Bug#1084323: pydoctor: FTBFS: Resource wordnet not found.

[Colin Watson]

reassign 1084235 python3-nltk 3.9.1-1
reassign 1084236 python3-nltk 3.9.1-1
reassign 1084237 python3-nltk 3.9.1-1
reassign 1084242 python3-nltk 3.9.1-1
reassign 1084249 python3-nltk 3.9.1-1
reassign 1084250 python3-nltk 3.9.1-1
reassign 1084291 python3-nltk 3.9.1-1
reassign 1084292 python3-nltk 3.9.1-1
reassign 1084294 python3-nltk 3.9.1-1
reassign 1084299 python3-nltk 3.9.1-1
reassign 1084392 python3-nltk 3.9.1-1
reassign 1084300 python3-nltk 3.9.1-1
reassign 1084306 python3-nltk 3.9.1-1
reassign 1084323 python3-nltk 3.9.1-1
reassign 1084332 python3-nltk 3.9.1-1
reassign 1084333 python3-nltk 3.9.1-1
reassign 1084334 python3-nltk 3.9.1-1
reassign 1084337 python3-nltk 3.9.1-1
reassign 1084338 python3-nltk 3.9.1-1
reassign 1084339 python3-nltk 3.9.1-1
reassign 1084341 python3-nltk 3.9.1-1
reassign 1084342 python3-nltk 3.9.1-1
reassign 1084344 python3-nltk 3.9.1-1
reassign 1084345 python3-nltk 3.9.1-1
reassign 1084346 python3-nltk 3.9.1-1
reassign 1084349 python3-nltk 3.9.1-1
reassign 1084385 python3-nltk 3.9.1-1
reassign 1084386 python3-nltk 3.9.1-1
forcemerge 1084235 1084236 1084237 1084242 1084249 1084250 1084291 1084292 1084294 1084299 1084392 1084300 1084306 1084323 1084332 1084333 1084334 1084337 1084338 1084339 1084341 1084342 1084344 1084345 1084346 1084349 1084385 1084386
affects 1084235 src:a2d src:abydos src:aiodogstatsd src:blag src:djangorestframework-api-key src:djangorestframework src:libspng src:mailmanclient src:markdown-callouts src:mintpy src:mkdocs-literate-nav src:mkdocs-section-index src:nlopt src:pydoctor src:python-django-pgtrigger src:python-djangorestframework-yaml src:python-djantic src:python-igraph src:python-inline-snapshot src:python-jellyfish src:python-markdown src:python-mkdocs src:python-opt-einsum src:python-pipx src:python-respx src:python-uvicorn src:twisted src:typer
thanks

On Mon, Oct 07, 2024 at 11:19:03AM +0100, Colin Watson wrote:
> On Mon, Oct 07, 2024 at 10:38:13AM +0200, Santiago Vila wrote:
> > During a rebuild of all packages in unstable, your package failed to build:
> [...]
> > /usr/lib/python3/dist-packages/nltk/stem/__init__.py:34: in <module>
> >     from nltk.stem.wordnet import WordNetLemmatizer
> > /usr/lib/python3/dist-packages/nltk/stem/wordnet.py:13: in <module>
> >     class WordNetLemmatizer:
> > /usr/lib/python3/dist-packages/nltk/stem/wordnet.py:48: in WordNetLemmatizer
> >     morphy = wn.morphy
> > /usr/lib/python3/dist-packages/nltk/corpus/util.py:120: in __getattr__
> >     self.__load()
> > /usr/lib/python3/dist-packages/nltk/corpus/util.py:86: in __load
> >     raise e
> > /usr/lib/python3/dist-packages/nltk/corpus/util.py:81: in __load
> >     root = nltk.data.find(f"{self.subdir}/{self.__name}")
> > /usr/lib/python3/dist-packages/nltk/data.py:579: in find
> >     raise LookupError(resource_not_found)
> > E   LookupError:
> > E   **********************************************************************
> > E     Resource wordnet not found.
> > E     Please use the NLTK Downloader to obtain the resource:
> > E
> > E     >>> import nltk
> > E     >>> nltk.download('wordnet')
> > E     
> > E     For more information see: https://www.nltk.org/data.html
> > E
> > E     Attempted to load corpora/wordnet
> > E
> > E     Searched in:
> > E       - '/<<PKGBUILDDIR>>/.pybuild/cpython3_3.12_pydoctor/nltk_data'
> > E       - '/usr/nltk_data'
> > E       - '/usr/share/nltk_data'
> > E       - '/usr/lib/nltk_data'
> > E       - '/usr/share/nltk_data'
> > E       - '/usr/local/share/nltk_data'
> > E       - '/usr/lib/nltk_data'
> > E       - '/usr/local/lib/nltk_data'
> > E   **********************************************************************
> 
> I assume this is because some downloadable data went away, though I'm
> not certain.  Still, we obviously shouldn't have an implicit dependency
> on downloaded data during package builds.
> 
> Carsten, what would you think of this patch to python-lunr, which fixes
> both pydoctor and twisted (and I suspect probably a bunch of other
> packages, since mkdocs also depends on python3-lunr)?

Cancel this - we don't need to change python-lunr.  Sorry to bother you,
Carsten.

I tracked this down to a regression in nltk instead.  This is
https://github.com/nltk/nltk/issues/3308, fixed in
https://github.com/nltk/nltk/pull/3309.  

Mo, could we please apply the attached patch to nltk?  I've test-built
all the affected packages against this.  python-igraph has uninstallable
build-dependencies (indirectly due to https://bugs.debian.org/1084781, I
think), while python-uvicorn fails in an unrelated way (it looks as
though it may be fixed by the changes to ProxyHeadersMiddleware in
0.31.0); but everything else from the list of affected packages above
builds cleanly again after applying this patch.

Thanks,

-- 
Colin Watson (he/him)                              [cjwatson at debian.org]
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Don-t-read-the-WordNet-corpus-before-it-is-needed.patch
Type: text/x-diff
Size: 5540 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/debian-science-maintainers/attachments/20241008/4bcb143c/attachment.patch>