We produce PDFs using Apache FOP 0.20.5. When we create tags for accessibility in these PDFs the text in the tags has all its spaces removed, and the screen reader doesn't read it properly. See attachment. It shows a tag from a PDF that didn't originate from FOP, and one that did originate from FOP.
BTW we almost always do our tagging "manually". The autotag tool makes such a mess, it takes more time to fix it up than to tag manually. But we have noticed that the autotag tool does something to the PDF that cures the space-stripping problem.