I don't love the `Any` in that `Callable`, but I can't find a way to fix
it. In practice, it's either going to be `str` or `None`, but we pass an
`options` that's typed as `List[str] | str | None`, and one of the
functions has a header with a strict `str`. I tried various unions of
things and it wasn't working and I decided it's not worth it.
```
$ mypy --strict certbot-apache/certbot_apache/_internal/configurator.py
Success: no issues found in 1 source file
```
We could also leave the `jose.JWKRSA` call as-is and add
`--implicit-reexport`, or explicitly export `JWKRSA` in `josepy`, but I
think just cleaning the calls up is nice.
```
$ mypy --strict certbot-apache/certbot_apache/_internal/tests/util.py
Success: no issues found in 1 source file
```
```
$ mypy --strict certbot-dns-google/certbot_dns_google
Success: no issues found in 5 source files
```
I can change the `setattr`s to `# type: ignore [method-assign,
unused-ignore]` if that's preferred.
```
$ mypy --strict certbot-dns-cloudflare/certbot_dns_cloudflare
Success: no issues found in 5 source files
```
If we do `type: ignore` but don't set `--strict`, mypy gets mad. Flake8
doesn't like this but luckily we don't use that here (yet?). The other
option is to add `# type: ignore [method-assign, unused-ignore]`; I can
change it to that if that's preferred.
```
$ mypy --strict certbot-dns-rfc2136/certbot_dns_rfc2136
Success: no issues found in 5 source files
```
`dnspython` would be perfectly happy to accept a string once the
algorithm is passed through, but our `_RFC2136Client` object will only
accept `dns.name.Name` objects, so let's make it happy.
Decided that imports should be in ascii, not caseless, order. If we've
done otherwise elsewhere I'll change it.
The cast isn't the best but that's what we get for using a dict holding
various objects -- the thing marked by `vhost` is going to be a
virtualhost so 🤷
`re.findall` can
[return](https://docs.python.org/3/library/re.html#re.findall) either a
list of strings or a list of tuples of strings, depending what you
specify in the regex. In our case it's going to be a list of strings.
```
$ mypy --strict certbot-apache/certbot_apache/_internal/http_01.py
Success: no issues found in 1 source file
$ mypy --strict certbot-apache/certbot_apache/_internal/obj.py
Success: no issues found in 1 source file
```
PEP 526 says to declare types of unpacked tuples beforehand:
https://peps.python.org/pep-0526/#global-and-local-variable-annotations.
Could have just declared it in apache, but improved the acme return type
while I was at it.
Once again, `typing.Pattern` is deprecated in favor of `re.Pattern` so
changing that while parametrizing the type
Fixes#10000.
To create this PR, I looked through `constants.py` for defaults set to
`None`. If the action for the cli flag was `store_true` and there wasn't
other custom manual default specification, I changed it to report
`False`, and added a comment in `constants.py`. Adding `(default:` in
the help text suppresses listing of the actual default (done by
`cli_utils.py:CustomHelpFormatter`). Also added a comment for `redirect`
which is described manually since I noticed it while I was going
through.
---------
Co-authored-by: Brad Warren <bmw@users.noreply.github.com>
If we do `type: ignore` but don't set `--strict`, mypy gets mad. Flake8
doesn't like this but luckily we don't use that here (yet?). The other
option is to add `# type: ignore [method-assign, unused-ignore]`; I can
change it to that if that's preferred.
```
$ mypy --strict certbot-dns-digitalocean/certbot_dns_digitalocean
Success: no issues found in 5 source files
```
Fixes#10011
When we take a server block with no ssl addresses in and and enable ssl,
if it has any listens on the http port, use those host addresses when
creating a directive to listen on ssl. Addresses with no port and on
other ports will be ignored.
---------
Co-authored-by: Will Greenberg <willg@eff.org>
`typing.Type` is deprecated in favor of built-in `type`. In strict
mode,`find_ancestors` needs to be more specific about what it actually
returns, due to covariance and generics and such.
```
$ mypy --strict certbot-apache/certbot_apache/_internal/dualparser.py
Success: no issues found in 1 source file
```
Pasted from the old one. Maybe we can just rename it but this is what
github's web interface led me to create.
I want to make sure that they at least create the template so that they
read it. If they then choose to ignore it that's fine, but it should
always pop up. Basically I want to keep the old behavior. Open to
alternatives.
We could also play around with the new issue forms:
https://docs.github.com/en/communities/using-templates-to-encourage-useful-issues-and-pull-requests/syntax-for-issue-forms
Or label this one the "bug" template, and create a second one that is
blank but has the header text paragraph. I haven't seen a way to make
something appear in all templates, including the "blank" one, other than
just turning off blank templates.
* restore incorrect regex changes to CHANGELOG.md
* Update _release.sh regex to switch only first instance of main in changelog
(cherry picked from commit 0e225dcba2)
macOS-12 is [being deprecated](https://github.com/actions/runner-images/issues/10721) on Azure, so update to the latest available version.
* Upgrade macOS azure tests to use macOS-15
* switch standard azure tests to using python 3.12
* restore mac and linux cover tests to oldest and newest version style, and add explanation that that's what we're doing.
* Make reconfigure use staging server
* lint and imports
* Unset the account if it's been set in preparation for a dry run
* Add unit tests for checking we switch to staging and don't accidentally modify anything else
* add docstring
* Add test to make sure a requested new account id is saved
* update changelog
* set noninteractive mode for dry run
* error when account or server is set by the user
* switch to checking for changed values in account and server
* recommend using renew instead of certonly for forbidden fields
* change link to renew-reconfiguration
* Add deprecation warning when update_symlinks is run
* Remove information about update_symlinks from help
* ignore our own warning and remove unused imports
* Update changelog
* Add command to update config files without issuing/renewing cert
* toss up a vague untested skeleton
* remove duplicated code
* set certname in config
* consistent name, no zope
* import copy
* reconsitute is in renewal
* import renewal
* import cli
* fix lint errors
* call choose_configurator_plugins for its side effect of writing to config
* Set certonly in choose config plugins as we do for renew
* rewrite by piggybacking on existing side effects of a dry run instead
* do not allow domains to be set while reconfiguring
* remove unused cert_manager.reconfigure
* remove unused imports
* Add comments and messages
* add cli information
* start adding tests
* remove test code
* get certname before setting up plugins
* get plugin from lineage if not set on cli
* import copy
* always reconstitute
* only load cert once
* add error message
* improve comment
* mock everything out for tests
* test functionality is working!
* add tests for adding and modifying hooks
* test that we don't modify the config if the dry run fails
* improve documentation
* add webroot to reconfigure common options
* lint and clean up intermediate artifacts
* mock validate_hooks for windows
* print success message with updated parameters
* Improve success message
* add message for no changes have been made
* improve changed message to show before as well
* syntax
* Add changes will apply at the next renewal message
* lint
* lint really likes dict.items() for some reason
* run the deploy hook
* turn off dry run to test deploy hook
* patch list_hooks call for tests
* factor out reporting results code
* Remove reporting of which values were changed
* add flag to run deploy hook despite doing a dry run, and recommend setting that to yes when running reconfigure and modifying the deploy hook
* missing () around multi-line string
* test if the two dicts are equal instead of finding the actual changes, thus avoiding having to deal with webroot_map being a list
* refer to --deploy-hook instead of deploy hook
* use renewal configuration instead of configuration information
* mention that the deploy hook will use the active cert not the test one
* disable lint and remove new from language asking about running a deploy hook
* pluralize run deploy hook(s)
* Add test for reporting results when there is a webroot map
* update changelog
* Update error message about modifying domains on the certificate
* update changelog
* Add basic integration tests
* Just set -a rather than redoing the whole testing infrastructure
* used webroot in integration test since it's already installed
* file contents are accessed twice now
---------
Co-authored-by: Alex Zorin <alex@zorin.au>
* try the easy thing of just doing what the error message says
* temporarily add deploy stage to extended tests to see if it uploads properly
* follow instructions on https://forum.snapcraft.io/t/snapcraft-authentication-options/30473
* just run the packaging jobs for speed
* fix formatting
* import changes from test- branch and revert temporary changes
* Update instructions in deploy-stage.yml
Also, update `dev-cli.ini` example to use new flag.
Although https://github.com/bw2/ConfigArgParse/pull/216 allowed setting a `count` action value in a config file, our default detection system won't let us use that functionality. While we should eventually fix that, for now, let developers have a cli.ini with a higher logging level by adding this flag.
Note that this flag is intended to work the same way adding `-vvv`s does; that is, as a modifier to the pre-set level, rather than setting the absolute level. The number it is set to is equivalent to the number of `v`s that would otherwise have been passed, with "2" as the current maximum effective number of levels (warning --> info --> debug).
* Add --verbose-level flag for devs to set in cli.ini
* Update dev-cli.ini to use new flag
* Add instructions for setting up a cronjob in the docs
* Be more specific about where the cron entry will be created
Co-authored-by: alexzorin <alex@zorin.id.au>
* Correct &s to &s
Co-authored-by: alexzorin <alex@zorin.id.au>
* Correct other & to &
Co-authored-by: alexzorin <alex@zorin.id.au>
* De-weasel the double-scheduled-task comment
Co-authored-by: alexzorin <alex@zorin.id.au>
* Have users create directory hooks instead of command line hooks
* Use sudo in command
Co-authored-by: alexzorin <alex@zorin.id.au>
* tell windows users to ignore these instructions instead of telling them they won't work
* Use the same commands that we have in the general instructions
Co-authored-by: alexzorin <alex@zorin.id.au>
