/// <summary>
/// 合并表
/// </summary>
/// <param name="doc"></param>
/// <param name="NextTableId"></param>
public static void MergeTable(AnnouceDocument doc, int NextTableId)
{
var table = new HTMLTable(doc.root.TableList[NextTableId - 1]);
string[] pos;
string[] tablerec;
string value;
var offset = table.RowCount;
//修改第二张表格的数据
foreach (var Nextitem in doc.root.TableList[NextTableId])
{
tablerec = Nextitem.Split("|");
pos = tablerec[0].Split(",");
value = tablerec[1];
var newtablerec = (NextTableId - 1) + "," + (offset + int.Parse(pos[1])) + "," + pos[2] + "|" + value;
doc.root.TableList[NextTableId - 1].Add(newtablerec);
}
doc.root.TableList[NextTableId].Clear();
for (int i = 0; i < doc.root.Children.Count; i++)
{
for (int j = 0; j < doc.root.Children[i].Children.Count; j++)
{
var node = doc.root.Children[i].Children[j];
if (node.TableId == NextTableId)
{
node.TableId = -1;
}
}
}
}
/// <summary>
/// /// 分页表格的修复
/// </summary>
/// <param name="root"></param>
public static void FixSpiltTable(MyRootHtmlNode root, AnnouceDocument doc)
{
for (int NextTableId = 2; NextTableId <= doc.root.TableList.Count; NextTableId++)
{
foreach (var item in doc.root.TableList[NextTableId])
{
var FirstTablePos = -1;
var SecondTablePos = -1;
foreach (var p in root.Children)
{
foreach (var s in p.Children)
{
if (s.TableId == NextTableId - 1)
{
FirstTablePos = s.PositionId;
}
if (s.TableId == NextTableId)
{
SecondTablePos = s.PositionId;
}
}
}
if (SecondTablePos - FirstTablePos > 200)
{
continue;
}
var tablerec = item.Split("|");
var pos = tablerec[0].Split(",");
var value = tablerec[1];
var row = int.Parse(pos[1]);
//第二张表，第一行存在NULL
if (row == 1 && value == strNullValue)
{
var table = new HTMLTable(doc.root.TableList[NextTableId - 1]);
var nexttable = new HTMLTable(doc.root.TableList[NextTableId]);
if (table.ColumnCount != nexttable.ColumnCount)
{
continue;
}
//合并表
var offset = table.RowCount;
//修改第二张表格的数据
foreach (var Nextitem in root.TableList[NextTableId])
{
tablerec = Nextitem.Split("|");
pos = tablerec[0].Split(",");
value = tablerec[1];
var newtablerec = (NextTableId - 1) + "," + (offset + int.Parse(pos[1])) + "," + pos[2] + "|" + value;
root.TableList[NextTableId - 1].Add(newtablerec);
}
root.TableList[NextTableId].Clear();
for (int i = 0; i < root.Children.Count; i++)
{
for (int j = 0; j < root.Children[i].Children.Count; j++)
{
var node = root.Children[i].Children[j];
if (node.TableId == NextTableId)
{
node.TableId = -1;
}
}
}
break;
}
}
}
//1.是否存在连续表格 NextBrother
for (int i = 0; i < root.Children.Count; i++)
{
for (int j = 0; j < root.Children[i].Children.Count; j++)
{
var node = root.Children[i].Children[j];
if (node.TableId != -1)
{
if (node.NextBrother != null)
{
if (node.NextBrother.TableId != -1)
{
var nextnode = node.NextBrother;
var table = new HTMLTable(root.TableList[node.TableId]);
var nexttable = new HTMLTable(root.TableList[nextnode.TableId]);
//Console.WriteLine("First Table:" + table.RowCount + "X" + table.ColumnCount);
//Console.WriteLine("Second Table:" + nexttable.RowCount + "X" + nexttable.ColumnCount);
if (table.ColumnCount != nexttable.ColumnCount)
{
continue;
}
//Console.WriteLine("Two Tables Has Same Column Count!");
//2.连续表格的后一个，往往是有<NULL>的行
bool hasnull = false;
for (int nullcell = 1; nullcell <= table.ColumnCount; nullcell++)
{
if (nexttable.CellValue(1, nullcell) == HTMLTable.strNullValue)
{
hasnull = true;
break;
}
}
var ComboCompanyName = "";
var ComboCompanyNameColumnNo = -1;
var CompanyFullNameList = doc.companynamelist.Select((x) => { return(x.secFullName); }).Distinct().ToList();
//两表同列的元素，是否有能够合并成为公司名称的？注意，需要去除空格！！
int MaxColumn = table.ColumnCount;
for (int col = 1; col <= MaxColumn; col++)
{
int TableAMaxRow = table.RowCount;
int TableBMaxRow = nexttable.RowCount;
for (int RowCntA = 1; RowCntA < TableAMaxRow; RowCntA++)
{
for (int RowCntB = 1; RowCntB < TableBMaxRow; RowCntB++)
{
var valueA = table.CellValue(RowCntA, col).Replace(" ", "");
var valueB = nexttable.CellValue(RowCntB, col).Replace(" ", "");
if (valueA != "" && valueB != "")
{
var value = valueA + valueB;
if (CompanyFullNameList.Contains(value))
{
ComboCompanyName = value;
ComboCompanyNameColumnNo = col;
//Console.WriteLine("Found FullName:" + value);
break;
}
}
}
if (ComboCompanyNameColumnNo != -1)
{
break;
}
}
if (ComboCompanyNameColumnNo != -1)
{
break;
}
}
if (ComboCompanyNameColumnNo != -1)
{
//补完:注意，不能全部补！！A表以公司名开头，B表以公司名结尾
for (int k = 0; k < root.TableList[node.TableId].Count; k++)
{
var tablerec = root.TableList[node.TableId][k].Split("|");
var value = tablerec[1].Replace(" ", "");
//A表以公司名开头
if (ComboCompanyName.StartsWith(value))
{
root.TableList[node.TableId][k] = tablerec[0] + "|" + ComboCompanyName;
}
}
for (int k = 0; k < root.TableList[nextnode.TableId].Count; k++)
{
var tablerec = root.TableList[nextnode.TableId][k].Split("|");
var value = tablerec[1].Replace(" ", "");
//A表以公司名开头
if (ComboCompanyName.EndsWith(value))
{
root.TableList[nextnode.TableId][k] = tablerec[0] + "|" + ComboCompanyName;
}
}
}
//特殊业务处理:增减持
bool specaillogic = false;
var BuyMethod = new string[] { "集中竞价交易", "竞价交易", "大宗交易", "约定式购回" }.ToList();
if (doc.GetType() == typeof(StockChange))
{
//增减持无表头的特殊处理
for (int spCell = 1; spCell <= table.ColumnCount; spCell++)
{
if (BuyMethod.Contains(nexttable.CellValue(1, spCell)))
{
specaillogic = true;
break;
}
}
}
if (hasnull || ComboCompanyNameColumnNo != -1 || specaillogic)
{
var offset = table.RowCount;
//修改第二张表格的数据
foreach (var item in root.TableList[nextnode.TableId])
{
var tablerec = item.Split("|");
var pos = tablerec[0].Split(",");
var value = tablerec[1];
var newtablerec = node.TableId + "," + (offset + int.Parse(pos[1])) + "," + pos[2] + "|" + value;
root.TableList[node.TableId].Add(newtablerec);
}
root.TableList[nextnode.TableId].Clear();
nextnode.TableId = -1;
//Console.WriteLine("Found Split Tables!!");
}
}
}
}
}
}
}