123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483 |
- //
- // Copyright (c) 2009-2011 Artyom Beilis (Tonkikh)
- //
- // Distributed under 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)
- //
- #ifndef BOOST_LOCALE_BOUNDARY_SEGMENT_HPP_INCLUDED
- #define BOOST_LOCALE_BOUNDARY_SEGMENT_HPP_INCLUDED
- #include <boost/locale/config.hpp>
- #ifdef BOOST_MSVC
- # pragma warning(push)
- # pragma warning(disable : 4275 4251 4231 4660)
- #endif
- #include <locale>
- #include <string>
- #include <iosfwd>
- #include <iterator>
- namespace boost {
- namespace locale {
- namespace boundary {
- /// \cond INTERNAL
- namespace details {
- template<typename LeftIterator,typename RightIterator>
- int compare_text(LeftIterator l_begin,LeftIterator l_end,RightIterator r_begin,RightIterator r_end)
- {
- typedef LeftIterator left_iterator;
- typedef typename std::iterator_traits<left_iterator>::value_type char_type;
- typedef std::char_traits<char_type> traits;
- while(l_begin!=l_end && r_begin!=r_end) {
- char_type lchar = *l_begin++;
- char_type rchar = *r_begin++;
- if(traits::eq(lchar,rchar))
- continue;
- if(traits::lt(lchar,rchar))
- return -1;
- else
- return 1;
- }
- if(l_begin==l_end && r_begin==r_end)
- return 0;
- if(l_begin==l_end)
- return -1;
- else
- return 1;
- }
- template<typename Left,typename Right>
- int compare_text(Left const &l,Right const &r)
- {
- return compare_text(l.begin(),l.end(),r.begin(),r.end());
- }
-
- template<typename Left,typename Char>
- int compare_string(Left const &l,Char const *begin)
- {
- Char const *end = begin;
- while(*end!=0)
- end++;
- return compare_text(l.begin(),l.end(),begin,end);
- }
- template<typename Right,typename Char>
- int compare_string(Char const *begin,Right const &r)
- {
- Char const *end = begin;
- while(*end!=0)
- end++;
- return compare_text(begin,end,r.begin(),r.end());
- }
- }
- /// \endcond
- ///
- /// \addtogroup boundary
- /// @{
- ///
- /// \brief a segment object that represents a pair of two iterators that define the range where
- /// this segment exits and a rule that defines it.
- ///
- /// This type of object is dereferenced by the iterators of segment_index. Using a rule() member function
- /// you can get a specific rule this segment was selected with. For example, when you use
- /// word boundary analysis, you can check if the specific word contains Kana letters by checking (rule() & \ref word_kana)!=0
- /// For a sentence analysis you can check if the sentence is selected because a sentence terminator is found (\ref sentence_term) or
- /// there is a line break (\ref sentence_sep).
- ///
- /// This object can be automatically converted to std::basic_string with the same type of character. It is also
- /// valid range that has begin() and end() member functions returning iterators on the location of the segment.
- ///
- /// \see
- ///
- /// - \ref segment_index
- /// - \ref boundary_point
- /// - \ref boundary_point_index
- ///
- template<typename IteratorType>
- class segment : public std::pair<IteratorType,IteratorType> {
- public:
- ///
- /// The type of the underlying character
- ///
- typedef typename std::iterator_traits<IteratorType>::value_type char_type;
- ///
- /// The type of the string it is converted to
- ///
- typedef std::basic_string<char_type> string_type;
- ///
- /// The value that iterators return - the character itself
- ///
- typedef char_type value_type;
- ///
- /// The iterator that allows to iterate the range
- ///
- typedef IteratorType iterator;
- ///
- /// The iterator that allows to iterate the range
- ///
- typedef IteratorType const_iterator;
- ///
- /// The type that represent a difference between two iterators
- ///
- typedef typename std::iterator_traits<IteratorType>::difference_type difference_type;
- ///
- /// Default constructor
- ///
- segment() {}
- ///
- /// Create a segment using two iterators and a rule that represents this point
- ///
- segment(iterator b,iterator e,rule_type r) :
- std::pair<IteratorType,IteratorType>(b,e),
- rule_(r)
- {
- }
- ///
- /// Set the start of the range
- ///
- void begin(iterator const &v)
- {
- this->first = v;
- }
- ///
- /// Set the end of the range
- ///
- void end(iterator const &v)
- {
- this->second = v;
- }
- ///
- /// Get the start of the range
- ///
- IteratorType begin() const
- {
- return this->first;
- }
- ///
- /// Set the end of the range
- ///
- IteratorType end() const
- {
- return this->second;
- }
- ///
- /// Convert the range to a string automatically
- ///
- template <class T, class A>
- operator std::basic_string<char_type, T, A> ()const
- {
- return std::basic_string<char_type, T, A>(this->first, this->second);
- }
-
- ///
- /// Create a string from the range explicitly
- ///
- string_type str() const
- {
- return string_type(begin(),end());
- }
- ///
- /// Get the length of the text chunk
- ///
- size_t length() const
- {
- return std::distance(begin(),end());
- }
- ///
- /// Check if the segment is empty
- ///
- bool empty() const
- {
- return begin() == end();
- }
- ///
- /// Get the rule that is used for selection of this segment.
