perl_matcher_non_recursive.hpp

C++的一个好库。。。现在很流行

   const re_repeat* rep = static_cast<const re_repeat*>(pstate);

   // find out which of these two alternatives we need to take:
   bool take_first, take_second;
   if(position == last)
   {
      take_first = rep->can_be_null & mask_take;
      take_second = rep->can_be_null & mask_skip;
   }
   else
   {
      take_first = can_start(*position, rep->_map, (unsigned char)mask_take);
      take_second = can_start(*position, rep->_map, (unsigned char)mask_skip);
   }

   if(take_first || (next_count->get_id() != rep->id))
   {
      // we're moving to a different repeat from the last
      // one, so set up a counter object:
      push_repeater_count(rep->id, &next_count);
   }
   //
   // If we've had at least one repeat already, and the last one 
   // matched the NULL string then set the repeat count to
   // maximum:
   //
   next_count->check_null_repeat(position, rep->max);

   if(next_count->get_count() < rep->min)
   {
      // we must take the repeat:
      if(take_first)
      {
         // increase the counter:
         ++(*next_count);
         pstate = rep->next.p;
         return true;
      }
      return false;
   }

   bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);   
   if(greedy)
   {
      // try and take the repeat if we can:
      if((next_count->get_count() < rep->max) && take_first)
      {
         if(take_second)
         {
            // store position in case we fail:
            push_alt(rep->alt.p);
         }
         // increase the counter:
         ++(*next_count);
         pstate = rep->next.p;
         return true;
      }
      else if(take_second)
      {
         pstate = rep->alt.p;
         return true;
      }
      return false; // can't take anything, fail...
   }
   else // non-greedy
   {
      // try and skip the repeat if we can:
      if(take_second)
      {
         if((next_count->get_count() < rep->max) && take_first)
         {
            // store position in case we fail:
            push_non_greedy_repeat(rep->next.p);
         }
         pstate = rep->alt.p;
         return true;
      }
      if((next_count->get_count() < rep->max) && take_first)
      {
         // increase the counter:
         ++(*next_count);
         pstate = rep->next.p;
         return true;
      }
   }
   return false;
#ifdef __BORLANDC__
#pragma option pop
#endif
#ifdef BOOST_MSVC
#pragma warning(pop)
#endif
}

template <class BidiIterator, class Allocator, class traits>
bool perl_matcher<BidiIterator, Allocator, traits>::match_dot_repeat_slow()
{
   unsigned count = 0;
   const re_repeat* rep = static_cast<const re_repeat*>(pstate);
   re_syntax_base* psingle = rep->next.p;
   // match compulsary repeats first:
   while(count < rep->min)
   {
      pstate = psingle;
      if(!match_wild())
         return false;
      ++count;
   }
   bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);   
   if(greedy)
   {
      // repeat for as long as we can:
      while(count < rep->max)
      {
         pstate = psingle;
         if(!match_wild())
            break;
         ++count;
      }
      // remember where we got to if this is a leading repeat:
      if((rep->leading) && (count < rep->max))
         restart = position;
      // push backtrack info if available:
      if(count - rep->min)
         push_single_repeat(count, rep, position, saved_state_greedy_single_repeat);
      // jump to next state:
      pstate = rep->alt.p;
      return true;
   }
   else
   {
      // non-greedy, push state and return true if we can skip:
      if(count < rep->max)
         push_single_repeat(count, rep, position, saved_state_rep_slow_dot);
      pstate = rep->alt.p;
      return (position == last) ? (rep->can_be_null & mask_skip) : can_start(*position, rep->_map, mask_skip);
   }
}

template <class BidiIterator, class Allocator, class traits>
bool perl_matcher<BidiIterator, Allocator, traits>::match_dot_repeat_fast()
{
   if(m_match_flags & match_not_dot_null)
      return match_dot_repeat_slow();
   if((static_cast<const re_dot*>(pstate->next.p)->mask & match_any_mask) == 0)
      return match_dot_repeat_slow();

   const re_repeat* rep = static_cast<const re_repeat*>(pstate);
   bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);   
   unsigned count = static_cast<unsigned>((std::min)(static_cast<unsigned>(::boost::re_detail::distance(position, last)), static_cast<unsigned>(greedy ? rep->max : rep->min)));
   if(rep->min > count)
   {
      position = last;
      return false;  // not enough text left to match
   }
   std::advance(position, count);

