/*

 *

 * 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>

#include <climits>

#ifdef BOOST_MSVC

#pragma warning(push)

#pragma warning(disable: 4103)

#endif

#ifdef BOOST_HAS_ABI_HEADERS

#  include BOOST_ABI_PREFIX

#endif

#ifdef BOOST_MSVC

#pragma warning(pop)

#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 = static_cast<std::ptrdiff_t>(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)

   {

#ifndef BOOST_NO_EXCEPTIONS

      try{

#endif

         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);

#ifndef BOOST_NO_EXCEPTIONS

      }

      catch(...)

      {

         this->m_pmessages->close(cat);

         throw;

      }

#endif

   }

   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();

   }

   cpp_regex_traits_char_layer(const cpp_regex_traits_base<char>& l)

   : cpp_regex_traits_base<char>(l)

   {

      init();

   }

   regex_constants::syntax_type syntax_type(char c)const

   {

      return m_char_map[static_cast<unsigned char>(c)];

   }

   regex_constants::escape_syntax_type escape_syntax_type(char c) const

   {

      return m_char_map[static_cast<unsigned char>(c)];

   }

private:

   regex_constants::syntax_type m_char_map[1u << CHAR_BIT];

   void init();

};

#ifdef BOOST_REGEX_BUGGY_CTYPE_FACET

enum

{

   char_class_space=1<<0, 

   char_class_print=1<<1, 

   char_class_cntrl=1<<2, 

   char_class_upper=1<<3, 

   char_class_lower=1<<4,

   char_class_alpha=1<<5, 

   char_class_digit=1<<6, 

   char_class_punct=1<<7, 

   char_class_xdigit=1<<8,

   char_class_alnum=char_class_alpha|char_class_digit, 

   char_class_graph=char_class_alnum|char_class_punct,

   char_class_blank=1<<9,

   char_class_word=1<<10,

   char_class_unicode=1<<11

};

#endif

//

// class cpp_regex_traits_implementation:

// provides pimpl implementation for cpp_regex_traits.

//

template <class charT>

class cpp_regex_traits_implementation : public cpp_regex_traits_char_layer<charT>

{

public:

   typedef typename cpp_regex_traits<charT>::char_class_type char_class_type;

   typedef typename std::ctype<charT>::mask                  native_mask_type;

#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET

   BOOST_STATIC_CONSTANT(char_class_type, mask_blank = 1u << 24);

   BOOST_STATIC_CONSTANT(char_class_type, mask_word = 1u << 25);

   BOOST_STATIC_CONSTANT(char_class_type, mask_unicode = 1u << 26);

#endif

   typedef std::basic_string<charT> string_type;

   typedef charT char_type;

   //cpp_regex_traits_implementation();

   cpp_regex_traits_implementation(const std::locale& l)

      : cpp_regex_traits_char_layer<charT>(l)

   {

      init();

   }

   cpp_regex_traits_implementation(const cpp_regex_traits_base<charT>& l)

      : cpp_regex_traits_char_layer<charT>(l)

   {

      init();

   }

   std::string error_string(regex_constants::error_type n) const

   {

      if(!m_error_strings.empty())

      {

         std::map<int, std::string>::const_iterator p = m_error_strings.find(n);

         return (p == m_error_strings.end()) ? std::string(get_default_error_string(n)) : p->second;

      }

      return get_default_error_string(n);

   }

   char_class_type lookup_classname(const charT* p1, const charT* p2) const

   {

      char_class_type result = lookup_classname_imp(p1, p2);

      if(result == 0)

      {

         string_type temp(p1, p2);

         this->m_pctype->tolower(&*temp.begin(), &*temp.begin() + temp.size());

         result = lookup_classname_imp(&*temp.begin(), &*temp.begin() + temp.size());

      }

      return result;

   }

   string_type lookup_collatename(const charT* p1, const charT* p2) const;

   string_type transform_primary(const charT* p1, const charT* p2) const;

   string_type transform(const charT* p1, const charT* p2) const;

private:

   std::map<int, std::string>     m_error_strings;   // error messages indexed by numberic ID

   std::map<string_type, char_class_type>  m_custom_class_names; // character class names

   std::map<string_type, string_type>      m_custom_collate_names; // collating element names

   unsigned                       m_collate_type;    // the form of the collation string

   charT                          m_collate_delim;   // the collation group delimiter

   //

   // helpers:

   //

   char_class_type lookup_classname_imp(const charT* p1, const charT* p2) const;

   void init();

#ifdef BOOST_REGEX_BUGGY_CTYPE_FACET

public:

   bool isctype(charT c, char_class_type m)const;

#endif

};

#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET

#if !defined(BOOST_NO_INCLASS_MEMBER_INITIALIZATION)

template <class charT>

typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_blank;

template <class charT>

typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_word;

template <class charT>

typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_unicode;

#endif

#endif

template <class charT>

typename cpp_regex_traits_implementation<charT>::string_type 

   cpp_regex_traits_implementation<charT>::transform_primary(const charT* p1, const charT* p2) const

{

   //

   // PRECONDITIONS:

   //

   // A bug in gcc 3.2 (and maybe other versions as well) treats

   // p1 as a null terminated string, for efficiency reasons 

   // we work around this elsewhere, but just assert here that

   // we adhere to gcc's (buggy) preconditions...

   //

   BOOST_ASSERT(*p2 == 0);

   string_type result;

   //

   // swallowing all exceptions here is a bad idea

   // however at least one std lib will always throw

   // std::bad_alloc for certain arguments...

   //

   try{

      //

      // What we do here depends upon the format of the sort key returned by

      // sort key returned by this->transform:

      //

      switch(m_collate_type)

      {

      case sort_C:

      case sort_unknown:

         // the best we can do is translate to lower case, then get a regular sort key:

         {

            result.assign(p1, p2);

            this->m_pctype->tolower(&*result.begin(), &*result.begin() + result.size());

            result = this->m_pcollate->transform(&*result.begin(), &*result.begin() + result.size());

            break;

         }

      case sort_fixed:

         {

            // get a regular sort key, and then truncate it:

            result.assign(this->m_pcollate->transform(p1, p2));

            result.erase(this->m_collate_delim);

            break;

         }

      case sort_delim:

            // get a regular sort key, and then truncate everything after the delim:

            result.assign(this->m_pcollate->transform(p1, p2));