boost/regex/v4/cpp_regex_traits.hpp

   1 /*
   2  *
   3  * Copyright (c) 2004
   4  * John Maddock
   5  *
   6  * Use, modification and distribution are subject to the
   7  * Boost Software License, Version 1.0. (See accompanying file
   8  * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
   9  *
  10  */
  11
  12  /*
  13   *   LOCATION:    see http://www.boost.org for most recent version.
  14   *   FILE         cpp_regex_traits.hpp
  15   *   VERSION      see <boost/version.hpp>
  16   *   DESCRIPTION: Declares regular expression traits class cpp_regex_traits.
  17   */
  18
  19 #ifndef BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED
  20 #define BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED
  21
  22 #include <boost/config.hpp>
  23
  24 #ifndef BOOST_NO_STD_LOCALE
  25
  26 #ifndef BOOST_RE_PAT_EXCEPT_HPP
  27 #include <boost/regex/pattern_except.hpp>
  28 #endif
  29 #ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED
  30 #include <boost/regex/v4/regex_traits_defaults.hpp>
  31 #endif
  32 #ifdef BOOST_HAS_THREADS
  33 #include <boost/regex/pending/static_mutex.hpp>
  34 #endif
  35 #ifndef BOOST_REGEX_PRIMARY_TRANSFORM
  36 #include <boost/regex/v4/primary_transform.hpp>
  37 #endif
  38 #ifndef BOOST_REGEX_OBJECT_CACHE_HPP
  39 #include <boost/regex/pending/object_cache.hpp>
  40 #endif
  41
  42 #include <istream>
  43 #include <ios>
  44 #include <climits>
  45
  46 #ifdef BOOST_MSVC
  47 #pragma warning(push)
  48 #pragma warning(disable: 4103)
  49 #endif
  50 #ifdef BOOST_HAS_ABI_HEADERS
  51 #  include BOOST_ABI_PREFIX
  52 #endif
  53 #ifdef BOOST_MSVC
  54 #pragma warning(pop)
  55 #endif
  56
  57 #ifdef BOOST_MSVC
  58 #pragma warning(push)
  59 #pragma warning(disable:4786)
  60 #endif
  61
  62 namespace boost{
  63
  64 //
  65 // forward declaration is needed by some compilers:
  66 //
  67 template <class charT>
  68 class cpp_regex_traits;
  69
  70 namespace re_detail{
  71
  72 //
  73 // class parser_buf:
  74 // acts as a stream buffer which wraps around a pair of pointers:
  75 //
  76 template <class charT,
  77           class traits = ::std::char_traits<charT> >
  78 class parser_buf : public ::std::basic_streambuf<charT, traits>
  79 {
  80    typedef ::std::basic_streambuf<charT, traits> base_type;
  81    typedef typename base_type::int_type int_type;
  82    typedef typename base_type::char_type char_type;
  83    typedef typename base_type::pos_type pos_type;
  84    typedef ::std::streamsize streamsize;
  85    typedef typename base_type::off_type off_type;
  86 public:
  87    parser_buf() : base_type() { setbuf(0, 0); }
  88    const charT* getnext() { return this->gptr(); }
  89 protected:
  90    std::basic_streambuf<charT, traits>* setbuf(char_type* s, streamsize n);
  91    typename parser_buf<charT, traits>::pos_type seekpos(pos_type sp, ::std::ios_base::openmode which);
  92    typename parser_buf<charT, traits>::pos_type seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which);
  93 private:
  94    parser_buf& operator=(const parser_buf&);
  95    parser_buf(const parser_buf&);
  96 };
  97
  98 template<class charT, class traits>
  99 std::basic_streambuf<charT, traits>*
 100 parser_buf<charT, traits>::setbuf(char_type* s, streamsize n)
 101 {
 102    this->setg(s, s, s + n);
 103    return this;
 104 }
 105
 106 template<class charT, class traits>
 107 typename parser_buf<charT, traits>::pos_type
 108 parser_buf<charT, traits>::seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which)
 109 {
 110    if(which & ::std::ios_base::out)
 111       return pos_type(off_type(-1));
 112    std::ptrdiff_t size = this->egptr() - this->eback();
 113    std::ptrdiff_t pos = this->gptr() - this->eback();
 114    charT* g = this->eback();
 115    switch(way)
 116    {
 117    case ::std::ios_base::beg:
 118       if((off < 0) || (off > size))
 119          return pos_type(off_type(-1));
 120       else
 121          this->setg(g, g + off, g + size);
 122       break;
 123    case ::std::ios_base::end:
 124       if((off < 0) || (off > size))
 125          return pos_type(off_type(-1));
 126       else
 127          this->setg(g, g + size - off, g + size);
 128       break;
 129    case ::std::ios_base::cur:
 130    {
 131       std::ptrdiff_t newpos = static_cast<std::ptrdiff_t>(pos + off);
 132       if((newpos < 0) || (newpos > size))
 133          return pos_type(off_type(-1));
 134       else
 135          this->setg(g, g + newpos, g + size);
 136       break;
 137    }
 138    default: ;
 139    }
 140 #ifdef BOOST_MSVC
 141 #pragma warning(push)
 142 #pragma warning(disable:4244)
 143 #endif
 144    return static_cast<pos_type>(this->gptr() - this->eback());
 145 #ifdef BOOST_MSVC
 146 #pragma warning(pop)
 147 #endif
 148 }
 149
 150 template<class charT, class traits>
 151 typename parser_buf<charT, traits>::pos_type
 152 parser_buf<charT, traits>::seekpos(pos_type sp, ::std::ios_base::openmode which)
 153 {
 154    if(which & ::std::ios_base::out)
 155       return pos_type(off_type(-1));
 156    off_type size = static_cast<off_type>(this->egptr() - this->eback());
 157    charT* g = this->eback();
 158    if(off_type(sp) <= size)
 159    {
 160       this->setg(g, g + off_type(sp), g + size);
 161    }
 162    return pos_type(off_type(-1));
 163 }
 164
 165 //
 166 // class cpp_regex_traits_base:
 167 // acts as a container for locale and the facets we are using.
