v1 of media_file GET endpoint.

Media files are used for inference outputs, file uploads, Zero Shot
files, etc. We don't have any of the basic CRUD operations around them
yet, so this is a start.

This endpoint will need more metadata - moderation, file provenance,
file details, etc. But this will do for now.
This commit is contained in:
Brandon Thomas committed 2023-10-13 02:38:34 -04:00
1 parent 63b4a3180f
commit 138057e105
8 files changed
+481 -4

No files matched your search

@@ -37,7 +37,6 @@ and status IN ('pending', 'started', 'complete_failure', 'attempt_failed')
group by maybe_creator_user_token, maybe_creator_username, creator_ip_address
order by attempts desc;
-- Detailed report on most recent jobs, ordered by worst performing.
-- TODO: Determine if the problem is in downloading models. Make sure the job timer doesn't include
-- sections where models / files get downloaded.
@@ -71,8 +70,8 @@ from (
m.token = jobs.maybe_input_source_token
where
jobs.status != 'pending'
AND jobs.created_at > NOW() - INTERVAL 2000 MINUTE
AND jobs.maybe_model_type IN ('sad_talker')
AND jobs.created_at > NOW() - INTERVAL 20 MINUTE
AND jobs.maybe_model_type IN ('so_vits_svc')
order by id desc
limit 200
) as t