Feature upgrade apache pdfbox

[buchen]

First draft for #4449

It switches to PDFBox 3.0.3 and falls back - in case of errors - to PDFBox 1.8.

What I am still thinking about: How to learn if and where there are differences between the PDFBox versions.

Right now, I am printing a log message. But users will most likely not notice it let alone inform us.

I wonder if we should create both versions of the extracted text, and, if there are differences, put both files into the debug text. At least we see documents that are different.

[ZfT2]

Maybe just provide a separate "beta" Version of PP with (only) pdfbox 3, post it in the forums and let the community test it?
If after a time no (bigger) failures are reported, upgrade it also in the main release...

Advantage: we will (hopefully) only deal with one pdfbox Version in future.

[buchen]

Advantage: we will (hopefully) only deal with one pdfbox Version in future.

Yes. That is the goal.

I have now made two more changes to help us learn about differences.

The extracted debug text indicates if the version is different. At least we will learn if there are differences.

And I print the diff to the log file. In case we want to better understand what the differences are. The changes so far do not make a material difference (better extraction of special characters, additional line breaks in address, etc.)

[stoeggich]

Can you perhaps make it so that you can check many files at once? Then I would run everything through myself

[buchen]

I added a menu option to create diffs for multiple files.
It is only visible if you activated the "experimental features" in the settings.

What it does not support: anonymizing the data via mouse click. You might have to do that manually.

Please post meaningful deltas in this issue #4449

So far, I think the diffs are manageable. They do not look like material differences that have impact on the relevant regular expressions.

[Nirus2000]

I added a menu option to create diffs for multiple files. It is only visible if you activated the "experimental features" in the settings.

What it does not support: anonymizing the data via mouse click. You might have to do that manually.

Please post meaningful deltas in this issue #4449

So far, I think the diffs are manageable. They do not look like material differences that have impact on the relevant regular expressions.

Perhaps you should also add “Experimental”, as in the XML document menu item.
BTW. why is “XML documents” grayed out, although the “marked as experimental” is activated?

[buchen]

BTW. why is “XML documents” grayed out, although the “marked as experimental” is activated?

Because it is a "headline" - all supported XML documents should come afterwards. There is only 1 at the moment. Menus do not support a headline. They only support deactivated items.

the item is only visible when the experimental stuff is activated. I hope that is enough labelling. We'll remove it in a couple weeks anyway. (or maybe in years... who knows ;-))