public GMParagraph AddParagraph(BoundingBox boundingBox)
{
var paragraph = new GMParagraph(boundingBox);
AddParagraph(paragraph);
return(paragraph);
}
public static string AsText(this GMParagraph paragraph, bool useLF = true)
{
var buffer = new StringBuilder();
foreach (var line in paragraph.Lines())
{
if (buffer.Length > 0)
{
buffer.Append(useLF ? '\n' : ' ');
}
var words = line.Words().ToList();
if (words.Count > 0)
{
double sumWidth = 0;
var charCount = 0;
foreach (var word in words)
{
sumWidth += word.BoundingBox.Width;
charCount += word.Text.Length;
}
var averageWidth = sumWidth / charCount;
buffer.Append(words[0].Text);
for (var iIdx = 1; iIdx < words.Count; iIdx++)
{
var curWord = words[iIdx];
var distance = HDistance(words[iIdx - 1].BoundingBox, curWord.BoundingBox);
var spacesCount = (int)Math.Round(distance / averageWidth);
if (spacesCount == 0)
{
spacesCount = 1;
}
buffer.Append(new string(' ', spacesCount));
buffer.Append(curWord.Text);
}
}
}
return(buffer.ToString());
}
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 AddParagraph(GMParagraph para)
{
paragraphs.Add(para);
}
public static int MinAccuracy(this GMParagraph paragraph)
{
var minAccuracy = paragraph.Lines().SelectMany(line => line.Words()).Aggregate(100, (curMin, word) => (word.Accuracy < curMin ? word.Accuracy : curMin));
return(minAccuracy);
}