TableWidthInfo InitializeTableWidthInfo(WordDocHolder docHolder, Table table)
{
TableWidthInfo widthInfo = new TableWidthInfo();
TableProperties tProp = table.GetFirstChild <TableProperties>();
if (tProp != null)
{
if (tProp.TableWidth != null)
{
widthInfo.TableWidthInPixels = TableWidthInfo.TryReadWidth(
tProp.TableWidth.Width,
tProp.TableWidth.Type,
docHolder.DocumentPageSizeInPixels);
}
if (tProp.TableIndentation != null)
{
widthInfo.TableIndentionInPixels = TableWidthInfo.TryReadWidth(
tProp.TableIndentation.Width,
tProp.TableIndentation.Type,
docHolder.DocumentPageSizeInPixels);
}
widthInfo.TableIndentionInPixels += docHolder.DocumentPageLeftMaginInPixels;
}
else
{
widthInfo.TableWidthInPixels = docHolder.DocumentPageSizeInPixels;
}
TableGrid tGrid = table.GetFirstChild <TableGrid>();
if (tGrid != null)
{
widthInfo.ColumnWidths = new List <int>();
foreach (var col in tGrid.Elements <GridColumn>())
{
widthInfo.ColumnWidths.Add(
TableWidthInfo.TryReadWidth(
col.Width,
TableWidthUnitValues.Dxa,
widthInfo.TableWidthInPixels));
}
}
return(widthInfo);
}
void InitPageSize()
{
var docPart = WordDocument.MainDocumentPart;
var pageSize = docPart.Document.Descendants <PageSize>().FirstOrDefault();
int pageDxa = 11906; // letter size is ISO 216 A4 (210x297mm
if (pageSize != null)
{
pageDxa = (int)(uint)pageSize.Width;
}
DocumentPageSizeInPixels = TableWidthInfo.DxaToPixels(pageDxa);
var pageMargin = docPart.Document.Descendants <PageMargin>().FirstOrDefault();
int pageMarginDxa = 0; // letter size is ISO 216 A4 (210x297mm
if (pageMargin != null && pageMargin.Left != null)
{
pageMarginDxa = (int)(uint)pageMargin.Left;
}
DocumentPageLeftMaginInPixels = TableWidthInfo.DxaToPixels(pageMarginDxa);
}
void ProcessWordTable(WordDocHolder docHolder, Table table, int maxRowsToProcess)
{
var rows = table.Descendants <TableRow>().ToList();
TableWidthInfo widthInfo = InitializeTableWidthInfo(docHolder, table);
int saveRowsCount = TableRows.Count;
int maxCellsCount = 0;
TableBorders tblBorders = GetTableBorders(table);
for (int r = 0; r < rows.Count(); ++r)
{
OpenXmlTableRow newRow = new OpenXmlTableRow();
int sumspan = 0;
var tableRow = rows[r];
int rowGridBefore = GetRowGridBefore(tableRow);
bool isEmpty = true;
var row = tableRow.Elements <TableCell>().ToArray();
for (var i = 0; i < row.Length; ++i)
{
var c = new OpenXmlWordCell(docHolder, row, i, widthInfo, TableRows.Count, sumspan, tblBorders);
if (newRow.RowCells.Count == 0)
{
c.MergedColsCount += rowGridBefore;
}
if (newRow.RowCells.Count > 0 && !newRow.RowCells.Last().HasRightBorder)
{
newRow.RowCells.Last().Text += c.Text;
newRow.RowCells.Last().CellWidth += c.CellWidth;
newRow.RowCells.Last().MergedColsCount += c.MergedColsCount;
newRow.RowCells.Last().HasRightBorder = c.HasRightBorder;
sumspan += c.MergedColsCount;
}
else
{
newRow.RowCells.Add(c);
sumspan += c.MergedColsCount;
}
isEmpty = isEmpty && c.IsEmpty;
}
if (isEmpty)
{
continue;
}
maxCellsCount = Math.Max(newRow.RowCells.Count, maxCellsCount);
