Error in single page exits the ocrmypdf command - add option to continue with next page?

[a1ch3mist]

I'm processing a 60-page PDF. One of the pages keeps getting an error with tesseract (supposedly deskew, when the image is mostly blank with a few dots/marks).

My ocrpdf command exits, when I'd rather it just continue to the next page and leave no OCR-text on the bad page. I don't see an option for doing that though.

Is there a way already to proceed when a single-page gives an error (error comes via tesseract it seems from the verbose logs)? If not, please add this as an option.

[a1ch3mist]

WARNING ocrmypdf._exec.tesseract -   17  [tesseract] Warning: Invalid resolution 25 dpi. Using 70 instead.
   INFO ocrmypdf._exec.tesseract -   17  [tesseract] Estimating resolution as 258

  ERROR ocrmypdf._sync - ExitCodeException
Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/ocrmypdf/_exec/tesseract.py", line 220, in get_deskew
    p = run(args_tesseract, stdout=PIPE, stderr=STDOUT, timeout=timeout, check=True)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3/dist-packages/ocrmypdf/subprocess/__init__.py", line 63, in run
    proc = subprocess_run(args, env=env, check=check, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/subprocess.py", line 571, in run
    raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['tesseract', '-l', 'eng', '--psm', '2', '/tmp/ocrmypdf.io.5be2b5fo/000008_rasterize.png', 'stdout']' returned non-zero exit status 1.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/ocrmypdf/_sync.py", line 409, in run_pipeline
    optimize_messages = exec_concurrent(context, executor)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3/dist-packages/ocrmypdf/_sync.py", line 286, in exec_concurrent
    executor(
  File "/usr/lib/python3/dist-packages/ocrmypdf/_concurrent.py", line 86, in __call__
    self._execute(
  File "/usr/lib/python3/dist-packages/ocrmypdf/builtin_plugins/concurrency.py", line 138, in _execute
    result = future.result()
             ^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/concurrent/futures/_base.py", line 449, in result
    return self.__get_result()
           ^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
    raise self._exception
  File "/usr/lib/python3.12/concurrent/futures/thread.py", line 58, in run
    result = self.fn(*self.args, **self.kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3/dist-packages/ocrmypdf/_sync.py", line 200, in exec_page_sync
    ocr_image, preprocess_out = make_intermediate_images(
                                ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3/dist-packages/ocrmypdf/_sync.py", line 133, in make_intermediate_images
    ocr_image = preprocess_out = preprocess(
                                 ^^^^^^^^^^^
  File "/usr/lib/python3/dist-packages/ocrmypdf/_sync.py", line 112, in preprocess
    image = preprocess_deskew(image, page_context)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3/dist-packages/ocrmypdf/_pipeline.py", line 558, in preprocess_deskew
    deskew_angle_degrees = ocr_engine.get_deskew(input_file, page_context.options)
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3/dist-packages/ocrmypdf/builtin_plugins/tesseract_ocr.py", line 239, in get_deskew
    return tesseract.get_deskew(
           ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3/dist-packages/ocrmypdf/_exec/tesseract.py", line 231, in get_deskew
    raise SubprocessOutputError() from e
ocrmypdf.exceptions.SubprocessOutputError
Temporary working files retained at:
/tmp/ocrmypdf.io.5be2b5fo

[jbarlow83]

ocrmypdf normally handles this correctly - it may be that Tesseract changed its behavior in the case of skew detection failing, or the file triggered some new error not previously seen.

In the temporary working files folder
/tmp/ocrmypdf.io.5be2b5fo (may vary on future runs)
please run the same command
['tesseract', '-l', 'eng', '--psm', '2', '/tmp/ocrmypdf.io.5be2b5fo/000008_rasterize.png', 'stdout']

and paste the results and return code here (echo $?).

Alternately, upload the test file.

[a1ch3mist]

[a1ch3mist]

blank image example above. anything that is blank or near-blank seemed to be causing a problem. I just deleted the blanks, but it would be good to figure this out

[jbarlow83]

The issue is not that blank images are present, but that your installation of Tesseract is returning some combination of output and error codes that ocrmypdf has not seen before. I would like to understand where this comes from and why, since it is probably a version specific issue.

It would be ideal if you could provide an actual PDF that reproduces this issue. You could do something like qpdf --empty --pages mypdf.pdf 8 -- out.pdf to extract just page 8 from the input the PDF if the rest of the content is sensitive. It would also be helpful to include the full log output from ocrmypdf -v1 ... instead of the truncated version above.

Also please provide the information requested in the issue template:

• Tesseract version
• ocrmypdf version
• Linux distribution and version or whatever your platform
• how did you install ocrmypdf - PyPI, docker, etc.?

[snailcatcher]

I have the exact same issue inside paperless-ngx.
ocrmypdf fails because tesseract (presumably) fails on deskew.

Here is the empty Page of the PDF which likely causing the issue:
out-empty.pdf

@jbarlow83 Thanks for having a look into it.

PS: If you need some more test pdfs I can provide at least one more. But they are all either white or mostly empty pages.

[snailcatcher]

Ah btw, this is the command I came up with which breaks with the least amount of parameters:

ocrmypdf --deskew out-empty.pdf out.pdf

[snailcatcher]

Just to let you know.
I tested my documents again and they work in both, ocrmypdf and paperless.
Thank you!