Chris@16: //
Chris@16: //  Copyright (c) 2009-2011 Artyom Beilis (Tonkikh)
Chris@16: //
Chris@16: //  Distributed under the Boost Software License, Version 1.0. (See
Chris@16: //  accompanying file LICENSE_1_0.txt or copy at
Chris@16: //  http://www.boost.org/LICENSE_1_0.txt)
Chris@16: //
Chris@16: #ifndef BOOST_LOCALE_CONVERTER_HPP_INCLUDED
Chris@16: #define BOOST_LOCALE_CONVERTER_HPP_INCLUDED
Chris@16: 
Chris@16: #include <boost/locale/config.hpp>
Chris@16: #ifdef BOOST_MSVC
Chris@16: #  pragma warning(push)
Chris@16: #  pragma warning(disable : 4275 4251 4231 4660)
Chris@16: #endif
Chris@16: #include <locale>
Chris@16: 
Chris@16: 
Chris@16: namespace boost {
Chris@16:     namespace locale {
Chris@16:         
Chris@16:         ///
Chris@16:         /// \defgroup convert Text Conversions 
Chris@16:         ///
Chris@16:         ///  This module provides various function for string manipulation like Unicode normalization, case conversion etc.
Chris@16:         /// @{
Chris@16:         ///
Chris@16: 
Chris@16:         
Chris@16:         ///
Chris@16:         /// \brief This class provides base flags for text manipulation. It is used as base for converter facet.
Chris@16:         ///
Chris@16:         class converter_base {
Chris@16:         public:
Chris@16:             ///
Chris@16:             /// The flag used for facet - the type of operation to perform
Chris@16:             ///
Chris@16:             typedef enum {
Chris@16:                 normalization,  ///< Apply Unicode normalization on the text
Chris@16:                 upper_case,     ///< Convert text to upper case
Chris@16:                 lower_case,     ///< Convert text to lower case
Chris@16:                 case_folding,   ///< Fold case in the text
Chris@16:                 title_case      ///< Convert text to title case
Chris@16:             } conversion_type;
Chris@16:         };
Chris@16: 
Chris@16:         template<typename CharType>
Chris@16:         class converter;
Chris@16: 
Chris@16:         #ifdef BOOST_LOCALE_DOXYGEN
Chris@16:         ///
Chris@16:         /// \brief The facet that implements text manipulation
Chris@16:         ///
Chris@16:         /// It is used to performs text conversion operations defined by \ref conversion_type. It is specialized
Chris@16:         /// for four types of characters \c char, \c wchar_t, \c char16_t, \c char32_t
Chris@16:         ///
Chris@16:         template<typename Char>
Chris@16:         class BOOST_LOCALE_DECL converter: public converter_base, public std::locale::facet {
Chris@16:         public:
Chris@16:             /// Locale identification
Chris@16:             static std::locale::id id;
Chris@16: 
Chris@16:             /// Standard constructor
Chris@16:             converter(size_t refs = 0) : std::locale::facet(refs)
Chris@16:             {
Chris@16:             }
Chris@16:             ///
Chris@16:             /// Convert text in range [\a begin, \a end) according to conversion method \a how. Parameter
Chris@16:             /// \a flags is used for specification of normalization method like nfd, nfc etc.
