Currently, we trust in the exit signal form OCRmyPDF. We should add some form of test to see if the OCR file actually now contains text.