perl_matcher.hpp

C++的一个好库。。。现在很流行

/*
 *
 * Copyright (c) 2002
 * John Maddock
 *
 * Use, modification and distribution are subject to the 
 * Boost Software License, Version 1.0. (See accompanying file 
 * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
 *
 */

#ifndef BOOST_REGEX_MATCHER_HPP
#define BOOST_REGEX_MATCHER_HPP

#include <boost/regex/v4/iterator_category.hpp>

#ifdef BOOST_HAS_ABI_HEADERS
#  include BOOST_ABI_PREFIX
#endif

namespace boost{
namespace re_detail{

//
// error checking API:
//
BOOST_REGEX_DECL void BOOST_REGEX_CALL verify_options(boost::regex_constants::syntax_option_type ef, match_flag_type mf);
//
// function can_start:
//
template <class charT>
bool can_start(charT c, const unsigned char* map, unsigned char mask)
{
   return ((c < static_cast<charT>(0)) ? true : ((c >= static_cast<charT>(1 << CHAR_BIT)) ? true : map[c] & mask));
}
inline bool can_start(char c, const unsigned char* map, unsigned char mask)
{
   return map[(unsigned char)c] & mask;
}
inline bool can_start(signed char c, const unsigned char* map, unsigned char mask)
{
   return map[(unsigned char)c] & mask;
}
inline bool can_start(unsigned char c, const unsigned char* map, unsigned char mask)
{
   return map[c] & mask;
}
inline bool can_start(unsigned short c, const unsigned char* map, unsigned char mask)
{
   return ((c >= (1 << CHAR_BIT)) ? true : map[c] & mask);
}
#if !defined(__HP_aCC)
#if defined(WCHAR_MIN) && (WCHAR_MIN == 0) && !defined(BOOST_NO_INTRINSIC_WCHAR_T)
inline bool can_start(wchar_t c, const unsigned char* map, unsigned char mask)
{
   return ((c >= (1 << CHAR_BIT)) ? true : map[c] & mask);
}
#endif
#endif


//
// Unfortunately Rogue Waves standard library appears to have a bug
// in std::basic_string::compare that results in eroneous answers
// in some cases (tested with Borland C++ 5.1, Rogue Wave lib version
// 0x020101) the test case was:
// {39135,0} < {0xff,0}
// which succeeds when it should not.
//
#ifndef _RWSTD_VER
#if !BOOST_WORKAROUND(BOOST_MSVC, < 1310)
template <class C, class T, class A>
inline int string_compare(const std::basic_string<C,T,A>& s, const C* p)
{ 
   if(0 == *p)
   {
      if(s.empty() || ((s.size() == 1) && (s[0] == 0)))
         return 0;
   }
   return s.compare(p); 
}
#endif
#else
#if !BOOST_WORKAROUND(BOOST_MSVC, < 1310)
template <class C, class T, class A>
inline int string_compare(const std::basic_string<C,T,A>& s, const C* p)
{ 
   if(0 == *p)
   {
      if(s.empty() || ((s.size() == 1) && (s[0] == 0)))
         return 0;
   }
   return s.compare(p); 
}
#endif
inline int string_compare(const std::string& s, const char* p)
{ return std::strcmp(s.c_str(), p); }
# ifndef BOOST_NO_WREGEX
inline int string_compare(const std::wstring& s, const wchar_t* p)
{ return std::wcscmp(s.c_str(), p); }
#endif
#endif
template <class Seq, class C>
inline int string_compare(const Seq& s, const C* p)
{
   std::size_t i = 0;
   while((i < s.size()) && (p[i] == s[i]))
   {
      ++i;
   }
   return (i == s.size()) ? -p[i] : s[i] - p[i];
}
# define STR_COMP(s,p) string_compare(s,p)

template<class charT>
inline const charT* re_skip_past_null(const charT* p)
{
  while (*p != static_cast<charT>(0)) ++p;
  return ++p;
}

