Add script to OCR text in PDFs #7
Conversation
refs #8639.
|
This needs some more discussion/work; see the original ticket for details. Specifically, the problem is that the Imagemagick step responsible for converting the PNG to PDFs is producing unreadable, terrible-quality flat images. |
|
Updated to fix the scaling issue. This can get squashed down to one commit if the client approves. |
|
The tesseract line doesn't seem to love being called with subprocess.check_call from within Archivematica in my experience: 3d8b463#diff-1090aa5b96aa7be6b941205f17dabdccR16 Had it working, and then a Python update on an old 12.04 server (?) made it start throwing a kwargs issue (I'll reproduce the exact one from my other machine's search history later today), as though there was something wrong with the cmd args being reconstituted from the passed list. Jury-rigging it into a cmd = "command " + var + " command " + var worked fine, but I still haven't been able to figure out why. |
refs #8639.