static string GetContractName(MyRootHtmlNode root)
{
var Extractor = new ExtractProperty();
var MarkFeature = new ExtractProperty.struMarkFeature();
MarkFeature.MarkStartWith = "《";
MarkFeature.MarkEndWith = "》";
MarkFeature.InnerEndWith = "合同";
var MarkFeatureConfirm = new ExtractProperty.struMarkFeature();
MarkFeatureConfirm.MarkStartWith = "《";
MarkFeatureConfirm.MarkEndWith = "》";
MarkFeatureConfirm.InnerEndWith = "确认书";
Extractor.MarkFeature = new ExtractProperty.struMarkFeature[] { MarkFeature, MarkFeatureConfirm };
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("合同名称候补词（《XXX》）：[" + item + "]");
return(item);
}
Extractor = new ExtractProperty();
//这些关键字后面
Extractor.LeadingWordList = new string[] { "合同名称：" };
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("合同名称候补词(关键字)：[" + item + "]");
return(item);
}
//合同
Extractor = new ExtractProperty();
var StartArray = new string[] { "签署了" };
var EndArray = new string[] { "合同" };
Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray);
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("合同候补词(合同)：[" + item + "]");
return(item.Replace(" ", ""));
}
return("");
}
static string GetProjectName(MyRootHtmlNode root)
{
var Extractor = new ExtractProperty();
//这些关键字后面
Extractor.LeadingWordList = new string[] { "项目名称：" };
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("项目名称候补词(关键字)：[" + item + "]");
return(item.Replace(" ", ""));
}
var MarkFeature = new ExtractProperty.struMarkFeature();
MarkFeature.MarkStartWith = "“";
MarkFeature.MarkEndWith = "”";
MarkFeature.InnerEndWith = "标段";
var MarkFeatureConfirm = new ExtractProperty.struMarkFeature();
MarkFeatureConfirm.MarkStartWith = "“";
MarkFeatureConfirm.MarkEndWith = "”";
MarkFeatureConfirm.InnerEndWith = "标";
Extractor.MarkFeature = new ExtractProperty.struMarkFeature[] { MarkFeature, MarkFeatureConfirm };
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("工程名称候补词（《XXX》）：[" + item + "]");
return(item);
}
var list = BussinessLogic.GetProjectName(root);
if (list.Count > 0)
{
return(list[0]);
}
return("");
}