contrib/gcc-4.7/libstdc++-v3/include/bits/regex_compiler.h

   1 // class template regex -*- C++ -*-
   2
   3 // Copyright (C) 2010, 2011 Free Software Foundation, Inc.
   4 //
   5 // This file is part of the GNU ISO C++ Library.  This library is free
   6 // software; you can redistribute it and/or modify it under the
   7 // terms of the GNU General Public License as published by the
   8 // Free Software Foundation; either version 3, or (at your option)
   9 // any later version.
  10
  11 // This library is distributed in the hope that it will be useful,
  12 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 // GNU General Public License for more details.
  15
  16 // Under Section 7 of GPL version 3, you are granted additional
  17 // permissions described in the GCC Runtime Library Exception, version
  18 // 3.1, as published by the Free Software Foundation.
  19
  20 // You should have received a copy of the GNU General Public License and
  21 // a copy of the GCC Runtime Library Exception along with this program;
  22 // see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
  23 // <http://www.gnu.org/licenses/>.
  24
  25 /**
  26  *  @file bits/regex_compiler.h
  27  *  This is an internal header file, included by other library headers.
  28  *  Do not attempt to use it directly. @headername{regex}
  29  */
  30
  31 namespace std _GLIBCXX_VISIBILITY(default)
  32 {
  33 namespace __regex
  34 {
  35 _GLIBCXX_BEGIN_NAMESPACE_VERSION
  36
  37   struct _Scanner_base
  38   {
  39     typedef unsigned int _StateT;
  40
  41     static constexpr _StateT _S_state_at_start    = 1 << 0;
  42     static constexpr _StateT _S_state_in_brace    = 1 << 2;
  43     static constexpr _StateT _S_state_in_bracket  = 1 << 3;
  44
  45     virtual ~_Scanner_base() { };
  46   };
  47
  48   //
  49   // @brief Scans an input range for regex tokens.
  50   //
  51   // The %_Scanner class interprets the regular expression pattern in the input
  52   // range passed to its constructor as a sequence of parse tokens passed to
  53   // the regular expression compiler.  The sequence of tokens provided depends
  54   // on the flag settings passed to the constructor:  different regular
  55   // expression grammars will interpret the same input pattern in
  56   // syntactically different ways.
  57   //
  58   template<typename _InputIterator>
  59     class _Scanner: public _Scanner_base
  60     {
  61     public:
  62       typedef _InputIterator                                        _IteratorT;
  63       typedef typename std::iterator_traits<_IteratorT>::value_type _CharT;
  64       typedef std::basic_string<_CharT>                             _StringT;
  65       typedef regex_constants::syntax_option_type                   _FlagT;
  66       typedef const std::ctype<_CharT>                              _CtypeT;
  67
  68       // Token types returned from the scanner.
  69       enum _TokenT
  70       {
  71         _S_token_anychar,
  72         _S_token_backref,
  73         _S_token_bracket_begin,
  74         _S_token_bracket_end,
  75         _S_token_inverse_class,
  76         _S_token_char_class_name,
  77         _S_token_closure0,
  78         _S_token_closure1,
  79         _S_token_collelem_multi,
  80         _S_token_collelem_single,
  81         _S_token_collsymbol,
  82         _S_token_comma,
  83         _S_token_dash,
  84         _S_token_dup_count,
  85         _S_token_eof,
  86         _S_token_equiv_class_name,
  87         _S_token_interval_begin,
  88         _S_token_interval_end,
  89         _S_token_line_begin,
  90         _S_token_line_end,
  91         _S_token_opt,
  92         _S_token_or,
  93         _S_token_ord_char,
  94         _S_token_quoted_char,
  95         _S_token_subexpr_begin,
  96         _S_token_subexpr_end,
  97         _S_token_word_begin,
  98         _S_token_word_end,
  99         _S_token_unknown
 100       };
 101
 102     public:
 103       _Scanner(_IteratorT __begin, _IteratorT __end, _FlagT __flags,
 104                std::locale __loc)
 105       : _M_current(__begin) , _M_end(__end) , _M_flags(__flags),
 106         _M_ctype(std::use_facet<_CtypeT>(__loc)), _M_state(_S_state_at_start)
 107       { _M_advance(); }
 108
 109       void
 110       _M_advance();
 111
 112       _TokenT
 113       _M_token() const
 114       { return _M_curToken; }
 115
 116       const _StringT&
 117       _M_value() const
 118       { return _M_curValue; }
 119
 120 #ifdef _GLIBCXX_DEBUG
 121       std::ostream&
 122       _M_print(std::ostream&);
 123 #endif
 124
 125     private:
 126       void
 127       _M_eat_escape();
 128
 129       void
 130       _M_scan_in_brace();
 131
 132       void
 133       _M_scan_in_bracket();
 134
 135       void
 136       _M_eat_charclass();
 137
 138       void
 139       _M_eat_equivclass();
 140
 141       void
 142       _M_eat_collsymbol();
 143
 144     private:
 145       _IteratorT  _M_current;
 146       _IteratorT  _M_end;
 147       _FlagT      _M_flags;
 148       _CtypeT&    _M_ctype;
 149       _TokenT     _M_curToken;
 150       _StringT    _M_curValue;
 151       _StateT     _M_state;
 152     };
 153
 154   template<typename _InputIterator>
 155     void
 156     _Scanner<_InputIterator>::
 157     _M_advance()
 158     {
 159       if (_M_current == _M_end)
 160         {
 161           _M_curToken = _S_token_eof;
 162           return;
 163         }
 164
 165       _CharT __c = *_M_current;
 166       if (_M_state & _S_state_in_bracket)
 167         {
 168           _M_scan_in_bracket();
 169           return;
 170         }
 171       if (_M_state & _S_state_in_brace)
 172         {
 173           _M_scan_in_brace();
 174           return;
 175         }
 176 #if 0
 177       // TODO: re-enable line anchors when _M_assertion is implemented.