 168 //
 169 template <class charT>
 170 struct cpp_regex_traits_base
 171 {
 172    cpp_regex_traits_base(const std::locale& l)
 173    { imbue(l); }
 174    std::locale imbue(const std::locale& l);
 175
 176    std::locale m_locale;
 177    std::ctype<charT> const* m_pctype;
 178 #ifndef BOOST_NO_STD_MESSAGES
 179    std::messages<charT> const* m_pmessages;
 180 #endif
 181    std::collate<charT> const* m_pcollate;
 182
 183    bool operator<(const cpp_regex_traits_base& b)const
 184    {
 185       if(m_pctype == b.m_pctype)
 186       {
 187 #ifndef BOOST_NO_STD_MESSAGES
 188          if(m_pmessages == b.m_pmessages)
 189          {
 190          }
 191          return m_pmessages < b.m_pmessages;
 192 #else
 193          return m_pcollate < b.m_pcollate;
 194 #endif
 195       }
 196       return m_pctype < b.m_pctype;
 197    }
 198    bool operator==(const cpp_regex_traits_base& b)const
 199    {
 200       return (m_pctype == b.m_pctype)
 201 #ifndef BOOST_NO_STD_MESSAGES
 202          && (m_pmessages == b.m_pmessages)
 203 #endif
 204          && (m_pcollate == b.m_pcollate);
 205    }
 206 };
 207
 208 template <class charT>
 209 std::locale cpp_regex_traits_base<charT>::imbue(const std::locale& l)
 210 {
 211    std::locale result(m_locale);
 212    m_locale = l;
 213    m_pctype = &BOOST_USE_FACET(std::ctype<charT>, l);
 214 #ifndef BOOST_NO_STD_MESSAGES
 215    m_pmessages = &BOOST_USE_FACET(std::messages<charT>, l);
 216 #endif
 217    m_pcollate = &BOOST_USE_FACET(std::collate<charT>, l);
 218    return result;
 219 }
 220
 221 //
 222 // class cpp_regex_traits_char_layer:
 223 // implements methods that require specialisation for narrow characters:
 224 //
 225 template <class charT>
 226 class cpp_regex_traits_char_layer : public cpp_regex_traits_base<charT>
 227 {
 228    typedef std::basic_string<charT> string_type;
 229    typedef std::map<charT, regex_constants::syntax_type> map_type;
 230    typedef typename map_type::const_iterator map_iterator_type;
 231 public:
 232    cpp_regex_traits_char_layer(const std::locale& l)
 233       : cpp_regex_traits_base<charT>(l)
 234    {
 235       init();
 236    }
 237    cpp_regex_traits_char_layer(const cpp_regex_traits_base<charT>& b)
 238       : cpp_regex_traits_base<charT>(b)
 239    {
 240       init();
 241    }
 242    void init();
 243
 244    regex_constants::syntax_type syntax_type(charT c)const
 245    {
 246       map_iterator_type i = m_char_map.find(c);
 247       return ((i == m_char_map.end()) ? 0 : i->second);
 248    }
 249    regex_constants::escape_syntax_type escape_syntax_type(charT c) const
 250    {
 251       map_iterator_type i = m_char_map.find(c);
 252       if(i == m_char_map.end())
 253       {
 254          if(this->m_pctype->is(std::ctype_base::lower, c)) return regex_constants::escape_type_class;
 255          if(this->m_pctype->is(std::ctype_base::upper, c)) return regex_constants::escape_type_not_class;
 256          return 0;
 257       }
 258       return i->second;
 259    }
 260
 261 private:
 262    string_type get_default_message(regex_constants::syntax_type);
 263    // TODO: use a hash table when available!
 264    map_type m_char_map;
 265 };
 266
 267 template <class charT>
 268 void cpp_regex_traits_char_layer<charT>::init()
 269 {
 270    // we need to start by initialising our syntax map so we know which
 271    // character is used for which purpose:
 272 #ifndef BOOST_NO_STD_MESSAGES
 273 #ifndef __IBMCPP__
 274    typename std::messages<charT>::catalog cat = static_cast<std::messages<char>::catalog>(-1);
 275 #else
 276    typename std::messages<charT>::catalog cat = reinterpret_cast<std::messages<char>::catalog>(-1);
 277 #endif
 278    std::string cat_name(cpp_regex_traits<charT>::get_catalog_name());
 279    if(cat_name.size())
 280    {
 281       cat = this->m_pmessages->open(
 282          cat_name,
 283          this->m_locale);
 284       if((int)cat < 0)
 285       {
 286          std::string m("Unable to open message catalog: ");
 287          std::runtime_error err(m + cat_name);
 288          boost::re_detail::raise_runtime_error(err);
 289       }
 290    }
 291    //
 292    // if we have a valid catalog then load our messages:
 293    //
 294    if((int)cat >= 0)
 295    {
 296 #ifndef BOOST_NO_EXCEPTIONS
 297       try{
 298 #endif
 299          for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
 300          {
 301             string_type mss = this->m_pmessages->get(cat, 0, i, get_default_message(i));
 302             for(typename string_type::size_type j = 0; j < mss.size(); ++j)
 303             {
 304                m_char_map[mss[j]] = i;
 305             }
 306          }
 307          this->m_pmessages->close(cat);
 308 #ifndef BOOST_NO_EXCEPTIONS
 309       }
 310       catch(...)
 311       {
 312          this->m_pmessages->close(cat);
 313          throw;
 314       }
 315 #endif
 316    }
 317    else
 318    {
 319 #endif
 320       for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
 321       {
 322          const char* ptr = get_default_syntax(i);
 323          while(ptr && *ptr)
 324          {
 325             m_char_map[this->m_pctype->widen(*ptr)] = i;
 326             ++ptr;
 327          }
 328       }
 329 #ifndef BOOST_NO_STD_MESSAGES
 330    }
 331 #endif
 332 }
 333
 334 template <class charT>
 335 typename cpp_regex_traits_char_layer<charT>::string_type
 336    cpp_regex_traits_char_layer<charT>::get_default_message(regex_constants::syntax_type i)
 337 {
 338    const char* ptr = get_default_syntax(i);
 339    string_type result;
 340    while(ptr && *ptr)
 341    {
 342       result.append(1, this->m_pctype->widen(*ptr));
 343       ++ptr;
 344    }
 345    return result;
 346 }
 347
 348 //
 349 // specialised version for narrow characters:
 350 //
 351 template <>
 352 class BOOST_REGEX_DECL cpp_regex_traits_char_layer<char> : public cpp_regex_traits_base<char>
 353 {
 354    typedef std::string string_type;
 355 public:
 356    cpp_regex_traits_char_layer(const std::locale& l)
 357    : cpp_regex_traits_base<char>(l)
 358    {
 359       init();
 360    }
 361    cpp_regex_traits_char_layer(const cpp_regex_traits_base<char>& l)
 362    : cpp_regex_traits_base<char>(l)
 363    {
 364       init();
 365    }
 366
 367    regex_constants::syntax_type syntax_type(char c)const
 368    {
 369       return m_char_map[static_cast<unsigned char>(c)];
 370    }
 371    regex_constants::escape_syntax_type escape_syntax_type(char c) const
 372    {
 373       return m_char_map[static_cast<unsigned char>(c)];
 374    }
 375
 376 private:
 377    regex_constants::syntax_type m_char_map[1u << CHAR_BIT];
 378    void init();
 379 };
 380
 381 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET
 382 enum
 383 {
 384    char_class_space=1<<0,
 385    char_class_print=1<<1,
 386    char_class_cntrl=1<<2,
 387    char_class_upper=1<<3,
 388    char_class_lower=1<<4,
 389    char_class_alpha=1<<5,
 390    char_class_digit=1<<6,
 391    char_class_punct=1<<7,
 392    char_class_xdigit=1<<8,
 393    char_class_alnum=char_class_alpha|char_class_digit,
 394    char_class_graph=char_class_alnum|char_class_punct,
 395    char_class_blank=1<<9,
 396    char_class_word=1<<10,
 397    char_class_unicode=1<<11,
 398    char_class_horizontal_space=1<<12,
 399    char_class_vertical_space=1<<13
 400 };
 401
 402 #endif
 403
 404 //
 405 // class cpp_regex_traits_implementation:
 406 // provides pimpl implementation for cpp_regex_traits.
