public void TestMergeOverlappingRectangles()
{
TableRectangle one = new TableRectangle(new PdfRectangle(0, 0, 10, 10)); //0f, 0f, 10f, 10f);
TableRectangle two = new TableRectangle(new PdfRectangle(5, 5, 15, 15)); //5f, 5f, 10f, 10f);
one.Merge(two);
Assert.Equal(15f, one.Width, 0);
Assert.Equal(15f, one.Height, 0);
Assert.Equal(0f, one.Left, 0);
Assert.Equal(0f, one.Bottom, 0); // one.getTop()
Assert.Equal(15, one.Top, 0);
}
public void TestMergeNoOverlappingRectangles()
{
TableRectangle one = new TableRectangle(new PdfRectangle(0, 0, 10, 10)); //0f, 0f, 10f, 10f);
TableRectangle two = new TableRectangle(new PdfRectangle(10, 0, 20, 10)); //0f, 10f, 10f, 10f);
one.Merge(two);
Assert.Equal(20f, one.Width, 0);
Assert.Equal(10f, one.Height, 0);
Assert.Equal(0f, one.Left, 0);
Assert.Equal(10, one.Top, 0); //0f, one.getTop(), 0);
Assert.Equal(0, one.Bottom, 0); //10f, one.getBottom(), 0);
Assert.Equal(20f * 10f, one.Area, 0);
}
/// <summary>
/// Gets columns positions.
/// </summary>
/// <param name="lines">Must be an array of lines sorted by their +top+ attribute.</param>
/// <returns>a list of column boundaries (x axis).</returns>
public static List <double> ColumnPositions(IReadOnlyList <TableLine> lines)
{
List <TableRectangle> regions = new List <TableRectangle>();
foreach (TextChunk tc in lines[0].TextElements)
{
if (tc.IsSameChar(TableLine.WHITE_SPACE_CHARS))
{
continue;
}
TableRectangle r = new TableRectangle();
r.SetRect(tc);
regions.Add(r);
}
foreach (TableLine l in lines.SubList(1, lines.Count))
{
List <TextChunk> lineTextElements = new List <TextChunk>();
foreach (TextChunk tc in l.TextElements)
{
if (!tc.IsSameChar(TableLine.WHITE_SPACE_CHARS))
{
lineTextElements.Add(tc);
}
}
foreach (TableRectangle cr in regions)
{
List <TextChunk> overlaps = new List <TextChunk>();
foreach (TextChunk te in lineTextElements)
{
if (cr.HorizontallyOverlaps(te))
{
overlaps.Add(te);
}
}
foreach (TextChunk te in overlaps)
{
cr.Merge(te);
}
foreach (var rem in overlaps)
{
lineTextElements.Remove(rem);
}
}
// added by bobld
// We need more checks here
/*
* foreach (TextChunk te in lineTextElements)
* {
* TableRectangle r = new TableRectangle();
* r.setRect(te);
* regions.Add(r);
* }
*/
if (lineTextElements.Count > 0)
{
// because testExtractColumnsCorrectly3() fails
// need to check here if the remaining te in lineTextElements do overlap among themselves
// might happen with multiline cell
TableRectangle r = new TableRectangle();
r.SetRect(lineTextElements[0]);
foreach (var rem in lineTextElements.SubList(1, lineTextElements.Count))
{
if (r.HorizontallyOverlaps(rem))
{
// they overlap!
// so this is multiline cell
r.Merge(rem);
}
else
{
regions.Add(r); // do not overlap (anymore), so add it
r = new TableRectangle();
r.SetRect(rem);
//regions.Add(r);
}
}
regions.Add(r);
}
// end added
}
List <double> rv = new List <double>();
foreach (TableRectangle r in regions)
{
rv.Add(r.Right);
}
rv.Sort(); //Collections.sort(rv);
return(rv);
}
