private List <TablePos> GetTablePoses(LinePos tableNamePos, LinePos tableEndPos)
{
List <TablePos> tablePoses = new List <TablePos>();
tableNamePos.AxisValue = tableNamePos.AxisValueWithLineHeight;
//There is only one page
if (tableEndPos.PageNum == tableNamePos.PageNum)
{
TablePos tablePos = GetTablePos(tableNamePos.PageNum, tableNamePos, tableEndPos, false, false, null);
if (tablePos != null)
{
tablePoses.Add(tablePos);
}
}
//The page amount is over one
else
{
//Get start page
//Mark
TablePos startTablePos = null;
LinePos endPos = pdfTronHelper.GetBottomPosOfPage(tableNamePos.PageNum, false, -1, -1);
startTablePos = GetTablePos(tableNamePos.PageNum, tableNamePos, endPos, false, false, null);
if (startTablePos != null)
{
tablePoses.Add(startTablePos);
}
bool isStartTablePosNull = startTablePos == null;
//Get medial page
for (int i = tableNamePos.PageNum + 1; i < tableEndPos.PageNum; i++)
{
TablePos intervalTablePos = GetTablePos(i, null, pdfTronHelper.GetBottomPosOfPage(i, false, -1, -1),
false,
!isStartTablePosNull, isStartTablePosNull ? null : startTablePos.VerticalLines);
if (intervalTablePos != null)
{
tablePoses.Add(intervalTablePos);
}
}
//Get end page
int endPageNum = tableEndPos.PageNum;
TablePos endTablePos = GetTablePos(endPageNum, null, tableEndPos, true,
!isStartTablePosNull, isStartTablePosNull ? null : startTablePos.VerticalLines);
if (endTablePos != null)
{
tablePoses.Add(endTablePos);
}
}
return(tablePoses);
}
void CreateTablePosHighlight(TablePos tablePos, PDFDoc pdfDoc)
{
double lineWidth = 3;
Page page = pdfDoc.GetPage(tablePos.PageNum);
foreach (FormLine line in tablePos.HorizontialLines.Concat(tablePos.VerticalLines).SelectMany(pair => pair.Value))
{
lineWidth = line.IsExistent ? 3 : 1;
Rect rect = line.IsTransverseLine ? new Rect(line.StartPoint.x, line.StartPoint.y, line.EndPoint.x, line.StartPoint.y + lineWidth) :
new Rect(line.StartPoint.x, line.StartPoint.y, line.StartPoint.x + lineWidth, line.EndPoint.y);
PdfTronHelper pdfTronHelper = new PdfTronHelper(pdfDoc);
pdfTronHelper.RevertTransportRect(tablePos.PageNum, rect);
CreateHighlight(pdfDoc, page, rect);
}
}
private TablePos GetTablePos(int pageNum, LinePos startPos, LinePos endPos, bool isEndTablePos, bool isSubsequentPage, SortedDictionary <double, FormLineList> lastPageVerticalLines)
{
FormLineSearcher searcher = new FormLineSearcher(pdfDoc);
if (endPos != null)
{
endPos.AxisValue = endPos.AxisValueWithLineHeight;
}
SortedDictionary <double, FormLineList>[] formLines = searcher.GetFormLines(pageNum, startPos, endPos, isSubsequentPage, lastPageVerticalLines);
if (formLines == null)
{
return(null);
}
TablePos tablePos;
tablePos = new TablePos
{
PageNum = pageNum,
HorizontialLines = formLines[0],
VerticalLines = formLines[1]
};
//RevertAxisTransform(tablePos);
return(tablePos);
}
/// <summary>
/// extract pdf table data
/// </summary>
/// <param name="doc"></param>
/// <param name="tablePos"></param>
/// <returns></returns>
public static FreeTable Extract(pdftron.PDF.PDFDoc doc, TablePos tablePos)
{
PageTextExtractor pdfPageProcess = new PageTextExtractor(doc.GetPage(tablePos.PageNum));
var vLines = new List <System.Windows.Rect>(tablePos.VerticalLines.Count);
var hLines = new List <System.Windows.Rect>(tablePos.HorizontialLines.Count);
foreach (var kv in tablePos.VerticalLines)
{
var top = kv.Value.Min(n => n.StartPoint.y);
var bottom = kv.Value.Max(n => n.EndPoint.y);
vLines.Add(new System.Windows.Rect(kv.Key, top, 0, bottom - top));
}
foreach (var kv in tablePos.HorizontialLines)
{
var left = kv.Value.Min(n => n.StartPoint.x);
var right = kv.Value.Max(n => n.EndPoint.x);
hLines.Add(new System.Windows.Rect(left, kv.Key, right - left, 0));
}
if (vLines.Count == 0 || hLines.Count == 0)
{
throw new Exception("vertical or horizontal lines is empty");
}
hLines.Reverse();
FreeTable freeTable = new FreeTable();
for (int i = 1; i < hLines.Count; i++)
{
var y = hLines[i].Y;
var y2 = hLines[i - 1].Y;
var prej = 0;
var x = hLines[i - 1].Left;
while (prej < vLines.Count)
{
if (vLines[prej].X > x - 5)
{
break;
}
prej++;
}
var x1 = vLines[prej].X;
FreeTableRow row = new FreeTableRow();
for (int j = prej + 1; j < vLines.Count; j++)
{
var vline = vLines[j];
if (vline.Bottom - y > 5 || j == vLines.Count - 1)
{
int ii = i;
for (; ii < hLines.Count - 1; ii++)
{
if (vline.X - hLines[ii].Left > 5)
{
break;
}
}
var data = pdfPageProcess.SearchTextWithStrictMode(
new System.Windows.Rect(x1, hLines[ii].Y, vline.X - x1, y2 - hLines[ii].Y));
row.Add(new FreeTableCell()
{
Value = data, ColSpan = j - prej, RowSpan = ii - i + 1
});
prej = j;
x1 = vline.X;
}
}
freeTable.Add(row);
}
return(freeTable);
}