- Jun 01, 2020
-
-
Antoine R. Dumont authored
This allows to avoid the clash names when depositing multiple archives with the same name. The folder name is a timestamp to the microsecond so there should no longer be any clash (famous last work and all that). Related to T2393 Related to D3192
-
- May 20, 2020
-
-
Antoine R. Dumont authored
Deposit whose external_id column matches are excluded from the listing. Related to T2406
-
- May 19, 2020
-
-
Antoine R. Dumont authored
This removes references in the following: - deposit specs and docs - deposit status read api - deposit tests in general... - deposit private update status api - deposit client cli Related to T2412
-
- May 15, 2020
-
-
Antoine R. Dumont authored
Migrate both "recent" and "old" format deposits [1] to the new specification. That means the deposit swh_id* fields will be set to: - swh_id: directory SWHID (no context) - swh_id_context: directory SWHID (with context, origin, visit, anchor path) Optionally, those 2 fields will be kept (for now) and realigned where it was not set ("old" deposits) to: - swh_anchor_id: revision SWHID (no context) - swh_anchor_id_context: revision SWHID (context with only origin) It's expected some very "old" deposits won't be migrated as we cannot resolve those values. They will be rescheduled when it will be possible to do so (deploy [2]). [1] "recent" format means all swh_id fields are set: - swh_id: directory SWHID (no context) - swh_id_context: directory SWHID (context with only origin) - swh_anchor_id: revision SWHID (no context) - swh_anchor_id_context: revision SWHID (context with only origin) "old" format: - swh_id: revision SWHID (no context) - swh_id_context: not set - swh_anchor_id: not set - swh_anchor_id_context: not set [2] Related to D3141 Related to T2398
-
- May 14, 2020
-
-
Antoine R. Dumont authored
This also ensures the loader sends all the required information now (it did already but it was a bit laxed on checks for that part). Related to T2398
-
Antoine R. Dumont authored
It's a tad more colorful
-
Antoine R. Dumont authored
-
Antoine R. Dumont authored
-
- May 11, 2020
-
-
Antoine R. Dumont authored
Prior to this commit, this expected to parse properly the api response. It could not as the response was not in the proper format. This also fixes incomplete test committed by mistake in c64ee543
-
Antoine R. Dumont authored
Prior to this commit, the returned result unexpectedly as a xml message within a string.
-
Antoine R. Dumont authored
So that the deposit client can parse the result as well Related to T2401
-
Antoine R. Dumont authored
when related database is down for maintenance Related to T2401
-
Antoine R. Dumont authored
-
Antoine R. Dumont authored
-
- May 07, 2020
-
-
Antoine R. Dumont authored
-
Antoine R. Dumont authored
-
Antoine R. Dumont authored
-
- May 06, 2020
-
-
vlorentz authored
This allows the checker to use a configurable URL, in case it needs an URL different from the one used by the client. For example a client outside docker would use http://localhost:5080/deposit/ to access the API, but the checker cannot use this URL (it needs either nginx or swh-deposit as hostname).
-
- May 04, 2020
-
-
vlorentz authored
That code was not covered so far.
-
- Apr 29, 2020
-
-
Stefano Zacchiroli authored
-
- Apr 23, 2020
-
-
Antoine R. Dumont authored
This avoids: - avoids unnecessary redundancy: metadata was set in multiple places - sending no longer read information
-
Antoine Lambert authored
pytest-django and pytest-flask both declare a fixture named client and the wrong one is picked when running swh-deposit tests. So avoid the loading of pytest flask plugin to prevent errors when running tests.
-
- Apr 20, 2020
-
-
Antoine R. Dumont authored
Related to T2367
-
- Apr 17, 2020
-
-
Stefano Zacchiroli authored
-
- Apr 08, 2020
-
-
Nicolas Dandrimont authored
I noticed this because my virtualenv ended up with the djangorestframework-stubs package installed, which made the mypy pre-commit hook fail. This commit adds drf-stubs to the tox.ini, and does the relevant updates to our type annotations/signatures. Turns out that upstream DRF has all `request` parameters named that way, so we rename all our `req` parameters to match, which appeases mypy's signature checks.
-
David Douard authored
-
David Douard authored
- blackify all the python files, - enable black in pre-commit, - add a black tox environment.
-
- Jan 23, 2020
-
-
Antoine R. Dumont authored
New latest release on swh.core added the environment parameter. This will fix the current broken jenkins build [1] [1] https://jenkins.softwareheritage.org/job/DDEP/job/tests/658/console
-
- Jan 17, 2020
-
-
Antoine Lambert authored
Related to T2188
-
- Jan 16, 2020
-
-
Antoine Lambert authored
Those tests were broken since rDCORE2255767b9763cefe0841a8d0231f568adcec87ff.
-
- Jan 13, 2020
-
-
David Douard authored
-
- Jan 10, 2020
-
-
David Douard authored
For this we: - ensure we create the DepositRequest with str as raw_metadata (instead of a bytes), and - make atom_dataset return a str instead of a bytes Every time this chunk of data is sent on the wire (http req to the django app) this needs to be an str, otherwise we have things like: raw_metadata = "b'[...]'" especially when testing against django 2, which is the only django version available on debian sid, so this is needed to make it possible to build the deb file on sid *with* tests enabled. - add a py3-django2 environment in tox - bump the dependency to django <3 in requirements-server.txt (otherwise it will take precendence on the deps: field of the tox file, thus execute tests in py3-django2 with django 1.11). Note that no other tests than executing pytest have been done with django2, so do not expect it works flawlessly.
-
David Douard authored
-
- Jan 09, 2020
-
-
David Douard authored
This later no longer exists.
-
- Jan 08, 2020
-
-
David Douard authored
for the sake of consistency.
-
David Douard authored
this code was the only user for the swh.core.tarball.compress function. Doing so allows to reduce a bit the complexity and coupligin in the code. It will also allow to get rid of the swh.core.tarball.compress function.
-
David Douard authored
No need for this to be a subpackage.
-
- Jan 07, 2020
-
-
vlorentz authored
This reflects what is done in practice and what the code expects. Using an email is invalid because the value of --author is used in a `http://schema.org/name` property, not `http://schema.org/email`
-
- Dec 20, 2019
-
-
David Douard authored
This new option allows these cli commands (upload and status) to produce JSON or YAML output, so it's easier to test.
-
David Douard authored
so the scheduler interaction code is executed. Note that this does not test for correctness in these interactions yet. also move tests/__init__.py content in tests/conftest.py and adapt test code accordingly. This also ensures retries_left is set otherwise tests may fail when using the local sheduler.
-