libstdc++-v3/include/bits/regex_compiler.tcc

   1 // class template regex -*- C++ -*-
   2
   3 // Copyright (C) 2013 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.tcc
  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 // FIXME make comments doxygen format.
  32
  33 // This compiler refers to "Regular Expression Matching Can Be Simple And Fast"
  34 // (http://swtch.com/~rsc/regexp/regexp1.html"),
  35 // but doesn't strictly follow it.
  36 //
  37 // When compiling, states are *chained* instead of tree- or graph-constructed.
  38 // It's more like structured programs: there's if statement and loop statement.
  39 //
  40 // For alternative structure(say "a|b"), aka "if statement", two branchs should
  41 // be constructed. However, these two shall merge to an "end_tag" at the end of
  42 // this operator:
  43 //
  44 //                branch1
  45 //              /        \
  46 // => begin_tag            end_tag =>
  47 //              \        /
  48 //                branch2
  49 //
  50 // This is the difference between this implementation and that in Russ's
  51 // article.
  52 //
  53 // That's why we introduced dummy node here ------ "end_tag" is a dummy node.
  54 // All dummy node will be eliminated at the end of compiling process.
  55
  56 namespace std _GLIBCXX_VISIBILITY(default)
  57 {
  58 namespace __detail
  59 {
  60 _GLIBCXX_BEGIN_NAMESPACE_VERSION
  61
  62   template<typename _FwdIter, typename _CharT, typename _TraitsT>
  63     _Compiler<_FwdIter, _CharT, _TraitsT>::
  64     _Compiler(_FwdIter __b, _FwdIter __e,
  65               const _TraitsT& __traits, _FlagT __flags)
  66     : _M_flags((__flags
  67                 & (regex_constants::ECMAScript
  68                    | regex_constants::basic
  69                    | regex_constants::extended
  70                    | regex_constants::grep
  71                    | regex_constants::egrep
  72                    | regex_constants::awk))
  73                ? __flags
  74                : __flags | regex_constants::ECMAScript),
  75     _M_traits(__traits),
  76     _M_ctype(std::use_facet<std::ctype<_CharT>>(_M_traits.getloc())),
  77     _M_scanner(__b, __e, _M_flags, _M_traits.getloc()),
  78     _M_nfa(_M_flags)
  79     {
  80       _StateSeqT __r(_M_nfa, _M_nfa._M_start());
  81       __r._M_append(_M_nfa._M_insert_subexpr_begin());
  82       this->_M_disjunction();
  83       if (!_M_match_token(_ScannerT::_S_token_eof))
  84         __throw_regex_error(regex_constants::error_paren);
  85       __r._M_append(_M_pop());
  86       _GLIBCXX_DEBUG_ASSERT(_M_stack.empty());
  87       __r._M_append(_M_nfa._M_insert_subexpr_end());
  88       __r._M_append(_M_nfa._M_insert_accept());
  89       _M_nfa._M_eliminate_dummy();
  90     }
  91
  92   template<typename _FwdIter, typename _CharT, typename _TraitsT>
  93     void
  94     _Compiler<_FwdIter, _CharT, _TraitsT>::
  95     _M_disjunction()
  96     {
  97       this->_M_alternative();
  98       while (_M_match_token(_ScannerT::_S_token_or))
  99         {
 100           _StateSeqT __alt1 = _M_pop();
 101           this->_M_alternative();
 102           _StateSeqT __alt2 = _M_pop();
 103           auto __end = _M_nfa._M_insert_dummy();
 104           __alt1._M_append(__end);
 105           __alt2._M_append(__end);
 106           _M_stack.push(_StateSeqT(_M_nfa,
 107                                    _M_nfa._M_insert_alt(__alt1._M_start,
 108                                                         __alt2._M_start, false),
 109                                    __end));
 110         }
 111     }
 112
 113   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 114     void
 115     _Compiler<_FwdIter, _CharT, _TraitsT>::
 116     _M_alternative()
 117     {
 118       if (this->_M_term())
 119         {
 120           _StateSeqT __re = _M_pop();
 121           this->_M_alternative();
 122           __re._M_append(_M_pop());
 123           _M_stack.push(__re);
 124         }
 125       else
 126         _M_stack.push(_StateSeqT(_M_nfa, _M_nfa._M_insert_dummy()));
 127     }
 128
 129   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 130     bool
 131     _Compiler<_FwdIter, _CharT, _TraitsT>::
 132     _M_term()
 133     {
 134       if (this->_M_assertion())
 135         return true;
 136       if (this->_M_atom())
 137         {
 138           this->_M_quantifier();
 139           return true;
 140         }
 141       return false;
 142     }
 143
 144   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 145     bool
 146     _Compiler<_FwdIter, _CharT, _TraitsT>::
 147     _M_assertion()
 148     {
 149       if (_M_match_token(_ScannerT::_S_token_line_begin))
 150         _M_stack.push(_StateSeqT(_M_nfa, _M_nfa.
