Purge non-verified user

[sjanssen2]

Handle #3416 first!

Addressing #3378 as well.
This PR is based on user creation_timestamp logging and offers a new Admin only page that lists those users that
a) are non yet verifier (i.e. user_status_id==5) and
b) where registers more than 30 days ago (give a new user sufficient time to react)
Individual or all users can be selected in the website and removed by one click.

[antgonza]

@sjanssen2, could you pull from dev and update this branch? Thank you.

[antgonza]

One comment that I completely missed in my original review. Thanks.

[antgonza]

Sorry for not catching this before but could you move these lines to patches/test_db_sql/92.sql. The issue is that if we leave them here they will also be inserted in the main deployment but in patches/test_db_sql they will only be applied to the test environment.

[antgonza]

FWIW, I think this is the source of the error.

[sjanssen2]

I changed that, makes total sense to me. I also know a guy who complained about test data leftovers in a fresh production DB ;-)

[sjanssen2]

I am a bit puzzled about the current DB error raised in the tests. It looks to me like the portals test does not clean up properly for tearDown. In my attempt to reproduce this issue locally, all tests passed when executed once, but re-running those caused the same errors as in github. However, I wonder why this becomes an issue now and not already earlier. Any ideas @antgonza ?

[ackermag]

1634 Edgemont Avenue - the address of one of the original donors to UBiome - haha

…

On Thu, Jun 20, 2024 at 5:52 AM Stefan Janssen ***@***.***> wrote: I am a bit puzzled about the current DB error raised in the tests. It looks to me like the portals test does not clean up properly for tearDown. In my attempt to reproduce this issue locally, all tests passed when executed once, but re-running those caused the same errors as in github. However, I wonder why this becomes an issue now and not already earlier. Any ideas @antgonza <https://github.com/antgonza> ? — Reply to this email directly, view it on GitHub <#3417 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAPL3AFLGZZDQRD2IBQMWDTZILF7RAVCNFSM6AAAAABJSGQRXWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOBQGU4TKMBXGY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Gail Ackermann Lab Co-ordinator Knight Lab UCSD 303-889-9763 ***@***.*** ***@***.***>

[sjanssen2]

I am a bit puzzled about the current DB error raised in the tests. It looks to me like the portals test does not clean up properly for tearDown. In my attempt to reproduce this issue locally, all tests passed when executed once, but re-running those caused the same errors as in github. However, I wonder why this becomes an issue now and not already earlier. Any ideas @antgonza ?

I might got the problem. Adding three new test users caused more iterations of this loop

qiita/qiita_db/portal.py

Lines 89 to 94 in 8b3f717

	FOR eml IN
	SELECT email FROM qiita.qiita_user
	LOOP
	INSERT INTO qiita.analysis
	(email, name, description, dflt)
	VALUES (eml, eml || '-dflt', 'dflt', true)

when creating a portal. Thus the assertion

qiita/qiita_db/test/test_portal.py

Line 51 in 8b3f717

self.assertCountEqual(obs, exp)

failed and subsequent code to clean up the new portal was never executed, leading to follow up fails.

[coveralls]

coverage: 92.814% (-0.004%) from 92.818%
when pulling 71b97f3 on jlab:purge_nonverified_user
into 8b3f717 on qiita-spots:dev.