private static GMTextBlock TransformTextBlock(GMTextBlock textBlock, Func <BoundingBox, BoundingBox> boxConverter)
{
var targetTextBlock = new GMTextBlock(boxConverter(textBlock.BoundingBox));
foreach (var paragraph in textBlock.Paragraphs())
{
var targetParagraph = targetTextBlock.AddParagraph(boxConverter(paragraph.BoundingBox));
foreach (var line in paragraph.Lines())
{
var targetLine = targetParagraph.AddLine(boxConverter(line.BoundingBox));
foreach (var word in line.Words())
{
var targetWord = new GMWord(boxConverter(word.BoundingBox), word.Text, word.Accuracy);
targetLine.AddWord(targetWord);
}
}
}
foreach (var word in textBlock.StandaloneWords())
{
var targetWord = new GMWord(boxConverter(word.BoundingBox), word.Text, word.Accuracy);
targetTextBlock.AddStandaloneWord(targetWord);
}
return(targetTextBlock);
}
public GMTextBlock AddTextBlock([NotNull] BoundingBox boundingBox)
{
var result = new GMTextBlock(boundingBox);
textBlocks.Add(result);
return(result);
}
public GMTableCell([NotNull] GMTextBlock textBlock, int rowIndex, int colIndex, int rowSpan, int colSpan)
{
TextBlock = textBlock;
RowIndex = rowIndex;
ColIndex = colIndex;
RowSpan = rowSpan;
ColSpan = colSpan;
}
private static GMTableCell CreateTableCell(GMTextBlock textBlock, int[] hSeparators, int[] vSeparators)
{
var box = textBlock.BoundingBox;
int rowIndex = hSeparators.Count(y => y <= box.YMin) - 1;
int rowSpan = hSeparators.Count(y => y <= box.YMax) - rowIndex - 1;
int colIndex = vSeparators.Count(x => x <= box.XMin) - 1;
int colSpan = vSeparators.Count(x => x <= box.XMax) - colIndex - 1;
return(new GMTableCell(textBlock, rowIndex, colIndex, rowSpan, colSpan));
}
private static void AppendTextBlock(GMTextBlock textBlock, StringBuilder result)
{
var bbox = textBlock.BoundingBox;
result.Append(string.Format(HTMLDebugTextBoxTemplate, bbox.XMin, bbox.YMin, bbox.Width, bbox.Height));
foreach (var paragraph in textBlock.Paragraphs())
{
bbox = paragraph.BoundingBox;
result.Append(string.Format(HTMLDebugParagraphBoxTemplate, bbox.XMin, bbox.YMin, bbox.Width, bbox.Height));
foreach (var line in paragraph.Lines())
{
bbox = line.BoundingBox;
result.Append(string.Format(HTMLDebugLineBoxTemplate, bbox.XMin, bbox.YMin, bbox.Width, bbox.Height));
}
}
}
private static GMTextBlock RemoveEmptyElements(this GMTextBlock textBlock)
{
var textBlockNew = new GMTextBlock(textBlock.BoundingBox);
foreach (var paragraph in textBlock.Paragraphs())
{
var paragraphNew = new GMParagraph(paragraph.BoundingBox);
foreach (var line in paragraph.Lines())
{
var lineNew = new GMLine(line.BoundingBox);
foreach (var word in line.Words())
{
if (word.Text.Trim().Length > 0)
{
var wordNew = new GMWord(word.BoundingBox, word.Text, word.Accuracy);
lineNew.AddWord(wordNew);
}
}
if (lineNew.Words().Any())
{
paragraphNew.AddLine(lineNew);
}
}
if (paragraphNew.Lines().Any())
{
textBlockNew.AddParagraph(paragraphNew);
}
foreach (var word in textBlock.StandaloneWords())
{
if (word.Text.Trim().Length > 0)
{
var wordNew = new GMWord(word.BoundingBox, word.Text, word.Accuracy);
textBlockNew.AddStandaloneWord(wordNew);
}
}
}
return(textBlockNew);
}
public void AddTextBlock(GMTextBlock textBlock)
{
textBlocks.Add(textBlock);
}