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);
}
private static bool IsWordValidForAngleDetection(GMWord word)
{
var result = validWordPattern.IsMatch(word.Text);
// if (!result)
// {
// if (!invalidWordPattern.IsMatch(word.Text) && word.Text.Trim().Length > 0)
// {
// Console.Out.WriteLine("Rejected word: \"{0}\"", word.Text);
// }
// }
return(result);
}
/// <summary>
/// Assuming that firstWord and secondWord are representing bounding boxes for words of the same line
/// (edges of bounding boxes are aligned to X and Y axes) whis method detects angle between
/// line and positive direction of X-axis.
/// Angle is detected by direction of line connecting the centers of given bounding rectangles
/// </summary>
/// <param name="firstWord"></param>
/// <param name="secondWord"></param>
/// <returns>Angle in radians</returns>
private static double DetectAngle(GMWord firstWord, GMWord secondWord)
{
var firstBBox = firstWord.BoundingBox;
var secondBBox = secondWord.BoundingBox;
// Page coordinates have origin in top-left corner,
// positive direction of Y-axis is oriented to down,
// positive direction of X-axis is oriented to right
Func <BoundingBox, int> selectorX = (bbox => (bbox.XMin + bbox.XMax));
Func <BoundingBox, int> selectorY = (bbox => (bbox.YMin + bbox.YMax));
// Here are coordinates of vector of text direction
var dirX = selectorX(secondBBox) - selectorX(firstBBox);
var dirY = selectorY(secondBBox) - selectorY(firstBBox);
dirY = -dirY; // To invert Y-axis direction
double angle = (dirX >= 0) ? 0 : (dirY > 0) ? Math.PI : -Math.PI;
if (dirX < 0)
{
dirX = -dirX;
dirY = -dirY;
}
if (-dirX <= dirY && dirY <= dirX)
{
angle += Math.Atan(dirY / (double)dirX);
}
else if (dirY > 0)
{
angle += Math.PI / 2 - Math.Atan(dirX / (double)dirY);
}
else
{
angle += -Math.PI / 2 + Math.Atan(dirX / (double)dirY);
}
return(angle);
}
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 AddWord(GMWord word)
{
words.Add(word);
}
/// <summary>
/// Alternative approach to angle detection: this method uses line connecting those vertices of word's bounding rectangles which
/// are the closest to base line of text.
/// </summary>
/// <param name="firstWord"></param>
/// <param name="secondWord"></param>
/// <returns>Angle in radians</returns>
private static double DetectAngle3(GMWord firstWord, GMWord secondWord)
{
var firstBBox = firstWord.BoundingBox;
var secondBBox = secondWord.BoundingBox;
Func <BoundingBox, int> selectorX;
Func <BoundingBox, int> selectorY;
// Page coordinates have origin in top-left corner,
// positive direction of Y-axis is oriented to down,
// positive direction of X-axis is oriented to right
if (secondBBox.XMax < firstBBox.XMax)
{
// Page is rotated upside-down (words in line go from right to left)
if (firstBBox.YMin >= secondBBox.YMin)
{
// Text goes from down-right corner to upper-left corner
selectorX = (bbox => bbox.XMin);
selectorY = (bbox => bbox.YMin);
}
else
{
// Text goes from upper-right corner to down-left corner
selectorX = (bbox => bbox.XMax);
selectorY = (bbox => bbox.YMin);
}
}
else
{
// Page is properly rotated (words in linego from left to right)
if (firstBBox.YMax >= secondBBox.YMax)
{
// Text goes from down-left corner to upper-right corner
selectorX = (bbox => bbox.XMin);
selectorY = (bbox => bbox.YMax);
}
else
{
// Text goes from upper-left corner to down-right corner
selectorX = (bbox => bbox.XMax);
selectorY = (bbox => bbox.YMax);
}
}
// Here are coordinates of vector of text direction
var dirX = selectorX(secondBBox) - selectorX(firstBBox);
var dirY = selectorY(secondBBox) - selectorY(firstBBox);
dirY = -dirY;
var dirLength = Math.Sqrt(dirX * dirX + dirY * dirY);
//Console.Out.WriteLine("{0:N4}:{1:N4}", dirX / dirLength, dirY / dirLength);
double angle = 0;
if (dirX < 0)
{
if (dirY > 0)
{
angle = Math.PI;
}
else
{
angle = -Math.PI;
}
dirX = -dirX;
dirY = -dirY;
}
if (-dirX <= dirY && dirY <= dirX)
{
angle += Math.Atan(dirY / (double)dirX);
}
else if (dirY > 0)
{
angle += Math.PI / 2 - Math.Atan(dirX / (double)dirY);
}
else
{
angle += -Math.PI / 2 + Math.Atan(dirX / (double)dirY);
}
return(angle);
}
public void AddStandaloneWord(GMWord word)
{
words.Add(word);
}