cast<boost::regex_constants::error_type>(i + 1))
|
---|
| 451 | {
|
---|
| 452 | const char* p = get_default_error_string(i);
|
---|
| 453 | string_type default_message;
|
---|
| 454 | while(*p)
|
---|
| 455 | {
|
---|
| 456 | default_message.append(1, static_cast<charT>(*p));
|
---|
| 457 | ++p;
|
---|
| 458 | }
|
---|
| 459 | string_type s = ::boost::re_detail::w32_cat_get(cat, this->m_locale, i+200, default_message);
|
---|
| 460 | std::string result;
|
---|
| 461 | for(std::string::size_type j = 0; j < s.size(); ++j)
|
---|
| 462 | {
|
---|
| 463 | result.append(1, static_cast<char>(s[j]));
|
---|
| 464 | }
|
---|
| 465 | m_error_strings[i] = result;
|
---|
| 466 | }
|
---|
| 467 | //
|
---|
| 468 | // Custom class names:
|
---|
| 469 | //
|
---|
| 470 | static const char_class_type masks[14] =
|
---|
| 471 | {
|
---|
| 472 | 0x0104u, // C1_ALPHA | C1_DIGIT
|
---|
| 473 | 0x0100u, // C1_ALPHA
|
---|
| 474 | 0x0020u, // C1_CNTRL
|
---|
| 475 | 0x0004u, // C1_DIGIT
|
---|
| 476 | (~(0x0020u|0x0008u) & 0x01ffu) | 0x0400u, // not C1_CNTRL or C1_SPACE
|
---|
| 477 | 0x0002u, // C1_LOWER
|
---|
| 478 | (~0x0020u & 0x01ffu) | 0x0400, // not C1_CNTRL
|
---|
| 479 | 0x0010u, // C1_PUNCT
|
---|
| 480 | 0x0008u, // C1_SPACE
|
---|
| 481 | 0x0001u, // C1_UPPER
|
---|
| 482 | 0x0080u, // C1_XDIGIT
|
---|
| 483 | 0x0040u, // C1_BLANK
|
---|
| 484 | w32_regex_traits_implementation<charT>::mask_word,
|
---|
| 485 | w32_regex_traits_implementation<charT>::mask_unicode,
|
---|
| 486 | };
|
---|
| 487 | static const string_type null_string;
|
---|
| 488 | for(unsigned int j = 0; j <= 13; ++j)
|
---|
| 489 | {
|
---|
| 490 | string_type s(::boost::re_detail::w32_cat_get(cat, this->m_locale, j+300, null_string));
|
---|
| 491 | if(s.size())
|
---|
| 492 | this->m_custom_class_names[s] = masks[j];
|
---|
| 493 | }
|
---|
| 494 | }
|
---|
| 495 | //
|
---|
| 496 | // get the collation format used by m_pcollate:
|
---|
| 497 | //
|
---|
| 498 | m_collate_type = re_detail::find_sort_syntax(this, &m_collate_delim);
|
---|
| 499 | }
|
---|
| 500 |
|
---|
| 501 | template <class charT>
|
---|
| 502 | typename w32_regex_traits_implementation<charT>::char_class_type
|
---|
| 503 | w32_regex_traits_implementation<charT>::lookup_classname_imp(const charT* p1, const charT* p2) const
|
---|
| 504 | {
|
---|
| 505 | static const char_class_type masks[20] =
|
---|
| 506 | {
|
---|
| 507 | 0,
|
---|
| 508 | 0x0104u, // C1_ALPHA | C1_DIGIT
|
---|
| 509 | 0x0100u, // C1_ALPHA
|
---|
| 510 | 0x0040u, // C1_BLANK
|
---|
| 511 | 0x0020u, // C1_CNTRL
|
---|
| 512 | 0x0004u, // C1_DIGIT
|
---|
| 513 | 0x0004u, // C1_DIGIT
|
---|
| 514 | (~(0x0020u|0x0008u|0x0040) & 0x01ffu) | 0x0400u, // not C1_CNTRL or C1_SPACE or C1_BLANK
|
---|
| 515 | 0x0002u, // C1_LOWER
|
---|
| 516 | 0x0002u, // C1_LOWER
|
---|
| 517 | (~0x0020u & 0x01ffu) | 0x0400, // not C1_CNTRL
|
---|
| 518 | 0x0010u, // C1_PUNCT
|
---|
| 519 | 0x0008u, // C1_SPACE
|
---|
| 520 | 0x0008u, // C1_SPACE
|
---|
| 521 | 0x0001u, // C1_UPPER
|
---|
| 522 | w32_regex_traits_implementation<charT>::mask_unicode,
|
---|
| 523 | 0x0001u, // C1_UPPER
|
---|
| 524 | 0x0104u | w32_regex_traits_implementation<charT>::mask_word,
|
---|
| 525 | 0x0104u | w32_regex_traits_implementation<charT>::mask_word,
|
---|
| 526 | 0x0080u, // C1_XDIGIT
|
---|
| 527 | };
|
---|
| 528 | if(m_custom_class_names.size())
|
---|
| 529 | {
|
---|
| 530 | typedef typename std::map<std::basic_string<charT>, char_class_type>::const_iterator map_iter;
|
---|
| 531 | map_iter pos = m_custom_class_names.find(string_type(p1, p2));
|
---|
| 532 | if(pos != m_custom_class_names.end())
|
---|
| 533 | return pos->second;
|
---|
| 534 | }
|
---|
| 535 | std::size_t id = 1 + re_detail::get_default_class_id(p1, p2);
|
---|
| 536 | BOOST_ASSERT(id < sizeof(masks) / sizeof(masks[0]));
|
---|
| 537 | return masks[id];
|
---|
| 538 | }
|
---|
| 539 |
|
---|
| 540 |
|
---|
| 541 | template <class charT>
|
---|
| 542 | boost::shared_ptr<w32_regex_traits_implementation<charT> > create_w32_regex_traits(::boost::re_detail::lcid_type l BOOST_APPEND_EXPLICIT_TEMPLATE_TYPE(charT))
|
---|
| 543 | {
|
---|
| 544 | // TODO: create a cache for previously constructed objects.