Chris@16:             ///
Chris@16:             virtual std::basic_string<Char> convert(conversion_type how,Char const *begin,Char const *end,int flags = 0) const = 0;
Chris@16: #if defined (__SUNPRO_CC) && defined (_RWSTD_VER)
Chris@16:             std::locale::id& __get_id (void) const { return id; }
Chris@16: #endif
Chris@16:         };
Chris@16:         #else
Chris@16: 
Chris@16:         template<>
Chris@16:         class BOOST_LOCALE_DECL converter<char> : public converter_base, public std::locale::facet {
Chris@16:         public:
Chris@16:             static std::locale::id id;
Chris@16: 
Chris@16:             converter(size_t refs = 0) : std::locale::facet(refs)
Chris@16:             {
Chris@16:             }
Chris@16:             virtual std::string convert(conversion_type how,char const *begin,char const *end,int flags = 0) const = 0;
Chris@16: #if defined (__SUNPRO_CC) && defined (_RWSTD_VER)
Chris@16:             std::locale::id& __get_id (void) const { return id; }
Chris@16: #endif
Chris@16:         };
Chris@16: 
Chris@16:         template<>
Chris@16:         class BOOST_LOCALE_DECL converter<wchar_t> : public converter_base, public std::locale::facet {
Chris@16:         public:
Chris@16:             static std::locale::id id;
Chris@16:             converter(size_t refs = 0) : std::locale::facet(refs)
Chris@16:             {
Chris@16:             }
Chris@16:              virtual std::wstring convert(conversion_type how,wchar_t const *begin,wchar_t const *end,int flags = 0) const = 0;
Chris@16: #if defined (__SUNPRO_CC) && defined (_RWSTD_VER)
Chris@16:             std::locale::id& __get_id (void) const { return id; }
Chris@16: #endif
Chris@16:         };
Chris@16: 
Chris@16:         #ifdef BOOST_HAS_CHAR16_T
Chris@16:         template<>
Chris@16:         class BOOST_LOCALE_DECL converter<char16_t> : public converter_base, public std::locale::facet {
Chris@16:         public:
Chris@16:             static std::locale::id id;
Chris@16:             converter(size_t refs = 0) : std::locale::facet(refs)
Chris@16:             {
Chris@16:             }
Chris@16:             virtual std::u16string convert(conversion_type how,char16_t const *begin,char16_t const *end,int flags = 0) const = 0; 
Chris@16: #if defined (__SUNPRO_CC) && defined (_RWSTD_VER)
Chris@16:             std::locale::id& __get_id (void) const { return id; }
Chris@16: #endif
Chris@16:         };
Chris@16:         #endif
Chris@16: 
Chris@16:         #ifdef BOOST_HAS_CHAR32_T
Chris@16:         template<>
Chris@16:         class BOOST_LOCALE_DECL converter<char32_t> : public converter_base, public std::locale::facet {
Chris@16:         public:
Chris@16:             static std::locale::id id;
Chris@16:             converter(size_t refs = 0) : std::locale::facet(refs)
Chris@16:             {
Chris@16:             }
Chris@16:             virtual std::u32string convert(conversion_type how,char32_t const *begin,char32_t const *end,int flags = 0) const = 0;
Chris@16: #if defined (__SUNPRO_CC) && defined (_RWSTD_VER)
Chris@16:             std::locale::id& __get_id (void) const { return id; }
Chris@16: #endif
Chris@16:         };
Chris@16:         #endif
Chris@16: 
Chris@16:         #endif
Chris@16: 
Chris@16:         ///
Chris@16:         /// The type that defined <a href="http://unicode.org/reports/tr15/#Norm_Forms">normalization form</a>
Chris@16:         ///
Chris@16: 
Chris@16:         typedef enum {
Chris@16:             norm_nfd,   ///< Canonical decomposition
Chris@16:             norm_nfc,   ///< Canonical decomposition followed by canonical composition
Chris@16:             norm_nfkd,  ///< Compatibility decomposition
Chris@16:             norm_nfkc,  ///< Compatibility decomposition followed by canonical composition.
Chris@16:             norm_default = norm_nfc, ///< Default normalization - canonical decomposition followed by canonical composition
Chris@16:         } norm_type;
Chris@16:        
Chris@16:         ///
Chris@16:         /// Normalize Unicode string \a str according to \ref norm_type "normalization form" \a n
Chris@16:         ///
Chris@16:         /// Note: This function receives only Unicode strings, i.e.: UTF-8, UTF-16 or UTF-32. It does not take
Chris@16:         /// in account the locale encoding, because Unicode decomposition and composition are meaningless outside 
Chris@16:         /// of a Unicode character set.
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> normalize(std::basic_string<CharType> const &str,norm_type n=norm_default,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::normalization,str.data(),str.data() + str.size(),n);
Chris@16:         }
Chris@16: 
Chris@16:         ///
Chris@16:         /// Normalize NUL terminated Unicode string \a str according to \ref norm_type "normalization form" \a n
Chris@16:         ///
Chris@16:         /// Note: This function receives only Unicode strings, i.e.: UTF-8, UTF-16 or UTF-32. It does not take
Chris@16:         /// in account the locale encoding, because Unicode decomposition and composition are meaningless outside 
Chris@16:         /// of a Unicode character set.
