view release on metacpan or  search on metacpan

include/boost/regex/v4/cpp_regex_traits.hpp  view on Meta::CPAN

{
   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), m_is(&m_sbuf)
   {
      init();
   }
   cpp_regex_traits_implementation(const cpp_regex_traits_base<charT>& l)
      : cpp_regex_traits_char_layer<charT>(l), m_is(&m_sbuf)
   {
      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;
   re_detail::parser_buf<charT>   m_sbuf;            // buffer for parsing numbers.
   std::basic_istream<charT>      m_is;              // stream for parsing numbers.
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));
            std::size_t i;
            for(i = 0; i < result.size(); ++i)
            {
               if(result[i] == m_collate_delim)
                  break;
            }
            result.erase(i);
            break;
      }
   }catch(...){}

include/boost/regex/v4/cpp_regex_traits.hpp  view on Meta::CPAN

#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)
   {
      //
      // Error messages:
      //
      for(boost::regex_constants::error_type i = static_cast<boost::regex_constants::error_type>(0); 
         i <= boost::regex_constants::error_unknown; 
         i = static_cast<boost::regex_constants::error_type>(i + 1))
      {
         const char* p = get_default_error_string(i);
         string_type default_message;
         while(*p)
         {
            default_message.append(1, this->m_pctype->widen(*p));
            ++p;
         }
         string_type s = this->m_pmessages->get(cat, 0, i+200, default_message);
         std::string result;
         for(std::string::size_type j = 0; j < s.size(); ++j)
         {
            result.append(1, this->m_pctype->narrow(s[j], 0));
         }
         m_error_strings[i] = result;
      }
      //
      // Custom class names:
      //
#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
      static const char_class_type masks[14] = 
      {
         std::ctype<charT>::alnum,
         std::ctype<charT>::alpha,
         std::ctype<charT>::cntrl,
         std::ctype<charT>::digit,
         std::ctype<charT>::graph,
         std::ctype<charT>::lower,
         std::ctype<charT>::print,
         std::ctype<charT>::punct,
         std::ctype<charT>::space,
         std::ctype<charT>::upper,
         std::ctype<charT>::xdigit,
         cpp_regex_traits_implementation<charT>::mask_blank,
         cpp_regex_traits_implementation<charT>::mask_word,
         cpp_regex_traits_implementation<charT>::mask_unicode,
      };
#else
      static const char_class_type masks[14] = 
      {
         ::boost::re_detail::char_class_alnum,
         ::boost::re_detail::char_class_alpha,
         ::boost::re_detail::char_class_cntrl,
         ::boost::re_detail::char_class_digit,
         ::boost::re_detail::char_class_graph,
         ::boost::re_detail::char_class_lower,
         ::boost::re_detail::char_class_print,
         ::boost::re_detail::char_class_punct,
         ::boost::re_detail::char_class_space,
         ::boost::re_detail::char_class_upper,
         ::boost::re_detail::char_class_xdigit,
         ::boost::re_detail::char_class_blank,
         ::boost::re_detail::char_class_word,
         ::boost::re_detail::char_class_unicode,
      };
#endif
      static const string_type null_string;
      for(unsigned int j = 0; j <= 13; ++j)
      {
         string_type s(this->m_pmessages->get(cat, 0, j+300, null_string));
         if(s.size())
            this->m_custom_class_names[s] = masks[j];
      }
   }
#endif
   //
   // get the collation format used by m_pcollate:
   //
   m_collate_type = re_detail::find_sort_syntax(this, &m_collate_delim);
}

