Test test_iss1723 is broken

[sbourlon]

Test test_iss1723 is failing on Windows.

See demo in the PR #2441

Environment

Which environment were you using when you encountered the problem?

$ python -m platform
platform win32 -- Python 3.12.1, pytest-7.4.0, pluggy-1.2.0 -- C:\hostedtoolcache\windows\Python\3.12.1\x64\python.exe

$ python -c "import pypdf;print(pypdf._debug_versions)"
main branch

Code + PDF

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

pytest tests/test_writer.py -k "test_iss1723"

Traceback

This is the complete traceback I see:

================================== FAILURES ===================================
________________________________ test_iss1723 _________________________________
[gw2] win32 -- Python 3.12.1 C:\hostedtoolcache\windows\Python\3.12.1\x64\python.exe
IndexError: size = 4, index: 4

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

    @pytest.mark.enable_socket()
    def test_iss1723():
        # test of an annotation(link) directly stored in the /Annots in the page
        url = "https://github.com/py-pdf/pypdf/files/11015242/inputFile.pdf"
        reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
        writer = PdfWriter()
>       writer.append(reader, (3, 5))

tests\test_writer.py:1412: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
pypdf\_writer.py:2261: in append
    self.merge(
pypdf\_writer.py:2345: in merge
    pg = reader.pages[page]
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

self = <pypdf._page._VirtualList object at 0x000001A408C085C0>, index = 4

    def __getitem__(
        self, index: Union[int, slice]
    ) -> Union[PageObject, Sequence[PageObject]]:
        if isinstance(index, slice):
            indices = range(*index.indices(len(self)))
            cls = type(self)
            return cls(indices.__len__, lambda idx: self[indices[idx]])
        if not isinstance(index, int):
            raise TypeError("sequence indices must be integers") from TypeError(f"index = {index}")
        len_self = len(self)
        if index < 0:
            # support negative indexes
            index = len_self + index
        if index < 0 or index >= len_self:
>           raise IndexError("sequence index out of range") from IndexError(f"size = {len_self}, index: {index}")
E           IndexError: sequence index out of range

pypdf\_page.py:2225: IndexError
------------------------------ Captured log call ------------------------------
WARNING  pypdf._reader:_utils.py:477 Overwriting cache for 0 589

[stefan6419846]

I am not seeing these issues in my own fork, thus I would assume that we just have the usual network flakiness here. Re-running the job most likely fixes it.

[sbourlon]

Thank you Stefan for the answer. Do you have an idea of what did not go through the network? Would it be possible to catch such flakiness?

[stefan6419846]

It might be sufficient to add the file download to https://github.com/py-pdf/pypdf/blob/main/tests/example_files.yaml, see #2324.

[sbourlon]

Should be fixed with PR #2440

[stefan6419846]

Further investigation showed that this seems to indeed by a copy-and-paste error for the filename, thus it owuld depend on the execution order of the test runners whether this would pass or fail.