Remove Monkey Patch

[brianglass]

I have removed the monkey patches for _xpath_in_bbox and xpath_overlaps_bbox. It seems that they were not actually being used anywhere and were causing conflicts with lxml >= 3.

If I'm missing something here please let me know.

[jcushman]

Hmm hmm, thanks for passing this on. Yeah, those actually are used -- they're automagically turned into custom selectors for doing spatial searches:

https://github.com/jcushman/pdfquery/#custom-selectors

But it looks like lxml 3 switched to using the cssselect library under the hood, and now custom selectors are implemented in a different way (and hopefully a way that involves less monkeying around). Pyquery adds its custom selectors in JQueryTranslator here:

https://github.com/gawel/pyquery/blob/master/pyquery/cssselectpatch.py

So in_bbox and overlaps_bbox should be re-implemented the way those custom selectors are -- and then we have to get PyQuery using them. Looking at the pyquery source, I think the plan there is to subclass JQueryTranslator to PDFQueryTranslator with our custom selectors, and then instantiate our pyquery as PyQuery(css_translator=PDFQueryTranslator(xhtml=False))

I don't suppose there's any chance you feel like trying to do that? :) If not I'll put it on my list ... in the meantime I wouldn't mind pulling this if you could put it in a try block instead of deleting it, so people who still have the right version of lxml can still use it.

Thanks!
Jack

[brianglass]

Go ahead and put it on your list. I'll see if I can find some time to look at it, but don't hold high hopes for anything in the next couple weeks. ;) I'll push out a revised PR this week.

[brianglass]

jcushman, I have updated the PR as you requested. I have not had a chance to do it properly, but this is a stop-gap.