 407 //
 408 template <class charT>
 409 class cpp_regex_traits_implementation : public cpp_regex_traits_char_layer<charT>
 410 {
 411 public:
 412    typedef typename cpp_regex_traits<charT>::char_class_type char_class_type;
 413    typedef typename std::ctype<charT>::mask                  native_mask_type;
 414 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
 415    BOOST_STATIC_CONSTANT(char_class_type, mask_blank = 1u << 24);
 416    BOOST_STATIC_CONSTANT(char_class_type, mask_word = 1u << 25);
 417    BOOST_STATIC_CONSTANT(char_class_type, mask_unicode = 1u << 26);
 418    BOOST_STATIC_CONSTANT(char_class_type, mask_horizontal = 1u << 27);
 419    BOOST_STATIC_CONSTANT(char_class_type, mask_vertical = 1u << 28);
 420 #endif
 421
 422    typedef std::basic_string<charT> string_type;
 423    typedef charT char_type;
 424    //cpp_regex_traits_implementation();
 425    cpp_regex_traits_implementation(const std::locale& l)
 426       : cpp_regex_traits_char_layer<charT>(l)
 427    {
 428       init();
 429    }
 430    cpp_regex_traits_implementation(const cpp_regex_traits_base<charT>& l)
 431       : cpp_regex_traits_char_layer<charT>(l)
 432    {
 433       init();
 434    }
 435    std::string error_string(regex_constants::error_type n) const
 436    {
 437       if(!m_error_strings.empty())
 438       {
 439          std::map<int, std::string>::const_iterator p = m_error_strings.find(n);
 440          return (p == m_error_strings.end()) ? std::string(get_default_error_string(n)) : p->second;
 441       }
 442       return get_default_error_string(n);
 443    }
 444    char_class_type lookup_classname(const charT* p1, const charT* p2) const
 445    {
 446       char_class_type result = lookup_classname_imp(p1, p2);
 447       if(result == 0)
 448       {
 449          string_type temp(p1, p2);
 450          this->m_pctype->tolower(&*temp.begin(), &*temp.begin() + temp.size());
 451          result = lookup_classname_imp(&*temp.begin(), &*temp.begin() + temp.size());
 452       }
 453       return result;
 454    }
 455    string_type lookup_collatename(const charT* p1, const charT* p2) const;
 456    string_type transform_primary(const charT* p1, const charT* p2) const;
 457    string_type transform(const charT* p1, const charT* p2) const;
 458 private:
 459    std::map<int, std::string>     m_error_strings;   // error messages indexed by numberic ID
 460    std::map<string_type, char_class_type>  m_custom_class_names; // character class names
 461    std::map<string_type, string_type>      m_custom_collate_names; // collating element names
 462    unsigned                       m_collate_type;    // the form of the collation string
 463    charT                          m_collate_delim;   // the collation group delimiter
 464    //
 465    // helpers:
 466    //
 467    char_class_type lookup_classname_imp(const charT* p1, const charT* p2) const;
 468    void init();
 469 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET
 470 public:
 471    bool isctype(charT c, char_class_type m)const;
 472 #endif
 473 };
 474
 475 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
 476 #if !defined(BOOST_NO_INCLASS_MEMBER_INITIALIZATION)
 477
 478 template <class charT>
 479 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_blank;
 480 template <class charT>
 481 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_word;
 482 template <class charT>
 483 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_unicode;
 484 template <class charT>
 485 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_vertical;
 486 template <class charT>
 487 typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_horizontal;
 488
 489 #endif
 490 #endif
 491
 492 template <class charT>
 493 typename cpp_regex_traits_implementation<charT>::string_type
 494    cpp_regex_traits_implementation<charT>::transform_primary(const charT* p1, const charT* p2) const
 495 {
 496    //
 497    // PRECONDITIONS:
 498    //
 499    // A bug in gcc 3.2 (and maybe other versions as well) treats
 500    // p1 as a null terminated string, for efficiency reasons
 501    // we work around this elsewhere, but just assert here that
 502    // we adhere to gcc's (buggy) preconditions...
 503    //
 504    BOOST_ASSERT(*p2 == 0);
 505
 506    string_type result;
 507    //
 508    // swallowing all exceptions here is a bad idea
 509    // however at least one std lib will always throw
 510    // std::bad_alloc for certain arguments...
 511    //
 512    try{
 513       //
 514       // What we do here depends upon the format of the sort key returned by
 515       // sort key returned by this->transform:
 516       //
 517       switch(m_collate_type)
 518       {
 519       case sort_C:
 520       case sort_unknown:
 521          // the best we can do is translate to lower case, then get a regular sort key:
 522          {
 523             result.assign(p1, p2);
 524             this->m_pctype->tolower(&*result.begin(), &*result.begin() + result.size());
 525             result = this->m_pcollate->transform(&*result.begin(), &*result.begin() + result.size());
 526             break;
 527          }
 528       case sort_fixed:
 529          {
 530             // get a regular sort key, and then truncate it:
 531             result.assign(this->m_pcollate->transform(p1, p2));
 532             result.erase(this->m_collate_delim);
 533             break;
 534          }
 535       case sort_delim:
 536             // get a regular sort key, and then truncate everything after the delim:
 537             result.assign(this->m_pcollate->transform(p1, p2));
 538             std::size_t i;
 539             for(i = 0; i < result.size(); ++i)
 540             {
 541                if(result[i] == m_collate_delim)
 542                   break;
 543             }
 544             result.erase(i);
 545             break;
 546       }
 547    }catch(...){}
 548    while(result.size() && (charT(0) == *result.rbegin()))
 549       result.erase(result.size() - 1);
 550    if(result.empty())
 551    {
 552       // character is ignorable at the primary level:
 553       result = string_type(1, charT(0));
 554    }
 555    return result;
 556 }
 557
 558 template <class charT>
 559 typename cpp_regex_traits_implementation<charT>::string_type
 560    cpp_regex_traits_implementation<charT>::transform(const charT* p1, const charT* p2) const
 561 {
 562    //
 563    // PRECONDITIONS:
 564    //
 565    // A bug in gcc 3.2 (and maybe other versions as well) treats
 566    // p1 as a null terminated string, for efficiency reasons
 567    // we work around this elsewhere, but just assert here that
 568    // we adhere to gcc's (buggy) preconditions...
 569    //
 570    BOOST_ASSERT(*p2 == 0);
 571    //
 572    // swallowing all exceptions here is a bad idea
 573    // however at least one std lib will always throw
 574    // std::bad_alloc for certain arguments...
