Depending on the exact way the pdf file is written, iterating over
Annots can fail after the file has been closed. This is not the case
for matplotlib-generated pdfs, but is the case for mplcairo-generated
ones. As a simple repro of the different behavior between
"in-the-contextmanager" and "out-of-the-contextmanager":

frommatplotlibimportpyplotaspltplt.figtext(.5,.5,"hello, world",url="https://www.google.com")plt.savefig("/tmp/test.pdf",backend="pdf")importpikepdfwithpikepdf.Pdf.open("/tmp/test.pdf")aspdf:page=pdf.pages[0]print(repr(page.Annots))# within contextmanager: okwithpikepdf.Pdf.open("/tmp/test.pdf")aspdf:page=pdf.pages[0]print(repr(page.Annots))# after contextmanager: AttributeError

PR Summary

PR Checklist

• Has pytest style unit tests (andpytest passes).
• IsFlake 8 compliant (runflake8 on changed files to check).
• New features are documented, with examples if plot related.
• Documentation is sphinx and numpydoc compliant (the docs shouldbuild without error).
• Conforms to Matplotlib style conventions (installflake8-docstrings and runflake8 --docstring-convention=all).
• New features have an entry indoc/users/next_whats_new/ (follow instructions in README.rst there).
• API changes documented indoc/api/next_api_changes/ (follow instructions in README.rst there).