- ///
- rule_type rule() const
- {
- return rule_;
- }
- ///
- /// Set a rule that is used for segment selection
- ///
- void rule(rule_type r)
- {
- rule_ = r;
- }
- // make sure we override std::pair's operator==
- /// Compare two segments
- bool operator==(segment const &other)
- {
- return details::compare_text(*this,other) == 0;
- }
- /// Compare two segments
- bool operator!=(segment const &other)
- {
- return details::compare_text(*this,other) != 0;
- }
- private:
- rule_type rule_;
-
- };
-
- /// Compare two segments
- template<typename IteratorL,typename IteratorR>
- bool operator==(segment<IteratorL> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) == 0;
- }
- /// Compare two segments
- template<typename IteratorL,typename IteratorR>
- bool operator!=(segment<IteratorL> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) != 0;
- }
- /// Compare two segments
- template<typename IteratorL,typename IteratorR>
- bool operator<(segment<IteratorL> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) < 0;
- }
- /// Compare two segments
- template<typename IteratorL,typename IteratorR>
- bool operator<=(segment<IteratorL> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) <= 0;
- }
- /// Compare two segments
- template<typename IteratorL,typename IteratorR>
- bool operator>(segment<IteratorL> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) > 0;
- }
- /// Compare two segments
- template<typename IteratorL,typename IteratorR>
- bool operator>=(segment<IteratorL> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) >= 0;
- }
- /// Compare string and segment
- template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
- bool operator==(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) == 0;
- }
- /// Compare string and segment
- template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
- bool operator!=(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) != 0;
- }
- /// Compare string and segment
- template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
- bool operator<(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) < 0;
- }
- /// Compare string and segment
- template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
- bool operator<=(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) <= 0;
- }
- /// Compare string and segment
- template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
- bool operator>(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) > 0;
- }
- /// Compare string and segment
- template<typename CharType,typename Traits,typename Alloc,typename IteratorR>
- bool operator>=(std::basic_string<CharType,Traits,Alloc> const &l,segment<IteratorR> const &r)
- {
- return details::compare_text(l,r) >= 0;
- }
- /// Compare string and segment
- template<typename Iterator,typename CharType,typename Traits,typename Alloc>
- bool operator==(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
- {
- return details::compare_text(l,r) == 0;
- }
- /// Compare string and segment
- template<typename Iterator,typename CharType,typename Traits,typename Alloc>
- bool operator!=(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
- {
- return details::compare_text(l,r) != 0;
- }
- /// Compare string and segment
- template<typename Iterator,typename CharType,typename Traits,typename Alloc>
- bool operator<(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
- {
- return details::compare_text(l,r) < 0;
- }
- /// Compare string and segment
- template<typename Iterator,typename CharType,typename Traits,typename Alloc>
- bool operator<=(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
- {
- return details::compare_text(l,r) <= 0;
- }
- /// Compare string and segment
- template<typename Iterator,typename CharType,typename Traits,typename Alloc>
- bool operator>(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
- {
- return details::compare_text(l,r) > 0;
- }
- /// Compare string and segment
