* current workflow only needs to read git content
* if the workflow in the future does need write access, it's good to
see permissions explicitly changing
For context: "pull_request" runs never have write access anyway, so this
significantly changes only the "push" runs that happen when branches are
merged to develop.
Signed-off-by: Jussi Kukkonen <jkukkonen@vmware.com>
By default pylint does format checks:
https://pylint.pycqa.org/en/latest/technical_reference/features.html?highlight=format#format-checker
The problem is we also use black and isort who have format checkers as
well. This makes pylint format checks obsolete.
Also, it's possible that you would want to disable a warning and you
can end up in the situation where you will have to disable it for
two tools altogether.
Signed-off-by: Martin Vrachev <mvrachev@vmware.com>
Add one test with 1 subtests for various root key rotation situations.
The test data definition format is a bit tricky but I tried to document
that in the test function docstring.
Signed-off-by: Jussi Kukkonen <jkukkonen@vmware.com>
Store signers with their keyids so they are easier to remove.
The signers structure now looks like:
{
"role1": {
"keyidA": SSlibSigner,
"keyidB": SSlibSigner,
}
}
Add convenience method for adding a signer.
Signed-off-by: Jussi Kukkonen <jkukkonen@vmware.com>
Also tweak the docstrings: the "caching" target_dir usage is
presented in the module doc example: there should be no need for
additional comments in the methods themselves as long as the argument
docs are readable.
Signed-off-by: Jussi Kukkonen <jkukkonen@vmware.com>
This is slightly cosmetic but rename get_one_valid_targetinfo to
get_targetinfo:
* The function name is long without any reason: "one" and "valid" are
always implicit
* shortening makes code (incl. our examples and tests) easier to read
* We're also already changing updater API (compared to legacy) so this
alone does not break things -- it's also not a difficult "port".
Signed-off-by: Jussi Kukkonen <jkukkonen@vmware.com>
Remove updated_targets() as it doesn't fit the rest of the API.
In its stead add find_cached_target() which has a similar signature
as download_target(): both accept an optional local filepath as
argument and return full local filepath. In the
find_cached_target() case None is returned if the local file is not the
correct target file.
Updater constructor gets a new optional target_dir argument: This means
client can avoid giving a local filepath as an argument to
find_cached_target()/download_target() -- Updater will then generate a
filename within targets_dir.
A reasonable use pattern (when targets_dir is set in constructor):
info = updater.get_one_valid_targetinfo("targetname")
path = updater.find_cached_target(info)
if path is None:
path = updater.download_target(info)
Signed-off-by: Jussi Kukkonen <jkukkonen@vmware.com>
Validate that root role names are 4 and that they are exactly
"root", "snapshot", "targets" and "timestamp" as described in
the spec:
https://theupdateframework.github.io/specification/latest/#root-role
Additionally, fix the valid_roots dataset, so each of the cases contains
the top metadata role names inside the roles dictionary.
Signed-off-by: Martin Vrachev <mvrachev@vmware.com>
Our newly added metadata files in the
tests/repository_data/fishy_rolenames/metadata directory have an expiry
date until "2021-10-22T11:21:56Z" and today while running the tests on
develop branch I recived this error:
ExpiredMetadataError("Metadata X expired on Fri Oct 22 11:21:56 2021")
when running the tests in tests/test_updater.py file and more precisly
the TestUpdaterRolenames.test_unusual_rolenames() test.
That's why I decided to bump the expiration date to a random time in
the future (October 22-nd 2050) and I had to resign all of the metadata
files.
Signed-off-by: Martin Vrachev <mvrachev@vmware.com>
Our sdist has typically included everything from git apart from the CI
related files (.github/*, .fossa.yml, .readthedocs.yaml). Update our
MANIFEST.in and the check-manifest section of setup.cfg to be explicit
about this.
Signed-off-by: Joshua Lock <jlock@vmware.com>
Add a 'Documentation' entry to project_urls pointing to our stable docs
on readthedocs.io. This will result in a 'Documentation' entry under the
'Project links' section on PyPI.
Signed-off-by: Joshua Lock <jlock@vmware.com>
Invoking setup.py directly is deprecated, see:
https://blog.ganssle.io/articles/2021/10/setup-py-deprecated.html
Therefore:
* remove the executable bit from setup.py's permissions
* remove the shebang entry from setup.py
* update the comments in setup.py to recommend using build to create dists
and pip to install them
Signed-off-by: Joshua Lock <jlock@vmware.com>
Update the MANIFEST.in to be explicit about what we choose to ship in our
sdist. This _does not_ result in any additional files being included in
our sdist, but does remove warnings from build.
Signed-off-by: Joshua Lock <jlock@vmware.com>
* List license files in a new metadata section
* Remove .travis.yml from check-manifest section's ignore entry
Signed-off-by: Joshua Lock <jlock@vmware.com>
build, twine and wheel packages should all be installed in order to be
able to build and release python-tuf -- add those dependencies to
requirements-dev.txt
Signed-off-by: Joshua Lock <jlock@vmware.com>
Python 3.10 is released on October 4-th 2021 and it seems
logical to add support for it as it doesn't require any major effort
from the project.
For reference read:
https://www.python.org/downloads/release/python-3100/
Signed-off-by: Martin Vrachev <mvrachev@vmware.com>
When I tried adding support for Python3.10 we had CI errors due to
test failures: https://github.com/theupdateframework/python-tuf/pull/1610/checks?check_run_id=3861875325
The problem comes from the fact that we start a subprocess
executing simple_https_server.py, but then we fail to communicate the
message we expect from the server process to the main process actually
running the test. We expect our custom message to be the first line
printed from the server process, but instead, a deprecation warning is
printed first about the usage of ssl.wrap_socket(). Our custom message
is printed second.
As of Python 3.7 this function has been deprecated:
https://docs.python.org/3/library/ssl.html#ssl.wrap_socket and for
whatever the reason we didn't get a warning when using it before.
My fix does what is suggested in the warning and replaces the usage of
ssl.wrap_socket() by instantiating a ssl.SSLContext object and then
calling SSLContext.wrap_socket().
This removes the warning.
Signed-off-by: Martin Vrachev <mvrachev@vmware.com>
