static void Main(string[] args)
{
Logger = new StreamWriter("Log.log");
//全局编码
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
//公司全称简称曾用名字典
CompanyNameLogic.LoadCompanyName(@"Resources" + Path.DirectorySeparatorChar + "FDDC_announcements_company_name_20180531.json");
//增减持公告日期的读入
StockChange.ImportPublishTime();
//结巴分词的地名修正词典
PosNS.ImportNS(@"Resources" + Path.DirectorySeparatorChar + "ns.dict");
CIRecord = new StreamWriter("CI.log");
//预处理
Traning(); return;
Evaluator = new StreamWriter("Evaluator.log");
Score = new StreamWriter(@"Result" + Path.DirectorySeparatorChar + "Score" + Path.DirectorySeparatorChar + "score" + System.DateTime.Now.ToString("yyyyMMddHHmmss") + ".txt");
//new Contract(Program.DocBase + @"\FDDC_announcements_round1_train_20180518\round1_train_20180518\重大合同\html\1008828.html").Extract();return;
Extract();
CIRecord.Close();
Score.Close();
Evaluator.Close();
Logger.Close();
}
static void Main(string[] args)
{
if (Environment.OSVersion.Platform == System.PlatformID.Unix)
{
//静态变量已经定下来了，这里改不来了！
Console.WriteLine("Switch Doc Path To:" + DocBase);
}
//日志
Logger = new StreamWriter("Log.log");
//实体属性器日志设定
EntityProperty.Logger = Logger;
//全局编码
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
CIRecord = new StreamWriter("CI.log");
QuickTestArea(); return;
//PDFToTXT.GetPdf2TxtBatchFile();
//公司全称简称曾用名字典
CompanyNameLogic.LoadCompanyName("Resources" + Path.DirectorySeparatorChar + "FDDC_announcements_company_name_20180531.json");
//结巴分词的地名修正词典
PosNS.ImportNS("Resources" + Path.DirectorySeparatorChar + "ns.dict");
//预处理
Traning();
Evaluator = new StreamWriter("Evaluator.log");
Score = new StreamWriter("Result" + Path.DirectorySeparatorChar + "Score" + Path.DirectorySeparatorChar + "score" + System.DateTime.Now.ToString("yyyyMMddHHmmss") + ".txt");
Extract();
CIRecord.Close();
Score.Close();
Evaluator.Close();
Logger.Close();
}
/// <summary>
/// 快速测试区
/// </summary>
private static void QuickTestArea()
{
var plst = LTPTrainingNER.GetParagraghList(StockChangePath_TEST + "/ner/18877033.xml");
CompanyNameLogic.GetCompanyNameByNerInfo(plst);
return;
var s0 = "爱康科技向爱康实业、爱康国际、苏州度金、天地国际、钨业研究支付现金购买其合计持有爱康光电100%股权";
var pos = new PosSegmenter();
var words = pos.Cut(s0);
Evaluator = new StreamWriter("Evaluator.log");
Score = new StreamWriter("Result" + Path.DirectorySeparatorChar + "Score" + Path.DirectorySeparatorChar + "score" + System.DateTime.Now.ToString("yyyyMMddHHmmss") + ".txt");
//Evaluate.EvaluateReorganizationByFile(@"E:\WorkSpace2018\FDDC2018\FDDC_SRC\Result\chongzu_train.txt");
//Score.Close();
//Evaluator.Close();
//TraningDataset.InitReorganization();
ReOrganizationTraning.EvaluateMethodList = new string[] {
"收益法", "资产基础法", "市场法", "市场比较法", "估值法", "成本法", "现金流折现法", "现金流折现法", "剩余法",
"内含价值调整法", "可比公司市净率法", "重置成本法", "收益现值法", "基础资产法", "假设清偿法",
"成本逼近法", "单项资产加和法", "成本加和法", "基准地价修正法", "收益还原法", "现金流量法", "单项资产加总法", "折现现金流量法", "基准地价系数修正法"
}.ToList();
var t = new Reorganization();
t.Id = "748379";
t.HTMLFileName = ReorganizationPath_TEST + "/html/1759374.html";
//t.TextFileName = ContractPath_TEST + "/txt/128869.txt";
//t.NerXMLFileName = ContractPath_TEST + "/ner/128869.xml";
t.Init();
var recs = t.Extract();
var s1 = recs[0].ConvertToString();
}
/// <summary>
/// 公司名称的获得
/// </summary>
/// <param name="FullName"></param>
/// <param name="ShortName"></param>
/// <returns></returns>
public static (String FullName, String ShortName) NormalizeCompanyName(AnnouceDocument doc, string word)
{
if (String.IsNullOrEmpty(word))
{
return(String.Empty, String.Empty);
}
var fullname = word.Replace(" ", String.Empty);
var shortname = String.Empty;
foreach (var companyname in doc.companynamelist)
{
if (companyname.secFullName == fullname)
{
//注意：这里可能出现两个具有相同FullName，但是某个没有ShortName的可能性！
if (shortname == String.Empty && !String.IsNullOrEmpty(companyname.secShortName))
{
shortname = companyname.secShortName;
break;
}
}
if (companyname.secShortName == fullname)
{
fullname = companyname.secFullName;
shortname = companyname.secShortName;
break;
}
//如果进来的是简称，而提取的公司信息里面，只有全称，这里简单推断一下
//简称和全称的关系
if (companyname.secFullName.Contains(fullname) &&
companyname.secFullName.Length > fullname.Length)
{
fullname = companyname.secFullName;
shortname = word;
}
}
if (string.IsNullOrEmpty(shortname))
{
//字典
shortname = CompanyNameLogic.GetCompanyNameByFullName(fullname).secShortName;
}
if (string.IsNullOrEmpty(shortname))
{
//在原文中寻找该字符名称，然后看一下，其后是否有【简称】字样，
//简称后是否有引号字样“XXXX”有的话，差不多就是了
shortname = GetShortNameByFullName(fullname, doc);
if (!string.IsNullOrEmpty(shortname))
{
Console.WriteLine(fullname + ":" + shortname);
}
}
return(fullname, shortname);
}
public static (String FullName, String ShortName) NormalizeCompanyName(AnnouceDocument doc, string word)
{
if (String.IsNullOrEmpty(word))
{
return(String.Empty, String.Empty);
}
var fullname = word.Replace(" ", String.Empty);
var shortname = String.Empty;
foreach (var companyname in doc.companynamelist)
{
if (companyname.secFullName == fullname)
{
//注意：这里可能出现两个具有相同FullName，但是某个没有ShortName的可能性！
if (shortname == String.Empty && !String.IsNullOrEmpty(companyname.secShortName))
{
shortname = companyname.secShortName;
break;
}
}
if (companyname.secShortName == fullname)
{
fullname = companyname.secFullName;
shortname = companyname.secShortName;
break;
}
//如果进来的是简称，而提取的公司信息里面，只有全称，这里简单推断一下
//简称和全称的关系
if (companyname.secFullName.Contains(fullname) &&
companyname.secFullName.Length > fullname.Length)
{
fullname = companyname.secFullName;
shortname = word;
}
}
if (shortname == String.Empty)
{
shortname = CompanyNameLogic.GetCompanyNameByFullName(fullname).secShortName;
}
return(fullname, shortname);
}
ContractRec ExtractSingle()
{
contractType = String.Empty;
foreach (var paragrah in root.Children)
{
foreach (var item in paragrah.Children)
{
if (item.Content.Contains("中标"))
{
contractType = "中标";
break;
}
if (item.Content.Contains("合同"))
{
contractType = "合同";
break;
}
}
if (contractType != String.Empty)
{
break;
}
}
if (contractType == String.Empty)
{
Console.WriteLine("contractType Null:" + Id);
}
var contract = new ContractRec();
//公告ID
contract.Id = Id;
//乙方
contract.YiFang = GetYiFang();
if (contract.YiFang.Contains("本公司"))
{
contract.YiFang = string.Empty;
}
contract.YiFang = CompanyNameLogic.AfterProcessFullName(contract.YiFang).secFullName;
contract.YiFang = contract.YiFang.NormalizeTextResult();
//按照规定除去括号
contract.YiFang = RegularTool.TrimBrackets(contract.YiFang);
if (contract.YiFang.Length < 3)
{
contract.YiFang = string.Empty;
}
//甲方
contract.JiaFang = GetJiaFang(contract.YiFang);
if (contract.JiaFang.Contains("本公司"))
{
contract.JiaFang = string.Empty;
}
contract.JiaFang = CompanyNameLogic.AfterProcessFullName(contract.JiaFang).secFullName;
contract.JiaFang = contract.JiaFang.NormalizeTextResult();
if (contract.JiaFang.Contains("简称"))
{
contract.JiaFang = Utility.GetStringBefore(contract.JiaFang, "(");
}
//机构列表
if (Nerlist != null)
{
var NiList = Nerlist.Where((n) => n.Type == LTPTrainingNER.enmNerType.Ni).Select((m) => m.RawData);
if (!NiList.Contains(contract.JiaFang))
{
if (NiList.Contains("国家电网公司"))
{
contract.JiaFang = "国家电网公司";
}
}
}
//项目
contract.ProjectName = GetProjectName();
contract.ProjectName = contract.ProjectName.NormalizeTextResult();
if (contract.ProjectName.StartsWith("“") && contract.ProjectName.EndsWith("”"))
{
contract.ProjectName = contract.ProjectName.TrimStart("“".ToCharArray()).TrimEnd("”".ToCharArray());
}
if (contract.ProjectName.EndsWith("，签约双方"))
{
contract.ProjectName = Utility.GetStringAfter(contract.ProjectName, "，签约双方");
}
if (contract.ProjectName.Contains("(以下简称"))
{
contract.ProjectName = Utility.GetStringAfter(contract.ProjectName, "(以下简称");
}
if (contract.ProjectName.EndsWith(")"))
{
if (contract.ProjectName.Contains("(招标编号"))
{
contract.ProjectName = Utility.GetStringBefore(contract.ProjectName, "(招标编号");
}
if (contract.ProjectName.Contains("(合同编号"))
{
contract.ProjectName = Utility.GetStringBefore(contract.ProjectName, "(合同编号");
}
}
contract.ProjectName = contract.ProjectName.Replace("的推荐中标", "");
//特殊处理
contract.ProjectName = contract.ProjectName.Replace("<1>", "1、");
contract.ProjectName = contract.ProjectName.Replace("“", "");
contract.ProjectName = contract.ProjectName.Replace("”", "");
//合同名
contract.ContractName = GetContractName();
if (contract.ContractName.StartsWith("“") && contract.ContractName.EndsWith("”"))
{
contract.ContractName = contract.ContractName.TrimStart("“".ToCharArray()).TrimEnd("”".ToCharArray());
}
//去掉书名号
contract.ContractName = contract.ContractName.Replace("《", String.Empty).Replace("》", String.Empty);
contract.ContractName = contract.ContractName.NormalizeTextResult();
if (contract.ContractName.Contains("(以下简称"))
{
contract.ContractName = Utility.GetStringAfter(contract.ContractName, "(以下简称");
}
contract.ContractName = ExtendContractName(contract.ContractName);
//如果是采购协议，则工程名清空
if (contract.ContractName.Contains("采购"))
{
if (contract.ProjectName.Contains("标段"))
{
//TODO:
}
else
{
contract.ProjectName = string.Empty;
}
}
//金额
var money = GetMoney();
contract.ContractMoneyUpLimit = MoneyUtility.Format(money.MoneyAmount, String.Empty);
contract.ContractMoneyDownLimit = contract.ContractMoneyUpLimit;
//联合体
contract.UnionMember = GetUnionMember(contract);
contract.UnionMember = contract.UnionMember.NormalizeTextResult();
//按照规定除去括号
contract.UnionMember = RegularTool.TrimBrackets(contract.UnionMember);
var YiFangArray = contract.YiFang.Split(Utility.SplitChar);
if (YiFangArray.Length > 1)
{
contract.UnionMember = Utility.GetStringAfter(contract.YiFang, Utility.SplitChar);
contract.YiFang = YiFangArray[0];
Console.WriteLine("联合体：" + contract.UnionMember);
}
return(contract);
}
/// <summary>
/// 获得甲方
/// </summary>
/// <returns></returns>
public string GetJiaFang()
{
//最高置信度的抽取
EntityProperty e = new EntityProperty();
e.ExcludeContainsWordList = new string[] { "招标代理" };
e.LeadingColonKeyWordList = new string[] {
"甲方：", "合同买方：",
"发包人：", "发包单位：", "发包方：", "发包机构：", "发包人名称：",
"招标人：", "招标单位：", "招标方：", "招标机构：", "招标人名称：",
"业主：", "业主单位：", "业主方：", "业主机构：", "业主名称：",
"采购单位：", "采购单位名称：", "采购人：", "采购人名称：", "采购方：", "采购方名称："
};
e.CandidatePreprocess = (x =>
{
x = Normalizer.ClearTrailing(x);
return(CompanyNameLogic.AfterProcessFullName(x).secFullName);
});
e.MaxLength = ContractTraning.JiaFangES.MaxLength;
e.MaxLengthCheckPreprocess = Utility.TrimEnglish;
e.MinLength = 3;
e.Extract(this);
//这里不直接做Distinct，出现频次越高，则可信度越高
//多个甲方的时候，可能意味着没有甲方！
if (e.LeadingColonKeyWordCandidate.Distinct().Count() > 1)
{
foreach (var candidate in e.LeadingColonKeyWordCandidate)
{
Program.Logger.WriteLine("发现多个甲方：" + candidate);
}
}
if (e.LeadingColonKeyWordCandidate.Count > 0)
{
return(e.LeadingColonKeyWordCandidate[0]);
}
//招标
var Extractor = new ExtractPropertyByHTML();
var CandidateWord = new List <String>();
var StartArray = new string[] { "招标单位", "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 = CompanyNameLogic.AfterProcessFullName(item.Value.Trim());
if (JiaFang.secFullName.Contains("招标代理"))
{
continue; //特殊业务规则
}
JiaFang.secFullName = JiaFang.secFullName.Replace("业主", String.Empty).Trim();
JiaFang.secFullName = JiaFang.secFullName.Replace("招标单位", String.Empty).Trim();
if (Utility.TrimEnglish(JiaFang.secFullName).Length > ContractTraning.JiaFangES.MaxLength)
{
continue;
}
if (JiaFang.secFullName.Length < 3)
{
continue; //使用实际长度排除全英文的情况
}
if (!Program.IsMultiThreadMode)
{
Program.Logger.WriteLine("甲方候补词(招标)：[" + JiaFang.secFullName + "]");
}
CandidateWord.Add(JiaFang.secFullName);
}
//合同
Extractor = new ExtractPropertyByHTML();
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)
{
var JiaFang = CompanyNameLogic.AfterProcessFullName(item.Value.Trim());
JiaFang.secFullName = JiaFang.secFullName.Replace("业主", String.Empty).Trim();
if (JiaFang.secFullName.Contains("招标代理"))
{
continue; //特殊业务规则
}
if (Utility.TrimEnglish(JiaFang.secFullName).Length > ContractTraning.JiaFangES.MaxLength)
{
continue;
}
if (JiaFang.secFullName.Length < 3)
{
continue; //使用实际长度排除全英文的情况
}
if (!Program.IsMultiThreadMode)
{
Program.Logger.WriteLine("甲方候补词(合同)：[" + JiaFang.secFullName + "]");
}
CandidateWord.Add(JiaFang.secFullName);
}
return(CompanyNameLogic.MostLikeCompanyName(CandidateWord));
}
public static struCompanyName AfterProcessFullName(string FullName)
{
var ShortName = String.Empty;
var CompanyNameTrailingwords = new string[] {
"（以下简称", "（下称", "（以下称", "（简称", "(以下简称", "(下称", "(以下称", "(简称"
};
//暂时不做括号的正规化
foreach (var trailing in CompanyNameTrailingwords)
{
if (FullName.Contains(trailing))
{
//获取简称
var BracketsList = RegularTool.GetChineseBrackets(FullName);
foreach (var bracketItem in BracketsList)
{
var ShortNameList = RegularTool.GetChineseQuotation(bracketItem);
if (ShortNameList.Count > 0)
{
ShortName = ShortNameList.First();
if (!String.IsNullOrEmpty(ShortName))
{
ShortName = ShortName.Substring(1, ShortName.Length - 2);
}
}
}
FullName = Utility.GetStringBefore(FullName, trailing);
}
}
if (FullName.Contains("及其"))
{
FullName = Utility.GetStringBefore(FullName, "及其");
}
if (FullName.Contains("股东"))
{
FullName = Utility.GetStringAfter(FullName, "股东");
}
if (FullName.Contains("一致行动人"))
{
FullName = Utility.GetStringAfter(FullName, "一致行动人");
}
if (!String.IsNullOrEmpty(CompanyNameLogic.GetCompanyNameByShortName(FullName).secFullName))
{
FullName = CompanyNameLogic.GetCompanyNameByShortName(FullName).secFullName;
}
//删除前导
FullName = EntityWordAnlayzeTool.TrimLeadingUL(FullName);
FullName = CutOtherLeadingWords(FullName);
if (ShortName != String.Empty)
{
return(new struCompanyName()
{
secFullName = FullName, secShortName = ShortName, Score = 80
});
}
else
{
return(new struCompanyName()
{
secFullName = FullName, Score = 60
});
}
}
struContract ExtractSingle(MyRootHtmlNode root, String Id)
{
contractType = String.Empty;
foreach (var paragrah in root.Children)
{
foreach (var item in paragrah.Children)
{
if (item.Content.Contains("中标"))
{
contractType = "中标";
break;
}
if (item.Content.Contains("合同"))
{
contractType = "合同";
break;
}
}
if (contractType != String.Empty)
{
break;
}
}
if (contractType == String.Empty)
{
Console.WriteLine("contractType Null:" + Id);
}
var contract = new struContract();
//公告ID
contract.id = Id;
//甲方
contract.JiaFang = GetJiaFang();
contract.JiaFang = CompanyNameLogic.AfterProcessFullName(contract.JiaFang).secFullName;
contract.JiaFang = contract.JiaFang.NormalizeTextResult();
if (!Nerlist.Contains(contract.JiaFang))
{
//作为特殊单位，国家电网公司一般都是甲方
if (Nerlist.Contains("国家电网公司"))
{
contract.JiaFang = "国家电网公司";
}
}
//乙方
contract.YiFang = GetYiFang();
contract.YiFang = CompanyNameLogic.AfterProcessFullName(contract.YiFang).secFullName;
contract.YiFang = contract.YiFang.NormalizeTextResult();
//按照规定除去括号
contract.YiFang = RegularTool.TrimBrackets(contract.YiFang);
//项目
contract.ProjectName = GetProjectName();
if (contract.ProjectName.StartsWith("“") && contract.ProjectName.EndsWith("”"))
{
contract.ProjectName = contract.ProjectName.TrimStart("“".ToCharArray()).TrimEnd("”".ToCharArray());
}
if (contract.ProjectName.EndsWith("，签约双方"))
{
contract.ProjectName = Utility.GetStringAfter(contract.ProjectName, "，签约双方");
}
if (contract.ProjectName.Contains("（以下简称"))
{
contract.ProjectName = Utility.GetStringAfter(contract.ProjectName, "（以下简称");
}
contract.ProjectName = contract.ProjectName.NormalizeTextResult();
//合同
if (contractType == "中标")
{
//按照数据分析来看，应该工程名 在中标的时候填写，合同名在合同的时候填写
contract.ContractName = String.Empty;
}
else
{
contract.ContractName = GetContractName();
if (contract.ContractName.StartsWith("“") && contract.ContractName.EndsWith("”"))
{
contract.ContractName = contract.ContractName.TrimStart("“".ToCharArray()).TrimEnd("”".ToCharArray());
}
//去掉书名号
contract.ContractName = contract.ContractName.Replace("《", String.Empty).Replace("》", String.Empty);
if (contract.ContractName.Contains("（以下简称"))
{
contract.ContractName = Utility.GetStringAfter(contract.ContractName, "（以下简称");
}
contract.ContractName = contract.ContractName.NormalizeTextResult();
}
//金额
var money = GetMoney();
contract.ContractMoneyUpLimit = MoneyUtility.Format(money.MoneyAmount, String.Empty);
contract.ContractMoneyDownLimit = contract.ContractMoneyUpLimit;
//联合体
contract.UnionMember = GetUnionMember(contract.JiaFang, contract.YiFang);
contract.UnionMember = contract.UnionMember.NormalizeTextResult();
//按照规定除去括号
contract.UnionMember = RegularTool.TrimBrackets(contract.UnionMember);
return(contract);
}
/// <summary>
/// 根据表头标题抽取
/// </summary>
/// <param name="root"></param>
/// <param name="id"></param>
/// <returns></returns>
List <RecordBase> ExtractFromTable()
{
var StockHolderRule = new TableSearchTitleRule();
StockHolderRule.Name = "股东全称";
StockHolderRule.Title = new string[] { "股东名称", "名称", "增持主体", "增持人", "减持主体", "减持人", "姓名" }.ToList();
StockHolderRule.IsTitleEq = true;
StockHolderRule.IsRequire = true;
var ChangeDateRule = new TableSearchTitleRule();
ChangeDateRule.Name = "变动截止日期";
ChangeDateRule.Title = new string[] { "买卖时间", "日期", "减持期间", "增持期间", "减持股份期间", "增持股份期间",
"减持时间", "增持时间", "减持股份时间", "增持股份时间", "买入时间", "卖出时间" }.ToList();
ChangeDateRule.IsTitleEq = false;
ChangeDateRule.Normalize = NormailizeEndChangeDate;
var ChangePriceRule = new TableSearchTitleRule();
ChangePriceRule.Name = "变动价格";
ChangePriceRule.Title = new string[] { "买入均价", "卖出均价", "成交均价", "减持价格", "增持价格", "减持股均价", "增持股均价", "减持均", "增持均", "价格区间" }.ToList();
ChangePriceRule.IsTitleEq = false;
ChangePriceRule.Normalize = (x, y) =>
{
var prices = RegularTool.GetRegular(x, RegularTool.MoneyExpress);
if (prices.Count == 0)
{
if (x.Contains("元"))
{
return(Utility.GetStringBefore(x, "元"));
}
}
else
{
//增减持，区间的情况，取最高价,假设最后一个数字是最大的
return(prices.Last().RawData);
}
return(x);
};
var ChangeNumberRule = new TableSearchTitleRule();
ChangeNumberRule.Name = "变动数量";
ChangeNumberRule.Title = new string[] { "成交数量", "减持股数", "增持股数", "减持数量", "增持数量", "买入股份数", "卖出股份数", "股数" }.ToList();
ChangeNumberRule.IsTitleEq = false;
ChangeNumberRule.Normalize = NumberUtility.NormalizerStockNumber;
var Rules = new List <TableSearchTitleRule>();
Rules.Add(StockHolderRule);
Rules.Add(ChangeDateRule);
Rules.Add(ChangePriceRule);
Rules.Add(ChangeNumberRule);
var result = HTMLTable.GetMultiInfoByTitleRules(root, Rules, false);
if (result.Count == 0)
{
//没有抽取到任何数据
Rules.Clear();
ChangeDateRule.IsRequire = true;
Rules.Add(ChangeDateRule);
Rules.Add(ChangePriceRule);
Rules.Add(ChangeNumberRule);
result = HTMLTable.GetMultiInfoByTitleRules(root, Rules, false);
if (result.Count == 0)
{
return(new List <RecordBase>());
}
var NewResult = new List <CellInfo[]>();
var Name = GetHolderName();
if (String.IsNullOrEmpty(Name.FullName) && String.IsNullOrEmpty(Name.ShortName))
{
return(new List <RecordBase>());
}
foreach (var item in result)
{
NewResult.Add(new CellInfo[]
{ new CellInfo()
{
RawData = String.IsNullOrEmpty(Name.FullName)?Name.ShortName:Name.FullName
}, item[0], item[1], item[2] });
}
result = NewResult;
}
var holderafterlist = GetHolderAfter();
var stockchangelist = new List <RecordBase>();
foreach (var rec in result)
{
var stockchange = new StockChangeRec();
stockchange.Id = Id;
var ModifyName = rec[0].RawData;
//表格里面长的名字可能被分页切割掉
//这里使用合计表进行验证
if (!holderafterlist.Select((z) => { return(z.Name); }).ToList().Contains(ModifyName))