|
---|
| 545 | return boost::object_cache< ::boost::re_detail::lcid_type, w32_regex_traits_implementation<charT> >::get(l, 5);
|
---|
| 546 | }
|
---|
| 547 |
|
---|
| 548 | } // re_detail
|
---|
| 549 |
|
---|
| 550 | template <class charT>
|
---|
| 551 | class w32_regex_traits
|
---|
| 552 | {
|
---|
| 553 | public:
|
---|
| 554 | typedef charT char_type;
|
---|
| 555 | typedef std::size_t size_type;
|
---|
| 556 | typedef std::basic_string<char_type> string_type;
|
---|
| 557 | typedef ::boost::re_detail::lcid_type locale_type;
|
---|
| 558 | typedef boost::uint_least32_t char_class_type;
|
---|
| 559 |
|
---|
| 560 | struct boost_extensions_tag{};
|
---|
| 561 |
|
---|
| 562 | w32_regex_traits()
|
---|
| 563 | : m_pimpl(re_detail::create_w32_regex_traits<charT>(::boost::re_detail::w32_get_default_locale()))
|
---|
| 564 | { }
|
---|
| 565 | static size_type length(const char_type* p)
|
---|
| 566 | {
|
---|
| 567 | return std::char_traits<charT>::length(p);
|
---|
| 568 | }
|
---|
| 569 | regex_constants::syntax_type syntax_type(charT c)const
|
---|
| 570 | {
|
---|
| 571 | return m_pimpl->syntax_type(c);
|
---|
| 572 | }
|
---|
| 573 | regex_constants::escape_syntax_type escape_syntax_type(charT c) const
|
---|
| 574 | {
|
---|
| 575 | return m_pimpl->escape_syntax_type(c);
|
---|
| 576 | }
|
---|
| 577 | charT translate(charT c) const
|
---|
| 578 | {
|
---|
| 579 | return c;
|
---|
| 580 | }
|
---|
| 581 | charT translate_nocase(charT c) const
|
---|
| 582 | {
|
---|
| 583 | return this->m_pimpl->tolower(c);
|
---|
| 584 | }
|
---|
| 585 | charT translate(charT c, bool icase) const
|
---|
| 586 | {
|
---|
| 587 | return icase ? this->m_pimpl->tolower(c) : c;
|
---|
| 588 | }
|
---|
| 589 | charT tolower(charT c) const
|
---|
| 590 | {
|
---|
| 591 | return this->m_pimpl->tolower(c);
|
---|
| 592 | }
|
---|
| 593 | charT toupper(charT c) const
|
---|
| 594 | {
|
---|
| 595 | return ::boost::re_detail::w32_toupper(c, this->m_pimpl->m_locale);
|
---|
| 596 | }
|
---|
| 597 | string_type transform(const charT* p1, const charT* p2) const
|
---|
| 598 | {
|
---|
| 599 | return ::boost::re_detail::w32_transform(this->m_pimpl->m_locale, p1, p2);
|
---|
| 600 | }
|
---|
| 601 | string_type transform_primary(const charT* p1, const charT* p2) const
|
---|
| 602 | {
|
---|
| 603 | return m_pimpl->transform_primary(p1, p2);
|
---|
| 604 | }
|
---|
| 605 | char_class_type lookup_classname(const charT* p1, const charT* p2) const
|
---|
| 606 | {
|
---|
| 607 | return m_pimpl->lookup_classname(p1, p2);
|
---|
| 608 | }
|
---|
| 609 | string_type lookup_collatename(const charT* p1, const charT* p2) const
|
---|
| 610 | {
|
---|
| 611 | return m_pimpl->lookup_collatename(p1, p2);
|
---|
| 612 | }
|
---|
| 613 | bool isctype(charT c, char_class_type f) const
|
---|
| 614 | {
|
---|
| 615 | if((f & re_detail::w32_regex_traits_implementation<charT>::mask_base)
|
---|
| 616 | && (this->m_pimpl->isctype(f & re_detail::w32_regex_traits_implementation<charT>::mask_base, c)))
|
---|
| 617 | return true;
|
---|
| 618 | else if((f & re_detail::w32_regex_traits_implementation<charT>::mask_unicode) && re_detail::is_extended(c))
|
---|
| 619 | return true;
|
---|
| 620 | else if((f & re_detail::w32_regex_traits_implementation<charT>::mask_word) && (c == '_'))
|
---|
| 621 | return true;
|
---|
| 622 | return false;
