static List <struIncreaseStock> GetMultiTarget(HTMLEngine.MyRootHtmlNode root,
struIncreaseStock SampleincreaseStock)
{
var BuyerRule = new TableSearchRule();
BuyerRule.Name = "认购对象";
//"投资者名称","股东名称"
BuyerRule.Rule = new string[] { "发行对象", "认购对象", "发行对象名称" }.ToList();
BuyerRule.IsEq = true;
var BuyNumber = new TableSearchRule();
BuyNumber.Name = "增发数量";
BuyNumber.Rule = new string[] { "配售股数", "认购数量", "认购股份数" }.ToList();
BuyNumber.IsEq = false; //包含即可
BuyNumber.Normalize = Normalizer.NormalizerStockNumber;
var BuyMoney = new TableSearchRule();
BuyMoney.Name = "增发金额";
BuyMoney.Rule = new string[] { "配售金额", "认购金额" }.ToList();
BuyMoney.IsEq = false; //包含即可
BuyMoney.Normalize = Normalizer.NormalizerMoney;
var FreezeYear = new TableSearchRule();
FreezeYear.Name = "锁定期";
FreezeYear.Rule = new string[] { "锁定期", "限售期" }.ToList();
FreezeYear.IsEq = false; //包含即可
FreezeYear.Normalize = NormalizerFreezeYear;
var Rules = new List <TableSearchRule>();
Rules.Add(BuyerRule);
Rules.Add(BuyNumber);
Rules.Add(BuyMoney);
Rules.Add(FreezeYear);
var result = HTMLTable.GetMultiInfo(root, Rules, true);
var increaseStocklist = new List <struIncreaseStock>();
foreach (var item in result)
{
var increase = new struIncreaseStock();
increase.id = SampleincreaseStock.id;
increase.BuyMethod = SampleincreaseStock.BuyMethod;
increase.PublishTarget = item[0].RawData;
if (String.IsNullOrEmpty(increase.PublishTarget))
{
continue;
}
increase.IncreaseNumber = item[1].RawData;
increase.IncreaseMoney = item[2].RawData;
increase.FreezeYear = item[3].RawData;
increaseStocklist.Add(increase);
}
return(increaseStocklist);
}
static List <struStockChange> ExtractFromTable(HTMLEngine.MyRootHtmlNode root, string id)
{
var StockHolderRule = new TableSearchRule();
StockHolderRule.Name = "股东全称";
StockHolderRule.Rule = new string[] { "股东名称" }.ToList();
StockHolderRule.IsEq = true;
var ChangeDateRule = new TableSearchRule();
ChangeDateRule.Name = "变动截止日期";
ChangeDateRule.Rule = new string[] { "减持期间", "增持期间", "减持时间", "增持时间" }.ToList();
ChangeDateRule.IsEq = false;
ChangeDateRule.Normalize = Normalizer.NormailizeDate;
var ChangePriceRule = new TableSearchRule();
ChangePriceRule.Name = "变动价格";
ChangePriceRule.Rule = new string[] { "减持均价", "增持均价" }.ToList();
ChangePriceRule.IsEq = false;
ChangePriceRule.Normalize = (x, y) =>
{
if (x.Contains("元"))
{
return(Utility.GetStringBefore(x, "元"));
}
return(x);
};
var ChangeNumberRule = new TableSearchRule();
ChangeNumberRule.Name = "变动数量";
ChangeNumberRule.Rule = new string[] { "减持股数", "增持股数" }.ToList();
ChangeNumberRule.IsEq = false;
ChangeNumberRule.Normalize = Normalizer.NormalizerStockNumber;
var Rules = new List <TableSearchRule>();
Rules.Add(StockHolderRule);
Rules.Add(ChangeDateRule);
Rules.Add(ChangePriceRule);
Rules.Add(ChangeNumberRule);
var result = HTMLTable.GetMultiInfo(root, Rules, false);
//只写在最后一条记录的地方,不过必须及时过滤掉不存在的记录
result.Reverse();
var stockchangelist = new List <struStockChange>();
foreach (var rec in result)
{
var stockchange = new struStockChange();
stockchange.id = id;
var Name = NormalizeCompanyName(rec[0].RawData);
stockchange.HolderFullName = Name.Item1;
stockchange.HolderShortName = Name.Item2;
stockchange.ChangeEndDate = rec[1].RawData;
stockchange.ChangePrice = rec[2].RawData;
stockchange.ChangeNumber = rec[3].RawData;
var holderafterlist = GetHolderAfter(root);
for (int i = 0; i < holderafterlist.Count; i++)
{
var after = holderafterlist[i];
if (after.Used)
{
continue;
}
if (after.Name == stockchange.HolderFullName || after.Name == stockchange.HolderShortName)
{