 575    //
 576    string_type result;
 577    try{
 578       result = this->m_pcollate->transform(p1, p2);
 579       //
 580       // Borland's STLPort version returns a NULL-terminated
 581       // string that has garbage at the end - each call to
 582       // std::collate<wchar_t>::transform returns a different string!
 583       // So as a workaround, we'll truncate the string at the first NULL
 584       // which _seems_ to work....
 585 #if BOOST_WORKAROUND(__BORLANDC__, < 0x580)
 586       result.erase(result.find(charT(0)));
 587 #else
 588       //
 589       // some implementations (Dinkumware) append unnecessary trailing \0's:
 590       while(result.size() && (charT(0) == *result.rbegin()))
 591          result.erase(result.size() - 1);
 592 #endif
 593       BOOST_ASSERT(std::find(result.begin(), result.end(), charT(0)) == result.end());
 594    }
 595    catch(...)
 596    {
 597    }
 598    return result;
 599 }
 600
 601
 602 template <class charT>
 603 typename cpp_regex_traits_implementation<charT>::string_type
 604    cpp_regex_traits_implementation<charT>::lookup_collatename(const charT* p1, const charT* p2) const
 605 {
 606    typedef typename std::map<string_type, string_type>::const_iterator iter_type;
 607    if(m_custom_collate_names.size())
 608    {
 609       iter_type pos = m_custom_collate_names.find(string_type(p1, p2));
 610       if(pos != m_custom_collate_names.end())
 611          return pos->second;
 612    }
 613 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
 614                && !BOOST_WORKAROUND(BOOST_MSVC, < 1300)\
 615                && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
 616    std::string name(p1, p2);
 617 #else
 618    std::string name;
 619    const charT* p0 = p1;
 620    while(p0 != p2)
 621       name.append(1, char(*p0++));
 622 #endif
 623    name = lookup_default_collate_name(name);
 624 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
 625                && !BOOST_WORKAROUND(BOOST_MSVC, < 1300)\
 626                && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
 627    if(name.size())
 628       return string_type(name.begin(), name.end());
 629 #else
 630    if(name.size())
 631    {
 632       string_type result;
 633       typedef std::string::const_iterator iter;
 634       iter b = name.begin();
 635       iter e = name.end();
 636       while(b != e)
 637          result.append(1, charT(*b++));
 638       return result;
 639    }
 640 #endif
 641    if(p2 - p1 == 1)
 642       return string_type(1, *p1);
 643    return string_type();
 644 }
 645
 646 template <class charT>
 647 void cpp_regex_traits_implementation<charT>::init()
 648 {
 649 #ifndef BOOST_NO_STD_MESSAGES
 650 #ifndef __IBMCPP__
 651    typename std::messages<charT>::catalog cat = static_cast<std::messages<char>::catalog>(-1);
 652 #else
 653    typename std::messages<charT>::catalog cat = reinterpret_cast<std::messages<char>::catalog>(-1);
 654 #endif
 655    std::string cat_name(cpp_regex_traits<charT>::get_catalog_name());
 656    if(cat_name.size())
 657    {
 658       cat = this->m_pmessages->open(
 659          cat_name,
 660          this->m_locale);
 661       if((int)cat < 0)
 662       {
 663          std::string m("Unable to open message catalog: ");
 664          std::runtime_error err(m + cat_name);
 665          boost::re_detail::raise_runtime_error(err);
 666       }
 667    }
 668    //
 669    // if we have a valid catalog then load our messages:
 670    //
 671    if((int)cat >= 0)
 672    {
 673       //
 674       // Error messages:
 675       //
 676       for(boost::regex_constants::error_type i = static_cast<boost::regex_constants::error_type>(0);
 677          i <= boost::regex_constants::error_unknown;
 678          i = static_cast<boost::regex_constants::error_type>(i + 1))
 679       {
 680          const char* p = get_default_error_string(i);
 681          string_type default_message;
 682          while(*p)
 683          {
 684             default_message.append(1, this->m_pctype->widen(*p));
 685             ++p;
 686          }
 687          string_type s = this->m_pmessages->get(cat, 0, i+200, default_message);
 688          std::string result;
 689          for(std::string::size_type j = 0; j < s.size(); ++j)
 690          {
 691             result.append(1, this->m_pctype->narrow(s[j], 0));
 692          }
 693          m_error_strings[i] = result;
 694       }
 695       //
 696       // Custom class names:
 697       //
 698 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
 699       static const char_class_type masks[16] =
 700       {
 701          std::ctype<charT>::alnum,
 702          std::ctype<charT>::alpha,
 703          std::ctype<charT>::cntrl,
 704          std::ctype<charT>::digit,
 705          std::ctype<charT>::graph,
 706          cpp_regex_traits_implementation<charT>::mask_horizontal,
 707          std::ctype<charT>::lower,
