We will be modeling future RPC implementations on an OpenAPI spec.
While this commit does not completely cohere to OpenAPI in terms
of response data, this is a good start and will allow us to cleanly
document these openapi routes in the current and future.
This commit brings in the new RPC routes:
- GET /rpc/v5/info/{pkgname}
- GET /rpc/v5/info?arg[]=pkg1&arg[]=pkg2
- POST /rpc/v5/info with JSON data `{"arg": ["pkg1", "pkg2"]}`
- GET /rpc/v5/search?arg=keywords&by=valid-by-value
- POST /rpc/v5/search with JSON data `{"by": "valid-by-value", "arg": "keywords"}`
Signed-off-by: Kevin Morris <kevr@0cost.org>
This commit fixes two of our tests in test_templates.py to go along
with our new template modifications, as well as a new test in
test_packages_routes.py which constructs two packages belonging
to the same package base, then tests that viewing their pages
produces their independent descriptions.
Signed-off-by: Kevin Morris <kevr@0cost.org>
Derived off of original work done by Leonidas Spyropoulos
at https://gitlab.archlinux.org/archlinux/aurweb/-/merge_requests/503
This revision of that original work finishes off the inconsistencies
mentioned in the original MR and adds a small bit of testing for more
regression checks.
Fixes: #360
Signed-off-by: Kevin Morris <kevr@0cost.org>
Found along with the previous commit to be a security hole in our
implementation. This commit resolves an issue regarding comment editing.
Signed-off-by: Kevin Morris <kevr@0cost.org>
This addresses a severe security issue, which is omitted from this
git message for obscurity purposes.
Otherwise, it allows co-maintainers to see the keyword form when
viewing a package they co-maintain.
Closes#378
Signed-off-by: Kevin Morris <kevr@0cost.org>
This was incorrectly using the particular Package record's name
to format options.snapshot_uri in order to produce URLPath.
It should, instead, use the PackageBase record's name, which
this commit resolves.
Bug reported by thomy2000
Closes#382
Signed-off-by: Kevin Morris <kevr@0cost.org>
Speeds up SSHPubKeys.PubKey searches in a larger database.
Fixed form of the original commit which was reverted,
1a7f6e1fa9
Signed-off-by: Kevin Morris <kevr@0cost.org>
This reverts commit 1a7f6e1fa9.
This commit broke account creation in some way. We'd still like to
do this, but we need to ensure it does not intrude on other facets.
Extra: We should really work out how this even passed tests; it
should not have.
Remove all extra whitespace when parsing Keywords to ensure we don't add
empty keywords in the DB.
Closes: #332
Signed-off-by: Leonidas Spyropoulos <artafinde@archlinux.org>
This patch does not include a javascript implementating, but
provides a pure HTML/HTTP method of paging through these lists.
Also fixes erroneous limiting. We now use a hardcoded limit of 20
by default.
Signed-off-by: Kevin Morris <kevr@0cost.org>
This patch brings in two new features:
- when viewing proposal listings, there is a new Statistics section,
containing the total and active number of Trusted Users found in the
database.
- when viewing a proposal directly, the number of active trusted users
assigned when the proposal was added is now displayed in the details
section.
Closes#323
Signed-off-by: Kevin Morris <kevr@0cost.org>
Previously, Python code was looking for suggestions based on
`%<keyword>%`. This was inconsistent with PHP's suggestion
implementation and cause more records to be bundled with a suggestion,
along with supplying misleading suggestions.
Closes#343
Signed-off-by: Kevin Morris <kevr@0cost.org>
A new option has been added for configuration of SMTP timeout:
- notifications.smtp-timeout
During tests, we can change this timeout to be small, so we aren't
depending on hardware-based RNG to pass the timeout.
Without a timeout, users can run into a long-running test for no
particular reason.
Signed-off-by: Kevin Morris <kevr@0cost.org>
People can change comaintainer ownership to suspended users if they
want to.
Suspended users cannot login, so there is no breach of security
here. It does make sense to allow ownership to be changed, imo.
Closes#339
Signed-off-by: Kevin Morris <kevr@0cost.org>
As repeats of these traceback notifications were annoying some of
the devops staff, and it took coordination to share tracebacks with
developers, this commit removes that responsibility off of devops
by reporting tracebacks to Gitlab repositories in the form of issues.
- removed ServerErrorNotification
- removed notifications.postmaster configuration option
- added notifications.gitlab-instance option
- added notifications.error-project option
- added notifications.error-token option
- added aurweb.exceptions.handle_form_exceptions, a POST route decorator
Issues are filed confidentially. This change will need updates
in infrastructure's ansible configuration before this can be
applied to aur.archlinux.org.
Signed-off-by: Kevin Morris <kevr@0cost.org>
This change brings some new additions to our archives:
- SHA-256 .sha256 hexdigests
- We construct our archives in a tmpdir now and move them to the
archive destination when all are completed. This removes some
corrupted downloading when archiving is in-process.
Signed-off-by: Kevin Morris <kevr@0cost.org>
There was one blazing issue with the previous implementation regardless
of the multiple records: we were generating fingerprints by storing
the key into a file and reading it with ssh-keygen. This is absolutely
terrible and was not meant to be left around (it was forgotten, my bad).
Took this opportunity to clean up a few things:
- simplify pubkey validation
- centralize things a bit better
Signed-off-by: Kevin Morris <kevr@0cost.org>
By implicitly joining, sqlalchemy joined on
`TUVote.UsersID = TUVoteInfo.SubmitterID`. This should be joining on
`TUVote.VoteID = TUVoteInfo.ID` instead to include all TUVote instances
found in the database.
Closes#266
Signed-off-by: Kevin Morris <kevr@0cost.org>
- the "Flagged Out-of-date on ..." link in the package action panel does
not contain a timezone specifier.
Signed-off-by: Kevin Morris <kevr@0cost.org>
It was found in the aur.al database that some records have
a non-null flagger, but are not flagged. Using the flagger
relationship, we were false redirecting away from the flag
page.
Signed-off-by: Kevin Morris <kevr@0cost.org>
This causes an issue that should have been obvious from the get-go:
if a package request is up in the AUR, but the package has already
been picked up by an official repository, we would end up returning
a 404 here, leading a TU to not be able to perform an action for
a request's target.
Signed-off-by: Kevin Morris <kevr@0cost.org>
For tests, we only care about emails having a valid syntax.
I don't think we should verify this at all, as aurweb.scripts.notify
will timeout if it cant deliver via sendmail/smtp.
Signed-off-by: Kevin Morris <kevr@0cost.org>