- template<typename Iterator,typename CharType,typename Traits,typename Alloc>
- bool operator>=(segment<Iterator> const &l,std::basic_string<CharType,Traits,Alloc> const &r)
- {
- return details::compare_text(l,r) >= 0;
- }
- /// Compare C string and segment
- template<typename CharType,typename IteratorR>
- bool operator==(CharType const *l,segment<IteratorR> const &r)
- {
- return details::compare_string(l,r) == 0;
- }
- /// Compare C string and segment
- template<typename CharType,typename IteratorR>
- bool operator!=(CharType const *l,segment<IteratorR> const &r)
- {
- return details::compare_string(l,r) != 0;
- }
- /// Compare C string and segment
- template<typename CharType,typename IteratorR>
- bool operator<(CharType const *l,segment<IteratorR> const &r)
- {
- return details::compare_string(l,r) < 0;
- }
- /// Compare C string and segment
- template<typename CharType,typename IteratorR>
- bool operator<=(CharType const *l,segment<IteratorR> const &r)
- {
- return details::compare_string(l,r) <= 0;
- }
- /// Compare C string and segment
- template<typename CharType,typename IteratorR>
- bool operator>(CharType const *l,segment<IteratorR> const &r)
- {
- return details::compare_string(l,r) > 0;
- }
- /// Compare C string and segment
- template<typename CharType,typename IteratorR>
- bool operator>=(CharType const *l,segment<IteratorR> const &r)
- {
- return details::compare_string(l,r) >= 0;
- }
- /// Compare C string and segment
- template<typename Iterator,typename CharType>
- bool operator==(segment<Iterator> const &l,CharType const *r)
- {
- return details::compare_string(l,r) == 0;
- }
- /// Compare C string and segment
- template<typename Iterator,typename CharType>
- bool operator!=(segment<Iterator> const &l,CharType const *r)
- {
- return details::compare_string(l,r) != 0;
- }
- /// Compare C string and segment
- template<typename Iterator,typename CharType>
- bool operator<(segment<Iterator> const &l,CharType const *r)
- {
- return details::compare_string(l,r) < 0;
- }
- /// Compare C string and segment
- template<typename Iterator,typename CharType>
- bool operator<=(segment<Iterator> const &l,CharType const *r)
- {
- return details::compare_string(l,r) <= 0;
- }
- /// Compare C string and segment
- template<typename Iterator,typename CharType>
- bool operator>(segment<Iterator> const &l,CharType const *r)
- {
- return details::compare_string(l,r) > 0;
- }
- /// Compare C string and segment
- template<typename Iterator,typename CharType>
- bool operator>=(segment<Iterator> const &l,CharType const *r)
- {
- return details::compare_string(l,r) >= 0;
- }
- typedef segment<std::string::const_iterator> ssegment; ///< convenience typedef
- typedef segment<std::wstring::const_iterator> wssegment; ///< convenience typedef
- #ifdef BOOST_LOCALE_ENABLE_CHAR16_T
- typedef segment<std::u16string::const_iterator> u16ssegment;///< convenience typedef
- #endif
- #ifdef BOOST_LOCALE_ENABLE_CHAR32_T
- typedef segment<std::u32string::const_iterator> u32ssegment;///< convenience typedef
- #endif
-
- typedef segment<char const *> csegment; ///< convenience typedef
- typedef segment<wchar_t const *> wcsegment; ///< convenience typedef
- #ifdef BOOST_LOCALE_ENABLE_CHAR16_T
- typedef segment<char16_t const *> u16csegment; ///< convenience typedef
- #endif
- #ifdef BOOST_LOCALE_ENABLE_CHAR32_T
- typedef segment<char32_t const *> u32csegment; ///< convenience typedef
- #endif
-
-
- ///
- /// Write the segment to the stream character by character
- ///
- template<typename CharType,typename TraitsType,typename Iterator>
- std::basic_ostream<CharType,TraitsType> &operator<<(
- std::basic_ostream<CharType,TraitsType> &out,
- segment<Iterator> const &tok)
- {
- for(Iterator p=tok.begin(),e=tok.end();p!=e;++p)
- out << *p;
- return out;
- }
- /// @}
- } // boundary
- } // locale
- } // boost
- #ifdef BOOST_MSVC
- #pragma warning(pop)
- #endif
- #endif
- // vim: tabstop=4 expandtab shiftwidth=4 softtabstop=4
|