regex_compile.hpp

来自「CGAL is a collaborative effort of severa」· HPP 代码 · 共 1,861 行 · 第 1/5 页

               started = true;
               l = last_none;
            }
            break;
         case traits_type::syntax_dot:
            //
            // we have a collating element [.collating-name.]
            //
            if(traits_inst.lookup_collatename(s, base+2, first-2))
            {
               --first;
               if(s.size() > 1)
                  has_digraphs = true;
               if(s.size())goto char_set_literal;
            }
            fail(REG_ECOLLATE);
            return 0;
         case traits_type::syntax_equal:
            //
            // we have an equivalence class [=collating-name=]
            //
            if(traits_inst.lookup_collatename(s, base+2, first-2))
            {
               std::size_t len = s.size();
               if(len)
               {
                  unsigned i = 0;
                  while(i < len)
                  {
                     s[i] = traits_inst.translate(s[i], (_flags & regbase::icase));
                     ++i;
                  }
                  traits_string_type s2;
                  traits_inst.transform_primary(s2, s);
                  equivalents.push(s2);
                  started = true;
                  l = last_none;
                  break;
               }
            }
            fail(REG_ECOLLATE);
            return 0;
         case traits_type::syntax_left_word:
            if((started == false) && (traits_inst.syntax_type((traits_size_type)(traits_uchar_type)*first) == traits_type::syntax_close_set))
            {
               ++first;
               return add_simple(0, re_detail::syntax_element_word_start);
            }
            fail(REG_EBRACK);
            return 0;
         case traits_type::syntax_right_word:
            if((started == false) && (traits_inst.syntax_type((traits_size_type)(traits_uchar_type)*first) == traits_type::syntax_close_set))
            {
               ++first;
               return add_simple(0, re_detail::syntax_element_word_end);
            }
            fail(REG_EBRACK);
            return 0;
         default:
            if(started == false)
            {
               unsigned int t = traits_inst.syntax_type((traits_size_type)(traits_uchar_type)*(base+1));
               if((t != traits_type::syntax_colon) && (t != traits_type::syntax_dot) && (t != traits_type::syntax_equal))
               {
                  first = base;
                  s = (charT)c;
                  goto char_set_literal;
               }
            }
            fail(REG_EBRACK);
            return 0;
         }
         if(first == last)
         {
            fail(REG_EBRACK);
            return 0;
         }
         continue;
      }
      case traits_type::syntax_close_set:
         if(started == false)
         {
            s = (charT)c;
            goto char_set_literal;
         }
         done = true;
         break;
      case traits_type::syntax_dash:
         if(!started)
         {
            s = (charT)c;
            goto char_set_literal;
         }
         ++first;
         if(traits_inst.syntax_type((traits_size_type)(traits_uchar_type)*first) == traits_type::syntax_close_set)
         {
            --first;
            s = (charT)c;
            goto char_set_literal;
         }
         if((singles.empty() == true) || (l != last_single))
         {
            fail(REG_ERANGE);
            return 0;
         }
         ranges.push(singles.peek());
         if(singles.peek().size() <= 1)  // leave digraphs and ligatures in place
            singles.pop();
         l = last_dash;
         continue;
      case traits_type::syntax_slash:
         if(_flags & regbase::escape_in_lists)
         {
            ++first;
            if(first == last)
               continue;
            traits_size_type c = (traits_size_type)(traits_uchar_type)*first;
            // this is only used for the switch(), but cannot be folded in
            // due to a bug in Comeau 4.2.44beta3
            traits_size_type syntax = traits_inst.syntax_type(c);
            switch(syntax)
            {
            case traits_type::syntax_w:
               if(l == last_dash)
               {
                  fail(REG_ERANGE);
                  return 0;
               }
               classes.push(traits_type::char_class_word);
               started = true;
               l = last_none;
               ++first;
               continue;
            case traits_type::syntax_d:
               if(l == last_dash)
               {
                  fail(REG_ERANGE);
                  return 0;
               }
               classes.push(traits_type::char_class_digit);
               started = true;
               l = last_none;
               ++first;
               continue;
            case traits_type::syntax_s:
               if(l == last_dash)
               {
                  fail(REG_ERANGE);
                  return 0;
               }
               classes.push(traits_type::char_class_space);
               started = true;
               l = last_none;
               ++first;
               continue;
            case traits_type::syntax_l:
               if(l == last_dash)
               {
                  fail(REG_ERANGE);
                  return 0;
               }
               classes.push(traits_type::char_class_lower);
               started = true;
               l = last_none;
               ++first;
               continue;
            case traits_type::syntax_u:
               if(l == last_dash)
               {
                  fail(REG_ERANGE);
                  return 0;
               }
               classes.push(traits_type::char_class_upper);
               started = true;
               l = last_none;
               ++first;
               continue;
            case traits_type::syntax_W:
            case traits_type::syntax_D:
            case traits_type::syntax_S:
            case traits_type::syntax_U:
            case traits_type::syntax_L:
               fail(REG_EESCAPE);
               return 0;
            default:
               c = parse_escape(first, last);
               --first;
               s = (charT)c;
               goto char_set_literal;
            }
         }
         else
         {
            s = (charT)c;
            goto char_set_literal;
         }
      default:
         s = (charT)c;
         char_set_literal:
         unsigned i = 0;
         // get string length to stop us going past the end of string (DWA)
         std::size_t len = s.size();
         while(i < len)
         {
            s[i] = traits_inst.translate(s[i], (_flags & regbase::icase));
            ++i;
         }
         started = true;
         if(l == last_dash)
         {
            ranges.push(s);
            l = last_none;
            if(s.size() > 1)   // add ligatures to singles list as well
               singles.push(s);
         }
         else
         {
            singles.push(s);
            l = last_single;
         }
      }
      ++first;
   }
   if(!done)
      return 0;

