New GetText() option: NegativeGapAsWhitespace

[Kizaemon]

When parsing PDF files with tables containing multiple lines in a cell or "merged" cells, the separate words can appear out of horizontal order. This option can better predict the spaces between the words.

[Kizaemon]

Solves [#951]

[Kizaemon]

This is a sample screenshot from the PDF, where the next word starts to the left of the previous word, making the negative gaps.

[BobLd]

@Kizaemon thanks for sharing the information here.

Table extraction is generally tricky to do in pdf documents.

Can I ask you to look into tabula sharp which is a package based on PdfPig but specialised in table extraction?

[Kizaemon]

@BobLd Thank you for recommending tabula sharp, I will surely take a look at it.

As for this pull request, I still feel it could be useful to have an option to detect the negative gaps as whitespaces.
The current implementation does not do anything with negative letter distances.

[BobLd]

@Kizaemon I tend to agree here, reviewing your code I started to wonder if we should not simply always use the absolute value.

I'll give it a proper look later today

[Kizaemon]

There is a possibility that this change (unconditional absolute value) can break the existing parsing patterns for the current users.
Therefore I have suggested to add an option for backwards compatibility.

You will know better the typical usage patterns by the existing users and may just absolute in all cases.