   if(greedy)
   {
      if((rep->leading) && (count < rep->max))
         restart = position;
      // push backtrack info if available:
      if(count - rep->min)
         push_single_repeat(count, rep, position, saved_state_greedy_single_repeat);
      // jump to next state:
      pstate = rep->alt.p;
      return true;
   }
   else
   {
      // non-greedy, push state and return true if we can skip:
      if(count < rep->max)
         push_single_repeat(count, rep, position, saved_state_rep_fast_dot);
      pstate = rep->alt.p;
      return (position == last) ? (rep->can_be_null & mask_skip) : can_start(*position, rep->_map, mask_skip);
   }
}

template <class BidiIterator, class Allocator, class traits>
bool perl_matcher<BidiIterator, Allocator, traits>::match_char_repeat()
{
#ifdef BOOST_MSVC
#pragma warning(push)
#pragma warning(disable:4127)
#endif
#ifdef __BORLANDC__
#pragma option push -w-8008 -w-8066 -w-8004
#endif
   const re_repeat* rep = static_cast<const re_repeat*>(pstate);
   BOOST_ASSERT(1 == static_cast<const re_literal*>(rep->next.p)->length);
   const char_type what = *reinterpret_cast<const char_type*>(static_cast<const re_literal*>(rep->next.p) + 1);
   std::size_t count = 0;
   //
   // start by working out how much we can skip:
   //
   bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);   
   std::size_t desired = greedy ? rep->max : rep->min;
   if(::boost::is_random_access_iterator<BidiIterator>::value)
   {
      BidiIterator end = position;
      std::advance(end, (std::min)((std::size_t)::boost::re_detail::distance(position, last), desired));
      BidiIterator origin(position);
      while((position != end) && (traits_inst.translate(*position, icase) == what))
      {
         ++position;
      }
      count = (unsigned)::boost::re_detail::distance(origin, position);
   }
   else
   {
      while((count < desired) && (position != last) && (traits_inst.translate(*position, icase) == what))
      {
         ++position;
         ++count;
      }
   }

   if(count < rep->min)
      return false;

   if(greedy)
   {
      if((rep->leading) && (count < rep->max))
         restart = position;
      // push backtrack info if available:
      if(count - rep->min)
         push_single_repeat(count, rep, position, saved_state_greedy_single_repeat);
      // jump to next state:
      pstate = rep->alt.p;
      return true;
   }
   else
   {
      // non-greedy, push state and return true if we can skip:
      if(count < rep->max)
         push_single_repeat(count, rep, position, saved_state_rep_char);
      pstate = rep->alt.p;
      return (position == last) ? (rep->can_be_null & mask_skip) : can_start(*position, rep->_map, mask_skip);
   }
#ifdef __BORLANDC__
#pragma option pop
#endif
#ifdef BOOST_MSVC
#pragma warning(pop)
#endif
}

template <class BidiIterator, class Allocator, class traits>
bool perl_matcher<BidiIterator, Allocator, traits>::match_set_repeat()
{
#ifdef BOOST_MSVC
#pragma warning(push)
#pragma warning(disable:4127)
#endif
#ifdef __BORLANDC__
#pragma option push -w-8008 -w-8066 -w-8004
#endif
   const re_repeat* rep = static_cast<const re_repeat*>(pstate);
   const unsigned char* map = static_cast<const re_set*>(rep->next.p)->_map;
   std::size_t count = 0;
   //
   // start by working out how much we can skip:
   //
   bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);   
   std::size_t desired = greedy ? rep->max : rep->min;
   if(::boost::is_random_access_iterator<BidiIterator>::value)
   {
      BidiIterator end = position;
      std::advance(end, (std::min)((std::size_t)::boost::re_detail::distance(position, last), desired));
      BidiIterator origin(position);
      while((position != end) && map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])
      {
         ++position;
      }
      count = (unsigned)::boost::re_detail::distance(origin, position);
   }
   else
   {
      while((count < desired) && (position != last) && map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])
      {
         ++position;
         ++count;
      }
   }

   if(count < rep->min)
      return false;

   if(greedy)
   {
      if((rep->leading) && (count < rep->max))
         restart = position;
      // push backtrack info if available:
      if(count - rep->min)
         push_single_repeat(count, rep, position, saved_state_greedy_single_repeat);
      // jump to next state:
      pstate = rep->alt.p;
      return true;
   }
   else
   {
      // non-greedy, push state and return true if we can skip:
      if(count < rep->max)
         push_single_repeat(count, rep, position, saved_state_rep_short_set);
      pstate = rep->alt.p;
      return (position == last) ? (rep->can_be_null & mask_skip) : can_start(*position, rep->_map, mask_skip);
   }
#ifdef __BORLANDC__
#pragma option pop
#endif
#ifdef BOOST_MSVC
#pragma warning(pop)
#endif
}

