Fixed with D7748

Fix commit messsage as @vlorentz requested

As I suspected, api_swhid_known in swh/web/api/views/identifiers.py does some keying by hashes, and that logic is not quite right :).

Ah. and I guess there needs to be another round of tagging and whatnot to trigger a website redeploy?

(Post merge, the diff that shows up is just the first commit, but they all are on master.)

Eh, I'll just assume, given the timezone latency, that going down to one review was intentional. (With the directory algorithm factored out into swh-storage, this patch is now simpler and more "just do the obvious thing" too.)

@vlorentz is anyone else (@anlambert?) going to review, or should I push this once it passes again?

Rebase as master changed since

Use new swh.storage.algos.directory.directory_get

Update commit message in the way @anlambert requested

Oh awesome! Thank so much!!

Do you think it would be possible to land this, and a version of the main one that depends on this, and then circle back to add the additional tests for malformed objects?

Include raw manifest in directory object

(Mark notes to fix copyright as done)

Fix copyright dates

In D7192#201166, @vlorentz wrote:

In D7192#201132, @Ericson2314 wrote:

Do you all know what is causing these CI failures?

just Cypress tests timing out as usual

Kick off CI again after I saw swh-model's wor

Do you all know what is causing these CI failures?

Push same stuff again to restart job

I'll wait for now pending @anlambert review, but let me know if I should push again to to trigger a new build after the last one seems to have died spuriously (if I am not misunderstanding).

Remove dead code, remove 400 from docs

It would seem I did a no-op revise before

Remove unused imports

Address some review comments

It works! Clean up commit history and stop filtering the tests

Try to fix without with_data

Try to fix content test with with_data

Try again to fix tests

Fix import of snapshot_get_all_branches

Fix some issues, use @vlorentz's suggestion to vastly simplify!

@vlorentz Thanks! That is much nice!

My version of black was too old!

Try to fix tests

Fix typo, I wasn't running flake8 on enough files!

• Follow @vlorentz's advice to hopfully get mocking to work

Reuse storage from archive

Use storage directly rather than archive

Fix flake8 errors, fix messed up imports

Try new swh_model, unparse object_id

Fix url_args keys

-k in pytest.ini

Use git-style not phab-style commit message

@vlorentz do you mean repeat the title in the summary field?

On a diferrent note, once D7653 lands is there any synchronization work, or can the next version of this just start using content_git_object right away?

Make URL param name and python param name align

Yeah I am afraid I haven not yet gotten the container setup instructions from @plt-amy (we are both on NixOS) and so I am flying blind (ab)using CI. I was already planning on doing that today, but in the meantime I might see if I can temp change CI just to run the one test :).

Ah thanks so much, @vlorentz! I was even looking for something along those lines, but just totally failed to find it.

Add swh/web/api/urls.py import, fix positional param name

django.urls.exceptions.NoReverseMatch: Reverse for 'api-1-raw-object' not found. 'api-1-raw-object' is not a valid view function or pattern name.

Fix flake8 asking for more blank lines

Fix api_route decoration, factor out SWHID_RE

Fix typo

Fix typo

Tweak some things

BTW, is there a way to convert one of these back to a "draft" state?

Fix flake8 issues

Fix CI failures

Try adding tests, simplify implementation

Thanks @vlorentz for all your advice and help. I think this is good now!

I suppose a next step is testing this?

Fix missing imports

Convert to follow mailing list plan

So in https://github.com/obsidiansystems/go-ipfs-swh-plugin/blob/66272b4a7d903171e2a1dd3d75309b3c1cad5ece/bridge/bridge.go#L119-L155 you can see how we are using the content route today. Hopefully seeing some working code makes things clearer!