public void Extract(MyRootHtmlNode root)
{
CandidateWord.Clear();
//先导词列表
if (LeadingColonKeyWordList.Length > 0)
{
ExtractByColonKeyWord(root);
}
//结尾词列表
if (TrailingWordList.Length > 0)
{
ExtractByTrailingKeyWord(root);
}
//是否有符号包裹特征
if (MarkFeature.Length > 0)
{
ExtractByMarkFeature(root);
}
//开始字符结束字符
if (StartEndFeature.Length > 0)
{
ExtractByStartEndStringFeature(root);
}
//正则表达式检索
if (RegularExpressFeature.Length > 0)
{
ExtractByRegularExpressFeature(root);
}
}
//候选词
#region 常规文本
public void ExtractFromTextFile(string filename)
{
if (!File.Exists(filename))
{
return;
}
CandidateWord.Clear();
if (LeadingColonKeyWordList.Length > 0)
{
ExtractTextByColonKeyWord(filename);
}
if (StartEndFeature.Length > 0)
{
ExtractByStartEndStringFeature(filename);
}
if (LeadingColonKeyWordListInChineseBrackets.Length > 0)
{
ExtractTextByInChineseBracketsColonKeyWord(filename);
}
}
