开通博客赚积分
发布资源赚积分
分类

源码开发语言/平台
当前位置：首页 > 源码/资料 > 多国语言处理 > 查看源码
DoubleSplitter.cs
资源名称：2.rar　[点击查看]
上传用户：hshongkong
上传日期：2021-11-20
资源大小：10241k
文件大小：9k
源码类别：

多国语言处理

开发平台：

C#

DoubleSplitter.cs：源码内容
  1. using System;
  2. using System.Collections;
  3. using System.Collections.Generic;
  4. using System.Text;
  6. namespace ChineseSplitter
  7. {
  8.     /// <summary>
  9.     /// 双字切分
  10.     /// </summary>
  11.     public class DoubleSplitter : BaseSplitter
  12.     {
  13.         TwoSplit myTwoSplit;
  14.         /// <summary>
  15.         /// 构造函数
  16.         /// </summary>
  17.         public DoubleSplitter()
  18.         {
  19.             myTwoSplit = new TwoSplit(); 
  20.         }
  21.         public override string Split(string inputStr)
  22.         {
  23.             return myTwoSplit.ReviseString(inputStr);
  24.         }
  25.         /// <summary>
  26.         /// 释放内存
  27.         /// </summary>
  28.         public override void Dispose()
  29.         {
  31.         }
  32.     }
  34.     class TwoSplit
  35.     {
  36.         string SplitChar = "  "; //分隔符
  37.         string nums = "０１２３４５６７８９＋－％．ａｂｃｄｅｆｇｈｉｊｋｌｍｎｏｐｑｒｓｔｕｖｗｘｙｚＡＢＣＤＥＦＧＨＩＪＫＬＭＮＯＰＱＲＳＴＵＶＷＸＹＺ";
  38.         string fnums = "0123456789+-%.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
  40.         /// <summary>
  41.         /// 构造函数
  42.         /// </summary>
  43.         public TwoSplit()
  44.         {
  45.         }
  46.         
  47.         /// <summary>
  48.         /// 字符串是否全为中文
  49.         /// </summary>
  50.         /// <param name="inputString">字符串</param>
  51.         /// <return>返回值</return></returns>
  52.         public bool IsChinese(string inputString)
  53.         {
  54.             if (Encoding.UTF8.GetByteCount(inputString) == inputString.Length * 3)
  55.             {
  56.                 return true;
  57.             }
  58.             else
  59.             {
  60.                 return false;
  61.             }
  62.         }
  63.         /// <summary>
  64.         /// 切分字符串
  65.         /// </summary>
  66.         /// <param name="str">字符串</param>
  67.         /// <returns>返回值</returns>
  68.         public string ReviseString(string str)
  69.         {
  70.             string spc = SplitChar.ToString();
  71.             int slen = str.Length;
  72.             if (slen == 0) return "";
  73.             StringBuilder okstr = new StringBuilder();
  74.             int[] chartypes = GetWordTypes(str);
  75.             int lasttype = 0;
  76.             string lastchinese = "";
  77.             int chineseCount = 0;
  78.             for (int i = 0; i < chartypes.Length; i++)
  79.             {
  80.                 char alabchar = ' ';
  81.                 switch (chartypes[i])
  82.                 {
  83.                     case 1:
  84.                         if (chineseCount == 1)
  85.                         {
  86.                             okstr.Append(spc + lastchinese);
  87.                         }
  88.                         alabchar = str[i];
  89.                         switch (lasttype)
  90.                         {
  91.                             case 1:
  92.                                 okstr.Append(alabchar);
  93.                                 break;
  94.                             case 2:
  95.                                 okstr.Append(alabchar);
  96.                                 break;
  97.                             case 3:
  98.                                 okstr.Append(spc + alabchar);
  99.                                 break;
  100.                             default:
  101.                                 okstr.Append(spc + alabchar);
  102.                                 break;
  103.                         }
  104.                         lastchinese = "";
  105.                         chineseCount = 0;
  106.                         lasttype = 1;
  107.                         break;
  108.                     case 2:
  109.                         if (chineseCount == 1)
  110.                         {
  111.                             okstr.Append(spc + lastchinese);
  112.                         }
  113.                         alabchar = GetAlabNum(str[i]);
  114.                         switch (lasttype)
  115.                         {
  116.                             case 1:
  117.                                 okstr.Append(alabchar);
  118.                                 break;
  119.                             case 2:
  120.                                 okstr.Append(alabchar);
  121.                                 break;
  122.                             case 3:
  123.                                 okstr.Append(spc + alabchar);
  124.                                 break;
  125.                             default:
  126.                                 okstr.Append(spc + alabchar);
  127.                                 break;
  128.                         }
