public void GetBlocks(string name, string[] expected)
{
if (name == "90 180 270 rotated.pdf")
{
// The 'TimesNewRomanPSMT' font is used by this particular document. Thus, results cannot be trusted on
// platforms where this font isn't generally available (e.g. OSX, Linux, etc.), so we skip it!
var font = SystemFontFinder.Instance.GetTrueTypeFont("TimesNewRomanPSMT");
Skip.If(font == null, "Skipped because the font TimesNewRomanPSMT could not be found in the execution environment.");
}
var options = new DocstrumBoundingBoxes.DocstrumBoundingBoxesOptions()
{
LineSeparator = " "
};
using (var document = PdfDocument.Open(DlaHelper.GetDocumentPath(name)))
{
var page = document.GetPage(1);
var words = NearestNeighbourWordExtractor.Instance.GetWords(page.Letters);
var blocks = DocstrumBoundingBoxes.Instance.GetBlocks(words, options);
Assert.Equal(expected.Length, blocks.Count);
var orderedBlocks = blocks.OrderBy(b => b.BoundingBox.BottomLeft.X)
.ThenByDescending(b => b.BoundingBox.BottomLeft.Y).ToList();
for (int i = 0; i < orderedBlocks.Count; i++)
{
Assert.Equal(expected[i], orderedBlocks[i].Text);
}
}
}
public void GetBlocks(string name, string[] expected)
{
var options = new DocstrumBoundingBoxes.DocstrumBoundingBoxesOptions()
{
LineSeparator = " "
};
using (var document = PdfDocument.Open(DlaHelper.GetDocumentPath(name)))
{
var page = document.GetPage(1);
var words = NearestNeighbourWordExtractor.Instance.GetWords(page.Letters);
var blocks = DocstrumBoundingBoxes.Instance.GetBlocks(words, options);
Assert.Equal(expected.Length, blocks.Count);
var orderedBlocks = blocks.OrderBy(b => b.BoundingBox.BottomLeft.X)
.ThenByDescending(b => b.BoundingBox.BottomLeft.Y).ToList();
for (int i = 0; i < orderedBlocks.Count; i++)
{
Assert.Equal(expected[i], orderedBlocks[i].Text);
}
}
}
public void GetBlocksStatic(string name, string[] expected)
{
if (name == "90 180 270 rotated.pdf")
{
// The 'TimesNewRomanPSMT' font is used by this particular document. Thus, results cannot be trusted on
// platforms where this font isn't generally available (e.g. OSX, Linux, etc.), so we skip it!
var font = SystemFontFinder.Instance.GetTrueTypeFont("TimesNewRomanPSMT");
Skip.If(font == null, "Skipped because the font TimesNewRomanPSMT could not be found in the execution environment.");
}
var options = new DocstrumBoundingBoxes.DocstrumBoundingBoxesOptions()
{
LineSeparator = " "
};
using (var document = PdfDocument.Open(DlaHelper.GetDocumentPath(name)))
{
var page = document.GetPage(1);
var words = NearestNeighbourWordExtractor.Instance.GetWords(page.Letters).ToList();
// Docstrum using static methods
// Filter out white spaces
words = words.Where(w => !string.IsNullOrWhiteSpace(w.Text)).ToList();
var wlBounds = options.WithinLineBounds;
var wlBinSize = options.WithinLineBinSize;
var wlMultiplier = options.WithinLineMultiplier;
var blBounds = options.BetweenLineBounds;
var blBinSize = options.BetweenLineBinSize;
var blMultiplier = options.BetweenLineMultiplier;
var maxDegreeOfParallelism = options.MaxDegreeOfParallelism;
var angularDifferenceBounds = options.AngularDifferenceBounds;
var wordSeparator = options.WordSeparator;
var lineSeparator = options.LineSeparator;
var epsilon = options.Epsilon;
// 1. Estimate within line and between line spacing
if (!DocstrumBoundingBoxes.GetSpacingEstimation(words, wlBounds, wlBinSize, blBounds, blBinSize,
maxDegreeOfParallelism,
out double withinLineDistance, out double betweenLineDistance))
{
if (double.IsNaN(withinLineDistance))
{
withinLineDistance = 0;
}
if (double.IsNaN(betweenLineDistance))
{
betweenLineDistance = 0;
}
}
// 2. Determination of Text Lines
double maxWithinLineDistance = wlMultiplier * withinLineDistance;
var lines = DocstrumBoundingBoxes.GetLines(words, maxWithinLineDistance, wlBounds, wordSeparator, maxDegreeOfParallelism).ToArray();
// 3. Structural Block Determination
double maxBetweenLineDistance = blMultiplier * betweenLineDistance;
var blocks = DocstrumBoundingBoxes.GetStructuralBlocks(lines, maxBetweenLineDistance, angularDifferenceBounds, epsilon, lineSeparator, maxDegreeOfParallelism).ToList();
Assert.Equal(expected.Length, blocks.Count);
var orderedBlocks = blocks.OrderBy(b => b.BoundingBox.BottomLeft.X)
.ThenByDescending(b => b.BoundingBox.BottomLeft.Y).ToList();
for (int i = 0; i < orderedBlocks.Count; i++)
{
Assert.Equal(expected[i], orderedBlocks[i].Text);
}
}
}
public static void Run(string filePath)
{
var sb = new StringBuilder();
using (var document = PdfDocument.Open(filePath))
{
foreach (var page in document.GetPages())
{
// 0. Preprocessing
var letters = page.Letters; // no preprocessing
// 1. Extract words
var wordExtractor = NearestNeighbourWordExtractor.Instance;
var wordExtractorOptions = new NearestNeighbourWordExtractor.NearestNeighbourWordExtractorOptions()
{
Filter = (pivot, candidate) =>
{
// check if white space (default implementation of 'Filter')
if (string.IsNullOrWhiteSpace(candidate.Value))
{
// pivot and candidate letters cannot belong to the same word
// if candidate letter is null or white space.
// ('FilterPivot' already checks if the pivot is null or white space by default)
return(false);
}
// check for height difference
var maxHeight = Math.Max(pivot.PointSize, candidate.PointSize);
var minHeight = Math.Min(pivot.PointSize, candidate.PointSize);
if (minHeight != 0 && maxHeight / minHeight > 2.0)
{
// pivot and candidate letters cannot belong to the same word
// if one letter is more than twice the size of the other.
return(false);
}
// check for colour difference
var pivotRgb = pivot.Color.ToRGBValues();
var candidateRgb = candidate.Color.ToRGBValues();
if (!pivotRgb.Equals(candidateRgb))
{
// pivot and candidate letters cannot belong to the same word
// if they don't have the same colour.
return(false);
}
return(true);
}
};
var words = wordExtractor.GetWords(letters, wordExtractorOptions);
// 2. Segment page
var pageSegmenter = DocstrumBoundingBoxes.Instance;
var pageSegmenterOptions = new DocstrumBoundingBoxes.DocstrumBoundingBoxesOptions()
{
};
var textBlocks = pageSegmenter.GetBlocks(words, pageSegmenterOptions);
// 3. Postprocessing
var readingOrder = UnsupervisedReadingOrderDetector.Instance;
var orderedTextBlocks = readingOrder.Get(textBlocks);
// 4. Extract text
foreach (var block in orderedTextBlocks)
{
sb.Append(block.Text.Normalize(NormalizationForm.FormKC)); // normalise text
sb.AppendLine();
}
sb.AppendLine();
}
}
Console.WriteLine(sb.ToString());
}
