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_BOUNDARY_SEGMENT_HPP_INCLUDED
Chris@16: #define BOOST_LOCALE_BOUNDARY_SEGMENT_HPP_INCLUDED
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: #include <string>
Chris@16: #include <iosfwd>
Chris@16: #include <iterator>
Chris@16: 
Chris@16: 
Chris@16: namespace boost {
Chris@16: namespace locale {
Chris@16: namespace boundary {
Chris@16:     /// \cond INTERNAL
Chris@16:     namespace details {
Chris@16:         template<typename LeftIterator,typename RightIterator>
Chris@16:         int compare_text(LeftIterator l_begin,LeftIterator l_end,RightIterator r_begin,RightIterator r_end)
Chris@16:         {
Chris@16:             typedef LeftIterator left_iterator;
Chris@16:             typedef RightIterator right_iterator;
Chris@16:             typedef typename std::iterator_traits<left_iterator>::value_type char_type;
Chris@16:             typedef std::char_traits<char_type> traits;
Chris@16:             while(l_begin!=l_end && r_begin!=r_end) {
Chris@16:                 char_type lchar = *l_begin++;
Chris@16:                 char_type rchar = *r_begin++;
Chris@16:                 if(traits::eq(lchar,rchar))
Chris@16:                     continue;
Chris@16:                 if(traits::lt(lchar,rchar))
Chris@16:                     return -1;
Chris@16:                 else
Chris@16:                     return 1;
Chris@16:             }
Chris@16:             if(l_begin==l_end && r_begin==r_end)
Chris@16:                 return 0;
Chris@16:             if(l_begin==l_end)
Chris@16:                 return -1;
Chris@16:             else
Chris@16:                 return 1;
Chris@16:         }
Chris@16: 
Chris@16: 
Chris@16:         template<typename Left,typename Right>
Chris@16:         int compare_text(Left const &l,Right const &r)
Chris@16:         {
Chris@16:             return compare_text(l.begin(),l.end(),r.begin(),r.end());
Chris@16:         }
Chris@16:         
Chris@16:         template<typename Left,typename Char>
Chris@16:         int compare_string(Left const &l,Char const *begin)
Chris@16:         {
Chris@16:             Char const *end = begin;
Chris@16:             while(*end!=0)
Chris@16:                 end++;
Chris@16:             return compare_text(l.begin(),l.end(),begin,end);
Chris@16:         }
Chris@16: 
Chris@16:         template<typename Right,typename Char>
Chris@16:         int compare_string(Char const *begin,Right const &r)
Chris@16:         {
Chris@16:             Char const *end = begin;
Chris@16:             while(*end!=0)
Chris@16:                 end++;
Chris@16:             return compare_text(begin,end,r.begin(),r.end());
Chris@16:         }
Chris@16: 
Chris@16:     }
Chris@16:     /// \endcond
Chris@16: 
Chris@16:     ///
Chris@16:     /// \addtogroup boundary
Chris@16:     /// @{
Chris@16: 
Chris@16:     ///
Chris@16:     /// \brief a segment object that represents a pair of two iterators that define the range where
Chris@16:     /// this segment exits and a rule that defines it.
Chris@16:     ///
Chris@16:     /// This type of object is dereferenced by the iterators of segment_index. Using a rule() member function
Chris@16:     /// you can get a specific rule this segment was selected with. For example, when you use
Chris@16:     /// word boundary analysis, you can check if the specific word contains Kana letters by checking (rule() & \ref word_kana)!=0
Chris@16:     /// For a sentence analysis you can check if the sentence is selected because a sentence terminator is found (\ref sentence_term) or
Chris@16:     /// there is a line break (\ref sentence_sep).
Chris@16:     ///
Chris@16:     /// This object can be automatically converted to std::basic_string with the same type of character. It is also
Chris@16:     /// valid range that has begin() and end() member functions returning iterators on the location of the segment.