if (r == 0 && TableRows.Count > 0 &&
BigramsHolder.CheckMergeRow(
TableRows.Last().RowCells.ConvertAll(x => x.Text),
newRow.RowCells.ConvertAll(x => x.Text)))
{
MergeRow(TableRows.Last().RowCells, newRow.RowCells);
}
else
{
TableRows.Add(newRow);
}
if ((maxRowsToProcess != -1) && (TableRows.Count >= maxRowsToProcess))
{
break;
}
}
if ((TableRows.Count > 0) && !TableHeaderRecognizer.IsNamePositionAndIncomeTable(GetDataCells(0)))
{
if (maxCellsCount <= 4 || CheckNameColumnIsEmpty(saveRowsCount))
{
//remove this suspicious table
TableRows.RemoveRange(saveRowsCount, TableRows.Count - saveRowsCount);
}
}
}
public OpenXmlWordCell(WordDocHolder docHolder, TableCell[] row, int cellIndexInRow, TableWidthInfo tableWidth,
int rowIndexInTable, int unmergedColumnIndex, TableBorders tblBorders)
{
TableCell inputCell = row[cellIndexInRow];
InitTextProperties(docHolder, inputCell);
if (inputCell?.TableCellProperties?.TableCellBorders != null)
{
var borders = inputCell.TableCellProperties.TableCellBorders;
HasBottomBorder = WordDocHolder.BorderIsVisible(borders.BottomBorder);
HasTopBorder = WordDocHolder.BorderIsVisible(borders.TopBorder);
HasRightBorder = WordDocHolder.BorderIsVisible(borders.RightBorder);
if (!HasRightBorder && cellIndexInRow + 1 < row.Length && row[cellIndexInRow + 1].TableCellProperties?.TableCellBorders != null)
{
HasRightBorder = WordDocHolder.BorderIsVisible(row[cellIndexInRow + 1].TableCellProperties?.TableCellBorders?.LeftBorder);
}
if (!HasRightBorder && tblBorders?.InsideVerticalBorder != null && (uint)tblBorders.InsideVerticalBorder.Size > 0)
{
HasRightBorder = true;
}
}
var vmerge = inputCell.TableCellProperties.GetFirstChild <VerticalMerge>();
VerticallyMerged = null;
if (vmerge != null)
{
if ((vmerge.Val == null) || (vmerge.Val == MergedCellValues.Continue))
{
// null -> MergedCellValues.Continue
VerticallyMerged = MergedCellValues.Continue;
}
else if (vmerge.Val == MergedCellValues.Restart)
{
VerticallyMerged = MergedCellValues.Restart;
}
}
if (tblBorders?.InsideHorizontalBorder != null && (uint)tblBorders.InsideHorizontalBorder.Size > 0)
{
TableHasInsideHorizontalBorders = true;
}
var gridSpan = inputCell.TableCellProperties.GetFirstChild <GridSpan>();
IsMerged = gridSpan != null && gridSpan.Val > 1;
FirstMergedRow = -1; // init afterwards
MergedRowsCount = -1; // init afterwards
MergedColsCount = (gridSpan == null) ? 1 : (int)gridSpan.Val;
Row = rowIndexInTable;
Col = unmergedColumnIndex;
if (inputCell.TableCellProperties != null &&
inputCell.TableCellProperties.TableCellWidth != null &&
inputCell.TableCellProperties.TableCellWidth.Type != null &&
inputCell.TableCellProperties.TableCellWidth.Type != TableWidthUnitValues.Auto
)
{
CellWidth = TableWidthInfo.TryReadWidth(
inputCell.TableCellProperties.TableCellWidth.Width,
inputCell.TableCellProperties.TableCellWidth.Type,
tableWidth.TableWidthInPixels);
}
else
{
if (Col < tableWidth.ColumnWidths.Count)
{
CellWidth = tableWidth.ColumnWidths[Col];
}
}
AdditTableIndention = tableWidth.TableIndentionInPixels;
}