The only impactful change is the major version itself, where I've
fixed the breaking changes due to the upgrade of PyPDF2 [^1] and
checked there are no deprecation warnings when I run the tests.
[^1]: https://github.com/alphagov/notifications-utils/pull/973
`jsonschema[format]` includes all the formatting dependencies of
jsonschema, meaning that we don't have to specify `rfc3339-validator`
and `rfc3987` ourselves in the requirements.in file. This also has the
benefit of meaning that if the underlying formatting packages of
jsonschema change, we will be covered and won't accidentally miss the
fact that we need to change a package.
`charset-normalizer` is now used by default if installed instead of
`chardet` (https://pyup.io/changelogs/beautifulsoup4/#4.11.0). We do
have `charset-normalizer` installed because it's a subdependency of the
requests library, so it is being used.
This caused the `test_content_too_long_returns_400` to fail since it
now thought that the encoding of `ŵ` is `{'encoding': 'Big5',
'language': 'Chinese', 'confidence': 1.0}`.
There are two options for fixing this
- change the test content so that it doesn't just contain a single
letter - the docs state that you shouldn't run character detection on
very tiny content
- add `chardet` as a requirement, so that the code functions exactly the
same as before
I've chose the first option, since this avoids adding a dependency and
we should never have messages consisting of a single character.
The big breaking change for our code (not mentioned in the changelog) is
that the built-in validator for the `date-time` format now requires the
`rfc3339-validator` package instead of the `strict-rfc3339` package.
This updates the requirements file to use `rfc3339-validator`. Without
this change, wrong `date-time` formats would always silently pass validation.
There's no changelog for this, but I've looked through all the commits
and can't see any reason why this needed a major version bump or
anything that should cause us issues.