template <class iterator, class charT, class traits_type, class char_classT>
iterator BOOST_REGEX_CALL re_is_set_member(iterator next, 
                          iterator last, 
                          const re_set_long<char_classT>* set_, 
                          const regex_data<charT, traits_type>& e, bool icase)
{   
   const charT* p = reinterpret_cast<const charT*>(set_+1);
   iterator ptr;
   unsigned int i;
   //bool icase = e.m_flags & regex_constants::icase;

   if(next == last) return next;

   typedef typename traits_type::string_type traits_string_type;
   const ::boost::regex_traits_wrapper<traits_type>& traits_inst = *(e.m_ptraits);
   
   // dwa 9/13/00 suppress incorrect MSVC warning - it claims this is never
   // referenced
   (void)traits_inst;

   // try and match a single character, could be a multi-character
   // collating element...
   for(i = 0; i < set_->csingles; ++i)
   {
      ptr = next;
      if(*p == static_cast<charT>(0))
      {
         // treat null string as special case:
         if(traits_inst.translate(*ptr, icase) != *p)
         {
            while(*p == static_cast<charT>(0))++p;
            continue;
         }
         return set_->isnot ? next : (ptr == next) ? ++next : ptr;
      }
      else
      {
         while(*p && (ptr != last))
         {
            if(traits_inst.translate(*ptr, icase) != *p)
               break;
            ++p;
            ++ptr;
         }

         if(*p == static_cast<charT>(0)) // if null we've matched
            return set_->isnot ? next : (ptr == next) ? ++next : ptr;

         p = re_skip_past_null(p);     // skip null
      }
   }

   charT col = traits_inst.translate(*next, icase);


   if(set_->cranges || set_->cequivalents)
   {
      traits_string_type s1;
      //
      // try and match a range, NB only a single character can match
      if(set_->cranges)
      {
         if((e.m_flags & regex_constants::collate) == 0)
            s1.assign(1, col);
         else
         {
            charT a[2] = { col, charT(0), };
            s1 = traits_inst.transform(a, a + 1);
         }
         for(i = 0; i < set_->cranges; ++i)
         {
            if(STR_COMP(s1, p) >= 0)
            {
               do{ ++p; }while(*p);
               ++p;
               if(STR_COMP(s1, p) <= 0)
                  return set_->isnot ? next : ++next;
            }
            else
            {
               // skip first string
               do{ ++p; }while(*p);
               ++p;
            }
            // skip second string
            do{ ++p; }while(*p);
            ++p;
         }
      }
      //
      // try and match an equivalence class, NB only a single character can match
      if(set_->cequivalents)
      {
         charT a[2] = { col, charT(0), };
         s1 = traits_inst.transform_primary(a, a +1);
         for(i = 0; i < set_->cequivalents; ++i)
         {
            if(STR_COMP(s1, p) == 0)
               return set_->isnot ? next : ++next;
            // skip string
            do{ ++p; }while(*p);
            ++p;
         }
      }
   }
   if(traits_inst.isctype(col, set_->cclasses) == true)
      return set_->isnot ? next : ++next;
   if((set_->cnclasses != 0) && (traits_inst.isctype(col, set_->cnclasses) == false))
      return set_->isnot ? next : ++next;
   return set_->isnot ? ++next : next;
}

template <class BidiIterator>
class repeater_count
{
   repeater_count** stack;
   repeater_count* next;
   int id;
   std::size_t count;        // the number of iterations so far
   BidiIterator start_pos;   // where the last repeat started
public:
   repeater_count(repeater_count** s)
   {
      stack = s;
      next = 0;
      id = -1;
      count = 0;
   }
   repeater_count(int i, repeater_count** s, BidiIterator start)
      : start_pos(start)
   {
      id = i;
      stack = s;
      next = *stack;
      *stack = this;
      if(id > next->id)
         count = 0;
      else
      {
         repeater_count* p = next;