Data extraction from PDFs is not necessarily a trivial solution; however, it's a good time to have this problem in that there are quite a few options based upon what "type" of user you are. Probably the best overall software package (which I've used before to great success) is Tabula:
http://tabula.technology/
... it's open source, and I'm a full stack guy; so it's great from a dev's perspective (as well as being easily in the wheelhouse for a non-dev.) Tabula allows you to extract that data into a CSV or Microsoft Excel spreadsheet using a simple, easy-to-use interface. From there, it's trivial to export it to XML with either Microsoft Office or Libre Office... (as well as a myriad of online tools.) Although, since you mentioned it's accounting software; all accounting software I've ever worked with will allow you to import directly from a CSV.
However, sometimes people want a solution with "support", i.e. a commercial application... Here my recommendation is :
https://pdftables.com/
Finally, if you want more "control" over the flow; you can always look at creating your own "solution" by combining a couple of different Open Source items:
http://www.unixuser.org/~euske/python/pdfminer/
https://poppler.freedesktop.org/
http://www.pdfparser.org/
http://www.foolabs.com/xpdf/portsntools.html