{
foreach (var item in holderafterlist)
{
if (item.Name.EndsWith("先生"))
{
break; //特殊处理，没有逻辑可言
}
if (item.Name.StartsWith(ModifyName) && !item.Name.Equals(ModifyName))
{
ModifyName = item.Name;
break;
}
if (item.Name.EndsWith(ModifyName) && !item.Name.Equals(ModifyName))
{
ModifyName = item.Name;
break;
}
}
}
var Name = CompanyNameLogic.NormalizeCompanyName(this, ModifyName);
stockchange.HolderFullName = Name.FullName.NormalizeTextResult();
stockchange.HolderShortName = Name.ShortName;
if (stockchange.HolderFullName.Contains("简称"))
{
stockchange.HolderShortName = Utility.GetStringAfter(stockchange.HolderFullName, "简称");
stockchange.HolderShortName = stockchange.HolderShortName.Replace(")", String.Empty).Replace("“", String.Empty).Replace("”", String.Empty);
stockchange.HolderFullName = Utility.GetStringBefore(stockchange.HolderFullName, "(");
}
stockchange.ChangeEndDate = rec[1].RawData;
DateTime x;
if (!DateTime.TryParse(stockchange.ChangeEndDate, out x))
{
//无法处理的情况
if (!Program.IsDebugMode)
{
//非调试模式
stockchange.ChangeEndDate = String.Empty;
}
}
if (!String.IsNullOrEmpty(rec[2].RawData))
{
//股价区间化的去除
if (!(rec[2].RawData.Contains("-") || rec[2].RawData.Contains("~") || rec[2].RawData.Contains("至")))
{
stockchange.ChangePrice = rec[2].RawData.Replace(" ", String.Empty);
stockchange.ChangePrice = stockchange.ChangePrice.Replace("*", "");
stockchange.ChangePrice = stockchange.ChangePrice.NormalizeNumberResult();
}
}
if (!RegularTool.IsUnsign(stockchange.ChangePrice))
{
if (!String.IsNullOrEmpty(stockchange.ChangePrice))
{
Console.WriteLine("Error ChangePrice:[" + stockchange.ChangePrice + "]");
}
stockchange.ChangePrice = String.Empty;
}
if (!String.IsNullOrEmpty(rec[3].RawData))
{
stockchange.ChangeNumber = rec[3].RawData.Replace(" ", String.Empty);
stockchange.ChangeNumber = stockchange.ChangeNumber.NormalizeNumberResult();
if (!RegularTool.IsUnsign(stockchange.ChangeNumber))
{
if (!String.IsNullOrEmpty(stockchange.ChangeNumber))
{
Console.WriteLine("Error ChangeNumber:[" + stockchange.ChangeNumber + "]");
}
stockchange.ChangeNumber = String.Empty;
}
}
//基本上所有的有效记录都有股东名和截至日期，所以，这里这么做，可能对于极少数没有截至日期的数据有伤害，但是对于整体指标来说是好的
if (string.IsNullOrEmpty(stockchange.HolderFullName) || string.IsNullOrEmpty(stockchange.ChangeEndDate))
{
continue;
}
if (stockchange.ChangeNumber == "0" || stockchange.ChangePrice == "0")
{
continue;
}
stockchangelist.Add(stockchange);
}
//寻找所有的股东全称
var namelist = stockchangelist.Select(x => ((StockChangeRec)x).HolderFullName).Distinct().ToList();
var newRec = new List <StockChangeRec>();
foreach (var name in namelist)
{
var stocklist = stockchangelist.Where((x) => { return(((StockChangeRec)x).HolderFullName == name); }).ToList();
stocklist.Sort((x, y) => { return(((StockChangeRec)x).ChangeEndDate.CompareTo(((StockChangeRec)x).ChangeEndDate)); });
var last = (StockChangeRec)stocklist.Last();
for (int i = 0; i < holderafterlist.Count; i++)
{
var after = holderafterlist[i];
after.Name = after.Name.Replace(" ", "");
if (after.Name == last.HolderFullName || after.Name == last.HolderShortName)
{
stockchangelist.Remove(last); //结构体，无法直接修改！！使用删除，增加的方法
last.HoldNumberAfterChange = after.Count;