   typedef typename re_detail::is_byte<charT>::width_type width_type;
   
   re_detail::re_syntax_base* result;
   if(has_digraphs)
      result = compile_set_aux(singles, ranges, classes, equivalents, isnot, re_detail::_wide_type());
   else
      result = compile_set_aux(singles, ranges, classes, equivalents, isnot, width_type());
   #ifdef __BORLANDC__
   // delayed throw:
   if((result == 0) && (_flags & regbase::use_except))
      fail(error_code());
   #endif
   return result;
}

template <class charT, class traits, class Allocator>
re_detail::re_syntax_base* BOOST_REGEX_CALL reg_expression<charT, traits, Allocator>::compile_set_aux(re_detail::jstack<traits_string_type, Allocator>& singles, re_detail::jstack<traits_string_type, Allocator>& ranges, re_detail::jstack<boost::uint_fast32_t, Allocator>& classes, re_detail::jstack<traits_string_type, Allocator>& equivalents, bool isnot, const re_detail::_wide_type&)
{
   size_type base = data.size();
   data.extend(sizeof(re_detail::re_set_long));
   unsigned int csingles = 0;
   unsigned int cranges = 0;
   boost::uint_fast32_t cclasses = 0;
   unsigned int cequivalents = 0;
   bool nocollate_state = flags() & regbase::nocollate;