template <class BidiIterator, class Allocator, class traits>
bool perl_matcher<BidiIterator, Allocator, traits>::match_long_set_repeat()
{
#ifdef BOOST_MSVC
#pragma warning(push)
#pragma warning(disable:4127)
#endif
#ifdef __BORLANDC__
#pragma option push -w-8008 -w-8066 -w-8004
#endif
   typedef typename traits::char_class_type mask_type;
   const re_repeat* rep = static_cast<const re_repeat*>(pstate);
   const re_set_long<mask_type>* set = static_cast<const re_set_long<mask_type>*>(pstate->next.p);
   std::size_t count = 0;
   //
   // start by working out how much we can skip:
   //
   bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);   
   std::size_t desired = greedy ? rep->max : rep->min;
   if(::boost::is_random_access_iterator<BidiIterator>::value)
   {
      BidiIterator end = position;
      std::advance(end, (std::min)((std::size_t)::boost::re_detail::distance(position, last), desired));
      BidiIterator origin(position);
      while((position != end) && (position != re_is_set_member(position, last, set, re.get_data(), icase)))
      {
         ++position;
      }
      count = (unsigned)::boost::re_detail::distance(origin, position);
   }
   else
   {
      while((count < desired) && (position != last) && (position != re_is_set_member(position, last, set, re.get_data(), icase)))
      {
         ++position;
         ++count;
      }
   }

   if(count < rep->min)
      return false;

   if(greedy)
   {
      if((rep->leading) && (count < rep->max))
         restart = position;
      // push backtrack info if available:
      if(count - rep->min)
         push_single_repeat(count, rep, position, saved_state_greedy_single_repeat);
      // jump to next state:
      pstate = rep->alt.p;
      return true;
   }
   else
   {
      // non-greedy, push state and return true if we can skip:
      if(count < rep->max)
         push_single_repeat(count, rep, position, saved_state_rep_long_set);
      pstate = rep->alt.p;
      return (position == last) ? (rep->can_be_null & mask_skip) : can_start(*position, rep->_map, mask_skip);
   }
#ifdef __BORLANDC__
#pragma option pop
#endif
#ifdef BOOST_MSVC
#pragma warning(pop)
#endif
}

/****************************************************************************

Unwind and associated proceedures follow, these perform what normal stack
unwinding does in the recursive implementation.

****************************************************************************/

template <class BidiIterator, class Allocator, class traits>
bool perl_matcher<BidiIterator, Allocator, traits>::unwind(bool have_match)
{
   static unwind_proc_type const s_unwind_table[14] = 
   {
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_end,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_paren,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_recursion_stopper,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_assertion,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_alt,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_repeater_counter,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_extra_block,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_greedy_single_repeat,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_slow_dot_repeat,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_fast_dot_repeat,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_char_repeat,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_short_set_repeat,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_long_set_repeat,
      &perl_matcher<BidiIterator, Allocator, traits>::unwind_non_greedy_repeat,
   };

   m_recursive_result = have_match;
   unwind_proc_type unwinder;
   bool cont;
   //
   // keep unwinding our stack until we have something to do:
   //
   do
   {
      unwinder = s_unwind_table[m_backup_state->id];
      cont = (this->*unwinder)(m_recursive_result);
   }while(cont);
   //
   // return true if we have more states to try:
   //
   return pstate ? true : false;
}

template <class BidiIterator, class Allocator, class traits>
bool perl_matcher<BidiIterator, Allocator, traits>::unwind_end(bool)
{
   pstate = 0;   // nothing left to search
   return false; // end of stack nothing more to search
}

template <class BidiIterator, class Allocator, class traits>
bool perl_matcher<BidiIterator, Allocator, traits>::unwind_paren(bool have_match)
{
   saved_matched_paren<BidiIterator>* pmp = static_cast<saved_matched_paren<BidiIterator>*>(m_backup_state);
   // restore previous values if no match was found:
   if(have_match == false)
   {
      m_presult->set_first(pmp->sub.first, pmp->index);
      m_presult->set_second(pmp->sub.second, pmp->index, pmp->sub.matched);
   }
#ifdef BOOST_REGEX_MATCH_EXTRA
   //
   // we have a match, push the capture information onto the stack:
   //
   else if(pmp->sub.matched && (match_extra & m_match_flags))
      ((*m_presult)[pmp->index]).get_captures().push_back(pmp->sub);
#endif