  129.                         lastchinese = "";
  130.                         chineseCount = 0;
  131.                         lasttype = 2;
  132.                         break;
  133.                     case 3:
  134.                         alabchar = str[i];
  135.                         switch (lasttype)
  136.                         {
  137.                             case 3:
  138.                                 okstr.Append(spc + lastchinese + alabchar);
  139.                                 break;
  140.                             default:
  141.                                 break;
  142.                         }
  143.                         chineseCount++;
  144.                         lastchinese = alabchar.ToString();
  145.                         lasttype = 3;
  146.                         break;
  147.                     case 4:
  148.                         if (chineseCount == 1)
  149.                         {
  150.                             okstr.Append(spc + lastchinese);
  151.                         }
  152.                         lastchinese = "";
  153.                         chineseCount = 0;
  154.                         lasttype = 4;
  155.                         break;
  156.                     default:
  157.                         if (chineseCount == 1)
  158.                         {
  159.                             okstr.Append(spc + lastchinese);
  160.                         }
  161.                         alabchar = str[i];
  162.                         okstr.Append(spc + alabchar);
  163.                         chineseCount = 0;
  164.                         lastchinese = "";
  165.                         lasttype = 5;
  166.                         break;
  167.                 }
  168.             }
  169.             if (chineseCount == 1)
  170.             {
  171.                 okstr.Append(spc + lastchinese);
  172.             }
  173.             return okstr.ToString();
  174.         }
  175.         
  176.         /// <summary>
  177.         /// 把全角数字或英文单词转为半角
  178.         /// </summary>
  179.         /// <param name="fnum">字符</param>
  180.         /// <returns>返加</returns>
  181.         private char GetAlabNum(char fnum)
  182.         {
  183.             int place = nums.IndexOf(fnum);
  184.             if (place != -1)
  185.             {
  186.                 return fnums[place];
  187.             }
  188.             return fnum;
  189.         }
  191.         /// <summary>
  192.         /// 分析字符类型
  193.         /// </summary>
  194.         /// <param name="words">字符串</param>
  195.         /// <returns>返回类型数组</returns>
  196.         private int[] GetWordTypes(string words)
  197.         {
  198.             byte[] unicode = Encoding.Unicode.GetBytes(words);
  199.             int[] revalue = new int[unicode.Length / 2];
  200.             for (int i = 0; i < revalue.Length; i++)
  201.             {
  202.                 int nowvalue = BitConverter.ToUInt16(unicode, i * 2);
  203.                 if (nowvalue < 0x4E00)
  204.                 {
  205.                     if (nowvalue <= 0x0039 && nowvalue >= 0x0030)
  206.                     {
  207.                         revalue[i] = 1;
  208.                     }
  209.                     else if ((nowvalue <= 0x005A && nowvalue >= 0x0041) || (nowvalue <= 0x007A && nowvalue >= 0x0061))
  210.                     {
  211.                         revalue[i] = 1;
  212.                     }
  213.                     else if ((nowvalue <= 0x301B && nowvalue >= 0x3000) || nowvalue <= 0x00BB || (nowvalue <= 0x2FFF && nowvalue >= 0x2000))
  214.                     {
  215.                         revalue[i] = 4;
  216.                     }
  217.                     else
  218.                     {
  219.                         revalue[i] = 5;
  220.                     }
  221.                 }
  222.                 else if (nowvalue <= 0xEFFF && nowvalue >= 0x4E00)
  223.                 {
  224.                     revalue[i] = 3;
  225.                 }
  226.                 else
  227.                 {
  228.                     if (nowvalue <= 0xFF19 && nowvalue >= 0xFF10)
  229.                     {
  230.                         revalue[i] = 2;
  231.                     }
  232.                     else if ((nowvalue <= 0xFF3A && nowvalue >= 0xFF21) || (nowvalue <= 0xFF5A && nowvalue >= 0xFF41))
  233.                     {
  234.                         revalue[i] = 2;
  235.                     }
  236.                     else if (nowvalue <= 0xFF65 && nowvalue >= 0xFF00)
  237.                     {
  238.                         revalue[i] = 4;
  239.                     }
  240.                     else
  241.                     {
  242.                         revalue[i] = 5;
  243.                     }
  244.                 }
  245.             }
  246.             return revalue;
  247.         }
  248.     }
  249. }