   while(singles.empty() == false)
   {
      ++csingles;
      const traits_string_type& s = singles.peek();
      std::size_t len = (s.size() + 1) * sizeof(charT);
      std::memcpy(reinterpret_cast<charT*>(data.extend(len)), s.c_str(), len);
      singles.pop();
   }
   while(ranges.empty() == false)
   {
      traits_string_type c1, c2;
      if(nocollate_state)
         c1 = ranges.peek();
      else
         traits_inst.transform(c1, ranges.peek());
      ranges.pop();
      if(nocollate_state)
         c2 = ranges.peek();
      else
         traits_inst.transform(c2, ranges.peek());
      ranges.pop();
      if(c1 < c2)
      {
         // for some reason bc5 crashes when throwing exceptions
         // from here - probably an EH-compiler bug, but hard to
         // be sure...
         // delay throw to later:
         #ifdef __BORLANDC__
         boost::uint_fast32_t f = _flags;
         _flags &= ~regbase::use_except;
         #endif
         fail(REG_ERANGE);
         #ifdef __BORLANDC__
         _flags = f;
         #endif
         return 0;
      }
      ++cranges;
      std::size_t len = (re_detail::re_strlen(c1.c_str()) + 1) * sizeof(charT);
      std::memcpy(data.extend(len), c1.c_str(), len);
      len = (re_detail::re_strlen(c2.c_str()) + 1) * sizeof(charT);
      std::memcpy(data.extend(len), c2.c_str(), len);
   }
   while(classes.empty() == false)
   {
      cclasses |= classes.peek();
      classes.pop();
   }
   while(equivalents.empty() == false)
   {
      ++cequivalents;
      const traits_string_type& s = equivalents.peek();
      std::size_t len = (re_detail::re_strlen(s.c_str()) + 1) * sizeof(charT);
      std::memcpy(reinterpret_cast<charT*>(data.extend(len)), s.c_str(), len);
      equivalents.pop();
   }

   re_detail::re_set_long* dat = reinterpret_cast<re_detail::re_set_long*>(reinterpret_cast<unsigned char*>(data.data()) + base);
   dat->type = re_detail::syntax_element_long_set;
   dat->csingles = csingles;
   dat->cranges = cranges;
   dat->cclasses = cclasses;
   dat->cequivalents = cequivalents;
   dat->isnot = isnot;
   dat->next.i = 0;
   return dat;
}

template <class charT, class traits, class Allocator>
re_detail::re_syntax_base* BOOST_REGEX_CALL reg_expression<charT, traits, Allocator>::compile_set_aux(re_detail::jstack<traits_string_type, Allocator>& singles, re_detail::jstack<traits_string_type, Allocator>& ranges, re_detail::jstack<boost::uint_fast32_t, Allocator>& classes, re_detail::jstack<traits_string_type, Allocator>& equivalents, bool isnot, const re_detail::_narrow_type&)
{
   re_detail::re_set* dat = reinterpret_cast<re_detail::re_set*>(data.extend(sizeof(re_detail::re_set)));
   std::memset(dat, 0, sizeof(re_detail::re_set));

   while(singles.empty() == false)
   {
      dat->_map[(traits_size_type)(traits_uchar_type)*(singles.peek().c_str())] = re_detail::mask_all;
      singles.pop();
   }
   while(ranges.empty() == false)
   {
      traits_string_type c1, c2, c3, c4;

      if(flags() & regbase::nocollate)
         c1 = ranges.peek();
      else
         traits_inst.transform(c1, ranges.peek());
      ranges.pop();
      if(flags() & regbase::nocollate)
         c2 = ranges.peek();
      else
         traits_inst.transform(c2, ranges.peek());
      ranges.pop();

      if(c1 < c2)
      {
         // for some reason bc5 crashes when throwing exceptions
         // from here - probably an EH-compiler bug, but hard to
         // be sure...
         // delay throw to later:
         #ifdef __BORLANDC__
         boost::uint_fast32_t f = _flags;
         _flags &= ~regbase::use_except;
         #endif
         fail(REG_ERANGE);
         #ifdef __BORLANDC__
         _flags = f;
         #endif
         return 0;
      }
      for(unsigned int i = 0; i < 256; ++i)
      {
         c4 = (charT)i;
         if(flags() & regbase::nocollate)
            c3 = c4;
         else
            traits_inst.transform(c3, c4);
         if((c3 <= c1) && (c3 >= c2))
            dat->_map[i] = re_detail::mask_all;
      }
   }