template <class charT>
typename cpp_regex_traits_implementation<charT>::char_class_type 
   cpp_regex_traits_implementation<charT>::lookup_classname_imp(const charT* p1, const charT* p2) const
{
#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
   static const char_class_type masks[20] = 
   {
      0,
      std::ctype<char>::alnum, 
      std::ctype<char>::alpha,
      cpp_regex_traits_implementation<charT>::mask_blank,
      std::ctype<char>::cntrl,
      std::ctype<char>::digit,
      std::ctype<char>::digit,
      std::ctype<char>::graph,
      std::ctype<char>::lower,
      std::ctype<char>::lower,
      std::ctype<char>::print,
      std::ctype<char>::punct,
      std::ctype<char>::space,
      std::ctype<char>::space,
      std::ctype<char>::upper,
      cpp_regex_traits_implementation<charT>::mask_unicode,
      std::ctype<char>::upper,
      std::ctype<char>::alnum | cpp_regex_traits_implementation<charT>::mask_word, 
      std::ctype<char>::alnum | cpp_regex_traits_implementation<charT>::mask_word, 
      std::ctype<char>::xdigit,
   };
#else
   static const char_class_type masks[20] = 
   {
      0,
      ::boost::re_detail::char_class_alnum, 
      ::boost::re_detail::char_class_alpha,
      ::boost::re_detail::char_class_blank,
      ::boost::re_detail::char_class_cntrl,
      ::boost::re_detail::char_class_digit,
      ::boost::re_detail::char_class_digit,
      ::boost::re_detail::char_class_graph,
      ::boost::re_detail::char_class_lower,
      ::boost::re_detail::char_class_lower,
      ::boost::re_detail::char_class_print,
      ::boost::re_detail::char_class_punct,
      ::boost::re_detail::char_class_space,
      ::boost::re_detail::char_class_space,
      ::boost::re_detail::char_class_upper,
      ::boost::re_detail::char_class_unicode,
      ::boost::re_detail::char_class_upper,
      ::boost::re_detail::char_class_alnum | ::boost::re_detail::char_class_word, 
      ::boost::re_detail::char_class_alnum | ::boost::re_detail::char_class_word, 
      ::boost::re_detail::char_class_xdigit,
   };
#endif
   if(m_custom_class_names.size())
   {
      typedef typename std::map<std::basic_string<charT>, char_class_type>::const_iterator map_iter;
      map_iter pos = m_custom_class_names.find(string_type(p1, p2));
      if(pos != m_custom_class_names.end())
         return pos->second;
   }
   std::size_t id = 1 + re_detail::get_default_class_id(p1, p2);
   BOOST_ASSERT(id < sizeof(masks) / sizeof(masks[0]));
   return masks[id];
}

#ifdef BOOST_REGEX_BUGGY_CTYPE_FACET
template <class charT>
bool cpp_regex_traits_implementation<charT>::isctype(const charT c, char_class_type mask) const
{
   return
      ((mask & ::boost::re_detail::char_class_space) && (m_pctype->is(std::ctype<charT>::space, c)))
      || ((mask & ::boost::re_detail::char_class_print) && (m_pctype->is(std::ctype<charT>::print, c)))
      || ((mask & ::boost::re_detail::char_class_cntrl) && (m_pctype->is(std::ctype<charT>::cntrl, c)))
      || ((mask & ::boost::re_detail::char_class_upper) && (m_pctype->is(std::ctype<charT>::upper, c)))
      || ((mask & ::boost::re_detail::char_class_lower) && (m_pctype->is(std::ctype<charT>::lower, c)))
      || ((mask & ::boost::re_detail::char_class_alpha) && (m_pctype->is(std::ctype<charT>::alpha, c)))
      || ((mask & ::boost::re_detail::char_class_digit) && (m_pctype->is(std::ctype<charT>::digit, c)))
      || ((mask & ::boost::re_detail::char_class_punct) && (m_pctype->is(std::ctype<charT>::punct, c)))
      || ((mask & ::boost::re_detail::char_class_xdigit) && (m_pctype->is(std::ctype<charT>::xdigit, c)))
      || ((mask & ::boost::re_detail::char_class_blank) && (m_pctype->is(std::ctype<charT>::space, c)) && !::boost::re_detail::is_separator(c))
      || ((mask & ::boost::re_detail::char_class_word) && (c == '_'))
      || ((mask & ::boost::re_detail::char_class_unicode) && ::boost::re_detail::is_extended(c));
}
#endif


template <class charT>
boost::shared_ptr<cpp_regex_traits_implementation<charT> > create_cpp_regex_traits(const std::locale& l BOOST_APPEND_EXPLICIT_TEMPLATE_TYPE(charT))
{
   cpp_regex_traits_base<charT> key(l);
   return ::boost::object_cache<cpp_regex_traits_base<charT>, cpp_regex_traits_implementation<charT> >::get(key, 5);
}

