internal static TableLine RemoveRepeatedCharacters(TableLine line, char c, int minRunLength)
{
TableLine rv = new TableLine();
foreach (TextChunk t in line.TextElements)
{
foreach (TextChunk r in t.Squeeze(c, minRunLength))
{
rv.AddTextChunk(r);
}
}
return(rv);
}
public static List <TableLine> GroupByLines(List <TextChunk> textChunks)
{
Utils.Sort(textChunks); // added by bobLd: force re-sorting
List <TableLine> lines = new List <TableLine>();
if (textChunks.Count == 0)
{
return(lines);
}
double bbwidth = BoundingBoxOf(textChunks).Width;
TableLine l = new TableLine();
l.AddTextChunk(textChunks[0]);
textChunks.RemoveAt(0);
lines.Add(l);
TableLine last = lines[lines.Count - 1];
foreach (TextChunk te in textChunks)
{
if (last.VerticalOverlapRatio(te) < 0.1)
{
if (last.Width / bbwidth > 0.9 && AllSameChar(last.TextElements))
{
lines.RemoveAt(lines.Count - 1);
}
lines.Add(new TableLine());
last = lines[lines.Count - 1];
}
last.AddTextChunk(te);
}
if (last.Width / bbwidth > 0.9 && AllSameChar(last.TextElements))
{
lines.RemoveAt(lines.Count - 1);
}
List <TableLine> rv = new List <TableLine>(lines.Count);
foreach (TableLine line in lines)
{
rv.Add(TableLine.RemoveRepeatedCharacters(line, ' ', 3));
}
return(rv);
}