vcswatch reports that
this package seems to have new commits in its VCS but has
not yet updated debian/changelog. You should consider updating
the Debian changelog and uploading this new version into the archive.
Here are the relevant commit logs:
commit 60977c7baee9761c6ed2f015c1c02c0bb517d6da
Author: Antoine Beaupré <anarcat@debian.org>
Date: Tue Oct 26 21:33:20 2021 -0400
we only support python3, mark setup.py as such
commit 5d7063ae60e43cab2595d1299ad54e54993f82d3
Merge: be20f1c 02e2b64
Author: Antoine Beaupré <anarcat+gitlab@anarc.at>
Date: Sat Oct 23 13:33:46 2021 +0000
Merge branch 'cleanups' into 'main'
Fix flake8 complaints introduced by Wayback plugin changes
See merge request anarcat/feed2exec!14
commit 02e2b640c8741869298e82b07959e8fc570864f1
Author: Paul Wise <pabs3@bonedaddy.net>
Date: Mon Oct 18 10:09:54 2021 +0800
Wrap and sort Debian packaging
Makes diffs easier to read and makes dependencies easier to write.
Use trailing commas instead of prefixed commas, which are uglier.
Changes-by: wrap-and-sort --short-indent --wrap-always --sort-binary-packages --trailing-comma
commit 4bfc9a0dc18c861fea78745a24483d0bcac2c77c
Author: Paul Wise <pabs3@bonedaddy.net>
Date: Mon Oct 18 10:00:01 2021 +0800
Require feedparser 5 as minimum version for Python 3
Earlier feedparser versions crash when using Python 3.
commit 11b1f7ecff24c3624882884207a324caa53d5165
Author: Paul Wise <pabs3@bonedaddy.net>
Date: Mon Oct 18 08:10:52 2021 +0800
Add missing dependency on lxml for the wayback plugin
Fixes: commit 0deb1efab43f30f213e6f5f728a422e52a1cd1c2
commit 0a58133f8b17c776f0cb95802288ac51fefd2f0b
Author: Paul Wise <pabs3@bonedaddy.net>
Date: Mon Oct 18 07:51:07 2021 +0800
Fix flake8 complaints introduced by Wayback plugin changes
Fixes: commit 0deb1efab43f30f213e6f5f728a422e52a1cd1c2
commit be20f1c927c5fb28fe6077c66b60905296ae5f93
Author: Antoine Beaupré <anarcat@debian.org>
Date: Sat Oct 23 09:31:05 2021 -0400
allow failures on mypy
It turns out it actually fails on a clean repo right now, because of
this error:
feed2exec/__init__.py:6: error: Cannot find implementation or library stub for module named "feed2exec._version"
I hadn't found this problem because I have the file locally. I should
probably just create a stub for it and get it over with.
commit 247a3d9db1e270c7ce2855c0a303a52b38d454ec
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 19:59:19 2021 -0400
mention build module, which may be necessary for newer setuptools_scm
commit 9ba5521b1a11874bca4bb1e90ca43ec1d7f0c913
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 19:53:39 2021 -0400
fix rst syntax since we are not hacking it anymore
commit 9010701ea826c9567eeb0039a7b543583846246c
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 14:56:56 2021 -0400
switch to setup.cfg as much as possible
This mostly builds with `python -m build` now, but it's unclear if
metadata (or content!) is correct, needs further testing but
test.pypi.org is down..
https://github.com/pypa/packaging-problems/issues/551
Note that a lot of this comes from:
https://github.com/pypa/build/blob/main/setup.cfg
... which, interestingly, still has a setup.cfg even though it has a
pyproject.toml and should be the "state of the art"...
commit 7e88e167645cca0863c60d06c845d99cc41f409f
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 13:48:24 2021 -0400
first pass at type checking with mypy
We use a pyproject.toml even though we have a setup.cfg because (a) it
seems everyone is converging towards that, for better or for worse
and (b) the mypy in debian doesn't find my overrides in the setup.cfg.
commit 823a86c5e3d53da38d41cbb7a33c09a0bfb9ddf5
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 13:13:49 2021 -0400
enforce black in tox, but not in CI
This will show that black fails to pass in CI, while keeping pipelines
green for now.
commit 2f086c05c07cb0ad841decaba032b689a9bc5c47
Merge: a827bba dfe2f32
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 12:59:55 2021 -0400
Merge branch 'fp6'
commit dfe2f32a470a90491e05b61ab14b390e203501ae
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 12:47:58 2021 -0400
Revert "enable SAST audit on GitLab.com"
We don't look at the results and there are too many false positives.
