Make ParadeDB search typo-tolerant via configurable fuzzy distance and document the new setting. Add coverage for fuzzy query SQL and id parsing, and keep integration test stability under connection load.
Re-apply public Create visibility checks when resolving backend hits, improve UUID/FlakeId id parsing, and add tests for poisoned hits and UUID-string ids. Also tighten the docker integration test with localhost-only binding and config restoration.
Use uuid ids in the ParadeDB documents table to satisfy BM25 key_field requirements and adjust indexing/search to dump FlakeIds consistently. Also fix the ParadeDB backend unit test to avoid invalid pinning on struct fields.
Remove warnings that break --warnings-as-errors, add TCP readiness check, and start the ParadeDB Repo with a small pool to reduce connection churn while booting the dockerized backend.
Add a skipped-by-default integration test that boots a ParadeDB container, creates the BM25 index, indexes posts, and verifies search results against the real backend.
Introduce an optional search backend that indexes public/unlisted Notes into a separate ParadeDB (pg_search) Postgres and queries it for status search results.
Drop follow_redirect/force_redirect from the HTTP options used when warming MediaProxy, relying on Tesla middleware instead (Hackney redirect handling can crash behind CONNECT proxies).
Also add a regression assertion in the policy test and document the upstream Hackney issues in ReverseProxy redirect handling.
Exercises Pleroma.ReverseProxy.Client.Hackney with follow_redirect enabled behind an HTTPS CONNECT proxy, ensuring the client follows a relative redirect and can stream the final body.
Also ensure we always pass an absolute URL to Hackney when parsing a redirect response
(cherry picked from commit 00ac6bce8d244eec7e2460358296619e5cacba6b)
Hackney 1.25 crashes when follow_redirect is enabled behind an HTTPS CONNECT proxy and the Location header is relative (hackney_http_connect transport).
This test demonstrates the failure and verifies Tesla-level redirects work when hackney redirects are disabled.
Reproduces the Hackney 1.25 pooled redirect cleanup issue which can surface as :req_not_found when the adapter returns a Ref and the body is later fetched.
Hackney 1.25.x has redirect handling issues behind CONNECT proxies and with pools.
Disable hackney-level redirects and rely on Tesla.Middleware.FollowRedirects instead.
Also default to with_body: true so redirects can be followed reliably.