Chris@16:     ///
Chris@16:     /// \see
Chris@16:     ///
Chris@16:     /// - \ref segment_index
Chris@16:     /// - \ref boundary_point 
Chris@16:     /// - \ref boundary_point_index 
Chris@16:     ///
Chris@16:     template<typename IteratorType>
Chris@16:     class segment : public std::pair<IteratorType,IteratorType> {
Chris@16:     public:
Chris@16:         ///
Chris@16:         /// The type of the underlying character 
Chris@16:         ///
Chris@16:         typedef typename std::iterator_traits<IteratorType>::value_type char_type;
Chris@16:         ///
Chris@16:         /// The type of the string it is converted to
Chris@16:         ///
Chris@16:         typedef std::basic_string<char_type> string_type;
Chris@16:         ///
Chris@16:         /// The value that iterators return  - the character itself
Chris@16:         ///
Chris@16:         typedef char_type value_type;
Chris@16:         ///
Chris@16:         /// The iterator that allows to iterate the range
Chris@16:         ///
Chris@16:         typedef IteratorType iterator;
Chris@16:         ///
Chris@16:         /// The iterator that allows to iterate the range
Chris@16:         ///
Chris@16:         typedef IteratorType const_iterator;
Chris@16:         ///
Chris@16:         /// The type that represent a difference between two iterators
Chris@16:         ///
Chris@16:         typedef typename std::iterator_traits<IteratorType>::difference_type difference_type;
Chris@16: 
Chris@16:         ///
Chris@16:         /// Default constructor
Chris@16:         ///
Chris@16:         segment() {}
Chris@16:         ///
Chris@16:         /// Create a segment using two iterators and a rule that represents this point
Chris@16:         ///
Chris@16:         segment(iterator b,iterator e,rule_type r) :
Chris@16:             std::pair<IteratorType,IteratorType>(b,e),
Chris@16:             rule_(r)
Chris@16:         {
Chris@16:         }
Chris@16:         ///
Chris@16:         /// Set the start of the range
Chris@16:         ///
Chris@16:         void begin(iterator const &v)
Chris@16:         {
Chris@16:             this->first = v;
Chris@16:         }
Chris@16:         ///
Chris@16:         /// Set the end of the range
Chris@16:         ///
Chris@16:          void end(iterator const &v)
Chris@16:         {
Chris@16:             this->second = v;
Chris@16:         }
Chris@16: 
Chris@16:         ///
Chris@16:         /// Get the start of the range
Chris@16:         ///
Chris@16:         IteratorType begin() const 
Chris@16:         {
Chris@16:             return this->first;
Chris@16:         }
Chris@16:         ///
Chris@16:         /// Set the end of the range
Chris@16:         ///
Chris@16:         IteratorType end() const
Chris@16:         {
Chris@16:             return this->second;
Chris@16:         }
Chris@16: 
Chris@16:         ///
Chris@16:         /// Convert the range to a string automatically
Chris@16:         ///
Chris@16:         template <class T, class A>
Chris@16:         operator std::basic_string<char_type, T, A> ()const
Chris@16:         {
Chris@16:             return std::basic_string<char_type, T, A>(this->first, this->second);
Chris@16:         }
Chris@16:         
Chris@16:         ///
Chris@16:         /// Create a string from the range explicitly
Chris@16:         ///
Chris@16:         string_type str() const
Chris@16:         {
Chris@16:             return string_type(begin(),end());
Chris@16:         }
Chris@16: 
Chris@16:         ///
Chris@16:         /// Get the length of the text chunk
Chris@16:         ///
Chris@16: 
Chris@16:         size_t length() const
Chris@16:         {
Chris@16:             return std::distance(begin(),end());
Chris@16:         }
Chris@16: 
Chris@16:         ///
Chris@16:         /// Check if the segment is empty
Chris@16:         ///
Chris@16:         bool empty() const
Chris@16:         {
Chris@16:             return begin() == end();
Chris@16:         }
Chris@16: 
Chris@16:         ///
Chris@16:         /// Get the rule that is used for selection of this segment.