This reverts commit ad66dde120aac18b0a9c7b56f98fee39ee89233c.
commit fb76186b326c49d36d546cb6bc405cb68e0dad9b
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 12:46:34 2021 -0400
Revert "fix another test failure in newer feedparser"
I cannot reproduce those test failures in GitLab CI nor inside a
Python 3.9 Docker container. I have no idea what is going on here, but
I'm going to pretend this problem does not exist and move on.
This reverts commit ada7aaabd5f0f873d4c38a5d59549b6bd173bdcc.
commit 4536b099c1e0a7aa13528aced52cbbaa08cd28eb
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 12:41:47 2021 -0400
drop support for python 3.5, add 3.9
commit 4b47186360380d96c5b9e1e88d2f6af6eaf566c4
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 12:22:07 2021 -0400
avoid dateparser bug workaround in newer versions
This is unrelated with the previous commit: dateparser does not parse
the date better or worse with or without this. It's just glue so that
we use the upstream code correctly and so that we can eventually drop
our hack reliably.
commit ada7aaabd5f0f873d4c38a5d59549b6bd173bdcc
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 12:08:44 2021 -0400
fix another test failure in newer feedparser
It seems like feedparser found a new way to default certain dates. We
were previously correctly parsing that date before, but for some
obscure reason feedparser is now defaulting to the feed's
updated_parsed field instead.
Typically, this was being parsed by dateparser, but it seems this is
either failing (which I cannot confirm locally) or ignored by
feedparser:
anarcat@angela:feed2exec(fp6)$ python3
Python 3.9.2 (default, Feb 28 2021, 17:03:44)
[GCC 10.2.1 20210110] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from dateparser import parse, __version__
>>> parse("Sun, 15 Feb 2015 00:00:00")
datetime.datetime(2015, 2, 15, 0, 0)
>>> __version__
'1.0.0'
It's not clear why this is failing, therefore. The result is wrong,
but I want tests to pass, so let's just get this over with.
commit f1ec2a334163e2c82c3b6f8d585acbf37e62a2af
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 11:44:20 2021 -0400
catchup with html2text again (2020.1.16)
This was generated by copying the file from:
/tmp/pytest-of-anarcat/pytest-1/planet0/Mail/planet-debian.mbx
And just committing it. We change the User-Agent now because I am
tired of patching this, but it shouldn't matter.
commit e589a9741dfa8f97612bd9b97d383c0f38a5e396
Author: Antoine Beaupré <anarcat@debian.org>
Date: Fri Oct 22 11:36:19 2021 -0400
fix feedparser dependencies for newer python
As @pabs3 found out in MR !12, newer feedparser (6.0.0) releases
dropped support for older Python releases, which broke our CI
significantly. We were avoiding 6.0.0 because of a change in the data
structure, but that was fixed in 6.0.0b3, so let's depend on that for
newer releases.
commit a827bbaccb7eee0531a3d31fbd0935cb93d9035c
Merge: dead021 0deb1ef
Author: Antoine Beaupré <anarcat+gitlab@anarc.at>
Date: Sun Oct 17 17:54:23 2021 +0000
Merge branch 'spn2' into 'main'
Add support for saving page resources to the Wayback Machine too
See merge request anarcat/feed2exec!11
commit 0deb1efab43f30f213e6f5f728a422e52a1cd1c2
Author: Paul Wise <pabs3@bonedaddy.net>
Date: Fri Sep 3 07:42:44 2021 +0800
Add support for saving page resources to the Wayback Machine too
The SPN2 API used by this is not well documented and does
not return proper status codes or easily usable error text,
so make it enabled by default but add an option to disable it
and disable it in the tests, which need the right return codes.
The SPN1 API does return more specific error text in HTML in
the same way as SPN2 in some cases, so process that too.
See-also: https://blog.archive.org/2019/10/23/the-wayback-machines-save-page-now-is-new-and-improved/
commit dead021e59a9bea11450c2f45cc3fde4b92a872c
Author: Antoine Beaupré <anarcat@debian.org>
Date: Tue Sep 7 11:44:17 2021 -0400
exclude dependencies from scan
We're getting a *lot* of noise from the .eggs directory,
unfortunately. Just one example:
> Use of insecure MD2, MD4, MD5, or SHA1 hash function. .eggs/requests-2.26.0-py3.7.egg/requests/auth.py:154
commit 9c2ff0ce5dc518c7de23c5ac181fa77c430be865
Author: Antoine Beaupré <anarcat@debian.org>
Date: Tue Sep 7 11:41:44 2021 -0400
disable eslint analyser
This is not (currently) a web app, and the eslint analyser currently
just fails to run.
commit ad66dde120aac18b0a9c7b56f98fee39ee89233c
Author: Antoine Beaupré <anarcat@debian.org>
Date: Tue Sep 7 11:35:30 2021 -0400
enable SAST audit on GitLab.com
Let's see what this does.
