/// <summary>
/// 股数
/// </summary>
/// <param name="root"></param>
/// <returns></returns>
public static List <LocAndValue <String> > LocateStockNumber(HTMLEngine.MyRootHtmlNode root)
{
var targetRegular = new ExtractProperyBase.struRegularExpressFeature()
{
RegularExpress = @"\d+(,\d+)+",
TrailingWordList = new string[] { "股" }.ToList()
};
var list = new List <LocAndValue <String> >();
foreach (var paragrah in root.Children)
{
foreach (var sentence in paragrah.Children)
{
var ExpResult = ExtractPropertyByHTML.RegularExFinder(sentence.PositionId, sentence.Content, targetRegular, "|");
list.AddRange(ExpResult);
}
}
return(list);
}
/// <summary>
/// 从释义表抽取数据
/// </summary>
/// <param name="Target"></param>
/// <param name="Comany"></param>
/// <returns></returns>
private List <(string Target, string Company)> ExtractTargetFromExplainTable(List <struCompanyName> CompanyAtExplainTable, string[] ExplainKeys)
{
var AllCompanyName = new List <String>();
foreach (var item in CompanyAtExplainTable)
{
if (!String.IsNullOrEmpty(item.secShortName))
{
AllCompanyName.Add(item.secShortName);
}
if (!String.IsNullOrEmpty(item.secFullName))
{
AllCompanyName.Add(item.secFullName);
}
}
//股份的抽取
var targetRegular = new ExtractProperyBase.struRegularExpressFeature()
{
LeadingWordList = AllCompanyName,
RegularExpress = RegularTool.PercentExpress,
TrailingWordList = new string[] { "的股权", "股权", "的权益", "权益", "的股份", "股份" }.ToList()
};
//其他标的
var OtherTargets = new string[] { "资产及负债", "资产和负债",
"主要资产和部分负债", "主要资产及部分负债",
"经营性资产及负债", "经营性资产和负债", "应收账款和其他应收款",
"负债", "债权", "全部权益", "经营性资产", "非股权类资产", "资产、负债、业务",
"直属资产", "普通股股份", "土地使用权", "使用权", "房产" };
var TargetAndCompanyList = new List <(string Target, string Comany)>();
foreach (var Rplkey in ExplainKeys)
{
//可能性最大的排在最前
foreach (var ExplainDictItem in ExplainDict)
{
var keys = ExplainDictItem.Key.Split(Utility.SplitChar);
var keys2 = ExplainDictItem.Key.Split(new char[] { '／', '/' });
if (keys.Length == 1 && keys2.Length > 1)
{
keys = keys2;
}
var values = ExplainDictItem.Value.Split(Utility.SplitChar);
var values2 = ExplainDictItem.Value.Split("；");
if (values.Length == 1 && values2.Length > 1)
{
values = values2;
}
//keys里面可能包括【拟】字需要去除
var SearchKey = keys.Select((x) => { return(x.StartsWith("拟") ? x.Substring(1) : x); });
SearchKey = SearchKey.Select(x => x.Trim()).ToArray();
if (SearchKey.Contains(Rplkey))
{
if (Rplkey.Equals("交易标的") || Rplkey.Equals("标的资产") || Rplkey.Equals("标的公司"))
{
foreach (var cn in companynamelist)
{
if (ExplainDictItem.Value.Equals(cn.secFullName) ||
ExplainDictItem.Value.Equals(cn.secShortName))
{
var extra = ("100%股权", ExplainDictItem.Value);
TargetAndCompanyList.Add(extra);
Console.WriteLine(Id + ":100%股权" + ExplainDictItem.Value);
return(TargetAndCompanyList);
}
}
}
foreach (var targetRecordItem in values)
{
var SingleItemList = Utility.CutByPOSConection(targetRecordItem);
foreach (var SingleItem in SingleItemList)
{
var targetAndcompany = SingleItem.Trim().Replace(" ", "");
targetAndcompany = targetAndcompany.Trim().Replace("合计", "");
if (targetAndcompany.Contains("持有的"))
{
targetAndcompany = Utility.GetStringAfter(targetAndcompany, "持有的");
}
if (targetAndcompany.Contains("持有"))
{
targetAndcompany = Utility.GetStringAfter(targetAndcompany, "持有");
