T351642: Discard canary events in Hashtags

[theresnotime]

Discard canary events in Hashtags

Bug: T351642

[jsnshrmn]

does what it says on the tin; suggest sorting out ci for scripts dir tests separately, but this works for now.

[MusikAnimal]

I had to do some research, but it looks like if we add an empty __init__.py to the scripts/ directory, it will then be treated as a Python package and Django will look for the tests. However because the tests are ran from the root directory, it will think common is a top-level package when it's only part of scripts. The fix there seems to be to add a parent-relative import:

Suggested change

	from common import is_canary
	from .common import is_canary

Then docker compose exec app python manage.py test (and CI) should run the test. I think however the other files in scripts/ need the same treatment to import locally, which doesn't look a huge change:

• collect_hashtags.py – from common → from .common, import db → import .db
• load_old_db.py – from common → from .common

Should we go ahead and try that?

[MusikAnimal]

Note again I'm not very experienced with Python so this may not be the right way to go about it. If there's any disagreement, I'm happy to defer to @jsnshrmn and sort out running the tests in a separate PR.

[theresnotime]

The collector is started directly as python collect_hashtags.py, so Python doesn't treat it as part of a package, and from .common import ... only works inside a package so things would break if we did that :) getting the test working isn't a big change but might be worth following up with instead I think

[jsnshrmn]

Agreed; let's merge for now and consider this a coverage concern for future ci work.