See https://docs.gitlab.com/ee/user/application_security/sast/index.html
commit 7502f7104afc6255793efab4cb83b7ec4004552e
Merge: bbb1a39 11b69ed
Author: Antoine Beaupré <anarcat+gitlab@anarc.at>
Date: Thu Aug 26 16:14:38 2021 +0000
Merge branch 'lintian-fixes' into 'main'
Fix some issues reported by lintian
See merge request anarcat/feed2exec!10
commit 11b69ed9d19de8652211d8ab48a573732d9aed72
Author: Debian Janitor <janitor@jelmer.uk>
Date: Tue Jun 22 19:19:35 2021 +0000
Update standards version to 4.5.1, no changes needed.
Changes-By: lintian-brush
Fixes: lintian: out-of-date-standards-version
See-also: https://lintian.debian.org/tags/out-of-date-standards-version.html
commit d51ab3d5d25c53643b8516009e74c35fec83a422
Author: Debian Janitor <janitor@jelmer.uk>
Date: Tue Jun 22 19:19:35 2021 +0000
Use canonical URL in Vcs-Git.
Changes-By: lintian-brush
Fixes: lintian: vcs-field-not-canonical
See-also: https://lintian.debian.org/tags/vcs-field-not-canonical.html
commit 1fa44e0c62142beb6f6b90f96620bb6ae513c62d
Author: Debian Janitor <janitor@jelmer.uk>
Date: Tue Jun 22 19:19:31 2021 +0000
Use secure copyright file specification URI.
Changes-By: lintian-brush
Fixes: lintian: insecure-copyright-format-uri
See-also: https://lintian.debian.org/tags/insecure-copyright-format-uri.html
commit 1ca679b4ab72574c8704bc8f32d21996e85b36aa
Author: Debian Janitor <janitor@jelmer.uk>
Date: Tue Jun 22 19:19:31 2021 +0000
Trim trailing whitespace.
Changes-By: lintian-brush
Fixes: lintian: trailing-whitespace
See-also: https://lintian.debian.org/tags/trailing-whitespace.html
commit bbb1a3995dc6bb9c392a0295c2fe85ea5915aca9
Author: Antoine Beaupré <anarcat@debian.org>
Date: Mon May 24 09:45:28 2021 -0400
leave Freenode for oftc.net
This is becoming a long story, but basically Freenode was hijacked,
policies were changed to allow racism and bigotry, trolls have been
promoted to opers, and opers are hijacking channels. It's a shame,
really, but nothing we can do it seems:
https://gist.github.com/joepie91/df80d8d36cd9d1bde46ba018af497409
Moving to OFTC since that's where most of my other channels (Debian,
Tor) are.
commit c7b6f73686178725eb348c67d0cf0b58fa308f65
Author: Antoine Beaupré <anarcat@debian.org>
Date: Thu Feb 25 09:50:08 2021 -0500
fix crash in feed2exec parse
For some reason, I was getting this backtrace here when running a test
feed:
plugin generated exception: expected str, bytes or os.PathLike object, not NoneType, skipping
Traceback (most recent call last):
File "/home/anarcat/src/feed2exec/feed2exec/plugins/__init__.py", line 105, in output
return plugin.output(*args, feed=feed, item=item, lock=lock)
File "/home/anarcat/src/feed2exec/feed2exec/plugins/maildir.py", line 42, in __init__
prefix = os.path.expanduser(feed.get('mailbox', '~/Maildir'))
File "/usr/lib/python3.7/posixpath.py", line 235, in expanduser
path = os.fspath(path)
TypeError: expected str, bytes or os.PathLike object, not NoneType
This was caused by the maildir plugin constructor, which was looking
up the `mailbox` field in the field. Because that is unset in my
feed2exec.ini (because I was testing against a new feed), that
`mailbox` key was just not present. The feedparser dict behavior in
this case is to return None, instead of the default, which I find
strange and unusual.
Tests still seem to pass here after the patch, so I guess this should
work in the future, and does un-break the feed test.
I can't quite figure out how to come up with a unit test for that one,
but now I can't help but think that feed configuration should just be
split up from the feedparser dict, because so much of those issues
come from that (strange) coupling.
commit a44497c1ef055d6891accfad47c549a6f6d31ac1
Author: Antoine Beaupré <anarcat@debian.org>
Date: Mon Feb 8 15:49:17 2021 -0500
mention (lack of) Gemini and other protocol support
![[bug history graph]](https://qa.debian.org/data/bts/graphs/f/feed2exec.png){kind=link}