Chris@16:         ///
Chris@16:         rule_type rule() const
Chris@16:         {
Chris@16:             return rule_;
Chris@16:         }
Chris@16:         ///
Chris@16:         /// Set a rule that is used for segment selection
Chris@16:         ///
Chris@16:         void rule(rule_type r)
Chris@16:         {
Chris@16:             rule_ = r;
Chris@16:         }
Chris@16: 
Chris@16:         // make sure we override std::pair's operator==
Chris@16: 
Chris@16:         /// Compare two segments
Chris@16:         bool operator==(segment const &other)
Chris@16:         {
Chris@16:             return details::compare_text(*this,other) == 0;
Chris@16:         }
Chris@16: 
Chris@16:         /// Compare two segments
Chris@16:         bool operator!=(segment const &other)
Chris@16:         {
Chris@16:             return details::compare_text(*this,other) != 0;
Chris@16:         }
Chris@16: 
Chris@16:     private:
Chris@16:         rule_type rule_;
Chris@16:        
Chris@16:     };
Chris@16: 
Chris@16:    
Chris@16:     /// Compare two segments
Chris@16:     template<typename IteratorL,typename IteratorR>
Chris@16:     bool operator==(segment<IteratorL> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) == 0; 
Chris@16:     }
Chris@16:     /// Compare two segments
Chris@16:     template<typename IteratorL,typename IteratorR>
Chris@16:     bool operator!=(segment<IteratorL> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) != 0; 
Chris@16:     }
Chris@16: 
Chris@16:     /// Compare two segments
Chris@16:     template<typename IteratorL,typename IteratorR>
Chris@16:     bool operator<(segment<IteratorL> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) < 0; 
Chris@16:     }
Chris@16:     /// Compare two segments
Chris@16:     template<typename IteratorL,typename IteratorR>
Chris@16:     bool operator<=(segment<IteratorL> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) <= 0; 
Chris@16:     }
Chris@16:     /// Compare two segments
Chris@16:     template<typename IteratorL,typename IteratorR>
Chris@16:     bool operator>(segment<IteratorL> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) > 0; 
Chris@16:     }
Chris@16:     /// Compare two segments
Chris@16:     template<typename IteratorL,typename IteratorR>
Chris@16:     bool operator>=(segment<IteratorL> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) >= 0; 
Chris@16:     }
Chris@16: 
Chris@16:     /// Compare string and segment
Chris@16:     template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
Chris@16:     bool operator==(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) == 0; 
Chris@16:     }
Chris@16:     /// Compare string and segment
Chris@16:     template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
Chris@16:     bool operator!=(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) != 0; 
Chris@16:     }
Chris@16: 
Chris@16:     /// Compare string and segment
Chris@16:     template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
Chris@16:     bool operator<(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) < 0; 
Chris@16:     }
Chris@16:     /// Compare string and segment
Chris@16:     template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
Chris@16:     bool operator<=(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) <= 0; 
Chris@16:     }
Chris@16:     /// Compare string and segment
Chris@16:     template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
Chris@16:     bool operator>(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) > 0; 
Chris@16:     }
Chris@16:     /// Compare string and segment
Chris@16:     template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
Chris@16:     bool operator>=(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) >= 0; 
Chris@16:     }
Chris@16: 
Chris@16:     /// Compare string and segment
Chris@16:     template<typename Iterator,typename CharType,typename Traits,typename Alloc>
Chris@16:     bool operator==(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) == 0; 
Chris@16:     }
Chris@16:     /// Compare string and segment
Chris@16:     template<typename Iterator,typename CharType,typename Traits,typename Alloc>
Chris@16:     bool operator!=(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) != 0; 
Chris@16:     }
Chris@16: 
Chris@16:     /// Compare string and segment
Chris@16:     template<typename Iterator,typename CharType,typename Traits,typename Alloc>
Chris@16:     bool operator<(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) < 0; 
Chris@16:     }
Chris@16:     /// Compare string and segment
Chris@16:     template<typename Iterator,typename CharType,typename Traits,typename Alloc>
Chris@16:     bool operator<=(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) <= 0; 
Chris@16:     }
Chris@16:     /// Compare string and segment
Chris@16:     template<typename Iterator,typename CharType,typename Traits,typename Alloc>
Chris@16:     bool operator>(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) > 0; 
Chris@16:     }
Chris@16:     /// Compare string and segment