last.HoldPercentAfterChange = after.Percent;
newRec.Add(last);
}
}
}
if (holderafterlist.Count != namelist.Count)
{
if (!Program.IsMultiThreadMode)
{
Program.Logger.WriteLine("增持者数量确认！");
}
}
stockchangelist.AddRange(newRec);
return(stockchangelist);
}
/// <summary>
/// 获得甲方
/// </summary>
/// <returns></returns>
string GetJiaFang(String YiFang)
{
//最高置信度的抽取
EntityProperty e = new EntityProperty();
e.ExcludeContainsWordList = new string[] { "招标代理" };
e.LeadingColonKeyWordList = new string[] {
"甲方：", "合同买方：",
"发包人：", "发包单位：", "发包方：", "发包机构：", "发包人名称：",
"招标人：", "招标单位：", "招标方：", "招标机构：", "招标人名称：", "项目招标人：",
"业主：", "业主单位：", "业主方：", "业主机构：", "业主名称：",
"采购单位：", "采购单位名称：", "采购人：", "采购人名称：", "采购方：", "采购方名称："
};
e.CandidatePreprocess = (x =>
{
x = Normalizer.ClearTrailing(x);
return(CompanyNameLogic.AfterProcessFullName(x).secFullName);
});
e.MaxLength = 32;
e.MaxLengthCheckPreprocess = Utility.TrimEnglish;
e.MinLength = 3;
e.Extract(this);
//这里不直接做Distinct，出现频次越高，则可信度越高
//多个甲方的时候，可能意味着没有甲方！
if (e.LeadingColonKeyWordCandidate.Distinct().Count() > 1)
{
foreach (var candidate in e.LeadingColonKeyWordCandidate)
{
Program.Logger.WriteLine("发现多个甲方：" + candidate);
}
}
if (e.LeadingColonKeyWordCandidate.Count > 0)
{
return(e.LeadingColonKeyWordCandidate[0]);
}
var ner = SearchJiaFang();
var NerJia = String.Empty;
if (!String.IsNullOrEmpty(ner))
{
foreach (var cn in companynamelist)
{
if (cn.secShortName == ner)
{
ner = cn.secFullName;
}
}
if (String.IsNullOrEmpty(YiFang))
{
NerJia = ner;
}
if (!YiFang.Equals(ner))
{
NerJia = ner;
}
}
//招标
var Extractor = new ExtractPropertyByHTML();
var CandidateWord = new List <String>();
var StartArray = new string[] { "招标单位", "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 = CompanyNameLogic.AfterProcessFullName(item.Value.Trim());
if (JiaFang.secFullName.Contains("招标代理"))
{
continue; //特殊业务规则
}
JiaFang.secFullName = JiaFang.secFullName.Replace("业主", String.Empty).Trim();
JiaFang.secFullName = JiaFang.secFullName.Replace("招标单位", String.Empty).Trim();
if (Utility.TrimEnglish(JiaFang.secFullName).Length > 32)
{
continue;
}
if (JiaFang.secFullName.Length < 3)
{
continue; //使用实际长度排除全英文的情况
}
if (!Program.IsMultiThreadMode)
{
Program.Logger.WriteLine("甲方候补词(招标)：[" + JiaFang.secFullName + "]");
}
CandidateWord.Add(JiaFang.secFullName);
}
//合同
Extractor = new ExtractPropertyByHTML();
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)
{
var JiaFang = CompanyNameLogic.AfterProcessFullName(item.Value.Trim());
JiaFang.secFullName = JiaFang.secFullName.Replace("业主", String.Empty).Trim();
if (JiaFang.secFullName.Contains("招标代理"))
{
continue; //特殊业务规则
}
if (Utility.TrimEnglish(JiaFang.secFullName).Length > 32)
{
continue;
}
if (JiaFang.secFullName.Length < 3)
{
continue; //使用实际长度排除全英文的情况
}
if (!Program.IsMultiThreadMode)
{
Program.Logger.WriteLine("甲方候补词(合同)：[" + JiaFang.secFullName + "]");
}
CandidateWord.Add(JiaFang.secFullName);
}
if (!String.IsNullOrEmpty(NerJia))
{
//原则上，有NER中提取的甲方，则使用甲方
foreach (var c in CandidateWord)
{
//但是，这里有可能是正确的解答，例如
//NER：(集团)有限公司 实际上应该是 XXXX(集团)有限公司
if (c.EndsWith(NerJia))
{
return(c);
}
}
return(NerJia);
}
else
{
return(CompanyNameLogic.MostLikeCompanyName(CandidateWord));
}
}