stockchange.HoldNumberAfterChange = after.Count;
stockchange.HoldPercentAfterChange = after.Percent;
after.Used = true;
break;
}
}
//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;
}
stockchangelist.Add(stockchange);
}
//合并记录
for (int i = 0; i < stockchangelist.Count; i++)
{
var x = stockchangelist[i];
for (int j = i + 1; j < stockchangelist.Count; j++)
{
var y = stockchangelist[j];
if (x.GetKey() == y.GetKey())
{
if (string.IsNullOrEmpty(x.HoldNumberAfterChange) &&
!string.IsNullOrEmpty(y.HoldNumberAfterChange))
{
x.id = "";
}
if (!string.IsNullOrEmpty(x.HoldNumberAfterChange) &&
string.IsNullOrEmpty(y.HoldNumberAfterChange))
{
y.id = "";
}
}
}
}
return(stockchangelist.Where((x) => { return !String.IsNullOrEmpty(x.id); }).ToList());
}
List <struHoldAfter> GetHolderAfter3rdStep()
{
var HoldList = new List <struHoldAfter>();
var StockHolderRule = new TableSearchRule();
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" }.ToList();
StockHolderRule.IsTitleEq = true;
StockHolderRule.IsRequire = true;
var HoldNumberAfterChangeRule = new TableSearchRule();
HoldNumberAfterChangeRule.Name = "变动后持股数";
HoldNumberAfterChangeRule.IsRequire = true;
HoldNumberAfterChangeRule.SuperTitle = new string[] { "减持后", "增持后" }.ToList();
HoldNumberAfterChangeRule.IsSuperTitleEq = false;
HoldNumberAfterChangeRule.Title = new string[] {
"持股股数", "持股股数",
"持股数量", "持股数量",
"持股总数", "持股总数", "股数"
}.ToList();
HoldNumberAfterChangeRule.IsTitleEq = false;
var HoldPercentAfterChangeRule = new TableSearchRule();
HoldPercentAfterChangeRule.Name = "变动后持股数比例";
HoldPercentAfterChangeRule.IsRequire = true;
HoldPercentAfterChangeRule.SuperTitle = HoldNumberAfterChangeRule.SuperTitle;
HoldPercentAfterChangeRule.IsSuperTitleEq = false;
HoldPercentAfterChangeRule.Title = new string[] { "比例" }.ToList();
HoldPercentAfterChangeRule.IsTitleEq = false;
var Rules = new List <TableSearchRule>();
Rules.Add(StockHolderRule);
Rules.Add(HoldNumberAfterChangeRule);
Rules.Add(HoldPercentAfterChangeRule);
var result = HTMLTable.GetMultiInfo(root, Rules, false);
if (result.Count != 0)
{
foreach (var item in result)
{
var HolderName = item[0].RawData;
var strHolderCnt = item[1].RawData;
strHolderCnt = Normalizer.NormalizeNumberResult(strHolderCnt);
string HolderCnt = getAfterstock(item[1].Title, strHolderCnt);
var StrPercent = item[2].RawData;
var HodlerPercent = getAfterpercent(StrPercent);
//Console.WriteLine("GetHolderAfter3rdStep:" + HolderName);
HoldList.Add(new struHoldAfter()
{
Name = HolderName,
Count = HolderCnt,
Percent = HodlerPercent,
Used = false
});
}
}
else
{
StockHolderRule.SuperTitle = StockHolderRule.Title;
StockHolderRule.IsSuperTitleEq = true;
StockHolderRule.Title = null;
StockHolderRule.IsTitleEq = false;
Rules = new List <TableSearchRule>();
Rules.Add(HoldNumberAfterChangeRule);
Rules.Add(HoldPercentAfterChangeRule);
Rules.Add(StockHolderRule);
result = HTMLTable.GetMultiInfo(root, Rules, false);
if (result.Count != 0)
{
foreach (var item in result)
{
var HolderName = item[2].RawData;
var strHolderCnt = item[0].RawData;
strHolderCnt = Normalizer.NormalizeNumberResult(strHolderCnt);
string HolderCnt = getAfterstock(item[1].Title, strHolderCnt);
var StrPercent = item[1].RawData;
var HodlerPercent = getAfterpercent(StrPercent);
//Console.WriteLine("GetHolderAfter4thStep:" + HolderName);
HoldList.Add(new struHoldAfter()
{
Name = HolderName,
Count = HolderCnt,
Percent = HodlerPercent,
Used = false
});
}
}
}
return(HoldList);
}
List <struStockChange> ExtractFromTable(HTMLEngine.MyRootHtmlNode root, string id)
{
var StockHolderRule = new TableSearchRule();