|
---|
| 623 | }
|
---|
| 624 | int toi(const charT*& p1, const charT* p2, int radix)const
|
---|
| 625 | {
|
---|
| 626 | return ::boost::re_detail::global_toi(p1, p2, radix, *this);
|
---|
| 627 | }
|
---|
| 628 | int value(charT c, int radix)const
|
---|
| 629 | {
|
---|
| 630 | int result = ::boost::re_detail::global_value(c);
|
---|
| 631 | return result < radix ? result : -1;
|
---|
| 632 | }
|
---|
| 633 | locale_type imbue(locale_type l)
|
---|
| 634 | {
|
---|
| 635 | ::boost::re_detail::lcid_type result(getloc());
|
---|
| 636 | m_pimpl = re_detail::create_w32_regex_traits<charT>(l);
|
---|
| 637 | return result;
|
---|
| 638 | }
|
---|
| 639 | locale_type getloc()const
|
---|
| 640 | {
|
---|
| 641 | return m_pimpl->m_locale;
|
---|
| 642 | }
|
---|
| 643 | std::string error_string(regex_constants::error_type n) const
|
---|
| 644 | {
|
---|
| 645 | return m_pimpl->error_string(n);
|
---|
| 646 | }
|
---|
| 647 |
|
---|
| 648 | //
|
---|
| 649 | // extension:
|
---|
| 650 | // set the name of the message catalog in use (defaults to "boost_regex").
|
---|
| 651 | //
|
---|
| 652 | static std::string catalog_name(const std::string& name);
|
---|
| 653 | static std::string get_catalog_name();
|
---|
| 654 |
|
---|
| 655 | private:
|
---|
| 656 | boost::shared_ptr<re_detail::w32_regex_traits_implementation<charT> > m_pimpl;
|
---|
| 657 | //
|
---|
| 658 | // catalog name handler:
|
---|
| 659 | //
|
---|
| 660 | static std::string& get_catalog_name_inst();
|
---|
| 661 |
|
---|
| 662 | #ifdef BOOST_HAS_THREADS
|
---|
| 663 | static static_mutex& get_mutex_inst();
|
---|
| 664 | #endif
|
---|
| 665 | };
|
---|
| 666 |
|
---|
| 667 | template <class charT>
|
---|
| 668 | std::string w32_regex_traits<charT>::catalog_name(const std::string& name)
|
---|
| 669 | {
|
---|
| 670 | #ifdef BOOST_HAS_THREADS
|
---|
| 671 | static_mutex::scoped_lock lk(get_mutex_inst());
|
---|
| 672 | #endif
|
---|
| 673 | std::string result(get_catalog_name_inst());
|
---|
| 674 | get_catalog_name_inst() = name;
|
---|
| 675 | return result;
|
---|
| 676 | }
|
---|
| 677 |
|
---|
| 678 | template <class charT>
|
---|
| 679 | std::string& w32_regex_traits<charT>::get_catalog_name_inst()
|
---|
| 680 | {
|
---|
| 681 | static std::string s_name;
|
---|
| 682 | return s_name;
|
---|
| 683 | }
|
---|
| 684 |
|
---|
| 685 | template <class charT>
|
---|
| 686 | std::string w32_regex_traits<charT>::get_catalog_name()
|
---|
| 687 | {
|
---|
| 688 | #ifdef BOOST_HAS_THREADS
|
---|
| 689 | static_mutex::scoped_lock lk(get_mutex_inst());
|
---|
| 690 | #endif
|
---|
| 691 | std::string result(get_catalog_name_inst());
|
---|
| 692 | return result;
|
---|
| 693 | }
|
---|
| 694 |
|
---|
| 695 | #ifdef BOOST_HAS_THREADS
|
---|
| 696 | template <class charT>
|
---|
| 697 | static_mutex& w32_regex_traits<charT>::get_mutex_inst()
|
---|
| 698 | {
|
---|
| 699 | static static_mutex s_mutex = BOOST_STATIC_MUTEX_INIT;
|
---|
| 700 | return s_mutex;
|
---|
| 701 | }
|
---|
| 702 | #endif
|
---|
| 703 |
|
---|
| 704 |
|
---|
| 705 | } // boost
|
---|
| 706 |
|
---|
| 707 | #ifdef BOOST_MSVC
|
---|
| 708 | #pragma warning(pop)
|
---|
| 709 | #endif
|
---|
| 710 |
|
---|
| 711 | #ifdef BOOST_HAS_ABI_HEADERS
|
---|
| 712 | # include BOOST_ABI_SUFFIX
|
---|
| 713 | #endif
|
---|
| 714 |
|
---|
| 715 | #endif
|
---|