Just a couple of nitpicks.

Could you add a short docstring to the test, explaining the same thing as the diff's description?

Could you add a test for this? (eg. using Cypress to mock calls to the counter API)

DULWICH_TARGET_TYPES are already the right enum values.

This is going in the right direction, but I don't like the CassObject thing. I think you made it because you are conflating two uses of my "Frankenstein's model objects":

yay

This remove test coverage of swh.storage.converters.date_to_db.

Why didn't mypy catch this?

You should rewrite all the object-generating strategies:

In D2863#70027, @DanSeraf wrote:

Thanks! I will check a solution for that; anyways, i didn't have problem testing directories that have the same name, could you provide me an example, so i can replicate your test?

Sorry I didn't do it earlier, but I finally got around to actually running this code, and I found a bug: if two directories have the same name (or if there's one named "total"), then the swh-scanner process exits early, and nothing is shown in the browser.

(Landing the diff nonetheless, because the build fails because of a pending change by @ardumont)

Note: this is needed because we will make this argument mandatory

This time I reviewed plot.py.

It tests for the missing url, so it's still relevant.

Current plan:

remove annotation.

In D2872#69251, @olasd wrote:

I think there's general value in keeping exception arguments inambiguous and human-readable.

Having to fish out and copy/paste a repr'd bytes object is really annoying when you want to quickly check whether something is in the database or not, for instance. Also, I don't trust our exception handling utilities (be that logging, sentry, serialization in the RPC protocol, ...) to handle arbitrary bytes unharmed.

I understand this is to work around the sentry issue, but I think it's a bad idea. We don't want parts of the API to use hex encoding while everything else uses bytes.

Needs a test, though

fix docstring.

As the cassandra cluster is paused indefinitely, I'm landing this diff now

rebase

see mandatory fields for citation

Now it does

I do like the idea of adding a role to author, but the way schema.org is organized, is by adding properties with specific roles

Taxon map: