src/romanization.hxx

   1 #ifndef _ROMANIZATION_HXX
   2 #define _ROMANIZATION_HXX
   3
   4 #include <string>
   5 #include <vector>
   6 #include "utils.hxx"
   7
   8 /*! \file romanization.hxx
   9 * \todo refactor everything
  10 * \note http://www.unicode.org/charts/unihangridindex.html
  11 * \note <b>ic_cjk*()</b> functions: https://en.wikipedia.org/wiki/CJK_Unified_Ideographs#CJK_Unified_Ideographs
  12 */
  13
  14
  15 using std::string;
  16 using std::vector;
  17
  18 /*!
  19 * Class responsible for converting romaji to hiragana/katakana.
  20 * Uses wapuro- style romanization (currently mostly Hepburn).
  21 * \todo kunrei-shiki and nihon-shiki
  22 */
  23 class Romanization
  24 {
  25   public:
  26     Romanization();
  27     ~Romanization(){};
  28
  29     //! One element in romanization table.
  30     struct RmnElement {
  31       vector<string> romaji;
  32       string hiragana;
  33       string katakana;
  34     };
  35     inline static bool is_cjk_base (unsigned int ucs) {return ucs>=0x4e00 && ucs<=0x9fff;}
  36     inline static bool is_cjk_a    (unsigned int ucs) {return ucs>=0x3400 && ucs<=0x4dbf;}
  37     inline static bool is_cjk_b    (unsigned int ucs) {return ucs>=0x20000 && ucs<=0x2a6df;}
  38     inline static bool is_cjk_c    (unsigned int ucs) {return ucs>=0x2a700 && ucs<=0x2b73f;}
  39     inline static bool is_cjk_d    (unsigned int ucs) {return ucs>=0x2b740 && ucs<=0x2b81f;}
  40     // range U+1B000 - U+1B0FF is both hiragana ans katakana
  41     inline static bool is_hiragana (unsigned int ucs) {return ucs>=0x3040 && ucs<=0x309f;}
  42     inline static bool is_katakana (unsigned int ucs) {return ucs>=0x30a0 && ucs<=0xffef;}
  43     inline static bool is_kana     (unsigned int ucs) {return is_hiragana(ucs) || is_katakana(ucs);}
  44     inline static bool is_cjk      (unsigned int ucs)
  45       {return is_cjk_base(ucs) || is_cjk_a(ucs) || is_cjk_b(ucs) || is_cjk_c(ucs) || is_cjk_d(ucs);}
  46
  47
  48     string romaji_to_kana ( const string &rmj, bool to_hiragana = true );
  49     inline string romaji_to_kana ( const char *rmj, bool to_hiragana = true )
  50       { return romaji_to_kana(string(rmj),to_hiragana); };
  51
  52     inline string romaji_to_hiragana ( const char *s )
  53       { return romaji_to_kana(string(s),true);  };
  54     inline string romaji_to_hiragana ( const string &s )
  55       { return romaji_to_kana(s,true);  };
  56
  57     inline string romaji_to_katakana ( const char *s )
  58       { return romaji_to_kana(string(s),false); };
  59     inline string romaji_to_katakana ( const string &s )
  60       { return romaji_to_kana(s,false); };
  61
  62     inline string hiragana_to_katakana ( const string &s );
  63
  64     string kana_to_romaji ( const string &k );
  65
  66     //! Returns true if at least one letter in <i>s</i> is a kanji.
  67     static inline bool contains_kanji ( const char *s )
  68     {
  69       for( unsigned int code: utils::utf8_to_ints(s) )
  70         if ( is_cjk(code) )
  71           return true;
  72       return false;
  73     };
  74
  75     //! Returns true if all letters in <i>s</i> are kanji.
  76     static inline bool is_kanji ( const char *s )
  77     {
  78       for ( unsigned int code: utils::utf8_to_ints(s) )
  79         if ( !is_cjk(code) )
  80           return false;
  81       return true;
  82     };
  83
  84     static inline bool is_kanji ( int code ) { return is_cjk(code);};
  85
  86   private:
  87     vector<RmnElement> table_;
  88     size_t current_romaji_;
  89     string tsu_h_;
  90     string tsu_k_;
  91     vector<string> tsus_;
  92
  93 };
  94
  95
  96 #endif // _ROMANIZATION_HXX