ROB: guard empty link destination arrays during page transfer

[metsw24-max]

Empty link destination arrays crash page transfer

A /Link annotation whose /Dest (or /GoTo action /D) is an empty array makes page transfer read the target page from d[0] with no length check, so append()/merge() raise IndexError in _insert_filtered_annotations, and add_page() followed by write() raises it from DirectReferenceLink.find_referenced_page via _resolve_links; a non-array /Dest trips TypeError likewise. Both sites now skip a destination that is not a non-empty array, matching the existing drop-when-unresolvable handling and the None path already present in _resolve_links.

Repro:

from io import BytesIO
from pypdf import PdfReader, PdfWriter
from pypdf.generic import ArrayObject, DictionaryObject, NameObject

src = PdfWriter()
src.add_blank_page(width=72, height=72)
link = DictionaryObject({
    NameObject("/Subtype"): NameObject("/Link"),
    NameObject("/Dest"): ArrayObject([]),  # empty destination
})
src.pages[0][NameObject("/Annots")] = ArrayObject([src._add_object(link)])
buf = BytesIO(); src.write(buf); buf.seek(0)

writer = PdfWriter()
writer.append(PdfReader(buf))   # IndexError before this change
writer.write(BytesIO())

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.86%. Comparing base (5286c38) to head (5ffea49).
⚠️ Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3896      +/-   ##
==========================================
+ Coverage   97.83%   97.86%   +0.03%     
==========================================
  Files          55       57       +2     
  Lines       10609    10742     +133     
  Branches     1985     2009      +24     
==========================================
+ Hits        10379    10513     +134     
  Misses        127      127              
+ Partials      103      102       -1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[stefan6419846]

If we check for an ArrayObject here, could we avoid the explicit casting?

[metsw24-max]

Good point. Switched both branches to check isinstance(d, ArrayObject) directly, so the explicit cast is gone and mypy still narrows fine.

[stefan6419846]

See above.

[metsw24-max]

Same change here, checks for ArrayObject directly now.

[stefan6419846]

Why do we need an external helper if it is only used in one other function?

[metsw24-max]

Fair, it was only used once. Inlined it into the test.

[stefan6419846]

Please provide meaningful IDs.

[metsw24-max]

Added ids: empty-dest-array, empty-goto-action-dest, non-array-dest.

[metsw24-max]

Pushed the review changes: both branches now check isinstance(d, ArrayObject) directly (cast dropped), the test helper is inlined, and the parametrised cases have meaningful ids.

[metsw24-max]

any update?

[stefan6419846]

As already mentioned in your three other PRs, I recommend you to have a look at #3900. Until then, pinging for updates might have the effect of moving it to the end of my queue and even further delaying it.