src/LibNMeCab/Core/CharProperty.cs

   1 //  MeCab -- Yet Another Part-of-Speech and Morphological Analyzer
   2 //
   3 //  Copyright(C) 2001-2006 Taku Kudo <taku@chasen.org>
   4 //  Copyright(C) 2004-2006 Nippon Telegraph and Telephone Corporation
   5 using System;
   6 using System.Collections.Generic;
   7 using System.Text;
   8 using System.IO;
   9
  10 namespace NMeCab.Core
  11 {
  12     public class CharProperty
  13     {
  14         #region Const/Field/Property
  15
  16         private const string CharPropertyFile = "char.bin";
  17
  18         private string[] cList;
  19
  20         private readonly CharInfo[] charInfoList = new CharInfo[0xFFFF];
  21
  22         public int Size
  23         {
  24             get { return this.cList.Length; }
  25         }
  26
  27         #endregion
  28
  29         #region Open
  30
  31         public void Open(string dicDir)
  32         {
  33             string fileName = Path.Combine(dicDir, CharPropertyFile);
  34
  35             using (var stream = new FileStream(fileName, FileMode.Open, FileAccess.Read))
  36             using (var reader = new BinaryReader(stream))
  37             {
  38                 this.Open(reader, fileName);
  39             }
  40         }
  41
  42         public void Open(BinaryReader reader, string fileName = null)
  43         {
  44             uint cSize = reader.ReadUInt32();
  45
  46             if (reader.BaseStream.CanSeek)
  47             {
  48                 long fSize = sizeof(uint) + 32 * cSize + sizeof(uint) * charInfoList.Length;
  49                 if (reader.BaseStream.Length != fSize)
  50                     throw new MeCabInvalidFileException("invalid file size", fileName);
  51             }
  52
  53             this.cList = new string[cSize];
  54             for (int i = 0; i < this.cList.Length; i++)
  55             {
  56                 this.cList[i] = StrUtils.GetString(reader.ReadBytes(32), Encoding.ASCII);
  57             }
  58
  59             for (int i = 0; i < this.charInfoList.Length; i++)
  60             {
  61                 this.charInfoList[i] = new CharInfo(reader.ReadUInt32());
  62             }
  63         }
  64
  65         #endregion
  66
  67         #region Get Infometion
  68
  69         public string Name(int i)
  70         {
  71             return this.cList[i];
  72         }
  73
  74         public unsafe char* SeekToOtherType(char* begin, char* end, CharInfo c, CharInfo* fail, int* cLen)
  75         {
  76             char* p = begin;
  77             *cLen = 0;
  78
  79             *fail = this.GetCharInfo(*p);
  80
  81             while (p != end && c.IsKindOf(*fail))
  82             {
  83                 p++;
  84                 (*cLen)++;
  85                 c = *fail;
  86
  87                 *fail = this.GetCharInfo(*p);
  88             }
  89
  90             return p;
  91         }
  92
  93         public CharInfo GetCharInfo(char c)
  94         {
  95             return this.charInfoList[c];
  96         }
  97
  98         #endregion
  99     }
 100 }