private PageShape ExtractResults(Rectangle pageBounds, int horizontalResolution, int verticalResolution)
{
List <TextBlockShape> textBlocks = new List <TextBlockShape>();
using (ResultIterator iterator = NativeMethods.TessBaseAPIGetIterator(this.handle))
{
TextBlockShape textBlock = ExtractTextBlock();
if (textBlock != null)
{
textBlocks.Add(textBlock);
}
TextBlockShape ExtractTextBlock()
{
Rectangle bounds = iterator.GetBoundingBox(PageIteratorLevel.TextBlock);
if (bounds.IsEmpty)
{
return(null);
}
List <Shape> shapes = new List <Shape>();
do
{
PolyBlockType type = iterator.GetBlockType();
switch (type)
{
case PolyBlockType.HorizontalLine:
case PolyBlockType.VerticalLine:
bounds = iterator.GetBoundingBox(PageIteratorLevel.TextBlock);
if (!bounds.IsEmpty)
{
LineShape line = type == PolyBlockType.HorizontalLine ?
new LineShape(bounds, bounds.Height, LineTypes.Horizontal) :
new LineShape(bounds, bounds.Width, LineTypes.Vertical);
shapes.Add(line);
}
break;
default:
do
{
ParagraphShape shape = ExtractParagraph();
if (shape != null)
{
shapes.Add(shape);
}
}while (!iterator.IsAtFinalElement(PageIteratorLevel.TextBlock, PageIteratorLevel.Paragraph) &&
iterator.Next(PageIteratorLevel.Paragraph));
break;
}
}while (iterator.Next(PageIteratorLevel.TextBlock));
return(shapes.Count > 0 ? new TextBlockShape(bounds, shapes) : null);
}
ParagraphShape ExtractParagraph()
{
Rectangle bounds = iterator.GetBoundingBox(PageIteratorLevel.Paragraph);
if (bounds.IsEmpty)
{
return(null);
}
List <TextLineShape> shapes = new List <TextLineShape>();
do
{
TextLineShape shape = ExtractTextLine();
if (shape != null)
{
shapes.Add(shape);
}
}while (!iterator.IsAtFinalElement(PageIteratorLevel.Paragraph, PageIteratorLevel.TextLine) &&
iterator.Next(PageIteratorLevel.TextLine));
return(shapes.Count > 0 ? new ParagraphShape(bounds, shapes) : null);
}
TextLineShape ExtractTextLine()
{
Rectangle bounds = iterator.GetBoundingBox(PageIteratorLevel.TextLine);
if (bounds.IsEmpty)
{
return(null);
}
List <TextShape> shapes = new List <TextShape>();
do
{
TextShape shape = ExtractWord();
if (shape != null)
{
shapes.Add(shape);
}
}while (!iterator.IsAtFinalElement(PageIteratorLevel.TextLine, PageIteratorLevel.Word) &&
iterator.Next(PageIteratorLevel.Word));
return(shapes.Count > 0 ? new TextLineShape(bounds, shapes) : null);
}
TextShape ExtractWord()
{
Rectangle bounds = iterator.GetBoundingBox(PageIteratorLevel.Word);
if (bounds.IsEmpty)
{
return(null);
}
/*List<CharacterAnswer> ^ characters = gcnew List<CharacterAnswer>();
* do
* {
* CharacterAnswer character;
* if (ExtractCharacter(iterator, character))
* {
* characters->Add(character);
* }
*
* if (iterator->IsAtFinalElement(level, RIL_SYMBOL))
* {
* break;
* }
* }
* while (iterator->Next(RIL_SYMBOL));
*
* if (!Enumerable::Any(characters))
* {
* return nullptr;
* }
*
* int confidence = MakeConfidence(iterator->Confidence(level));*/
string text = iterator.GetUTF8Text(PageIteratorLevel.Word);
if (string.IsNullOrWhiteSpace(text))
{
return(null);
}
float confidence = iterator.GetConfidence(PageIteratorLevel.Word) / 100.0f;
return(new TextShape(bounds, text, confidence));
}
}
PageShape page = new PageShape(pageBounds, horizontalResolution, verticalResolution);
page.AddShapes(textBlocks);
return(page);
}