Chris@16:         /// 
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> normalize(CharType const *str,norm_type n=norm_default,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             CharType const *end=str;
Chris@16:             while(*end)
Chris@16:                 end++;
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::normalization,str,end,n);
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Normalize Unicode string in range [begin,end) according to \ref norm_type "normalization form" \a n
Chris@16:         ///
Chris@16:         /// Note: This function receives only Unicode strings, i.e.: UTF-8, UTF-16 or UTF-32. It does not take
Chris@16:         /// in account the locale encoding, because Unicode decomposition and composition are meaningless outside 
Chris@16:         /// of a Unicode character set.
Chris@16:         /// 
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> normalize(  CharType const *begin,
Chris@16:                                                 CharType const *end,
Chris@16:                                                 norm_type n=norm_default,
Chris@16:                                                 std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::normalization,begin,end,n);
Chris@16:         }
Chris@16: 
Chris@16:         ///////////////////////////////////////////////////
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a string \a str to upper case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16: 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_upper(std::basic_string<CharType> const &str,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::upper_case,str.data(),str.data()+str.size());
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a NUL terminated string \a str to upper case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_upper(CharType const *str,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             CharType const *end=str;
Chris@16:             while(*end)
Chris@16:                 end++;
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::upper_case,str,end);
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a string in range [begin,end) to upper case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_upper(CharType const *begin,CharType const *end,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::upper_case,begin,end);
Chris@16:         }
Chris@16: 
Chris@16:         ///////////////////////////////////////////////////
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a string \a str to lower case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16: 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_lower(std::basic_string<CharType> const &str,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::lower_case,str.data(),str.data()+str.size());
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a NUL terminated string \a str to lower case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_lower(CharType const *str,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             CharType const *end=str;
Chris@16:             while(*end)
Chris@16:                 end++;
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::lower_case,str,end);
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a string in range [begin,end) to lower case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_lower(CharType const *begin,CharType const *end,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::lower_case,begin,end);
Chris@16:         }
Chris@16:         ///////////////////////////////////////////////////
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a string \a str to title case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16: 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_title(std::basic_string<CharType> const &str,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::title_case,str.data(),str.data()+str.size());
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a NUL terminated string \a str to title case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_title(CharType const *str,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             CharType const *end=str;
Chris@16:             while(*end)
Chris@16:                 end++;
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::title_case,str,end);
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Convert a string in range [begin,end) to title case according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> to_title(CharType const *begin,CharType const *end,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::title_case,begin,end);
Chris@16:         }
Chris@16: 
Chris@16:         ///////////////////////////////////////////////////
Chris@16:         
Chris@16:         ///
Chris@16:         /// Fold case of a string \a str according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16: 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> fold_case(std::basic_string<CharType> const &str,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::case_folding,str.data(),str.data()+str.size());
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Fold case of a NUL terminated string \a str according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> fold_case(CharType const *str,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             CharType const *end=str;
Chris@16:             while(*end)
Chris@16:                 end++;
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::case_folding,str,end);
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Fold case of a string in range [begin,end) according to locale \a loc
Chris@16:         ///
Chris@16:         /// \note throws std::bad_cast if loc does not have \ref converter facet installed
Chris@16:         /// 
Chris@16:         template<typename CharType>
Chris@16:         std::basic_string<CharType> fold_case(CharType const *begin,CharType const *end,std::locale const &loc=std::locale())
Chris@16:         {
Chris@16:             return std::use_facet<converter<CharType> >(loc).convert(converter_base::case_folding,begin,end);
Chris@16:         }
Chris@16: 
Chris@16:         ///
Chris@16:         ///@}
Chris@16:         ///
Chris@16:     } // locale
Chris@16: 
Chris@16: } // boost
Chris@16: 
Chris@16: #ifdef BOOST_MSVC
Chris@16: #pragma warning(pop)
Chris@16: #endif
Chris@16: 
Chris@16: 
Chris@16: #endif
Chris@16: 
Chris@16: ///
Chris@16: /// \example conversions.cpp
Chris@16: ///
Chris@16: /// Example of using various text conversion functions.
Chris@16: ///
Chris@16: /// \example wconversions.cpp
Chris@16: ///
Chris@16: /// Example of using various text conversion functions with wide strings.
Chris@16: ///
Chris@16: 
Chris@16: // vim: tabstop=4 expandtab shiftwidth=4 softtabstop=4
Chris@16: