If one is not restricted to Python only, then there are more options.
On the Java side, one could look into using Apache PDFBox or iText. I don't suppose anyone wrote wrapper RF libraries for them yet, but one could easily do that, following the RF documentation & the example docs for the Java PDF libraries. And run these over Jython, switching to Java remote server if Jython route is a no go.
I've used Apache PDFBox to extract out text from a PDF for test verification. But we don't use RF at my organization so I just had it as a regular Java wrapper class library that's not built for RF in mind.
FYI, there's a list of good PDF tools that one can evaluate for this job as well for integration with RF. I haven't checked which are Python and not.