NT2chView45/NTSimString.cs

   1 using System;\r
   2 using System.Collections.Generic;\r
   3 using System.Text;\r
   4 \r
   5 namespace NT2chView\r
   6 {\r
   7     class NTSimString\r
   8     {\r
   9                 List<string> mSplitedString;\r
  10                 \r
  11                 \r
  12                 const int TOKEN_SPACE = 0;\r
  13                 const int TOKEN_HIRAKANA = 1;\r
  14                 const int TOKEN_NUMALPHA = 2;\r
  15                 const int TOKEN_OTHER = 3;\r
  16                 \r
  17                 public NTSimString(String seed){\r
  18             mSplitedString = new List<string>();\r
  19                         int current = TOKEN_OTHER;\r
  20                         int len = seed.Length;\r
  21                         int start = 0;\r
  22                         int i = 0;\r
  23                         for(; i < len; i++){\r
  24                                 char c = seed[i];\r
  25                                 int token = getToken(c);\r
  26                                 if(token == TOKEN_SPACE){\r
  27                                         if(current != TOKEN_SPACE){\r
  28                                                 if(start < i){\r
  29                             mSplitedString.Add(seed.Substring(start, i - start));\r
  30                                                 }\r
  31                                                 current = TOKEN_SPACE;\r
  32                                         }\r
  33                                         start = i+1;\r
  34                                 }else{\r
  35                                         if(current != token){\r
  36                                                 if(start < i){\r
  37                             mSplitedString.Add(\r
  38                                 seed.Substring(start, i - start));\r
  39                                                 }\r
  40                                                 current = token;\r
  41                                                 start = i;\r
  42                                         }\r
  43                                 }\r
  44                         }\r
  45                         if(start < i && current != TOKEN_SPACE){\r
  46                 mSplitedString.Add(\r
  47                     seed.Substring(start, i - start));\r
  48                         }\r
  49                 }\r
  50                 \r
  51                 public int match(String source){\r
  52                         int match_len = 0;\r
  53             foreach (string line in mSplitedString)\r
  54             {\r
  55                 //int len;\r
  56                 //int idx = line.match(source, 0, out len);\r
  57                 //if (idx >= 0 && len > 1)\r
  58                 //{\r
  59                 //    match_len += len;\r
  60                 //}\r
  61                 if (0 <= source.IndexOf(line, StringComparison.CurrentCultureIgnoreCase))\r
  62                 {\r
  63                                         match_len += line.Length;\r
  64                                 }\r
  65                         }\r
  66                         return match_len;\r
  67                 }\r
  68                 public void log(){\r
  69                         //DebugUtil.log("SimString logging...");\r
  70                         //foreach(String line in mSplitedString){\r
  71                                 //DebugUtil.log(line);\r
  72                         //}\r
  73                 }\r
  74                 int getToken(char c){\r
  75                         for(int i = 0; i < mSeparatorChar.Length; i++){\r
  76                                 if(c == mSeparatorChar[i]){\r
  77                                         return TOKEN_SPACE;\r
  78                                 }\r
  79                         }\r
  80                         for(int i = 0; i < mHiraKakaChar.Length; i++){\r
  81                                 if(c == mHiraKakaChar[i]){\r
  82                                         return TOKEN_HIRAKANA;\r
  83                                 }\r
  84                         }\r
  85                         for(int i = 0; i < mNumAlphaChar.Length; i++){\r
  86                                 if(c == mNumAlphaChar[i]){\r
  87                                         return TOKEN_NUMALPHA;\r
  88                                 }\r
  89                         }\r
  90                         \r
  91                         return TOKEN_OTHER;\r
  92                 }\r
  93                 static char [] mSeparatorChar = {\r
  94                         ' ','　','(',')','【','】','（','）','「','」','{','}','｛','｝','〈','〉',\r
  95                         '《','》','｢','｣','『','』',',','.','、','。','★','※','*','?','!','？','！','・'};\r
  96                 static char [] mHiraKakaChar = {\r
  97                         'ー','ｰ','あ','ア','ぁ','ァ','い','イ','ぃ','ィ','う','ウ','ぅ','ゥ',\r
  98                         'え','エ','ぇ','ェ','お','オ','ぉ','ォ','を','ヲ',\r
  99                         'か','カ','ヵ','ヵ','き','キ','く','ク','け','ケ','ヶ','ヶ','こ','コ',\r
 100                         'さ','サ','し','シ','す','ス','せ','セ','そ','ソ',\r
 101                         'た','タ','ち','チ','つ','ツ','っ','ッ','て','テ','と','ト',\r
 102                         'な','ナ','に','ニ','ぬ','ヌ','ね','ネ','の','ノ',\r
 103                         'は','ハ','ワ','わ','ひ','ヒ','ふ','フ','へ','ヘ','ほ','ホ',\r
 104                         'ま','マ','み','ミ','む','ム','め','メ','も','モ',\r
 105                         'や','ヤ','ゃ','ャ','ゆ','ユ','ゅ','ュ','よ','ヨ','ょ','ョ',\r
 106                         'ラ','ら','ﾗ','り','リ','ﾘ','る','ル','ﾙ',\r
 107                         'れ','レ','ﾚ','ろ','ロ','ﾛ','ん','ン',\r
 108                         'が','ガ','ぎ','ギ','ぐ','グ','げ','ゲ','ご','ゴ',\r
 109                         'ざ','ザ','じ','ジ','ぢ','ヂ','ず','ズ','づ','ヅ',\r
 110                         'ぜ','ゼ','ぞ','ゾ','だ','ダ','で','デ','ど','ド',\r
 111                         'ば','バ','び','ビ','ぶ','ブ','べ','ベ','ぼ','ボ',\r
 112                         'ぱ','パ','ぴ','ピ','ぷ','プ','ぺ','ペ','ぽ','ポ'};\r
 113                 static char [] mNumAlphaChar = {\r
 114                         '-','_','－','＿',\r
 115                         '0','０','1','１','2','２','3','３','4','４',\r
 116                         '5','５','6','６','7','７','8','８','9','９',\r
 117                         'A','a','Ａ','ａ','B','b','Ｂ','ｂ','C','c','Ｃ','ｃ',\r
 118                         'D','d','Ｄ','ｄ','E','e','Ｅ','ｅ','F','f','Ｆ','ｆ',\r
 119                         'G','g','Ｇ','ｇ','H','h','Ｈ','ｈ','I','i','Ｉ','ｉ',\r
 120                         'J','j','Ｊ','ｊ','K','k','Ｋ','ｋ',        'L','l','Ｌ','ｌ',\r
 121                         'M','m','Ｍ','ｍ','N','n','Ｎ','ｎ','O','o','Ｏ','ｏ',\r
 122                         'P','p','Ｐ','ｐ','Q','q','Ｑ','ｑ','R','r','Ｒ','ｒ',\r
 123                         'S','s','Ｓ','ｓ','T','t','Ｔ','ｔ',        'U','u','Ｕ','ｕ',\r
 124                         'V','v','Ｖ','ｖ','W','w','Ｗ','ｗ','X','x','Ｘ','ｘ',\r
 125                         'Y','y','Ｙ','ｙ','Z','z','Ｚ','ｚ','%','％','$','＄','#','＃',\r
 126                         '@','＠','+','＋'};\r
 127         }\r
 128     \r
 129 }\r