UnicodeDecodeError 'utf-8' codec can't decode byte 0xac in position 0: invalid start byte

[nalin-udhaar]

While trying to merge 2 PDF's, I get the following error.

Environment

Which environment were you using when you encountered the problem?

$ python -m platform
Linux-5.4.0-144-generic-x86_64-with-glibc2.29

$ python -c "import pypdf;print(pypdf.__version__)"
3.7.0

Code + PDF

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

def merge_pdf(source_pdf, pdf_to_merge: list):
    """
    :param source_pdf:
    :param pdf_to_merge: list of pdf to append to source_pdf
    :return:
    """
    merger = PdfMerger(strict=False)
    items = [source_pdf]
    items.extend(pdf_to_merge)
    pdf_merged_buffer = io.BytesIO()
    for _pdf_file in items:
        # Append PDF files
        pdf_buffer = PdfReader(io.BytesIO(_pdf_file), strict=False)
        merger.append(pdf_buffer, import_outline=False)
    merger.write(pdf_merged_buffer)
    merger.close()

I can share the pdf over a private channel.

Traceback

This is the Traceback I see, only adding the part with pypdf:
  File "/home/nalin/workspace/project/utils/pdf_utils.py", line 185, in merge_pdf
    merger.write(pdf_merged_buffer)
  File "/home/nalin/.virtualenv/project/lib/python3.8/site-packages/pypdf/_merger.py", line 333, in write
    self.output.add_page(page.pagedata)
  File "/home/nalin/.virtualenv/project/lib/python3.8/site-packages/pypdf/_writer.py", line 359, in add_page
    return self._add_page(page, list.append, excluded_keys)
  File "/home/nalin/.virtualenv/project/lib/python3.8/site-packages/pypdf/_writer.py", line 311, in _add_page
    other = page_org.pdf.pdf_header
  File "/home/nalin/.virtualenv/project/lib/python3.8/site-packages/pypdf/_reader.py", line 363, in pdf_header
    pdf_file_version = self.stream.read(8).decode("utf-8")
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xac in position 0: invalid start byte

PDF can be opened with a PDF viewer.

[pubpub-zz]

You can send the file to @MartinThoma info@martin-thoma.de

[MartinThoma]

I think the fix might be rather easy .. .at least for the first issue: #1759

[pubpub-zz]

@nalin-udhaar, we are still waiting for your file

[nalin-udhaar]

I have shared the PDF over the the mentioned email.

[pubpub-zz]

I've reviewed your file and this file is damaged (you can check that with a binary editor) I've proposed an alternative robustness improvement

[nalin-udhaar]

do you want me to share that file again?

[pubpub-zz]

@nalin-udhaar
Your mail was good but if you look to the file with a binary editor you will not be able to find the standard "%PDF-" which is expected.
I agree that at least acrobat reader cope with this situation (even if some elements may not be displayed), that's why I've proposed a fix for pypdf to be more tolerant too.

[MartinThoma]

I just merged #1768 to main. It will be part of pypdf > 3.7.0 which will be released to PyPI on Sunday (9th of April 2023)

[MartinThoma]

@nalin-udhaar Thank you for letting us know! If you want, I put you on the list of contributors: https://pypdf.readthedocs.io/en/latest/meta/CONTRIBUTORS.html

[nalin-udhaar]

yeah, that will be great.