index out of bounds in pypdf._text_extraction.handle_tj

[rgwood-rely]

On decoding a pdf in the second line:

if orientation in orientations:
    if isinstance(operands[0], str):

len(operands) == 0 and it raises an ex.

Should change it to:

if orientation in orientations and len(operands) > 0:

Environment

Which environment were you using when you encountered the problem?

$ python -m platform
macOS-14.1.1-x86_64-i386-64bit

$ python -c "import pypdf;print(pypdf._debug_versions)"
pypdf==3.17.1, crypt_provider=('cryptography', '3.3.2'), PIL=10.1.0

Code + PDF

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

# sorry; PDF is confidential

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:

<our software>
    page_text = page_obj.extract_text()
  File "/Users/rgwood/mambaforge/envs/e_16730_w3/lib/python3.9/site-packages/pypdf/_page.py", line 2279, in extract_text
    return self._extract_text(
  File "/Users/rgwood/mambaforge/envs/e_16730_w3/lib/python3.9/site-packages/pypdf/_page.py", line 2115, in _extract_text
    process_operation(b"Tj", operands)
  File "/Users/rgwood/mambaforge/envs/e_16730_w3/lib/python3.9/site-packages/pypdf/_page.py", line 2075, in process_operation
    text, rtl_dir = handle_tj(
  File "/Users/rgwood/mambaforge/envs/e_16730_w3/lib/python3.9/site-packages/pypdf/_text_extraction/__init__.py", line 220, in handle_tj
    if isinstance(operands[0], str):
IndexError: list index out of range

[pubpub-zz]

Without the PDF we can not analyse anything. If you agree email it privately to @MartinThoma (info@martin-thoma.de)
We will not disclose it

[rgwood-rely]

Is there a way to cut specific pages from a pdf? I tried:

qpdf input.pdf --pages . 1-7 -- output.pdf

But the error was not present in the resultant pdf.

[pubpub-zz]

try:

import pypdf
w = pypdf.PdfWriter()
w.append("pdf_to_read.pdf",[0,2,5])  # the list of pages to extract
w.write("output.pdf")

[rgwood-rely]

That worked!

[rgwood-rely]

I sent a redacted pdf page to the above email instead of attaching here out of an abundance of caution.

[rgwood-rely]

Link to code position:

pypdf/pypdf/_text_extraction/__init__.py

Line 220 in 38795f5

if isinstance(operands[0], str):

[stefan6419846]

Have you considered submitting a corresponding PR for this (the offending line has already been part of your initial traceback)? I cannot debug this without a PDF file, but it looks like we can have an early return due to an empty operands list here.

[rgwood-rely]

Hi Stefan. I've sent a PDF to info@martin-thoma.de as requested and yes this should be an easy fix. Will try and create a PR if I find a moment.

[rgwood-rely]

PR: rgwood-rely:rgwood/2320_fix_index_out_of_bounds_in_handle_tj