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" }.ToList();
StockHolderRule.IsTitleEq = true;
StockHolderRule.IsRequire = true;
var ChangeDateRule = new TableSearchRule();
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" }.ToList();
ChangeDateRule.IsTitleEq = false;
ChangeDateRule.Normalize = NormailizeEndChangeDate;
var ChangePriceRule = new TableSearchRule();
ChangePriceRule.Name = "变动价格";
ChangePriceRule.Title = new string[] { "成交均价", "减持价格", "增持价格", "减持均", "增持均" }.ToList();
ChangePriceRule.IsTitleEq = false;
ChangePriceRule.Normalize = (x, y) =>
{
if (x.Contains("元"))
{
return(Utility.GetStringBefore(x, "元"));
}
return(x);
};
var ChangeNumberRule = new TableSearchRule();
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" }.ToList();
ChangeNumberRule.IsTitleEq = false;
ChangeNumberRule.Normalize = NumberUtility.NormalizerStockNumber;
var Rules = new List <TableSearchRule>();
Rules.Add(StockHolderRule);
Rules.Add(ChangeDateRule);
Rules.Add(ChangePriceRule);
Rules.Add(ChangeNumberRule);
var result = HTMLTable.GetMultiInfo(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.GetMultiInfo(root, Rules, false);
if (result.Count == 0)
{
return(new List <struStockChange>());
}
var NewResult = new List <CellInfo[]>();
var Name = GetHolderName(this.root);
if (String.IsNullOrEmpty(Name.FullName) && String.IsNullOrEmpty(Name.ShortName))
{
return(new List <struStockChange>());
}
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 <struStockChange>();
foreach (var rec in result)
{
var stockchange = new struStockChange();
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.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 => x.HolderFullName).Distinct().ToList();
var newRec = new List <struStockChange>();
foreach (var name in namelist)
{
var stocklist = stockchangelist.Where((x) => { return(x.HolderFullName == name); }).ToList();
stocklist.Sort((x, y) => { return(x.ChangeEndDate.CompareTo(y.ChangeEndDate)); });
var last = 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);
}
static List <struStockChange> ExtractFromTable(HTMLEngine.MyRootHtmlNode root, string id)
{
var StockHolderRule = new TableSearchRule();
StockHolderRule.Name = "股东全称";
StockHolderRule.Rule = new string[] { "股东名称" }.ToList();
StockHolderRule.IsEq = true;
var ChangeDateRule = new TableSearchRule();
ChangeDateRule.Name = "变动截止日期";
ChangeDateRule.Rule = new string[] { "减持期间", "增持期间", "减持时间", "增持时间" }.ToList();
ChangeDateRule.IsEq = false;
ChangeDateRule.Normalize = Normalizer.NormailizeDate;
var ChangePriceRule = new TableSearchRule();
ChangePriceRule.Name = "变动价格";
ChangePriceRule.Rule = new string[] { "减持均价", "增持均价" }.ToList();
ChangePriceRule.IsEq = false;
ChangePriceRule.Normalize = (x, y) =>
{
if (x.Contains("元"))
{
return(Utility.GetStringBefore(x, "元"));
}
return(x);
};
var ChangeNumberRule = new TableSearchRule();
ChangeNumberRule.Name = "变动数量";
ChangeNumberRule.Rule = new string[] { "减持股数", "增持股数" }.ToList();
ChangeNumberRule.IsEq = false;
ChangeNumberRule.Normalize = Normalizer.NormalizerStockNumber;
var Rules = new List <TableSearchRule>();
Rules.Add(StockHolderRule);
Rules.Add(ChangeDateRule);
Rules.Add(ChangePriceRule);
Rules.Add(ChangeNumberRule);
var result = HTMLTable.GetMultiInfo(root, Rules, false);
//只写在最后一条记录的地方,不过必须及时过滤掉不存在的记录
result.Reverse();
var stockchangelist = new List <struStockChange>();
foreach (var rec in result)
{
var stockchange = new struStockChange();
stockchange.id = id;
var Name = NormalizeCompanyName(rec[0].RawData);
stockchange.HolderFullName = Name.Item1;
