private RecognitionStatistics computeStatisticsForBlock(int pageIndex, FREngine.IBlock block)
{
RecognitionStatistics result = new RecognitionStatistics();
if (block.Type == FREngine.BlockTypeEnum.BT_Table)
{
FREngine.TableBlock tableBlock = block.GetAsTableBlock();
for (int iCell = 0; iCell < tableBlock.Cells.Count; iCell++)
{
result += computeStatisticsForBlock(pageIndex, tableBlock.Cells[iCell].Block);
}
}
else if (block.Type == FREngine.BlockTypeEnum.BT_Text)
{
FREngine.ITextBlock textBlock = block.GetAsTextBlock();
int paragraphsCount = textBlock.Text.Paragraphs.Count;
for (int iPar = 0; iPar < paragraphsCount; iPar++)
{
FREngine.IParagraph par = textBlock.Text.Paragraphs[iPar];
string text = par.Text;
result.TotalCharsCount += text.Length;
FREngine.CharParams charParams = engine.CreateCharParams();
for (int iChar = 0; iChar < text.Length; iChar++)
{
par.GetCharParams(iChar, charParams);
if (charParams.IsSuspicious)
{
result.SuspiciousCharsCount++;
}
}
}
}
return(result);
}
private CustomBlock buildTextFromBlock(FREngine.IBlock block, int iBlock)
{
if (block.Type == FREngine.BlockTypeEnum.BT_Table)
{
FREngine.ITableBlock tableBlock = block.GetAsTableBlock();
CustomTable cstable = new CustomTable();
cstable.X = tableBlock.Region.BoundingRectangle.Left;
cstable.Y = tableBlock.Region.BoundingRectangle.Top;
cstable.Width = tableBlock.Region.BoundingRectangle.Width;
cstable.Height = tableBlock.Region.BoundingRectangle.Height;
cstable.ParentIndex = Page.Index;
cstable.Index = tableIndex;
tableIndex++;
List <CustomBlock> blockItems = new List <CustomBlock>();
for (int iCell = 0; iCell < tableBlock.Cells.Count; iCell++)
{
FREngine.ITableCell cell = tableBlock.Cells[iCell];
CustomBlock csBlock = buildTextFromBlock(cell.Block, iCell);
if (csBlock != null)
{
blockItems.Add(csBlock);
//blockAll.Add(csBlock);
}
}
var rowItemsGroup = blockItems.GroupBy(x => x.Y);
int iRow = 0;
foreach (var item in rowItemsGroup)
{
CustomRow rowItem = new CustomRow()
{
Width = cstable.Width,
Height = item.Max(x => x.Height),
BlockItems = item.ToList(),
X = item.Min(x => x.X),
Y = item.Min(x => x.Y),
Value = string.Join("\t", item.Select(x => x.Value)),
Index = iRow,
ParentIndex = iBlock,
};
cstable.RowItems.Add(rowItem);
iRow++;
}
cstable.Value = string.Join("\n", cstable.RowItems.Select(x => x.Value));; //
//string.Join("|", cstable.BlockItems.Select(x => x.Value));
Page.TableItems.Add(cstable);
return(null);
}
if (block.Type != FREngine.BlockTypeEnum.BT_Text)
{
return(null);
}
CustomBlock result = new CustomBlock();
FREngine.ITextBlock textBlock = block.GetAsTextBlock();
result.Index = iBlock;
result.X = textBlock.Region.BoundingRectangle.Left;
result.Y = textBlock.Region.BoundingRectangle.Top;
result.Width = textBlock.Region.BoundingRectangle.Width;
result.Height = textBlock.Region.BoundingRectangle.Height;
for (int iPar = 0; iPar < textBlock.Text.Paragraphs.Count; iPar++)
{
FREngine.IParagraph par = textBlock.Text.Paragraphs[iPar];
result.LineItems.AddRange(buildTextFromParagraph(par, iBlock));
}
result.Value = string.Join("\n", result.LineItems.Select(x => x.Value));
return(result);
}
