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 List <CustomLine> buildTextFromParagraph(FREngine.IParagraph par, int iBlock)
{
// CustomParagraph result = new CustomParagraph();
List <CustomWord> wordItems = new List <CustomWord>();
string text = par.Text;
FREngine.CharParams charParams = par.Application.CreateCharParams();
int wordStart = 0;
int wordStartInText = Text.Count;
List <Character> charaterItems = new List <Character>();
for (int iChar = 0; iChar < text.Length; iChar++)
{
char ch = text[iChar];
par.GetCharParams(iChar, charParams);
bool isVisible = (!char.IsWhiteSpace(ch));
if (isVisible)
{
Text.Add(new Character(ch, charParams));
charaterItems.Add(new Character(ch, charParams));
}
// Splitting words by whitespaces
// TODO: line break, hyphen
if (char.IsWhiteSpace(ch))
{
if (wordStart < iChar)
{
int wordEndInText = Text.Count - 1;
words.Add(new Word(wordStartInText, wordEndInText, Text));
CustomWord word = new CustomWord();
//word.Word = new Word(wordStartInText, wordEndInText, Text);
word.X = charaterItems.Min(x => x.Rect.X);
word.Y = charaterItems.Min(x => x.Rect.Y);
word.Value = new Word(wordStartInText, wordEndInText, Text).ToString();
int xMax = charaterItems.Max(x => x.Rect.X);
word.Width = xMax + charaterItems.First(x => x.Rect.X == xMax).Rect.Width - word.X;
word.Height = charaterItems.Max(x => x.Rect.Height);
//word.CharaterItems = charaterItems;
wordItems.Add(word);
listWords.Add(word);
charaterItems = new List <Character>();
}
wordStart = iChar + 1;
wordStartInText = Text.Count;
}
}
// Add last word if necessary
if (wordStart < text.Length)
{
int wordEndInText = Text.Count - 1;
words.Add(new Word(wordStartInText, wordEndInText, Text));
CustomWord word = new CustomWord();
//word.Word = new Word(wordStartInText, wordEndInText, Text);
//word.CharaterItems = charaterItems;
word.Value = new Word(wordStartInText, wordEndInText, Text).ToString();
word.X = charaterItems.Min(x => x.Rect.X);
word.Y = charaterItems.Min(x => x.Rect.Y);
int xMax = charaterItems.Max(x => x.Rect.X);
word.Width = xMax + charaterItems.First(x => x.Rect.X == xMax).Rect.Width - word.X;
word.Height = charaterItems.Max(x => x.Rect.Height);
wordItems.Add(word);
listWords.Add(word);
charaterItems = new List <Character>();
}
List <CustomLine> csLineItems = new List <CustomLine>();
//if (result.WordItems.Any())
//{
// CustomLine line = new CustomLine()
// {
// X = par.Left,
// Y = par.Top,
// Width = par.Right - par.Left,
// Height = par.Bottom - par.Top,
// WordItems = result.WordItems,
// Index = 0,
// ParentBlockIndex = iBlock,
// Value = string.Join(" ", result.WordItems.Select(x => x.Word.ToString()).ToArray())
// }
// ;
// csLineItems.Add(line);
// lineAll.Add(line);
//}
FREngine.IParagraphLines lines = par.Lines;
if (wordItems.Any())
{
int iLine = 0;
foreach (FREngine.IParagraphLine item in lines)
{
CustomLine lineItem = new CustomLine()
{
X = item.Left,
Y = item.Top,
Height = item.Bottom - item.Top,
Width = item.Right - item.Left,
ParentIndex = 0,
Index = iLine,
};
int iWords = 0;
lineItem.WordItems = wordItems.Where(x => IsCollucion(lineItem, x)).Select(x => { x.Index = iWords; iWords++; x.ParentIndex = iLine; return(x); }).ToList();
lineItem.Value = string.Join(" ", lineItem.WordItems.OrderBy(x => x.X).Select(x => x.Value));
lineItem.ValueWithoutSpace = string.Join("", lineItem.WordItems.OrderBy(x => x.X).Select(x => x.Value));
csLineItems.Add(lineItem);
lineAll.Add(lineItem);
iLine++;
}
}
return(csLineItems.OrderBy(x => x.Index).ToList());
}
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);
}