}
if (targetAndcompany.Contains("所持"))
{
targetAndcompany = Utility.GetStringAfter(targetAndcompany, "所持");
}
//将公司名称和交易标的划分开来
var ExpResult = ExtractPropertyByHTML.RegularExFinder(0, targetAndcompany, targetRegular, "|");
if (ExpResult.Count == 0)
{
//其他类型的标的
if (!String.IsNullOrEmpty(GetOtherOwnerByExplainTable(targetAndcompany)))
{
var extra = (targetAndcompany, GetOtherOwnerByExplainTable(targetAndcompany));
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
}
else
{
foreach (var rc in CompanyAtExplainTable)
{
var IsFullNameHit = false;
//资产里面可能是带有公司名字的情况
if (!String.IsNullOrEmpty(rc.secFullName) && targetAndcompany.Contains(rc.secFullName))
{
foreach (var ot in OtherTargets)
{
if (targetAndcompany.Contains(ot))
{
IsFullNameHit = true;
TargetAndCompanyList.Add((ot, rc.secFullName));
break;
}
}
}
if (!IsFullNameHit)
{
if (!String.IsNullOrEmpty(rc.secShortName) && targetAndcompany.Contains(rc.secShortName))
{
foreach (var ot in OtherTargets)
{
if (targetAndcompany.Contains(ot))
{
IsFullNameHit = true;
TargetAndCompanyList.Add((ot, rc.secFullName));
break;
}
}
}
}
//XXXX持有的XXXX的形式，不过现在可能已经不用了
if (TargetAndCompanyList.Count == 0 && !String.IsNullOrEmpty(rc.secFullName) && targetAndcompany.StartsWith(rc.secFullName))
{
var extra = (targetAndcompany.Substring(rc.secFullName.Length), rc.secFullName);
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
break;
}
if (TargetAndCompanyList.Count == 0 && !String.IsNullOrEmpty(rc.secShortName) && targetAndcompany.StartsWith(rc.secShortName))
{
var extra = (targetAndcompany.Substring(rc.secShortName.Length), rc.secShortName);
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
break;
}
}
}
}
else
{
foreach (var r in ExpResult)
{
var arr = r.Value.Split("|");
var target = arr[1] + arr[2];
var targetCompany = arr[0];
if (targetCompany.Contains("持有的"))
{
targetCompany = Utility.GetStringAfter(targetCompany, "持有的");
}
if (targetCompany.Contains("持有"))
{
targetCompany = Utility.GetStringAfter(targetCompany, "持有");
}
if (targetCompany.Contains("所持"))
{
targetCompany = Utility.GetStringAfter(targetCompany, "所持");
}
var extra = (target.Replace(" ", ""), targetCompany.Replace(" ", ""));
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
}
}
}
}
if (TargetAndCompanyList.Count != 0)
{
return(TargetAndCompanyList);
}
}
}
}
return(TargetAndCompanyList);
}
private List <(string Target, string Company)> ExtractExtend(string[] ExplainKeys)
{
var targetRegular = new ExtractProperyBase.struRegularExpressFeature()
{
RegularExpress = RegularTool.PercentExpress,
TrailingWordList = new string[] { "的股权", "股权", "的权益", "权益", "的股份", "股份" }.ToList()
};
var Result = new List <(string Target, string Comany)>();
//可能性最大的排在最前
foreach (var item in ExplainDict)
{
var list = new List <(string Target, string Comany)>();
var keys = item.Key.Split(Utility.SplitChar);
var keys2 = item.Key.Split(new char[] { '／', '/' });
if (keys.Length == 1 && keys2.Length > 1)
{
keys = keys2;
}
var values = item.Value.Split(Utility.SplitChar);
var values2 = item.Value.Split("；");
if (values.Length == 1 && values2.Length > 1)
{
values = values2;
}
foreach (var ek in ExplainKeys)
{
if (keys.Contains(ek))
{
foreach (var value in values)
{
var serachWord = value.Replace(" ", string.Empty);
foreach (var words in serachWord.Split(Utility.SplitChar))
{
var SingleItemList = Utility.CutByPOSConection(words);
foreach (var SingleItem in SingleItemList)
{
var ExpResult = ExtractPropertyByHTML.RegularExFinder(0, SingleItem, targetRegular, "|");
foreach (var r in ExpResult)
{
var arr = r.Value.Split("|");
var target = arr[1] + arr[2];
var targetCompany = SingleItem.Substring(0, r.StartIdx);
if (targetCompany.Contains("持有的"))
{
targetCompany = Utility.GetStringAfter(targetCompany, "持有的");
}
if (targetCompany.Contains("持有"))
{
targetCompany = Utility.GetStringAfter(targetCompany, "持有");
}
if (targetCompany.Contains("所持"))
{
targetCompany = Utility.GetStringAfter(targetCompany, "所持");
}
var extra = (target, targetCompany);
list.Add(extra);
}
}
}
}
if (list.Count != 0)
{
return(list.Distinct().ToList());
}
}
}
}
return(Result);
}
/// <summary>
/// 从释义表格中抽取
/// </summary>
/// <returns></returns>
List <(string Target, string Comany)> getTargetListFromReplaceTable()
{
var ReplaceCompany = new List <String>();
foreach (var c in companynamelist)
{
if (c.positionId == -1)
{
//释义表
if (!String.IsNullOrEmpty(c.secShortName))
{
ReplaceCompany.Add(c.secShortName);
}
}
}
var TargetAndCompanyList = new List <(string Target, string Comany)>();
//股份的抽取
var targetRegular = new ExtractProperyBase.struRegularExpressFeature()
{
LeadingWordList = ReplaceCompany,
RegularExpress = RegularTool.PercentExpress,
TrailingWordList = new string[] { "的股权", "股权", "的权益", "权益" }.ToList()
};
var ReplacementKeys = new string[]
{
"交易标的", //09% 00303
"标的资产", //15% 00464
"本次交易", //12% 00369
"本次重组", //09% 00297
"拟购买资产", //07% 00221
"本次重大资产重组", //07% 00219
"置入资产", //03% 00107
"本次发行", //02% 00070
"拟注入资产", //02% 00068
"目标资产" //02% 00067
};
foreach (var Rplkey in ReplacementKeys)
{
//可能性最大的排在最前
foreach (var item in ReplacementDict)
{
var keys = item.Key.Split(Utility.SplitChar);
var keys2 = item.Key.Split("/");
if (keys.Length == 1 && keys2.Length > 1)
{
keys = keys2;
}
var values = item.Value.Split(Utility.SplitChar);
var values2 = item.Value.Split("；");
if (values.Length == 1 && values2.Length > 1)
{
values = values2;
}
if (keys.Contains(Rplkey))
{
foreach (var value in values)
{
var targetAndcompany = value.Trim();
//将公司名称和交易标的划分开来
var ExpResult = ExtractPropertyByHTML.RegularExFinder(0, value, targetRegular, "|");
if (ExpResult.Count == 0)
{
//其他类型的标的
foreach (var rc in ReplaceCompany)
{
if (targetAndcompany.StartsWith(rc))
{
var extra = (value.Substring(rc.Length), rc);
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
break;
}
}
}
else
{
foreach (var r in ExpResult)
{
var arr = r.Value.Split("|");
var extra = (arr[1] + arr[2], arr[0]);
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
}
}
}
if (TargetAndCompanyList.Count != 0)
{
return(TargetAndCompanyList);
}
}
}
}
return(TargetAndCompanyList);
}
/// <summary>
/// 从释义表抽取数据
/// </summary>
/// <param name="Target"></param>
/// <param name="Comany"></param>
/// <returns></returns>
private List <(string Target, string Comany)> ExtractFromExplainTable(List <struCompanyName> CompanyAtExplainTable, string[] ExplainKeys)
{
var AllCompanyName = new List <String>();
foreach (var item in CompanyAtExplainTable)
