cpp_regex_traits.hpp

C++的一个好库。。。现在很流行

/*
 *
 * Copyright (c) 2004
 * 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)
 *
 */
 
 /*
  *   LOCATION:    see http://www.boost.org for most recent version.
  *   FILE         cpp_regex_traits.hpp
  *   VERSION      see <boost/version.hpp>
  *   DESCRIPTION: Declares regular expression traits class cpp_regex_traits.
  */

#ifndef BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED
#define BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED

#include <boost/config.hpp>

#ifndef BOOST_NO_STD_LOCALE

#ifndef BOOST_RE_PAT_EXCEPT_HPP
#include <boost/regex/pattern_except.hpp>
#endif
#ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED
#include <boost/regex/v4/regex_traits_defaults.hpp>
#endif
#ifdef BOOST_HAS_THREADS
#include <boost/regex/pending/static_mutex.hpp>
#endif
#ifndef BOOST_REGEX_PRIMARY_TRANSFORM
#include <boost/regex/v4/primary_transform.hpp>
#endif
#ifndef BOOST_REGEX_OBJECT_CACHE_HPP
#include <boost/regex/pending/object_cache.hpp>
#endif

#include <istream>
#include <ios>

#ifdef BOOST_HAS_ABI_HEADERS
#  include BOOST_ABI_PREFIX
#endif

#ifdef BOOST_MSVC
#pragma warning(push)
#pragma warning(disable:4786)
#endif

namespace boost{ 

//
// forward declaration is needed by some compilers:
//
template <class charT>
class cpp_regex_traits;
   
namespace re_detail{

//
// class parser_buf:
// acts as a stream buffer which wraps around a pair of pointers:
//
template <class charT,
          class traits = ::std::char_traits<charT> >
class parser_buf : public ::std::basic_streambuf<charT, traits>
{
   typedef ::std::basic_streambuf<charT, traits> base_type;
   typedef typename base_type::int_type int_type;
   typedef typename base_type::char_type char_type;
   typedef typename base_type::pos_type pos_type;
   typedef ::std::streamsize streamsize;
   typedef typename base_type::off_type off_type;
public:
   parser_buf() : base_type() { setbuf(0, 0); }
   const charT* getnext() { return this->gptr(); }
protected:
   std::basic_streambuf<charT, traits>* setbuf(char_type* s, streamsize n);
   typename parser_buf<charT, traits>::pos_type seekpos(pos_type sp, ::std::ios_base::openmode which);
   typename parser_buf<charT, traits>::pos_type seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which);
private:
   parser_buf& operator=(const parser_buf&);
   parser_buf(const parser_buf&);
};

template<class charT, class traits>
std::basic_streambuf<charT, traits>*
parser_buf<charT, traits>::setbuf(char_type* s, streamsize n)
{
   this->setg(s, s, s + n);
   return this;
}

template<class charT, class traits>
typename parser_buf<charT, traits>::pos_type
parser_buf<charT, traits>::seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which)
{
   if(which & ::std::ios_base::out)
      return pos_type(off_type(-1));
   std::ptrdiff_t size = this->egptr() - this->eback();
   std::ptrdiff_t pos = this->gptr() - this->eback();
   charT* g = this->eback();
   switch(way)
   {
   case ::std::ios_base::beg:
      if((off < 0) || (off > size))
         return pos_type(off_type(-1));
      else
         this->setg(g, g + off, g + size);
      break;
   case ::std::ios_base::end:
      if((off < 0) || (off > size))
         return pos_type(off_type(-1));
      else
         this->setg(g, g + size - off, g + size);
      break;
   case ::std::ios_base::cur:
   {
      std::ptrdiff_t newpos = pos + off;
      if((newpos < 0) || (newpos > size))
         return pos_type(off_type(-1));
      else
         this->setg(g, g + newpos, g + size);
      break;
   }
   default: ;
   }
#ifdef BOOST_MSVC
#pragma warning(push)
#pragma warning(disable:4244)
#endif
   return static_cast<pos_type>(this->gptr() - this->eback());
#ifdef BOOST_MSVC
#pragma warning(pop)
#endif
}

template<class charT, class traits>
typename parser_buf<charT, traits>::pos_type
parser_buf<charT, traits>::seekpos(pos_type sp, ::std::ios_base::openmode which)
{
   if(which & ::std::ios_base::out)
      return pos_type(off_type(-1));
   off_type size = static_cast<off_type>(this->egptr() - this->eback());
   charT* g = this->eback();
   if(off_type(sp) <= size)
   {
      this->setg(g, g + off_type(sp), g + size);
   }
   return pos_type(off_type(-1));
}

