KeyError: '/Contents'

[feinerer]

$ docker run --rm -v "$(pwd):/home/docker"   ocrmypdf --skip-text input.pdf output.pdf
   INFO - Tesseract v4.x.alpha found. OCRmyPDF support is experimental.
  ERROR - Traceback (most recent call last):
  File "/appenv/lib/python3.5/site-packages/ruffus/task.py", line 751, in run_pooled_job_without_exceptions
    register_cleanup, touch_files_only)
  File "/appenv/lib/python3.5/site-packages/ruffus/task.py", line 567, in job_wrapper_io_files
    ret_val = user_defined_work_func(*params)
  File "/appenv/lib/python3.5/site-packages/ocrmypdf/pipeline.py", line 183, in repair_pdf
    pdfinfo = pdf_get_all_pageinfo(output_file)
  File "/appenv/lib/python3.5/site-packages/ocrmypdf/pageinfo.py", line 524, in pdf_get_all_pageinfo
    return [_pdf_get_pageinfo(infile, n) for n in range(pdf.numPages)]
  File "/appenv/lib/python3.5/site-packages/ocrmypdf/pageinfo.py", line 524, in <listcomp>
    return [_pdf_get_pageinfo(infile, n) for n in range(pdf.numPages)]
  File "/appenv/lib/python3.5/site-packages/ocrmypdf/pageinfo.py", line 496, in _pdf_get_pageinfo
    pageinfo['has_text'] = _page_has_text(pdf, page)
  File "/appenv/lib/python3.5/site-packages/ocrmypdf/pageinfo.py", line 468, in _page_has_text
    text = page.extractText()
  File "/appenv/lib/python3.5/site-packages/PyPDF2/pdf.py", line 2593, in extractText
    content = self["/Contents"].getObject()
  File "/appenv/lib/python3.5/site-packages/PyPDF2/generic.py", line 516, in __getitem__
    return dict.__getitem__(self, key).getObject()
KeyError: '/Contents'

This happens on a Debian Jessie system running the latest Docker container (see above command line).

Unfortunately I cannot include the corresponding PDF as it contains private information.

If you need further information, please give me instructions in order to help you debug this issue. Thank you!

[jbarlow83]

I'll add a check for this case in the next release.

The PDF is missing a data field that is strictly optional, but almost never omit, and the third party PyPDF2 library does not handle this.

Try re-frying the PDF with Ghostscript as this would likely insert the expected object. Note this constructs a visually identical PDF and will reencode JPEGs in the process.

gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -sOutputFile=out.pdf in.pdf

[feinerer]

Confirmed: using Ghostscript to rewrite the PDF suffices so that PyPDF2 can handle it.

A direct check in OCRmyPDF is appreciated to avoid the manual Ghostscript call.