 151               _M_insert_line_begin()));
 152       else if (_M_match_token(_ScannerT::_S_token_line_end))
 153         _M_stack.push(_StateSeqT(_M_nfa, _M_nfa.
 154               _M_insert_line_end()));
 155       else if (_M_match_token(_ScannerT::_S_token_word_bound))
 156         // _M_value[0] == 'n' means it's negtive, say "not word boundary".
 157         _M_stack.push(_StateSeqT(_M_nfa, _M_nfa.
 158               _M_insert_word_bound(_M_value[0] == 'n')));
 159       else if (_M_match_token(_ScannerT::_S_token_subexpr_lookahead_begin))
 160         {
 161           auto __neg = _M_value[0] == 'n';
 162           this->_M_disjunction();
 163           if (!_M_match_token(_ScannerT::_S_token_subexpr_end))
 164             __throw_regex_error(regex_constants::error_paren);
 165           auto __tmp = _M_pop();
 166           __tmp._M_append(_M_nfa._M_insert_accept());
 167           _M_stack.push(
 168               _StateSeqT(
 169                 _M_nfa,
 170                 _M_nfa._M_insert_lookahead(__tmp._M_start, __neg)));
 171         }
 172       else
 173         return false;
 174       return true;
 175     }
 176
 177   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 178     void
 179     _Compiler<_FwdIter, _CharT, _TraitsT>::
 180     _M_quantifier()
 181     {
 182       bool __neg = (_M_flags & regex_constants::ECMAScript);
 183       auto __init = [this, &__neg]()
 184         {
 185           if (_M_stack.empty())
 186             __throw_regex_error(regex_constants::error_badrepeat);
 187           __neg = __neg && _M_match_token(_ScannerT::_S_token_opt);
 188         };
 189       if (_M_match_token(_ScannerT::_S_token_closure0))
 190         {
 191           __init();
 192           auto __e = _M_pop();
 193           _StateSeqT __r(_M_nfa, _M_nfa._M_insert_alt(_S_invalid_state_id,
 194                                                       __e._M_start, __neg));
 195           __e._M_append(__r);
 196           _M_stack.push(__r);
 197         }
 198       else if (_M_match_token(_ScannerT::_S_token_closure1))
 199         {
 200           __init();
 201           auto __e = _M_pop();
 202           __e._M_append(_M_nfa._M_insert_alt(_S_invalid_state_id, __e._M_start,
 203                                              __neg));
 204           _M_stack.push(__e);
 205         }
 206       else if (_M_match_token(_ScannerT::_S_token_opt))
 207         {
 208           __init();
 209           auto __e = _M_pop();
 210           auto __end = _M_nfa._M_insert_dummy();
 211           _StateSeqT __r(_M_nfa, _M_nfa._M_insert_alt(_S_invalid_state_id,
 212                                                       __e._M_start, __neg));
 213           __e._M_append(__end);
 214           __r._M_append(__end);
 215           _M_stack.push(__r);
 216         }
 217       else if (_M_match_token(_ScannerT::_S_token_interval_begin))
 218         {
 219           if (_M_stack.empty())
 220             __throw_regex_error(regex_constants::error_badrepeat);
 221           if (!_M_match_token(_ScannerT::_S_token_dup_count))
 222             __throw_regex_error(regex_constants::error_badbrace);
 223           _StateSeqT __r(_M_pop());
 224           _StateSeqT __e(_M_nfa, _M_nfa._M_insert_dummy());
 225           long __min_rep = _M_cur_int_value(10);
 226           bool __infi = false;
 227           long __n;
 228
 229           // {3
 230           if (_M_match_token(_ScannerT::_S_token_comma))
 231             if (_M_match_token(_ScannerT::_S_token_dup_count)) // {3,7}
 232               __n = _M_cur_int_value(10) - __min_rep;
 233             else
 234               __infi = true;
 235           else
 236             __n = 0;
 237           if (!_M_match_token(_ScannerT::_S_token_interval_end))
 238             __throw_regex_error(regex_constants::error_brace);
 239
 240           __neg = __neg && _M_match_token(_ScannerT::_S_token_opt);
 241
 242           for (long __i = 0; __i < __min_rep; ++__i)
 243             __e._M_append(__r._M_clone());
 244
 245           if (__infi)
 246             {
 247               auto __tmp = __r._M_clone();
 248               _StateSeqT __s(_M_nfa,
 249                              _M_nfa._M_insert_alt(_S_invalid_state_id,
 250                                                   __tmp._M_start, __neg));
 251               __tmp._M_append(__s);
 252               __e._M_append(__s);
 253             }
 254           else
 255             {
 256               if (__n < 0)
 257                 __throw_regex_error(regex_constants::error_badbrace);
 258               auto __end = _M_nfa._M_insert_dummy();
 259               // _M_alt is the "match more" branch, and _M_next is the
 260               // "match less" one. Switch _M_alt and _M_next of all created
 261               // nodes. This is a hacking but IMO works well.
