(I forgot to add the logic for ://" character sequencemangled into ":, fixing it now)

I think it would be better to handle that process directly in the Archive.get_origin function, adding an utility function for a small set of instructions seems a bit overkill.

Rebase

More code simplification

Simplify some code

Rebase and remove redundant sorted call

I think the proper fix would be to add a specific index to the origin_visit table in PostgreSQL database (see T4198).

Rebase

Address @vlorentz comments

In D7976#207263, @vlorentz wrote:

why don't we also use this module for ZIP files?

Alternatively, we could use the zipfile standard Python module which seems to work in a similar way as the jar command, see below:

(swh) anlambert@carnavalet:/tmp/jar_test$ wget https://repo1.maven.org/maven2/org/pustefixframework/pustefix-archetype-basic/0.15.20/pustefix-archetype-basic-0.15.20-sources.jar
--2022-06-09 14:56:17--  https://repo1.maven.org/maven2/org/pustefixframework/pustefix-archetype-basic/0.15.20/pustefix-archetype-basic-0.15.20-sources.jar
Resolving repo1.maven.org (repo1.maven.org)... 151.101.120.209
Connecting to repo1.maven.org (repo1.maven.org)|151.101.120.209|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 45637 (45K) [application/java-archive]
Saving to: ‘pustefix-archetype-basic-0.15.20-sources.jar’

Looks good to me.

Remove debian folder staged by mistake.

In D7971#207146, @vlorentz wrote:

Why bother with swh-indexer 1.x? updating prod should be good enough

and I don't think you meant to stage debian/

Address @vlorentz comments

Looks good, thanks !

Remove file added by mistake

Bump swh-indexer in requirements-swh.txt

Rebase

In D7960#206850, @ardumont wrote:

ouch!

@vlorentz some impact from stopping the concurrency build part ¯\_(ツ)_/¯

In D7947#206811, @ardumont wrote:

/me closes his eyes and accepts the diff ¯\_(ツ)_/¯

Looks good to me. I will submit another diff to ensure the query_params parameter of the reverse function is of type Dict[str, Optional[str]] in order to avoid that kind of issues.

I also spotted the issue and could reproduce locally, looks like it is related to latest hypothesis release that makes some of our tests became flaky. I just pushed D7948 as a workaround to make them pass again.

Looks good to me.

In D7938#206258, @vlorentz wrote:

Can query_params[k] really be not a string? why?

Address @ardumont comment.