 708          std::ctype<charT>::print,
 709          std::ctype<charT>::punct,
 710          std::ctype<charT>::space,
 711          std::ctype<charT>::upper,
 712          cpp_regex_traits_implementation<charT>::mask_vertical,
 713          std::ctype<charT>::xdigit,
 714          cpp_regex_traits_implementation<charT>::mask_blank,
 715          cpp_regex_traits_implementation<charT>::mask_word,
 716          cpp_regex_traits_implementation<charT>::mask_unicode,
 717       };
 718 #else
 719       static const char_class_type masks[14] =
 720       {
 721          ::boost::re_detail::char_class_alnum,
 722          ::boost::re_detail::char_class_alpha,
 723          ::boost::re_detail::char_class_cntrl,
 724          ::boost::re_detail::char_class_digit,
 725          ::boost::re_detail::char_class_graph,
 726          ::boost::re_detail::char_class_horizontal_space,
 727          ::boost::re_detail::char_class_lower,
 728          ::boost::re_detail::char_class_print,
 729          ::boost::re_detail::char_class_punct,
 730          ::boost::re_detail::char_class_space,
 731          ::boost::re_detail::char_class_upper,
 732          ::boost::re_detail::char_class_vertical_space,
 733          ::boost::re_detail::char_class_xdigit,
 734          ::boost::re_detail::char_class_blank,
 735          ::boost::re_detail::char_class_word,
 736          ::boost::re_detail::char_class_unicode,
 737       };
 738 #endif
 739       static const string_type null_string;
 740       for(unsigned int j = 0; j <= 13; ++j)
 741       {
 742          string_type s(this->m_pmessages->get(cat, 0, j+300, null_string));
 743          if(s.size())
 744             this->m_custom_class_names[s] = masks[j];
 745       }
 746    }
 747 #endif
 748    //
 749    // get the collation format used by m_pcollate:
 750    //
 751    m_collate_type = re_detail::find_sort_syntax(this, &m_collate_delim);
 752 }
 753
 754 template <class charT>
 755 typename cpp_regex_traits_implementation<charT>::char_class_type
 756    cpp_regex_traits_implementation<charT>::lookup_classname_imp(const charT* p1, const charT* p2) const
 757 {
 758 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
 759    static const char_class_type masks[22] =
 760    {
 761       0,
 762       std::ctype<char>::alnum,
 763       std::ctype<char>::alpha,
 764       cpp_regex_traits_implementation<charT>::mask_blank,
 765       std::ctype<char>::cntrl,
 766       std::ctype<char>::digit,
 767       std::ctype<char>::digit,
 768       std::ctype<char>::graph,
 769       cpp_regex_traits_implementation<charT>::mask_horizontal,
 770       std::ctype<char>::lower,
 771       std::ctype<char>::lower,
 772       std::ctype<char>::print,
 773       std::ctype<char>::punct,
 774       std::ctype<char>::space,
 775       std::ctype<char>::space,
 776       std::ctype<char>::upper,
 777       cpp_regex_traits_implementation<charT>::mask_unicode,
 778       std::ctype<char>::upper,
 779       cpp_regex_traits_implementation<charT>::mask_vertical,
 780       std::ctype<char>::alnum | cpp_regex_traits_implementation<charT>::mask_word,
 781       std::ctype<char>::alnum | cpp_regex_traits_implementation<charT>::mask_word,
 782       std::ctype<char>::xdigit,
 783    };
 784 #else
 785    static const char_class_type masks[22] =
 786    {
 787       0,
 788       ::boost::re_detail::char_class_alnum,
 789       ::boost::re_detail::char_class_alpha,
 790       ::boost::re_detail::char_class_blank,
 791       ::boost::re_detail::char_class_cntrl,
 792       ::boost::re_detail::char_class_digit,
 793       ::boost::re_detail::char_class_digit,
 794       ::boost::re_detail::char_class_graph,
 795       ::boost::re_detail::char_class_horizontal_space,
 796       ::boost::re_detail::char_class_lower,
 797       ::boost::re_detail::char_class_lower,
 798       ::boost::re_detail::char_class_print,
 799       ::boost::re_detail::char_class_punct,
 800       ::boost::re_detail::char_class_space,
 801       ::boost::re_detail::char_class_space,
 802       ::boost::re_detail::char_class_upper,
 803       ::boost::re_detail::char_class_unicode,
 804       ::boost::re_detail::char_class_upper,
 805       ::boost::re_detail::char_class_vertical_space,
 806       ::boost::re_detail::char_class_alnum | ::boost::re_detail::char_class_word,
 807       ::boost::re_detail::char_class_alnum | ::boost::re_detail::char_class_word,
 808       ::boost::re_detail::char_class_xdigit,
 809    };
 810 #endif
 811    if(m_custom_class_names.size())
 812    {
 813       typedef typename std::map<std::basic_string<charT>, char_class_type>::const_iterator map_iter;
 814       map_iter pos = m_custom_class_names.find(string_type(p1, p2));
 815       if(pos != m_custom_class_names.end())
 816          return pos->second;
 817    }
 818    std::size_t state_id = 1 + re_detail::get_default_class_id(p1, p2);
 819    BOOST_ASSERT(state_id < sizeof(masks) / sizeof(masks[0]));
 820    return masks[state_id];
 821 }
 822
 823 #ifdef BOOST_REGEX_BUGGY_CTYPE_FACET
 824 template <class charT>
 825 bool cpp_regex_traits_implementation<charT>::isctype(const charT c, char_class_type mask) const
 826 {
 827    return
 828       ((mask & ::boost::re_detail::char_class_space) && (m_pctype->is(std::ctype<charT>::space, c)))
 829       || ((mask & ::boost::re_detail::char_class_print) && (m_pctype->is(std::ctype<charT>::print, c)))
 830       || ((mask & ::boost::re_detail::char_class_cntrl) && (m_pctype->is(std::ctype<charT>::cntrl, c)))
 831       || ((mask & ::boost::re_detail::char_class_upper) && (m_pctype->is(std::ctype<charT>::upper, c)))
 832       || ((mask & ::boost::re_detail::char_class_lower) && (m_pctype->is(std::ctype<charT>::lower, c)))
 833       || ((mask & ::boost::re_detail::char_class_alpha) && (m_pctype->is(std::ctype<charT>::alpha, c)))
 834       || ((mask & ::boost::re_detail::char_class_digit) && (m_pctype->is(std::ctype<charT>::digit, c)))
 835       || ((mask & ::boost::re_detail::char_class_punct) && (m_pctype->is(std::ctype<charT>::punct, c)))
 836       || ((mask & ::boost::re_detail::char_class_xdigit) && (m_pctype->is(std::ctype<charT>::xdigit, c)))
 837       || ((mask & ::boost::re_detail::char_class_blank) && (m_pctype->is(std::ctype<charT>::space, c)) && !::boost::re_detail::is_separator(c))
 838       || ((mask & ::boost::re_detail::char_class_word) && (c == '_'))
 839       || ((mask & ::boost::re_detail::char_class_unicode) && ::boost::re_detail::is_extended(c))
 840       || ((mask & ::boost::re_detail::char_class_vertical) && (is_separator(c) || (c == '\v')))
 841       || ((mask & ::boost::re_detail::char_class_horizontal) && m_pctype->is(std::ctype<charT>::space, c) && !(is_separator(c) || (c == '\v')));
 842 }
 843 #endif
 844
 845
 846 template <class charT>
 847 inline boost::shared_ptr<const cpp_regex_traits_implementation<charT> > create_cpp_regex_traits(const std::locale& l BOOST_APPEND_EXPLICIT_TEMPLATE_TYPE(charT))
 848 {
 849    cpp_regex_traits_base<charT> key(l);
 850    return ::boost::object_cache<cpp_regex_traits_base<charT>, cpp_regex_traits_implementation<charT> >::get(key, 5);
 851 }
 852
 853 } // re_detail
 854
 855 template <class charT>
 856 class cpp_regex_traits
 857 {
 858 private:
 859    typedef std::ctype<charT>            ctype_type;
 860 public:
 861    typedef charT                        char_type;
 862    typedef std::size_t                  size_type;
 863    typedef std::basic_string<char_type> string_type;
 864    typedef std::locale                  locale_type;
 865    typedef boost::uint_least32_t        char_class_type;
 866
 867    struct boost_extensions_tag{};
 868
 869    cpp_regex_traits()
 870       : m_pimpl(re_detail::create_cpp_regex_traits<charT>(std::locale()))
 871    { }
 872    static size_type length(const char_type* p)
 873    {
 874       return std::char_traits<charT>::length(p);
 875    }
 876    regex_constants::syntax_type syntax_type(charT c)const
 877    {
 878       return m_pimpl->syntax_type(c);
 879    }
 880    regex_constants::escape_syntax_type escape_syntax_type(charT c) const
 881    {
 882       return m_pimpl->escape_syntax_type(c);
 883    }
 884    charT translate(charT c) const
 885    {
 886       return c;
 887    }
 888    charT translate_nocase(charT c) const
 889    {
 890       return m_pimpl->m_pctype->tolower(c);
 891    }
 892    charT translate(charT c, bool icase) const
 893    {
 894       return icase ? m_pimpl->m_pctype->tolower(c) : c;
 895    }
 896    charT tolower(charT c) const
 897    {
 898       return m_pimpl->m_pctype->tolower(c);
 899    }
 900    charT toupper(charT c) const
 901    {
 902       return m_pimpl->m_pctype->toupper(c);
 903    }
 904    string_type transform(const charT* p1, const charT* p2) const
 905    {
 906       return m_pimpl->transform(p1, p2);
 907    }
 908    string_type transform_primary(const charT* p1, const charT* p2) const
 909    {
 910       return m_pimpl->transform_primary(p1, p2);
 911    }
 912    char_class_type lookup_classname(const charT* p1, const charT* p2) const
 913    {
 914       return m_pimpl->lookup_classname(p1, p2);
 915    }
 916    string_type lookup_collatename(const charT* p1, const charT* p2) const
 917    {
 918       return m_pimpl->lookup_collatename(p1, p2);
 919    }
 920    bool isctype(charT c, char_class_type f) const
 921    {
 922 #ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
 923       typedef typename std::ctype<charT>::mask ctype_mask;
 924
 925       static const ctype_mask mask_base =
 926          static_cast<ctype_mask>(
 927             std::ctype<charT>::alnum
 928             | std::ctype<charT>::alpha
 929             | std::ctype<charT>::cntrl
 930             | std::ctype<charT>::digit
 931             | std::ctype<charT>::graph
 932             | std::ctype<charT>::lower