{
if (!String.IsNullOrEmpty(item.secShortName))
{
AllCompanyName.Add(item.secShortName);
}
if (!String.IsNullOrEmpty(item.secFullName))
{
AllCompanyName.Add(item.secFullName);
}
}
//股份的抽取
var targetRegular = new ExtractProperyBase.struRegularExpressFeature()
{
LeadingWordList = AllCompanyName,
RegularExpress = RegularTool.PercentExpress,
TrailingWordList = new string[] { "的股权", "股权", "的权益", "权益" }.ToList()
};
var OtherTargets = new string[] { "资产及负债", "直属资产" };
var TargetAndCompanyList = new List <(string Target, string Comany)>();
foreach (var Rplkey in ExplainKeys)
{
//可能性最大的排在最前
foreach (var item in ExplainDict)
{
var keys = item.Key.Split(Utility.SplitChar);
var keys2 = item.Key.Split("/");
if (keys.Length == 1 && keys2.Length > 1)
{
keys = keys2;
}
var values = item.Value.Split(Utility.SplitChar);
var values2 = item.Value.Split("；");
if (values.Length == 1 && values2.Length > 1)
{
values = values2;
}
//keys里面可能包括【拟】字需要去除
var SearchKey = keys.Select((x) => { return(x.StartsWith("拟") ? x.Substring(1) : x); });
SearchKey = SearchKey.Select(x => x.Trim()).ToArray();
if (SearchKey.Contains(Rplkey))
{
foreach (var targetRecordItem in values)
{
//DEBUG:
var SingleItemList = Utility.CutByPOSConection(targetRecordItem);
if (SingleItemList.Count == 2)
{
//1.家和股份 和的问题
//2.空格问题
//3.置入和置出问题
//4.其他奇怪的问题
//5.资产和负债
//6.所拥有的，所持有的
//Console.WriteLine(Id + " 分割：");
//Console.WriteLine(Id + " 原词：" + targetRecordItem);
//Console.WriteLine(Id + " 分量1：" + SingleItemList[0]);
//Console.WriteLine(Id + " 分量2：" + SingleItemList[1]);
}
foreach (var SingleItem in SingleItemList)
{
var targetAndcompany = SingleItem.Trim().Replace(" ", "");
//将公司名称和交易标的划分开来
var ExpResult = ExtractPropertyByHTML.RegularExFinder(0, targetAndcompany, targetRegular, "|");
if (ExpResult.Count == 0)
{
//其他类型的标的
foreach (var rc in CompanyAtExplainTable)
{
var IsFullNameHit = false;
if (!String.IsNullOrEmpty(rc.secFullName) && targetAndcompany.Contains(rc.secFullName))
{
foreach (var ot in OtherTargets)
{
if (targetAndcompany.Contains(ot))
{
IsFullNameHit = true;
TargetAndCompanyList.Add((rc.secFullName, ot));
break;
}
}
}
if (!IsFullNameHit)
{
if (!String.IsNullOrEmpty(rc.secShortName) && targetAndcompany.Contains(rc.secShortName))
{
foreach (var ot in OtherTargets)
{
if (targetAndcompany.Contains(ot))
{
IsFullNameHit = true;
TargetAndCompanyList.Add((rc.secShortName, ot));
break;
}
}
}
}
if (TargetAndCompanyList.Count == 0 && !String.IsNullOrEmpty(rc.secFullName) && targetAndcompany.StartsWith(rc.secFullName))
{
var extra = (SingleItem.Substring(rc.secFullName.Length), rc.secFullName);
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
break;
}
if (TargetAndCompanyList.Count == 0 && !String.IsNullOrEmpty(rc.secShortName) && targetAndcompany.StartsWith(rc.secShortName))
{
var extra = (SingleItem.Substring(rc.secShortName.Length), rc.secShortName);
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
break;
}
}
}
else
{
foreach (var r in ExpResult)
{
var arr = r.Value.Split("|");
var extra = (arr[1] + arr[2], arr[0]);
if (!TargetAndCompanyList.Contains(extra))
{
TargetAndCompanyList.Add(extra);
}
}
}
}
}
if (TargetAndCompanyList.Count != 0)
{
return(TargetAndCompanyList);
}
}
}
}
return(TargetAndCompanyList);
}