stockchange.HolderShortName = Name.Item2;
stockchange.ChangeEndDate = rec[1].RawData;
stockchange.ChangePrice = rec[2].RawData;
stockchange.ChangeNumber = rec[3].RawData;
var holderafterlist = GetHolderAfter(root);
for (int i = 0; i < holderafterlist.Count; i++)
{
var after = holderafterlist[i];
if (after.Used)
{
continue;
}
if (after.Name == stockchange.HolderFullName || after.Name == stockchange.HolderShortName)
{
stockchange.HoldNumberAfterChange = after.Count;
stockchange.HoldPercentAfterChange = after.Percent;
after.Used = true;
break;
}
}
stockchangelist.Add(stockchange);
}
return(stockchangelist);
}
List <struIncreaseStock> GetMultiTarget(HTMLEngine.MyRootHtmlNode root, struIncreaseStock SampleincreaseStock)
{
var BuyerRule = new TableSearchRule();
BuyerRule.Name = "认购对象";
//"投资者名称","股东名称"
BuyerRule.Title = new string[] { "发行对象", "认购对象", "发行对象名称" }.ToList();
BuyerRule.IsTitleEq = true;
BuyerRule.IsRequire = true;
var BuyNumber = new TableSearchRule();
BuyNumber.Name = "增发数量";
BuyNumber.Title = new string[] { "配售股数", "认购数量", "认购股数", "认购股份数", "发行股份数", "配售数量" }.ToList();
BuyNumber.IsTitleEq = false; //包含即可
BuyNumber.Normalize = NumberUtility.NormalizerStockNumber;
var BuyMoney = new TableSearchRule();
BuyMoney.Name = "增发金额";
BuyMoney.Title = new string[] { "配售金额", "认购金额", "获配金额" }.ToList();
BuyMoney.IsTitleEq = false; //包含即可
BuyMoney.Normalize = MoneyUtility.Format;
var FreezeYear = new TableSearchRule();
FreezeYear.Name = "锁定期";
FreezeYear.Title = new string[] { "锁定期", "限售期" }.ToList();
FreezeYear.IsTitleEq = false; //包含即可
FreezeYear.Normalize = NormalizerFreezeYear;
var BuyPrice = new TableSearchRule();
BuyPrice.Name = "价格";
BuyPrice.Title = new string[] { "认购价格", "配售价格", "申购报价" }.ToList();
BuyPrice.IsTitleEq = false; //包含即可
BuyPrice.Normalize = MoneyUtility.Format;
var Rules = new List <TableSearchRule>();
Rules.Add(BuyerRule);
Rules.Add(BuyNumber);
Rules.Add(BuyMoney);
Rules.Add(FreezeYear);
Rules.Add(BuyPrice);
var result = HTMLTable.GetMultiInfo(root, Rules, true);
var increaseStocklist = new List <struIncreaseStock>();
foreach (var item in result)
{
var increase = new struIncreaseStock();
increase.id = SampleincreaseStock.id;
increase.BuyMethod = SampleincreaseStock.BuyMethod;
increase.PublishTarget = item[0].RawData;
if (String.IsNullOrEmpty(increase.PublishTarget))
{
continue;
}
increase.PublishTarget = increase.PublishTarget.NormalizeTextResult();
increase.IncreaseNumber = item[1].RawData;
if (!String.IsNullOrEmpty(increase.IncreaseNumber) && increase.IncreaseNumber.Equals("0"))
{
continue;
}
if (!String.IsNullOrEmpty(increase.IncreaseNumber) && increase.IncreaseNumber.Contains("|"))
{
increase.IncreaseNumber = increase.IncreaseNumber.Split("|").Last();
}
increase.IncreaseMoney = item[2].RawData;
if (!String.IsNullOrEmpty(increase.IncreaseMoney) && increase.IncreaseMoney.Equals("0"))
{
continue;
}
if (!String.IsNullOrEmpty(increase.IncreaseMoney) && increase.IncreaseMoney.Contains("|"))
{
increase.IncreaseMoney = increase.IncreaseMoney.Split("|").Last();
}
//手工计算金额
if (String.IsNullOrEmpty(increase.IncreaseMoney))
{
if (!String.IsNullOrEmpty(increase.IncreaseNumber))
{
if (!String.IsNullOrEmpty(item[4].RawData))
{
double price;
if (double.TryParse(item[4].RawData, out price))
{
double number;
if (double.TryParse(increase.IncreaseNumber, out number))
{
double money = price * number;
if (!Program.IsMultiThreadMode)
{
Program.Logger.WriteLine("通过计算获得金额：" + money.ToString());
}
}
}
}
}
}
increase.FreezeYear = item[3].RawData;
increaseStocklist.Add(increase);
}
return(increaseStocklist);
}