//
// class cpp_regex_traits_base:
// acts as a container for locale and the facets we are using.
//
template <class charT>
struct cpp_regex_traits_base
{
   cpp_regex_traits_base(const std::locale& l)
   { imbue(l); }
   std::locale imbue(const std::locale& l);

   std::locale m_locale;
   std::ctype<charT> const* m_pctype;
#ifndef BOOST_NO_STD_MESSAGES
   std::messages<charT> const* m_pmessages;
#endif
   std::collate<charT> const* m_pcollate;

   bool operator<(const cpp_regex_traits_base& b)const
   {
      if(m_pctype == b.m_pctype)
      {
#ifndef BOOST_NO_STD_MESSAGES
         if(m_pmessages == b.m_pmessages)
         {
         }
         return m_pmessages < b.m_pmessages;
#else
         return m_pcollate < b.m_pcollate;
#endif
      }
      return m_pctype < b.m_pctype;
   }
   bool operator==(const cpp_regex_traits_base& b)const
   {
      return (m_pctype == b.m_pctype) 
#ifndef BOOST_NO_STD_MESSAGES
         && (m_pmessages == b.m_pmessages) 
#endif
         && (m_pcollate == b.m_pcollate);
   }
};

template <class charT>
std::locale cpp_regex_traits_base<charT>::imbue(const std::locale& l)
{
   std::locale result(m_locale);
   m_locale = l;
   m_pctype = &BOOST_USE_FACET(std::ctype<charT>, l);
#ifndef BOOST_NO_STD_MESSAGES
   m_pmessages = &BOOST_USE_FACET(std::messages<charT>, l);
#endif
   m_pcollate = &BOOST_USE_FACET(std::collate<charT>, l);
   return result;
}

//
// class cpp_regex_traits_char_layer:
// implements methods that require specialisation for narrow characters:
//
template <class charT>
class cpp_regex_traits_char_layer : public cpp_regex_traits_base<charT>
{
   typedef std::basic_string<charT> string_type;
   typedef std::map<charT, regex_constants::syntax_type> map_type;
   typedef typename map_type::const_iterator map_iterator_type;
public:
   cpp_regex_traits_char_layer(const std::locale& l)
      : cpp_regex_traits_base<charT>(l)
   {
      init();
   }
   cpp_regex_traits_char_layer(const cpp_regex_traits_base<charT>& b)
      : cpp_regex_traits_base<charT>(b)
   {
      init();
   }
   void init();

   regex_constants::syntax_type syntax_type(charT c)const
   {
      map_iterator_type i = m_char_map.find(c);
      return ((i == m_char_map.end()) ? 0 : i->second);
   }
   regex_constants::escape_syntax_type escape_syntax_type(charT c) const
   {
      map_iterator_type i = m_char_map.find(c);
      if(i == m_char_map.end())
      {
         if(this->m_pctype->is(std::ctype_base::lower, c)) return regex_constants::escape_type_class;
         if(this->m_pctype->is(std::ctype_base::upper, c)) return regex_constants::escape_type_not_class;
         return 0;
      }
      return i->second;
   }

private:
   string_type get_default_message(regex_constants::syntax_type);
   // TODO: use a hash table when available!
   map_type m_char_map;
};

template <class charT>
void cpp_regex_traits_char_layer<charT>::init()
{
   // we need to start by initialising our syntax map so we know which
   // character is used for which purpose:
#ifndef BOOST_NO_STD_MESSAGES
#ifndef __IBMCPP__
   typename std::messages<charT>::catalog cat = static_cast<std::messages<char>::catalog>(-1);
#else
   typename std::messages<charT>::catalog cat = reinterpret_cast<std::messages<char>::catalog>(-1);
#endif
   std::string cat_name(cpp_regex_traits<charT>::get_catalog_name());
   if(cat_name.size())
   {
      cat = this->m_pmessages->open(
         cat_name, 
         this->m_locale);
      if((int)cat < 0)
      {
         std::string m("Unable to open message catalog: ");
         std::runtime_error err(m + cat_name);
         boost::re_detail::raise_runtime_error(err);
      }
   }
   //
   // if we have a valid catalog then load our messages:
   //
   if((int)cat >= 0)
   {
      try{
         for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
         {
            string_type mss = this->m_pmessages->get(cat, 0, i, get_default_message(i));
            for(typename string_type::size_type j = 0; j < mss.size(); ++j)
            {
               m_char_map[mss[j]] = i;
            }
         }
         this->m_pmessages->close(cat);
      }
      catch(...)
      {
         this->m_pmessages->close(cat);
         throw;
      }
   }
   else
   {
#endif
      for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
      {
         const char* ptr = get_default_syntax(i);
         while(ptr && *ptr)
         {
            m_char_map[this->m_pctype->widen(*ptr)] = i;
            ++ptr;
         }
      }
#ifndef BOOST_NO_STD_MESSAGES
   }
#endif
}

template <class charT>
typename cpp_regex_traits_char_layer<charT>::string_type 
   cpp_regex_traits_char_layer<charT>::get_default_message(regex_constants::syntax_type i)
{
   const char* ptr = get_default_syntax(i);
   string_type result;
   while(ptr && *ptr)
   {
      result.append(1, this->m_pctype->widen(*ptr));
      ++ptr;
   }
   return result;
}

//
// specialised version for narrow characters:
//
template <>
class BOOST_REGEX_DECL cpp_regex_traits_char_layer<char> : public cpp_regex_traits_base<char>
{
   typedef std::string string_type;
public:
   cpp_regex_traits_char_layer(const std::locale& l)
   : cpp_regex_traits_base<char>(l)
   {
      init();