 178       // See PR libstdc++/47724
 179       else if (_M_state & _S_state_at_start && __c == _M_ctype.widen('^'))
 180         {
 181           _M_curToken = _S_token_line_begin;
 182           ++_M_current;
 183           return;
 184         }
 185       else if (__c == _M_ctype.widen('$'))
 186         {
 187           _M_curToken = _S_token_line_end;
 188           ++_M_current;
 189           return;
 190         }
 191 #endif
 192       else if (__c == _M_ctype.widen('.'))
 193         {
 194           _M_curToken = _S_token_anychar;
 195           ++_M_current;
 196           return;
 197         }
 198       else if (__c == _M_ctype.widen('*'))
 199         {
 200           _M_curToken = _S_token_closure0;
 201           ++_M_current;
 202           return;
 203         }
 204       else if (__c == _M_ctype.widen('+'))
 205         {
 206           _M_curToken = _S_token_closure1;
 207           ++_M_current;
 208           return;
 209         }
 210       else if (__c == _M_ctype.widen('|'))
 211         {
 212           _M_curToken = _S_token_or;
 213           ++_M_current;
 214           return;
 215         }
 216       else if (__c == _M_ctype.widen('['))
 217         {
 218           _M_curToken = _S_token_bracket_begin;
 219           _M_state |= (_S_state_in_bracket | _S_state_at_start);
 220           ++_M_current;
 221           return;
 222         }
 223       else if (__c == _M_ctype.widen('\\'))
 224         {
 225           _M_eat_escape();
 226           return;
 227         }
 228       else if (!(_M_flags & (regex_constants::basic | regex_constants::grep)))
 229         {
 230           if (__c == _M_ctype.widen('('))
 231             {
 232               _M_curToken = _S_token_subexpr_begin;
 233               ++_M_current;
 234               return;
 235             }
 236           else if (__c == _M_ctype.widen(')'))
 237             {
 238               _M_curToken = _S_token_subexpr_end;
 239               ++_M_current;
 240               return;
 241             }
 242           else if (__c == _M_ctype.widen('{'))
 243             {
 244               _M_curToken = _S_token_interval_begin;
 245               _M_state |= _S_state_in_brace;
 246               ++_M_current;
 247               return;
 248             }
 249         }
 250
 251       _M_curToken = _S_token_ord_char;
 252       _M_curValue.assign(1, __c);
 253       ++_M_current;
 254     }
 255
 256
 257   template<typename _InputIterator>
 258     void
 259     _Scanner<_InputIterator>::
 260     _M_scan_in_brace()
 261     {
 262       if (_M_ctype.is(_CtypeT::digit, *_M_current))
 263         {
 264           _M_curToken = _S_token_dup_count;
 265           _M_curValue.assign(1, *_M_current);
 266           ++_M_current;
 267           while (_M_current != _M_end
 268                  && _M_ctype.is(_CtypeT::digit, *_M_current))
 269             {
 270               _M_curValue += *_M_current;
 271               ++_M_current;
 272             }
 273           return;
 274         }
 275       else if (*_M_current == _M_ctype.widen(','))
 276         {
 277           _M_curToken = _S_token_comma;
 278           ++_M_current;
 279           return;
 280         }
 281       if (_M_flags & (regex_constants::basic | regex_constants::grep))
 282         {
 283           if (*_M_current == _M_ctype.widen('\\'))
 284             _M_eat_escape();
 285         }
 286       else
 287         {
 288           if (*_M_current == _M_ctype.widen('}'))
 289             {
 290               _M_curToken = _S_token_interval_end;
 291               _M_state &= ~_S_state_in_brace;
 292               ++_M_current;
 293               return;
 294             }
 295         }
 296     }
 297
 298   template<typename _InputIterator>
 299     void
 300     _Scanner<_InputIterator>::
 301     _M_scan_in_bracket()
 302     {
 303       if (_M_state & _S_state_at_start && *_M_current == _M_ctype.widen('^'))
 304         {
 305           _M_curToken = _S_token_inverse_class;
 306           _M_state &= ~_S_state_at_start;
 307           ++_M_current;
 308           return;
 309         }
 310       else if (*_M_current == _M_ctype.widen('['))
 311         {
 312           ++_M_current;
 313           if (_M_current == _M_end)
 314             {
 315               _M_curToken = _S_token_eof;
 316               return;
 317             }
 318
 319           if (*_M_current == _M_ctype.widen('.'))
 320             {
 321               _M_curToken = _S_token_collsymbol;
 322               _M_eat_collsymbol();
 323               return;
 324             }
 325           else if (*_M_current == _M_ctype.widen(':'))
 326             {
 327               _M_curToken = _S_token_char_class_name;
 328               _M_eat_charclass();
 329               return;
 330             }
 331           else if (*_M_current == _M_ctype.widen('='))
 332             {
 333               _M_curToken = _S_token_equiv_class_name;
 334               _M_eat_equivclass();
 335               return;
 336             }
 337         }
 338       else if (*_M_current == _M_ctype.widen('-'))
 339         {
 340           _M_curToken = _S_token_dash;
 341           ++_M_current;
 342           return;
 343         }
 344       else if (*_M_current == _M_ctype.widen(']'))
 345         {
 346           if (!(_M_flags & regex_constants::ECMAScript)
 347               || !(_M_state & _S_state_at_start))
 348             {
 349               // special case: only if  _not_ chr first after
 350               // '[' or '[^' and if not ECMAscript
 351               _M_curToken = _S_token_bracket_end;
 352               ++_M_current;
 353               return;
 354             }
 355         }
 356       _M_curToken = _S_token_collelem_single;
 357       _M_curValue.assign(1, *_M_current);
 358       ++_M_current;
 359     }
 360
 361   template<typename _InputIterator>
 362     void
 363     _Scanner<_InputIterator>::
 364     _M_eat_escape()
 365     {
 366       ++_M_current;
 367       if (_M_current == _M_end)
 368         {
 369           _M_curToken = _S_token_eof;
 370           return;
 371         }
 372       _CharT __c = *_M_current;
 373       ++_M_current;
 374
 375       if (__c == _M_ctype.widen('('))
 376         {
 377           if (!(_M_flags & (regex_constants::basic | regex_constants::grep)))
 378             {
 379               _M_curToken = _S_token_ord_char;
 380               _M_curValue.assign(1, __c);
 381             }
 382           else
 383             _M_curToken = _S_token_subexpr_begin;
 384         }
 385       else if (__c == _M_ctype.widen(')'))
 386         {
 387           if (!(_M_flags & (regex_constants::basic | regex_constants::grep)))
 388             {
 389               _M_curToken = _S_token_ord_char;
 390               _M_curValue.assign(1, __c);
 391             }
 392           else
 393             _M_curToken = _S_token_subexpr_end;
 394         }
 395       else if (__c == _M_ctype.widen('{'))
 396         {
 397           if (!(_M_flags & (regex_constants::basic | regex_constants::grep)))
 398             {
 399               _M_curToken = _S_token_ord_char;
 400               _M_curValue.assign(1, __c);
 401             }
 402           else
 403             {
 404               _M_curToken = _S_token_interval_begin;
 405               _M_state |= _S_state_in_brace;
 406             }
 407         }
 408       else if (__c == _M_ctype.widen('}'))
 409         {
 410           if (!(_M_flags & (regex_constants::basic | regex_constants::grep)))
 411             {
 412               _M_curToken = _S_token_ord_char;
 413               _M_curValue.assign(1, __c);
 414             }
 415           else
 416             {
 417               if (!(_M_state && _S_state_in_brace))
 418                 __throw_regex_error(regex_constants::error_badbrace);
 419               _M_state &= ~_S_state_in_brace;
 420               _M_curToken = _S_token_interval_end;
 421             }
 422         }
 423       else if (__c == _M_ctype.widen('x'))
 424         {
 425           ++_M_current;
 426           if (_M_current == _M_end)
 427             {
 428               _M_curToken = _S_token_eof;
 429               return;
 430             }
 431           if (_M_ctype.is(_CtypeT::digit, *_M_current))
 432             {
 433               _M_curValue.assign(1, *_M_current);
 434               ++_M_current;
 435               if (_M_current == _M_end)
 436                 {
 437                   _M_curToken = _S_token_eof;
 438                   return;
 439                 }
 440               if (_M_ctype.is(_CtypeT::digit, *_M_current))
 441                 {
 442                   _M_curValue += *_M_current;
 443                   ++_M_current;
 444                   return;
 445                 }
 446             }
 447         }
 448       else if (__c == _M_ctype.widen('^')
 449                || __c == _M_ctype.widen('.')
 450                || __c == _M_ctype.widen('*')
 451                || __c == _M_ctype.widen('$')
 452                || __c == _M_ctype.widen('\\'))
 453         {
 454           _M_curToken = _S_token_ord_char;
 455           _M_curValue.assign(1, __c);
 456         }
 457       else if (_M_ctype.is(_CtypeT::digit, __c))
 458         {
 459           _M_curToken = _S_token_backref;
 460           _M_curValue.assign(1, __c);
 461         }
 462       else
 463         __throw_regex_error(regex_constants::error_escape);
 464     }
 465
 466
 467   // Eats a character class or throwns an exception.
 468   // current point to ':' delimiter on entry, char after ']' on return
 469   template<typename _InputIterator>
 470     void
 471     _Scanner<_InputIterator>::
 472     _M_eat_charclass()
 473     {
 474       ++_M_current; // skip ':'
 475       if (_M_current == _M_end)
 476         __throw_regex_error(regex_constants::error_ctype);
 477       for (_M_curValue.clear();
 478            _M_current != _M_end && *_M_current != _M_ctype.widen(':');
 479            ++_M_current)
 480         _M_curValue += *_M_current;
 481       if (_M_current == _M_end)
 482         __throw_regex_error(regex_constants::error_ctype);
 483       ++_M_current; // skip ':'
 484       if (*_M_current != _M_ctype.widen(']'))
 485         __throw_regex_error(regex_constants::error_ctype);
 486       ++_M_current; // skip ']'
 487     }
 488
 489
 490   template<typename _InputIterator>
 491     void
 492     _Scanner<_InputIterator>::
 493     _M_eat_equivclass()
 494     {
 495       ++_M_current; // skip '='
 496       if (_M_current == _M_end)
 497         __throw_regex_error(regex_constants::error_collate);
 498       for (_M_curValue.clear();
 499            _M_current != _M_end && *_M_current != _M_ctype.widen('=');
 500            ++_M_current)
 501         _M_curValue += *_M_current;
 502       if (_M_current == _M_end)
 503         __throw_regex_error(regex_constants::error_collate);
 504       ++_M_current; // skip '='
 505       if (*_M_current != _M_ctype.widen(']'))
 506         __throw_regex_error(regex_constants::error_collate);
 507       ++_M_current; // skip ']'
 508     }
 509
 510
 511   template<typename _InputIterator>
 512     void
 513     _Scanner<_InputIterator>::
 514     _M_eat_collsymbol()
 515     {
 516       ++_M_current; // skip '.'
 517       if (_M_current == _M_end)
 518         __throw_regex_error(regex_constants::error_collate);
 519       for (_M_curValue.clear();
 520            _M_current != _M_end && *_M_current != _M_ctype.widen('.');
 521            ++_M_current)
 522         _M_curValue += *_M_current;
 523       if (_M_current == _M_end)
 524         __throw_regex_error(regex_constants::error_collate);
 525       ++_M_current; // skip '.'
 526       if (*_M_current != _M_ctype.widen(']'))
 527         __throw_regex_error(regex_constants::error_collate);
 528       ++_M_current; // skip ']'
 529     }
 530
 531 #ifdef _GLIBCXX_DEBUG
 532   template<typename _InputIterator>
 533     std::ostream&
 534     _Scanner<_InputIterator>::
 535     _M_print(std::ostream& ostr)
 536     {
 537       switch (_M_curToken)
 538       {
 539         case _S_token_anychar:
 540           ostr << "any-character\n";
 541           break;
 542         case _S_token_backref:
 543           ostr << "backref\n";
 544           break;
 545         case _S_token_bracket_begin:
 546           ostr << "bracket-begin\n";
 547           break;
 548         case _S_token_bracket_end:
 549           ostr << "bracket-end\n";
 550           break;
 551         case _S_token_char_class_name:
 552           ostr << "char-class-name \"" << _M_curValue << "\"\n";
 553           break;
 554         case _S_token_closure0:
 555           ostr << "closure0\n";
 556           break;
 557         case _S_token_closure1:
 558           ostr << "closure1\n";
 559           break;
 560         case _S_token_collelem_multi:
 561           ostr << "coll-elem-multi \"" << _M_curValue << "\"\n";
 562           break;
 563         case _S_token_collelem_single:
 564           ostr << "coll-elem-single \"" << _M_curValue << "\"\n";
 565           break;
 566         case _S_token_collsymbol:
 567           ostr << "collsymbol \"" << _M_curValue << "\"\n";
 568           break;
 569         case _S_token_comma:
 570           ostr << "comma\n";
 571           break;
 572         case _S_token_dash:
 573           ostr << "dash\n";
 574           break;
 575         case _S_token_dup_count:
 576           ostr << "dup count: " << _M_curValue << "\n";
 577           break;
 578         case _S_token_eof:
 579           ostr << "EOF\n";
 580           break;
 581         case _S_token_equiv_class_name:
 582           ostr << "equiv-class-name \"" << _M_curValue << "\"\n";
 583           break;
 584         case _S_token_interval_begin:
 585           ostr << "interval begin\n";
 586           break;
 587         case _S_token_interval_end:
 588           ostr << "interval end\n";
 589           break;
 590         case _S_token_line_begin:
 591           ostr << "line begin\n";
 592           break;
 593         case _S_token_line_end:
 594           ostr << "line end\n";
 595           break;
 596         case _S_token_opt:
 597           ostr << "opt\n";
 598           break;
 599         case _S_token_or:
 600           ostr << "or\n";
 601           break;
 602         case _S_token_ord_char:
 603           ostr << "ordinary character: \"" << _M_value() << "\"\n";
 604           break;
 605         case _S_token_quoted_char:
 606           ostr << "quoted char\n";
 607           break;
 608         case _S_token_subexpr_begin:
 609           ostr << "subexpr begin\n";
 610           break;
 611         case _S_token_subexpr_end:
 612           ostr << "subexpr end\n";
 613           break;
 614         case _S_token_word_begin:
 615           ostr << "word begin\n";
 616           break;
 617         case _S_token_word_end:
 618           ostr << "word end\n";
 619           break;
 620         case _S_token_unknown:
 621           ostr << "-- unknown token --\n";
 622           break;
 623       }
 624       return ostr;
 625     }
 626 #endif
 627
 628   // Builds an NFA from an input iterator interval.
 629   template<typename _InIter, typename _TraitsT>
 630     class _Compiler
 631     {
 632     public:
 633       typedef _InIter                                            _IterT;
 634       typedef typename std::iterator_traits<_InIter>::value_type _CharT;
 635       typedef std::basic_string<_CharT>                          _StringT;
 636       typedef regex_constants::syntax_option_type                _FlagT;
 637
 638     public:
 639       _Compiler(const _InIter& __b, const _InIter& __e,
 640                 _TraitsT& __traits, _FlagT __flags);
 641
 642       const _Nfa&
 643       _M_nfa() const
 644       { return _M_state_store; }
 645
 646     private:
 647       typedef _Scanner<_InIter>                              _ScannerT;
 648       typedef typename _ScannerT::_TokenT                    _TokenT;
 649       typedef std::stack<_StateSeq, std::vector<_StateSeq> > _StackT;
 650       typedef _RangeMatcher<_InIter, _TraitsT>               _RMatcherT;
 651
 652       // accepts a specific token or returns false.
 653       bool
 654       _M_match_token(_TokenT __token);
 655
 656       void
 657       _M_disjunction();
 658
 659       bool
 660       _M_alternative();
 661
 662       bool
 663       _M_term();
 664
 665       bool
 666       _M_assertion();
 667
 668       bool
 669       _M_quantifier();
 670
 671       bool
 672       _M_atom();
 673
 674       bool
 675       _M_bracket_expression();
 676
 677       bool
 678       _M_bracket_list(_RMatcherT& __matcher);
 679
 680       bool
 681       _M_follow_list(_RMatcherT& __matcher);
 682
 683       bool
 684       _M_follow_list2(_RMatcherT& __matcher);
 685
 686       bool
 687       _M_expression_term(_RMatcherT& __matcher);
 688
 689       bool
 690       _M_range_expression(_RMatcherT& __matcher);
 691
 692       bool
 693       _M_start_range(_RMatcherT& __matcher);
 694
 695       bool
 696       _M_collating_symbol(_RMatcherT& __matcher);
 697
 698       bool
 699       _M_equivalence_class(_RMatcherT& __matcher);
 700
 701       bool
 702       _M_character_class(_RMatcherT& __matcher);
 703
 704       int
 705       _M_cur_int_value(int __radix);
 706
 707     private:
 708       _TraitsT&      _M_traits;
 709       _ScannerT      _M_scanner;
 710       _StringT       _M_cur_value;
 711       _Nfa           _M_state_store;
 712       _StackT        _M_stack;
 713     };
 714
 715   template<typename _InIter, typename _TraitsT>
 716     _Compiler<_InIter, _TraitsT>::
 717     _Compiler(const _InIter& __b, const _InIter& __e, _TraitsT& __traits,
 718               _Compiler<_InIter, _TraitsT>::_FlagT __flags)
 719     : _M_traits(__traits), _M_scanner(__b, __e, __flags, _M_traits.getloc()),
 720       _M_state_store(__flags)
 721     {
 722       typedef _StartTagger<_InIter, _TraitsT> _Start;
 723       typedef _EndTagger<_InIter, _TraitsT> _End;
 724
 725       _StateSeq __r(_M_state_store,
 726                     _M_state_store._M_insert_subexpr_begin(_Start(0)));
 727       _M_disjunction();
 728       if (!_M_stack.empty())
 729         {
 730           __r._M_append(_M_stack.top());
 731           _M_stack.pop();
 732         }
 733       __r._M_append(_M_state_store._M_insert_subexpr_end(0, _End(0)));
 734       __r._M_append(_M_state_store._M_insert_accept());
 735     }
 736
 737   template<typename _InIter, typename _TraitsT>
 738     bool
 739     _Compiler<_InIter, _TraitsT>::
 740     _M_match_token(_Compiler<_InIter, _TraitsT>::_TokenT token)
 741     {
 742       if (token == _M_scanner._M_token())
 743         {
 744           _M_cur_value = _M_scanner._M_value();
 745           _M_scanner._M_advance();
 746           return true;
 747         }
 748       return false;
 749     }
 750
 751   template<typename _InIter, typename _TraitsT>
 752     void
 753     _Compiler<_InIter, _TraitsT>::
 754     _M_disjunction()
 755     {
 756       this->_M_alternative();
 757       if (_M_match_token(_ScannerT::_S_token_or))
 758         {
 759           _StateSeq __alt1 = _M_stack.top(); _M_stack.pop();
 760           this->_M_disjunction();
 761           _StateSeq __alt2 = _M_stack.top(); _M_stack.pop();
 762           _M_stack.push(_StateSeq(__alt1, __alt2));
 763         }
 764     }
 765
 766   template<typename _InIter, typename _TraitsT>
 767     bool
 768     _Compiler<_InIter, _TraitsT>::
 769     _M_alternative()
 770     {
 771       if (this->_M_term())
 772         {
 773           _StateSeq __re = _M_stack.top(); _M_stack.pop();
 774           this->_M_alternative();
 775           if (!_M_stack.empty())
 776             {
 777               __re._M_append(_M_stack.top());
 778               _M_stack.pop();
 779             }
 780           _M_stack.push(__re);
 781           return true;
 782         }
 783       return false;
 784     }
 785
 786   template<typename _InIter, typename _TraitsT>
 787     bool
 788     _Compiler<_InIter, _TraitsT>::
 789     _M_term()
 790     {
 791       if (this->_M_assertion())
 792         return true;
 793       if (this->_M_atom())
 794         {
 795           this->_M_quantifier();
 796           return true;
 797         }
 798       return false;
 799     }
 800
 801   template<typename _InIter, typename _TraitsT>
 802     bool
 803     _Compiler<_InIter, _TraitsT>::
 804     _M_assertion()
 805     {
 806       if (_M_match_token(_ScannerT::_S_token_line_begin))
 807         {
 808           // __m.push(_Matcher::_S_opcode_line_begin);
 809           return true;
 810         }
 811       if (_M_match_token(_ScannerT::_S_token_line_end))
 812         {
 813           // __m.push(_Matcher::_S_opcode_line_end);
 814           return true;
 815         }
 816       if (_M_match_token(_ScannerT::_S_token_word_begin))
 817         {
 818           // __m.push(_Matcher::_S_opcode_word_begin);
 819           return true;
 820         }
 821       if (_M_match_token(_ScannerT::_S_token_word_end))
 822         {
 823           // __m.push(_Matcher::_S_opcode_word_end);
 824           return true;
 825         }
 826       return false;
 827     }
 828
 829   template<typename _InIter, typename _TraitsT>
 830     bool
 831     _Compiler<_InIter, _TraitsT>::
 832     _M_quantifier()
 833     {
 834       if (_M_match_token(_ScannerT::_S_token_closure0))
 835         {
 836           if (_M_stack.empty())
 837             __throw_regex_error(regex_constants::error_badrepeat);
 838           _StateSeq __r(_M_stack.top(), -1);
 839           __r._M_append(__r._M_front());
 840           _M_stack.pop();
 841           _M_stack.push(__r);
 842           return true;
 843         }
 844       if (_M_match_token(_ScannerT::_S_token_closure1))
 845         {
 846           if (_M_stack.empty())
 847             __throw_regex_error(regex_constants::error_badrepeat);
 848           _StateSeq __r(_M_state_store,
 849                         _M_state_store.
