static string GetYiFang(HTMLEngine.MyRootHtmlNode root)
{
var Extractor = new ExtractProperty();
//这些关键字后面
Extractor.LeadingWordList = new string[] { "供应商名称：" };
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("甲方候补词(关键字)：[" + item + "]");
return(item);
}
//乙方:"有限公司"
Extractor = new ExtractProperty();
//这些关键字后面
Extractor.TrailingWordList = new string[] { "有限公司董事会" };
Extractor.Extract(root);
Extractor.CandidateWord.Reverse();
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("乙方候补词(关键字)：[" + item + "有限公司]");
return(item + "有限公司");
}
if (companynamelist.Count > 0)
{
return(companynamelist[companynamelist.Count - 1].secFullName);
}
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("");
}
public static string GetCompanyFullName(HTMLEngine.MyRootHtmlNode root)
{
var Extractor = new ExtractProperty();
Extractor.TrailingWordList = new string[] { "公司董事会" };
Extractor.Extract(root);
Extractor.CandidateWord.Reverse();
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("全称：[" + item + "公司]");
return(item);
}
return("");
}
//变动截止日期
static string GetChangeEndDate(HTMLEngine.MyRootHtmlNode root)
{
var Extractor = new ExtractProperty();
var StartArray = new string[] { "截止", "截至" };
var EndArray = new string[] { "日" };
Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray);
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("候补变动截止日期：[" + item + "]");
return(Normalizer.NormailizeDate(item + "日"));
}
return("");
}
static string GetMoney(HTMLEngine.MyRootHtmlNode node)
{
var Money = "";
var Extractor = new ExtractProperty();
//这些关键字后面
Extractor.LeadingWordList = new string[] { "中标金额", "中标价", "合同金额", "合同总价", "订单总金额" };
Extractor.Extract(node);
foreach (var item in Extractor.CandidateWord)
{
Money = Utility.SeekMoney(item, "");
Program.Logger.WriteLine("金额候补词：[" + Money + "]");
}
return(Money);
}
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("");
}
//固定搭配
public static string GetCompanyShortName(HTMLEngine.MyRootHtmlNode root)
{
var companyList = new Dictionary <string, string>();
//从第一行开始找到 有限公司 有限责任公司, 如果有简称的话Value是简称
//股票简称：东方电气
//东方电气股份有限公司董事会
var Extractor = new ExtractProperty();
Extractor.LeadingWordList = new string[] { "股票简称", "证券简称" };
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
var ShortName = item.Replace(":", "").Replace("：", "").Trim();
if (Utility.GetStringBefore(ShortName, "、") != "")
{
ShortName = Utility.GetStringBefore(ShortName, "、");
}
if (Utility.GetStringBefore(ShortName, "）") != "")
{
ShortName = Utility.GetStringBefore(ShortName, "）");
}
if (Utility.GetStringBefore(ShortName, "公告") != "")
{
ShortName = Utility.GetStringBefore(ShortName, "公告");
}
if (Utility.GetStringBefore(ShortName, "股票") != "")
{
ShortName = Utility.GetStringBefore(ShortName, "股票");
}
if (Utility.GetStringBefore(ShortName, "证券") != "")
{
ShortName = Utility.GetStringBefore(ShortName, "证券");
}
if (Utility.GetStringBefore(ShortName, " ") != "")
{
ShortName = Utility.GetStringBefore(ShortName, " ");
}
FDDC.Program.Logger.WriteLine("简称:[" + ShortName + "]");
return(ShortName);
}
return("");
}
static string GetJiaFang(MyRootHtmlNode root)
{
var Extractor = new ExtractProperty();
//这些关键字后面
Extractor.LeadingWordList = new string[] { "发包人：", "招标人：", "业主方：", "业主：", "甲方：", "采购人：", "采购人名称：" };
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("甲方候补词(关键字)：[" + item + "]");
return(item);
}
//招标
Extractor = new ExtractProperty();
var StartArray = new string[] { "业主", "收到", "接到" };
var EndArray = new string[] { "发来", "发出", "的中标" };
Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray);
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
var JiaFang = item;
JiaFang = JiaFang.Replace("业主", "");
Program.Logger.WriteLine("甲方候补词(招标)：[" + item + "]");
return(item);
}
//合同
Extractor = new ExtractProperty();
StartArray = new string[] { "与", "与业主" };
EndArray = new string[] { "签署", "签订" };
Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray);
Extractor.Extract(root);
foreach (var item in Extractor.CandidateWord)
{
Program.Logger.WriteLine("甲方候补词(合同)：[" + item + "]");
return(item);
}
return("");
}
static string GetHolderFullName(HTMLEngine.MyRootHtmlNode root)
{
var Extractor = new ExtractProperty();
var StartArray = new string[] { "接到", "收到", "股东" };
var EndArray = new string[] { "的", "通知", "告知函", "减持", "增持", "《" };
Extractor.StartEndFeature = Utility.GetStartEndStringArray(StartArray, EndArray);
Extractor.Extract(root);
foreach (var word in Extractor.CandidateWord)
{
if (word.Contains("简称"))
{
return(word);
}
Program.Logger.WriteLine("候补股东全称修正：[" + word + "]");
}
if (Extractor.CandidateWord.Count > 0)
{
return(Extractor.CandidateWord[0]);
}
return("");
}