 933             | std::ctype<charT>::print
 934             | std::ctype<charT>::punct
 935             | std::ctype<charT>::space
 936             | std::ctype<charT>::upper
 937             | std::ctype<charT>::xdigit);
 938
 939       if((f & mask_base)
 940          && (m_pimpl->m_pctype->is(
 941             static_cast<ctype_mask>(f & mask_base), c)))
 942          return true;
 943       else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_unicode) && re_detail::is_extended(c))
 944          return true;
 945       else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_word) && (c == '_'))
 946          return true;
 947       else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_blank)
 948          && m_pimpl->m_pctype->is(std::ctype<charT>::space, c)
 949          && !re_detail::is_separator(c))
 950          return true;
 951       else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_vertical)
 952          && (::boost::re_detail::is_separator(c) || (c == '\v')))
 953          return true;
 954       else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_horizontal)
 955          && this->isctype(c, std::ctype<charT>::space) && !this->isctype(c, re_detail::cpp_regex_traits_implementation<charT>::mask_vertical))
 956          return true;
 957       return false;
 958 #else
 959       return m_pimpl->isctype(c, f);
 960 #endif
 961    }
 962    int toi(const charT*& p1, const charT* p2, int radix)const;
 963    int value(charT c, int radix)const
 964    {
 965       const charT* pc = &c;
 966       return toi(pc, pc + 1, radix);
 967    }
 968    locale_type imbue(locale_type l)
 969    {
 970       std::locale result(getloc());
 971       m_pimpl = re_detail::create_cpp_regex_traits<charT>(l);
 972       return result;
 973    }
 974    locale_type getloc()const
 975    {
 976       return m_pimpl->m_locale;
 977    }
 978    std::string error_string(regex_constants::error_type n) const
 979    {
 980       return m_pimpl->error_string(n);
 981    }
 982
 983    //
 984    // extension:
 985    // set the name of the message catalog in use (defaults to "boost_regex").
 986    //
 987    static std::string catalog_name(const std::string& name);
 988    static std::string get_catalog_name();
 989
 990 private:
 991    boost::shared_ptr<const re_detail::cpp_regex_traits_implementation<charT> > m_pimpl;
 992    //
 993    // catalog name handler:
 994    //
 995    static std::string& get_catalog_name_inst();
 996
 997 #ifdef BOOST_HAS_THREADS
 998    static static_mutex& get_mutex_inst();
 999 #endif
1000 };
1001
1002
1003 template <class charT>
1004 int cpp_regex_traits<charT>::toi(const charT*& first, const charT* last, int radix)const
1005 {
1006    re_detail::parser_buf<charT>   sbuf;            // buffer for parsing numbers.
1007    std::basic_istream<charT>      is(&sbuf);       // stream for parsing numbers.
1008
1009    // we do NOT want to parse any thousands separators inside the stream:
1010    last = std::find(first, last, BOOST_USE_FACET(std::numpunct<charT>, is.getloc()).thousands_sep());
1011
1012    sbuf.pubsetbuf(const_cast<charT*>(static_cast<const charT*>(first)), static_cast<std::streamsize>(last-first));
1013    is.clear();
1014    if(std::abs(radix) == 16) is >> std::hex;
1015    else if(std::abs(radix) == 8) is >> std::oct;
1016    else is >> std::dec;
1017    int val;
1018    if(is >> val)
1019    {
1020       first = first + ((last - first) - sbuf.in_avail());
1021       return val;
1022    }
1023    else
1024       return -1;
1025 }
1026
1027 template <class charT>
1028 std::string cpp_regex_traits<charT>::catalog_name(const std::string& name)
1029 {
1030 #ifdef BOOST_HAS_THREADS
1031    static_mutex::scoped_lock lk(get_mutex_inst());
1032 #endif
1033    std::string result(get_catalog_name_inst());
1034    get_catalog_name_inst() = name;
1035    return result;
1036 }
1037
1038 template <class charT>
1039 std::string& cpp_regex_traits<charT>::get_catalog_name_inst()
1040 {
1041    static std::string s_name;
1042    return s_name;
1043 }
1044
1045 template <class charT>
1046 std::string cpp_regex_traits<charT>::get_catalog_name()
1047 {
1048 #ifdef BOOST_HAS_THREADS
1049    static_mutex::scoped_lock lk(get_mutex_inst());
1050 #endif
1051    std::string result(get_catalog_name_inst());
1052    return result;
1053 }
1054
1055 #ifdef BOOST_HAS_THREADS
1056 template <class charT>
1057 static_mutex& cpp_regex_traits<charT>::get_mutex_inst()
1058 {
1059    static static_mutex s_mutex = BOOST_STATIC_MUTEX_INIT;
1060    return s_mutex;
1061 }
1062 #endif
1063
1064
1065 } // boost
1066
1067 #ifdef BOOST_MSVC
1068 #pragma warning(pop)
1069 #endif
1070
1071 #ifdef BOOST_MSVC
1072 #pragma warning(push)
1073 #pragma warning(disable: 4103)
1074 #endif
1075 #ifdef BOOST_HAS_ABI_HEADERS
1076 #  include BOOST_ABI_SUFFIX
1077 #endif
1078 #ifdef BOOST_MSVC
1079 #pragma warning(pop)
1080 #endif
1081
1082 #endif
1083
1084 #endif
1085
1086