Chris@16:     template<typename Iterator,typename CharType,typename Traits,typename Alloc>
Chris@16:     bool operator>=(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
Chris@16:     {
Chris@16:         return details::compare_text(l,r) >= 0; 
Chris@16:     }
Chris@16: 
Chris@16: 
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename CharType,typename IteratorR>
Chris@16:     bool operator==(CharType const *l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) == 0; 
Chris@16:     }
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename CharType,typename IteratorR>
Chris@16:     bool operator!=(CharType const *l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) != 0; 
Chris@16:     }
Chris@16: 
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename CharType,typename IteratorR>
Chris@16:     bool operator<(CharType const *l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) < 0; 
Chris@16:     }
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename CharType,typename IteratorR>
Chris@16:     bool operator<=(CharType const *l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) <= 0; 
Chris@16:     }
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename CharType,typename IteratorR>
Chris@16:     bool operator>(CharType const *l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) > 0; 
Chris@16:     }
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename CharType,typename IteratorR>
Chris@16:     bool operator>=(CharType const *l,segment<IteratorR> const &r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) >= 0; 
Chris@16:     }
Chris@16: 
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename Iterator,typename CharType>
Chris@16:     bool operator==(segment<Iterator> const &l,CharType const *r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) == 0; 
Chris@16:     }
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename Iterator,typename CharType>
Chris@16:     bool operator!=(segment<Iterator> const &l,CharType const *r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) != 0; 
Chris@16:     }
Chris@16: 
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename Iterator,typename CharType>
Chris@16:     bool operator<(segment<Iterator> const &l,CharType const *r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) < 0; 
Chris@16:     }
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename Iterator,typename CharType>
Chris@16:     bool operator<=(segment<Iterator> const &l,CharType const *r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) <= 0; 
Chris@16:     }
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename Iterator,typename CharType>
Chris@16:     bool operator>(segment<Iterator> const &l,CharType const *r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) > 0; 
Chris@16:     }
Chris@16:     /// Compare C string and segment
Chris@16:     template<typename Iterator,typename CharType>
Chris@16:     bool operator>=(segment<Iterator> const &l,CharType const *r)
Chris@16:     {
Chris@16:         return details::compare_string(l,r) >= 0; 
Chris@16:     }
Chris@16: 
Chris@16: 
Chris@16: 
Chris@16: 
Chris@16: 
Chris@16: 
Chris@16:     typedef segment<std::string::const_iterator> ssegment;      ///< convenience typedef
Chris@16:     typedef segment<std::wstring::const_iterator> wssegment;    ///< convenience typedef
Chris@16:     #ifdef BOOST_HAS_CHAR16_T
Chris@16:     typedef segment<std::u16string::const_iterator> u16ssegment;///< convenience typedef
Chris@16:     #endif
Chris@16:     #ifdef BOOST_HAS_CHAR32_T
Chris@16:     typedef segment<std::u32string::const_iterator> u32ssegment;///< convenience typedef
Chris@16:     #endif
Chris@16:    
Chris@16:     typedef segment<char const *> csegment;                     ///< convenience typedef
Chris@16:     typedef segment<wchar_t const *> wcsegment;                 ///< convenience typedef
Chris@16:     #ifdef BOOST_HAS_CHAR16_T
Chris@16:     typedef segment<char16_t const *> u16csegment;              ///< convenience typedef
Chris@16:     #endif
Chris@16:     #ifdef BOOST_HAS_CHAR32_T
Chris@16:     typedef segment<char32_t const *> u32csegment;              ///< convenience typedef
Chris@16:     #endif
Chris@16: 
Chris@16: 
Chris@16: 
Chris@16:  
Chris@16:     
Chris@16:     ///
Chris@16:     /// Write the segment to the stream character by character
Chris@16:     ///
Chris@16:     template<typename CharType,typename TraitsType,typename Iterator>
Chris@16:     std::basic_ostream<CharType,TraitsType> &operator<<(
Chris@16:             std::basic_ostream<CharType,TraitsType> &out,
Chris@16:             segment<Iterator> const &tok)
Chris@16:     {
Chris@16:         for(Iterator p=tok.begin(),e=tok.end();p!=e;++p)
Chris@16:             out << *p;
Chris@16:         return out;
Chris@16:     }
Chris@16: 
Chris@16:     /// @}
Chris@16: 
Chris@16: } // boundary
Chris@16: } // locale
Chris@16: } // boost
Chris@16: 
Chris@16: #ifdef BOOST_MSVC
Chris@16: #pragma warning(pop)
Chris@16: #endif
Chris@16: 
Chris@16: #endif
Chris@16: 
Chris@16: // vim: tabstop=4 expandtab shiftwidth=4 softtabstop=4