 850                         _M_insert_alt(_S_invalid_state_id,
 851                                       _M_stack.top()._M_front()));
 852           _M_stack.top()._M_append(__r);
 853           return true;
 854         }
 855       if (_M_match_token(_ScannerT::_S_token_opt))
 856         {
 857           if (_M_stack.empty())
 858           __throw_regex_error(regex_constants::error_badrepeat);
 859           _StateSeq __r(_M_stack.top(), -1);
 860           _M_stack.pop();
 861           _M_stack.push(__r);
 862           return true;
 863         }
 864       if (_M_match_token(_ScannerT::_S_token_interval_begin))
 865         {
 866           if (_M_stack.empty())
 867             __throw_regex_error(regex_constants::error_badrepeat);
 868           if (!_M_match_token(_ScannerT::_S_token_dup_count))
 869             __throw_regex_error(regex_constants::error_badbrace);
 870           _StateSeq __r(_M_stack.top());
 871           int __min_rep = _M_cur_int_value(10);
 872           for (int __i = 1; __i < __min_rep; ++__i)
 873             _M_stack.top()._M_append(__r._M_clone());
 874           if (_M_match_token(_ScannerT::_S_token_comma))
 875             if (_M_match_token(_ScannerT::_S_token_dup_count))
 876               {
 877                 int __n = _M_cur_int_value(10) - __min_rep;
 878                 if (__n < 0)
 879                   __throw_regex_error(regex_constants::error_badbrace);
 880                 for (int __i = 0; __i < __n; ++__i)
 881                   {
 882                     _StateSeq __r(_M_state_store,
 883                                   _M_state_store.
 884                                   _M_insert_alt(_S_invalid_state_id,
 885                                                 _M_stack.top()._M_front()));
 886                     _M_stack.top()._M_append(__r);
 887                   }
 888               }
 889             else
 890               {
 891                 _StateSeq __r(_M_stack.top(), -1);
 892                 __r._M_push_back(__r._M_front());
 893                 _M_stack.pop();
 894                 _M_stack.push(__r);
 895               }
 896           if (!_M_match_token(_ScannerT::_S_token_interval_end))
 897             __throw_regex_error(regex_constants::error_brace);
 898           return true;
 899         }
 900       return false;
 901     }
 902
 903   template<typename _InIter, typename _TraitsT>
 904     bool
 905     _Compiler<_InIter, _TraitsT>::
 906     _M_atom()
 907     {
 908       typedef _CharMatcher<_InIter, _TraitsT> _CMatcher;
 909       typedef _StartTagger<_InIter, _TraitsT> _Start;
 910       typedef _EndTagger<_InIter, _TraitsT> _End;
 911
 912       if (_M_match_token(_ScannerT::_S_token_anychar))
 913         {
 914           _M_stack.push(_StateSeq(_M_state_store,
 915                                   _M_state_store._M_insert_matcher
 916                                   (_AnyMatcher)));
 917           return true;
 918         }
 919       if (_M_match_token(_ScannerT::_S_token_ord_char))
 920         {
 921           _M_stack.push(_StateSeq(_M_state_store,
 922                                   _M_state_store._M_insert_matcher
 923                                   (_CMatcher(_M_cur_value[0], _M_traits))));
 924           return true;
 925         }
 926       if (_M_match_token(_ScannerT::_S_token_quoted_char))
 927         {
 928           // note that in the ECMA grammar, this case covers backrefs.
 929           _M_stack.push(_StateSeq(_M_state_store,
 930                                   _M_state_store._M_insert_matcher
 931                                   (_CMatcher(_M_cur_value[0], _M_traits))));
 932           return true;
 933         }
 934       if (_M_match_token(_ScannerT::_S_token_backref))
 935         {
 936           // __m.push(_Matcher::_S_opcode_ordchar, _M_cur_value);
 937           return true;
 938         }
 939       if (_M_match_token(_ScannerT::_S_token_subexpr_begin))
 940         {
 941           int __mark = _M_state_store._M_sub_count();
 942           _StateSeq __r(_M_state_store,
 943                         _M_state_store.
 944                         _M_insert_subexpr_begin(_Start(__mark)));
 945           this->_M_disjunction();
 946           if (!_M_match_token(_ScannerT::_S_token_subexpr_end))
 947             __throw_regex_error(regex_constants::error_paren);
 948           if (!_M_stack.empty())
 949             {
 950               __r._M_append(_M_stack.top());
 951               _M_stack.pop();
 952             }
 953           __r._M_append(_M_state_store._M_insert_subexpr_end
 954                         (__mark, _End(__mark)));
 955           _M_stack.push(__r);
 956           return true;
 957         }
 958       return _M_bracket_expression();
 959     }
 960
 961   template<typename _InIter, typename _TraitsT>
 962     bool
 963     _Compiler<_InIter, _TraitsT>::
 964     _M_bracket_expression()
 965     {
 966       if (_M_match_token(_ScannerT::_S_token_bracket_begin))
 967         {
 968           _RMatcherT __matcher(_M_match_token(_ScannerT::_S_token_line_begin),
 969                                _M_traits);
 970           if (!_M_bracket_list(__matcher)
 971               || !_M_match_token(_ScannerT::_S_token_bracket_end))
 972             __throw_regex_error(regex_constants::error_brack);
 973           _M_stack.push(_StateSeq(_M_state_store,
 974                                   _M_state_store._M_insert_matcher(__matcher)));
 975           return true;
 976         }
 977       return false;
 978     }
 979
 980   // If the dash is the last character in the bracket expression, it is not
 981   // special.
 982   template<typename _InIter, typename _TraitsT>
 983     bool
 984     _Compiler<_InIter, _TraitsT>::
 985     _M_bracket_list(_RMatcherT& __matcher)
 986     {
 987       if (_M_follow_list(__matcher))
 988         {
 989           if (_M_match_token(_ScannerT::_S_token_dash))
 990             __matcher._M_add_char(_M_cur_value[0]);
 991           return true;
 992         }
 993       return false;
 994     }
 995
 996   template<typename _InIter, typename _TraitsT>
 997     bool
 998     _Compiler<_InIter, _TraitsT>::
 999     _M_follow_list(_RMatcherT& __matcher)
1000     { return _M_expression_term(__matcher) && _M_follow_list2(__matcher); }
1001
1002   template<typename _InIter, typename _TraitsT>
1003     bool
1004     _Compiler<_InIter, _TraitsT>::
1005     _M_follow_list2(_RMatcherT& __matcher)
1006     {
1007       if (_M_expression_term(__matcher))
1008         return _M_follow_list2(__matcher);
1009       return true;
1010     }
1011
1012   template<typename _InIter, typename _TraitsT>
1013     bool
1014     _Compiler<_InIter, _TraitsT>::
1015     _M_expression_term(_RMatcherT& __matcher)
1016     {
1017       return (_M_collating_symbol(__matcher)
1018               || _M_character_class(__matcher)
1019               || _M_equivalence_class(__matcher)
1020               || (_M_start_range(__matcher)
1021                   && _M_range_expression(__matcher)));
1022     }
1023
1024   template<typename _InIter, typename _TraitsT>
1025     bool
1026     _Compiler<_InIter, _TraitsT>::
1027     _M_range_expression(_RMatcherT& __matcher)
1028     {
1029       if (!_M_collating_symbol(__matcher))
1030         if (!_M_match_token(_ScannerT::_S_token_dash))
1031           __throw_regex_error(regex_constants::error_range);
1032       __matcher._M_make_range();
1033       return true;
1034     }
1035
1036   template<typename _InIter, typename _TraitsT>
1037     bool
1038     _Compiler<_InIter, _TraitsT>::
1039     _M_start_range(_RMatcherT& __matcher)
1040     { return _M_match_token(_ScannerT::_S_token_dash); }
1041
1042   template<typename _InIter, typename _TraitsT>
1043     bool
1044     _Compiler<_InIter, _TraitsT>::
1045     _M_collating_symbol(_RMatcherT& __matcher)
1046     {
1047       if (_M_match_token(_ScannerT::_S_token_collelem_single))
1048         {
1049           __matcher._M_add_char(_M_cur_value[0]);
1050           return true;
1051         }
1052       if (_M_match_token(_ScannerT::_S_token_collsymbol))
1053         {
1054           __matcher._M_add_collating_element(_M_cur_value);
1055           return true;
1056         }
1057       return false;
1058     }
1059
1060   template<typename _InIter, typename _TraitsT>
1061     bool
1062     _Compiler<_InIter, _TraitsT>::
1063     _M_equivalence_class(_RMatcherT& __matcher)
1064     {
1065       if (_M_match_token(_ScannerT::_S_token_equiv_class_name))
1066         {
1067           __matcher._M_add_equivalence_class(_M_cur_value);
1068           return true;
1069         }
1070       return false;
1071     }
1072
1073   template<typename _InIter, typename _TraitsT>
1074     bool
1075     _Compiler<_InIter, _TraitsT>::
1076     _M_character_class(_RMatcherT& __matcher)
1077     {
1078       if (_M_match_token(_ScannerT::_S_token_char_class_name))
1079         {
1080           __matcher._M_add_character_class(_M_cur_value);
1081           return true;
1082         }
1083       return false;
1084     }
1085
1086   template<typename _InIter, typename _TraitsT>
1087     int
1088     _Compiler<_InIter, _TraitsT>::
1089     _M_cur_int_value(int __radix)
1090     {
1091       int __v = 0;
1092       for (typename _StringT::size_type __i = 0;
1093            __i < _M_cur_value.length(); ++__i)
1094         __v =__v * __radix + _M_traits.value(_M_cur_value[__i], __radix);
1095       return __v;
1096     }
1097
1098   template<typename _InIter, typename _TraitsT>
1099     _AutomatonPtr
1100     __compile(const _InIter& __b, const _InIter& __e, _TraitsT& __t,
1101               regex_constants::syntax_option_type __f)
1102     { return _AutomatonPtr(new _Nfa(_Compiler<_InIter, _TraitsT>(__b, __e, __t,
1103                                         __f)._M_nfa())); }
1104
1105 _GLIBCXX_END_NAMESPACE_VERSION
1106 } // namespace __regex
1107 } // namespace std
1108
1109 /* vim: set ts=8 sw=2 sts=2: */