Month returned by habanero.cn.content_negotiation(ids = doi) no longer in curly brackets

[tardigradus]

With Python 3.6.8 and habanero 0.7.4, I have been successfully downloading BibTeX files and parsing them with BibtexParser for the last couple of years. However, some time between the middle of September 2023 and now, the format of the BibTex file I get via habanero.cn.content_negotiation(ids = doi) changed: The value of the field month is no longer enclosed in curly brackets.

So whereas in the middle of September 2023 I got:

@article{Zupan2023,
 author = {Hana Zupan and Frederick Heinz and Bettina G. Keller},
 doi = {10.1139/cjc-2022-0282},
 journal = {Canadian Journal of Chemistry},
 month = {sep},
 number = {9},
 pages = {710--724},
 publisher = {Canadian Science Publishing},
 title = {Grid-based state space exploration for molecular binding},
 url = {https://doi.org/10.1139/cjc-2022-0282},
 volume = {101},
 year = {2023}
}

I now get

@article{Zupan2023,
 author = {Hana Zupan and Frederick Heinz and Bettina G. Keller},
 doi = {10.1139/cjc-2022-0282},
 journal = {Canadian Journal of Chemistry},
 month = sep,
 number = {9},
 pages = {710--724},
 publisher = {Canadian Science Publishing},
 title = {Grid-based state space exploration for molecular binding},
 url = {https://doi.org/10.1139/cjc-2022-0282},
 volume = {101},
 year = {2023}
}

This is then causing BibtexParser to fail.

I assume this is not directly a problem with habanero, but this a known issue?

Edit: doi2bib has the same issue: https://www.doi2bib.org/bib/10.1139/cjc-2022-0282

[sckott]

Thanks and sorry about the delay! Life events made it hard to do any paid or non-paid work. Having a look

[sckott]

I can replicate the issue myself.

It does seem like a bare string like sep shouldn't be allowed given the text in https://www.bibtex.org/Format/

However, I'm not sure it's within scope to fix these issues in this package. I"ll check if it can be done easily

[sckott]

watch https://github.com/sciunto-org/python-bibtexparser/issues?q=is%3Aissue+is%3Aopen+label%3Av2

[sckott]

@tardigradus Okay, you can try the fixed version on main branch here - install from github, not on pypi yet.

[tardigradus]

Sorry for not getting round to this earlier.

I assumed that version 1.2.6 contains the fixes you made, so I installed that with pip. I also had to update bibtexparser to version 2, but I now get the month in curlies, as desired. Thanks for fixing.

[sckott]

thanks @tardigradus yeah there's an optional install for bibtex - see https://github.com/sckott/habanero?tab=readme-ov-file#installation

[hseg]

Hi, can you please make the running of
test-content_negotiation::test_content_negotiation_bad_bibtex conditional on
if bibtexparser is installed? Given that you've made it an optional
dependency, it makes it harder to test an installation of habanero will work
with a bare python -m pytest.

At the moment, I need to tell the users of the AUR package to either install
python-bibtexparser for the tests, or to manually disable the tests, which
isn't a great situation.

[hseg]

It seems it'd be enough to mark that test as
@pytest.importorskip("bibtexparser"), see
https://docs.pytest.org/en/stable/reference.html#pytest-importorskip