} // re_detail

template <class charT>
class cpp_regex_traits
{
private:
   typedef std::ctype<charT>            ctype_type;
public:
   typedef charT                        char_type;
   typedef std::size_t                  size_type;
   typedef std::basic_string<char_type> string_type;
   typedef std::locale                  locale_type;
   typedef boost::uint_least32_t        char_class_type;

   struct boost_extensions_tag{};

   cpp_regex_traits()
      : m_pimpl(re_detail::create_cpp_regex_traits<charT>(std::locale()))
   { }
   static size_type length(const char_type* p)
   {
      return std::char_traits<charT>::length(p);
   }
   regex_constants::syntax_type syntax_type(charT c)const
   {
      return m_pimpl->syntax_type(c);
   }
   regex_constants::escape_syntax_type escape_syntax_type(charT c) const
   {
      return m_pimpl->escape_syntax_type(c);
   }
   charT translate(charT c) const
   {
      return c;
   }
   charT translate_nocase(charT c) const
   {
      return m_pimpl->m_pctype->tolower(c);
   }
   charT translate(charT c, bool icase) const
   {
      return icase ? m_pimpl->m_pctype->tolower(c) : c;
   }
   charT tolower(charT c) const
   {
      return m_pimpl->m_pctype->tolower(c);
   }
   charT toupper(charT c) const
   {
      return m_pimpl->m_pctype->toupper(c);
   }
   string_type transform(const charT* p1, const charT* p2) const
   {
      return m_pimpl->transform(p1, p2);
   }
   string_type transform_primary(const charT* p1, const charT* p2) const
   {
      return m_pimpl->transform_primary(p1, p2);
   }
   char_class_type lookup_classname(const charT* p1, const charT* p2) const
   {
      return m_pimpl->lookup_classname(p1, p2);
   }
   string_type lookup_collatename(const charT* p1, const charT* p2) const
   {
      return m_pimpl->lookup_collatename(p1, p2);
   }
   bool isctype(charT c, char_class_type f) const
   {
#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET
      typedef typename std::ctype<charT>::mask ctype_mask;

      static const ctype_mask mask_base = 
         static_cast<ctype_mask>(
            std::ctype<charT>::alnum 
            | std::ctype<charT>::alpha
            | std::ctype<charT>::cntrl
            | std::ctype<charT>::digit
            | std::ctype<charT>::graph
            | std::ctype<charT>::lower
            | std::ctype<charT>::print
            | std::ctype<charT>::punct
            | std::ctype<charT>::space
            | std::ctype<charT>::upper
            | std::ctype<charT>::xdigit);

      if((f & mask_base) 
         && (m_pimpl->m_pctype->is(
            static_cast<ctype_mask>(f & mask_base), c)))
         return true;
      else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_unicode) && re_detail::is_extended(c))
         return true;
      else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_word) && (c == '_'))
         return true;
      else if((f & re_detail::cpp_regex_traits_implementation<charT>::mask_blank) 
         && m_pimpl->m_pctype->is(std::ctype<charT>::space, c)
         && !re_detail::is_separator(c))
         return true;
      return false;
#else
      return m_pimpl->isctype(c, f);
#endif
   }
   int toi(const charT*& p1, const charT* p2, int radix)const;
   int value(charT c, int radix)const
   {
      const charT* pc = &c;
      return toi(pc, pc + 1, radix);
   }
   locale_type imbue(locale_type l)
   {
      std::locale result(getloc());
      m_pimpl = re_detail::create_cpp_regex_traits<charT>(l);
      return result;
   }
   locale_type getloc()const
   {
      return m_pimpl->m_locale;
   }
   std::string error_string(regex_constants::error_type n) const
   {
      return m_pimpl->error_string(n);
   }

   //
   // extension:
   // set the name of the message catalog in use (defaults to "boost_regex").
   //
   static std::string catalog_name(const std::string& name);
   static std::string get_catalog_name();

private:
   boost::shared_ptr<re_detail::cpp_regex_traits_implementation<charT> > m_pimpl;
   //
   // catalog name handler:
   //
   static std::string& get_catalog_name_inst();

#ifdef BOOST_HAS_THREADS
   static static_mutex& get_mutex_inst();
#endif
};


template <class charT>
int cpp_regex_traits<charT>::toi(const charT*& first, const charT* last, int radix)const
{
   // we do NOT want to parse any thousands separators inside the stream:
   last = std::find(first, last, BOOST_USE_FACET(std::numpunct<charT>, m_pimpl->m_is.getloc()).thousands_sep());
   m_pimpl->m_sbuf.pubsetbuf(const_cast<charT*>(static_cast<const charT*>(first)), static_cast<std::streamsize>(last-first));
   m_pimpl->m_is.clear();