regex_compile.hpp

boost库提供标准的C++ API 配合dev c++使用,功能更加强大

template <class charT, class traits, class Allocator>
re_detail::re_syntax_base* BOOST_REGEX_CALL reg_expression<charT, traits, Allocator>::compile_set_simple(re_detail::re_syntax_base* dat, unsigned long cls, bool isnot)
{
   typedef typename re_detail::is_byte<charT>::width_type width_type;
   re_detail::jstack<traits_string_type, Allocator> singles(64, data.allocator());
   re_detail::jstack<traits_string_type, Allocator> ranges(64, data.allocator());
   re_detail::jstack<boost::uint_fast32_t, Allocator> classes(64, data.allocator());
   re_detail::jstack<traits_string_type, Allocator> equivalents(64, data.allocator());
   classes.push(cls);
   if(dat)
   {
      data.align();
      dat->next.i = data.size();
   }
   return compile_set_aux(singles, ranges, classes, equivalents, isnot, width_type());
}

template <class charT, class traits, class Allocator>
re_detail::re_syntax_base* BOOST_REGEX_CALL reg_expression<charT, traits, Allocator>::compile_set(const charT*& arg_first, const charT* arg_last)
{
   re_detail::jstack<traits_string_type, Allocator> singles(64, data.allocator());
   re_detail::jstack<traits_string_type, Allocator> ranges(64, data.allocator());
   re_detail::jstack<boost::uint_fast32_t, Allocator> classes(64, data.allocator());
   re_detail::jstack<traits_string_type, Allocator> equivalents(64, data.allocator());
   bool has_digraphs = false;
   jm_assert(traits_inst.syntax_type((traits_size_type)(traits_uchar_type)*arg_first) == traits_type::syntax_open_set);
   ++arg_first;
   bool started = false;
   bool done = false;
   bool isnot = false;

   enum last_type
   {
      last_single,
      last_none,
      last_dash
   };

   unsigned l = last_none;
   traits_string_type s;

   while((arg_first != arg_last) && !done)
   {
      traits_size_type c = (traits_size_type)(traits_uchar_type)*arg_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_caret:
         if(!started && !isnot)
         {
            isnot = true;
         }
         else
         {
            s = (charT)c;
            goto char_set_literal;
         }
         break;
      case traits_type::syntax_open_set:
      {
         if((_flags & char_classes) == 0)
         {
            s = (charT)c;
            goto char_set_literal;
         }
         // check to see if we really have a class:
         const charT* base = arg_first;
         // this is only used for the switch(), but cannot be folded in
         // due to a bug in Comeau 4.2.44beta3
    unsigned int inner_set = parse_inner_set(arg_first, arg_last);
         switch(inner_set)
         {
         case traits_type::syntax_colon:
            {
               if(l == last_dash)
               {
                  fail(REG_ERANGE);
                  return 0;
               }
               boost::uint_fast32_t id = traits_inst.lookup_classname(base+2, arg_first-2);
               if(_flags & regex_constants::icase)
               {
                  if((id == traits_type::char_class_upper) || (id == traits_type::char_class_lower))
                  {
                     id = traits_type::char_class_alpha;
                  }
               }
               if(id == 0)
               {
                  fail(REG_ECTYPE);
                  return 0;
               }
               classes.push(id);
               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, arg_first-2))
            {
               --arg_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, arg_first-2))
            {
               std::size_t len = s.size();
               if(len)
               {
                  unsigned i = 0;
                  while(i < len)
                  {
                     s[i] = traits_inst.translate(s[i], (_flags & regex_constants::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)*arg_first) == traits_type::syntax_close_set))
            {
               ++arg_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)*arg_first) == traits_type::syntax_close_set))
            {
               ++arg_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))
               {
                  arg_first = base;
                  s = (charT)c;
                  goto char_set_literal;
               }
            }
            fail(REG_EBRACK);
            return 0;
         }
         if(arg_first == arg_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;
         }
         ++arg_first;
         if(traits_inst.syntax_type((traits_size_type)(traits_uchar_type)*arg_first) == traits_type::syntax_close_set)
         {
            --arg_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 & regex_constants::escape_in_lists)
         {
            ++arg_first;
            if(arg_first == arg_last)
               continue;
            /*traits_size_type*/ c = (traits_size_type)(traits_uchar_type)*arg_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 syntax4 = traits_inst.syntax_type(c);
            switch(syntax4)
            {
            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;
               ++arg_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;
               ++arg_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;
               ++arg_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;
               ++arg_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;
               ++arg_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(arg_first, arg_last);
               --arg_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 & regex_constants::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;
         }
      }
      ++arg_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 & regex_constants::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() & regex_constants::collate);
   bool singleton = true;

   while(singles.empty() == false)
   {
      ++csingles;
      const traits_string_type& s = singles.peek();
      std::size_t len = (s.size() + 1) * sizeof(charT);
      if(len > sizeof(charT))
         singleton = false;
      std::memcpy(reinterpret_cast<charT*>(data.extend(len)), s.c_str(), len);