iTextSharp parse table - pdf

Question

Using iTextSharp v5.5.13
I have a huge amount of PDF files I need to parse. About 5% of them have a table with data I also need.
The table looks like this:
Most of the time the line I need is parsed as
2 januari 15 januari € 49,49 € 21,57 € 15,09 € 34,39
I can work with that. I split by space and it works.
But sometimes the month name has an extra space: janu ari
I know I can override the strategies to get rid of these extra spaces. I'm already using it with the rest of the pdf (ITextExtractionStrategy), but for this table, I'm using a rectangle strategy:
var rect = new System.util.RectangleJ(70, 425, 460, 200);
RenderFilter[] filter = { new RegionTextRenderFilter(rect) };
ITextExtractionStrategy strategy =
new FilteredTextRenderListener(new MyLocationTextExtractionStrategy(), filter);
var lines = PdfTextExtractor.GetTextFromPage(reader, pageNumber, strategy).Split('\n');
My override looks like this:
public class MyLocationTextExtractionStrategy : LocationTextExtractionStrategy
{
protected override bool IsChunkAtWordBoundary(TextChunk chunk, TextChunk previousChunk)
{
var dist = chunk.DistanceFromEndOf(previousChunk);
return dist < -chunk.CharSpaceWidth || dist > chunk.CharSpaceWidth / 2.0f;
}
}
I found this Googling. But it doesn't solve my problem.
In the case of janu ari dist is larger than -chunk.CharSpaceWidth and I'm not sure what to do next.
Please let me know when I should not use a rectangle strategy for this table but a different approach.

Answer 1

If your data in this type of table is always going to be in the same format, then you could take a different approach: just accept whatever data your MyLocationTextExtractionStrategy is throwing at you, and then massage that data into a format that you can use.
In this case, you data is always:
2 groups of:
1 or 2 digits (day of the month)
some characters (name of the month)
4 groups of:
Euro symbol
some digits (at least one)
comma
2 digits
In 2 januari 15 januari € 49,49 € 21,57 € 15,09 € 34,39 the spaces are separation characters, but with such well structured data, you don't even need spaces. So just drop them, and then your data becomes 2januari15januari€49,49€21,57€15,09€34,39.
Now you can use a regular expression with some capture groups, to massage your data into something palatable.
2 groups of:
[0-9]{1,2}
[a-z]*
4 groups of:
€
[0-9]{1,}
,
[0-9]{2}
As you wrote yourself in the comments, one possible resulting regular expression could be:
new Regex(#"([0-9]{1,2})([a-z]*)([0-9]{1,2})([a-z]*)(€[0-9]{1,},[0-9]{2})(€[0-9]{1,},[0-9]{2})(€[0-9]{1,},[0-9]{2})(€[0-9]{1,},[0-9]{2})")