cannot merge PDF with nullobject articles

[Noah-Houghton]

When merging an older PDF with one we generate, we found that add_filtered_articles within merge_pdfs was failing with a NullObject has no attribute 'get' error. I found during local testing that modifying the logic to check explicitly for NullObject (and to continue if found) was enough to get things working again, but I don't know enough about the PDF spec to know if that's a valid state that should be supported. The PDFs do load normally in e.g. Adobe Acrobat or Firefox PDF reader.

I'm waiting for permission to share the erroring PDFs; happy to provide anything that could help diagnose in the meantime.

Here's the snippet I replaced in PdfWriter.add_filtered_articles:

    obj = a.get_object() # capture output of get_object
    if isinstance(obj, NullObject):  # check if it's null
        continue
    thr = obj.get("/T") # continue with existing logic

Environment

Which environment were you using when you encountered the problem?

$ python -m platform
Linux-6.6.87.2-microsoft-standard-WSL2-x86_64-with-glibc2.35

$ python -c "import pypdf;print(pypdf._debug_versions)"
pypdf==6.1.1, crypt_provider=('cryptography', '45.0.7'), PIL=11.3.0

Code + PDF

This is a minimal, complete example that shows the issue:

def append_pdfs(*pdfs):
    merger = PdfWriter()
    temp = BytesIO()
    for pdf in pdfs:
        merger.append(pdf)
    merger.write(temp)
    temp.seek(0)
    return temp

Share here the PDF file(s) that cause the issue. The smaller they are, the
better. Let us know if we may add them to our tests!

Traceback

This is the complete traceback I see:

  File "<project path>/utils/file_utils.py", line 102, in append_pdfs
    merger.append(pdf)
  File "<project path>/lib/python3.11/site-packages/pypdf/_writer.py", line 2692, in append
    self.merge(
  File "<project path>/lib/python3.11/site-packages/pypdf/_writer.py", line 2889, in merge
    self.add_filtered_articles("", srcpages, reader)
  File "<project path>/lib/python3.11/site-packages/pypdf/_writer.py", line 2980, in add_filtered_articles
    thr = a.get_object().get("/T")
AttributeError: 'NullObject' object has no attribute 'get'

[stefan6419846]

Thanks for the report. This looks like the correct fix, but we generally require a corresponding file as well. Once this is available, feel free to submit a corresponding test and PR.

[Noah-Houghton]

@stefan6419846 I have the document and I'm ready to make a PR. To clarify - do I need to fork the repository to submit a PR? I tried cloning the repo and making a branch but I don't have permissions to push anything.

[stefan6419846]

Yes, the correct workflow usually is:

• Fork the repository.
• Checkout a new branch on your fork.
• Push the changes to your fork.
• Go to the upstream repository (https://github.com/py-pdf/pypdf). You should see some notice at the top suggesting you to open a pull request. Click the button and open a pull request against main (should be default).

[Noah-Houghton]

@stefan6419846 Thanks for the instructions! I've made a PR here: #3509

[Noah-Houghton]

@stefan6419846 I'm noticing that the test without a cryptography library is failing when trying to load the PDF -- for some reason it's trying to decrypt the PDF even though I know it hasn't been encrypted (no password needed to view). I haven't been able to replicate this locally. Can you please take a look and let me know if I'm missing something obvious?

[stefan6419846]

PDF files can be encrypted without requiring a user password. There should be existing tests which skip such cases when a cryptography library is not available.

[Noah-Houghton]

@stefan6419846 Thanks for the insight - I didn't know that! I've updated the PR and it's now passing tests.