// the function LowerBound() implements a binary search,
// which in terms of operator < returns the first position
// that satisfies the following condition:
// ! ( points_ordered[pos] < point_in ) == true ;
//
// the computational complexity is O(log N),
// where N is the number of points in a dataset;
static protected int LowerBound
(
List <Point> points_ordered,
PointComparer pnt_comparer_in,
Point point_in
)
{
int i_low = 0;
int i_high = points_ordered.Count;
int i_mid = 0;
while (i_low < i_high)
{
i_mid = (i_low + i_high) / 2;
if (pnt_comparer_in.Compare(points_ordered[i_mid], point_in) < 0)
{
i_low = i_mid + 1;
}
else
{
i_high = i_mid;
}
}
return(i_low);
}
/// <summary>
/// Extract the <see cref="PageArea"/>, with its text elements (letters) and rulings (processed PdfPath and PdfSubpath).
/// </summary>
/// <param name="pageNumber">The page number to extract.</param>
public PageArea ExtractPage(int pageNumber)
{
if (pageNumber > this.pdfDocument.NumberOfPages || pageNumber < 1)
{
throw new IndexOutOfRangeException("Page number does not exist");
}
Page p = this.pdfDocument.GetPage(pageNumber);
//ObjectExtractorStreamEngine se = new ObjectExtractorStreamEngine(p);
//se.processPage(p);
/**************** ObjectExtractorStreamEngine(PDPage page)*******************/
var rulings = new List <Ruling>();
foreach (var image in p.GetImages())
{
if (image.TryGetPng(out var png))
{
}
}
foreach (var path in p.ExperimentalAccess.Paths)
{
if (!path.IsFilled && !path.IsStroked)
{
continue; // strokeOrFillPath operator => filter stroke and filled
}
foreach (var subpath in path)
{
if (!(subpath.Commands[0] is Move first))
{
// skip paths whose first operation is not a MOVETO
continue;
}
if (subpath.Commands.Any(c => c is BezierCurve))
{
// or contains operations other than LINETO, MOVETO or CLOSE
// bobld: skip at subpath or path level?
continue;
}
// TODO: how to implement color filter?
PdfPoint? start_pos = RoundPdfPoint(first.Location, rounding);
PdfPoint? last_move = start_pos;
PdfPoint? end_pos = null;
PdfLine line;
PointComparer pc = new PointComparer();
foreach (var command in subpath.Commands)
{
if (command is Line linePath)
{
end_pos = RoundPdfPoint(linePath.To, rounding);
if (!start_pos.HasValue || !end_pos.HasValue)
{
break;
}
line = pc.Compare(start_pos.Value, end_pos.Value) == -1 ? new PdfLine(start_pos.Value, end_pos.Value) : new PdfLine(end_pos.Value, start_pos.Value);
// already clipped
Ruling r = new Ruling(line.Point1, line.Point2);
if (r.Length > 0.01)
{
rulings.Add(r);
}
}
else if (command is Move move)
{
start_pos = RoundPdfPoint(move.Location, rounding);
end_pos = start_pos;
}
else if (command is Close)
{
// according to PathIterator docs:
// "the preceding subpath should be closed by appending a line
// segment
// back to the point corresponding to the most recent
// SEG_MOVETO."
if (!start_pos.HasValue || !end_pos.HasValue)
{
break;
}
line = pc.Compare(end_pos.Value, last_move.Value) == -1 ? new PdfLine(end_pos.Value, last_move.Value) : new PdfLine(last_move.Value, end_pos.Value);
// already clipped
Ruling r = new Ruling(line.Point1, line.Point2); //.intersect(this.currentClippingPath());
if (r.Length > 0.01)
{
rulings.Add(r);
}
}
start_pos = end_pos;
}
}
}
/****************************************************************************/
TextStripper pdfTextStripper = new TextStripper(this.pdfDocument, pageNumber);
pdfTextStripper.Process();
Utils.Sort(pdfTextStripper.textElements, new TableRectangle.ILL_DEFINED_ORDER());
return(new PageArea(p.CropBox.Bounds,
p.Rotation.Value,
pageNumber,
p,
this.pdfDocument,
pdfTextStripper.textElements,
rulings,
pdfTextStripper.minCharWidth,
pdfTextStripper.minCharHeight,
pdfTextStripper.spatialIndex));
}