hocr-clean

[zuphilip]

Go through all ocr-elements and delete empty elements and possibly also elements with spaces only. Either do this recursive or start with the top elements and look at the textContent.

[amitdo]

This should be fixed in tesseract.

[stweil]

Yes, we think so, too. That would fix new hOCR files. Nevertheless hocr-clean would be useful for existing hOCR files.