 262               std::stack<_StateIdT> __stack;
 263               for (long __i = 0; __i < __n; ++__i)
 264                 {
 265                   auto __tmp = __r._M_clone();
 266                   auto __alt = _M_nfa._M_insert_alt(__tmp._M_start,
 267                                                     __end, __neg);
 268                   __stack.push(__alt);
 269                   __e._M_append(_StateSeqT(_M_nfa, __alt, __tmp._M_end));
 270                 }
 271               __e._M_append(__end);
 272               while (!__stack.empty())
 273                 {
 274                   auto& __tmp = _M_nfa[__stack.top()];
 275                   __stack.pop();
 276                   swap(__tmp._M_next, __tmp._M_alt);
 277                 }
 278             }
 279           _M_stack.push(__e);
 280         }
 281     }
 282
 283   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 284     bool
 285     _Compiler<_FwdIter, _CharT, _TraitsT>::
 286     _M_atom()
 287     {
 288       if (_M_match_token(_ScannerT::_S_token_anychar))
 289         _M_stack.push(_StateSeqT(_M_nfa,
 290                                 _M_nfa._M_insert_matcher
 291                                 (_AnyMatcher<_CharT, _TraitsT>(_M_traits))));
 292       else if (_M_try_char())
 293         _M_stack.push(_StateSeqT(_M_nfa,
 294                                  _M_nfa._M_insert_matcher
 295                                  (_CharMatcher<_CharT, _TraitsT>(_M_value[0],
 296                                                                  _M_traits,
 297                                                                  _M_flags))));
 298       else if (_M_match_token(_ScannerT::_S_token_backref))
 299         _M_stack.push(_StateSeqT(_M_nfa, _M_nfa.
 300                                  _M_insert_backref(_M_cur_int_value(10))));
 301       else if (_M_match_token(_ScannerT::_S_token_quoted_class))
 302         {
 303           _GLIBCXX_DEBUG_ASSERT(_M_value.size() == 1);
 304           _BMatcherT __matcher(_M_ctype.is(_CtypeT::upper, _M_value[0]),
 305                                _M_traits, _M_flags);
 306           __matcher._M_add_character_class(_M_value);
 307           _M_stack.push(_StateSeqT(_M_nfa,
 308                 _M_nfa._M_insert_matcher(__matcher)));
 309         }
 310       else if (_M_match_token(_ScannerT::_S_token_subexpr_no_group_begin))
 311         {
 312           _StateSeqT __r(_M_nfa, _M_nfa._M_insert_dummy());
 313           this->_M_disjunction();
 314           if (!_M_match_token(_ScannerT::_S_token_subexpr_end))
 315             __throw_regex_error(regex_constants::error_paren);
 316           __r._M_append(_M_pop());
 317           _M_stack.push(__r);
 318         }
 319       else if (_M_match_token(_ScannerT::_S_token_subexpr_begin))
 320         {
 321           _StateSeqT __r(_M_nfa, _M_nfa._M_insert_subexpr_begin());
 322           this->_M_disjunction();
 323           if (!_M_match_token(_ScannerT::_S_token_subexpr_end))
 324             __throw_regex_error(regex_constants::error_paren);
 325           __r._M_append(_M_pop());
 326           __r._M_append(_M_nfa._M_insert_subexpr_end());
 327           _M_stack.push(__r);
 328         }
 329       else if (!_M_bracket_expression())
 330         return false;
 331       return true;
 332     }
 333
 334   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 335     bool
 336     _Compiler<_FwdIter, _CharT, _TraitsT>::
 337     _M_bracket_expression()
 338     {
 339       bool __neg =
 340         _M_match_token(_ScannerT::_S_token_bracket_neg_begin);
 341       if (!(__neg || _M_match_token(_ScannerT::_S_token_bracket_begin)))
 342         return false;
 343       _BMatcherT __matcher(__neg, _M_traits, _M_flags);
 344       while (!_M_match_token(_ScannerT::_S_token_bracket_end))
 345         _M_expression_term(__matcher);
 346       _M_stack.push(_StateSeqT(_M_nfa, _M_nfa._M_insert_matcher(__matcher)));
 347       return true;
 348     }
 349
 350   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 351     void
