protected WordSeg()
{
_MatchName = false;
_FilterStopWords = false;
_MatchDirection = T_Direction.LeftToRight;
_ExtractWords = new CExtractWords();
_ExtractWords.CompareByPosEvent = CompareByPos;
_POS = new CPOS();
_LastSaveTime = DateTime.Now;
InitRules();
}
/// <summary>
/// 合并英文专用词。
/// 如果字典中有英文专用词如U.S.A, C++.C#等
/// 需要对初步分词后的英文和字母进行合并
/// </summary>
protected virtual string MergeEnglishSpecialWord(CExtractWords extractWords, ArrayList words, int start, ref int end)
{
StringBuilder str = new StringBuilder();
int i;
for (i = start; i < words.Count; i++)
{
string word = (string)words[i];
//word 为空或者为空格回车换行等分割符号，中断扫描
if (word.Trim() == "")
{
break;
}
//如果遇到中文，中断扫描
if (word[0] >= 0x4e00 && word[0] <= 0x9fa5)
{
break;
}
str.Append(word);
}
String mergeString = str.ToString();
List <T_WordInfo> exWords = extractWords.ExtractFullText(mergeString);
if (exWords.Count == 1)
{
T_WordInfo info = (T_WordInfo)exWords[0];
if (info.Word.Length == mergeString.Length)
{
end = i;
return(mergeString);
}
}
return(null);
}