 352     _Compiler<_FwdIter, _CharT, _TraitsT>::
 353     _M_expression_term(_BMatcherT& __matcher)
 354     {
 355       if (_M_match_token(_ScannerT::_S_token_collsymbol))
 356         __matcher._M_add_collating_element(_M_value);
 357       else if (_M_match_token(_ScannerT::_S_token_equiv_class_name))
 358         __matcher._M_add_equivalence_class(_M_value);
 359       else if (_M_match_token(_ScannerT::_S_token_char_class_name))
 360         __matcher._M_add_character_class(_M_value);
 361       else if (_M_try_char()) // [a
 362         {
 363           auto __ch = _M_value[0];
 364           if (_M_try_char())
 365             {
 366               if (_M_value[0] == '-') // [a-
 367                 {
 368                   if (_M_try_char()) // [a-z]
 369                     {
 370                       __matcher._M_make_range(__ch, _M_value[0]);
 371                       return;
 372                     }
 373                   // If the dash is the last character in the bracket
 374                   // expression, it is not special.
 375                   if (_M_scanner._M_get_token()
 376                       != _ScannerT::_S_token_bracket_end)
 377                     __throw_regex_error(regex_constants::error_range);
 378                 }
 379               __matcher._M_add_char(_M_value[0]);
 380             }
 381           __matcher._M_add_char(__ch);
 382         }
 383       else
 384         __throw_regex_error(regex_constants::error_brack);
 385     }
 386
 387   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 388     bool
 389     _Compiler<_FwdIter, _CharT, _TraitsT>::
 390     _M_try_char()
 391     {
 392       bool __is_char = false;
 393       if (_M_match_token(_ScannerT::_S_token_oct_num))
 394         {
 395           __is_char = true;
 396           _M_value.assign(1, _M_cur_int_value(8));
 397         }
 398       else if (_M_match_token(_ScannerT::_S_token_hex_num))
 399         {
 400           __is_char = true;
 401           _M_value.assign(1, _M_cur_int_value(16));
 402         }
 403       else if (_M_match_token(_ScannerT::_S_token_ord_char))
 404         __is_char = true;
 405       return __is_char;
 406     }
 407
 408   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 409     bool
 410     _Compiler<_FwdIter, _CharT, _TraitsT>::
 411     _M_match_token(_TokenT token)
 412     {
 413       if (token == _M_scanner._M_get_token())
 414         {
 415           _M_value = _M_scanner._M_get_value();
 416           _M_scanner._M_advance();
 417           return true;
 418         }
 419       return false;
 420     }
 421
 422   template<typename _FwdIter, typename _CharT, typename _TraitsT>
 423     int
 424     _Compiler<_FwdIter, _CharT, _TraitsT>::
 425     _M_cur_int_value(int __radix)
 426     {
 427       long __v = 0;
 428       for (typename _StringT::size_type __i = 0;
 429            __i < _M_value.length(); ++__i)
 430         __v =__v * __radix + _M_traits.value(_M_value[__i], __radix);
 431       return __v;
 432     }
 433
 434   template<typename _CharT, typename _TraitsT>
 435     bool _BracketMatcher<_CharT, _TraitsT>::
 436     operator()(_CharT __ch) const
 437     {
 438       bool __ret = false;
 439       if (_M_traits.isctype(__ch, _M_class_set)
 440           || _M_char_set.count(_M_translate(__ch))
 441           || _M_equiv_set.count(_M_traits.transform_primary(&__ch, &__ch+1)))
 442         __ret = true;
 443       else
 444         {
 445           _StringT __s = _M_get_str(_M_flags & regex_constants::collate
 446                                     ? _M_translate(__ch) : __ch);
 447           for (auto& __it : _M_range_set)
 448             if (__it.first <= __s && __s <= __it.second)
 449               {
 450                 __ret = true;
 451                 break;
 452               }
 453         }
 454       if (_M_is_non_matching)
 455         return !__ret;
 456       else
 457         return __ret;
 458     }
 459
 460 _GLIBCXX_END_NAMESPACE_VERSION
 461 } // namespace __detail
 462 } // namespace