"""Patreon API contract test (native ingester build step 4). The drift tripwire the plan calls for: if the field-set we SEND or the parser we resolve responses against drifts from what real Patreon returns, this build goes RED — instead of the ingester silently importing nothing in production (the exact failure mode the native ingester exists to escape). Two halves: - the recorded `/api/posts` fixture must still parse end-to-end (no drift), and - the request params must still carry every field the parser depends on (the fixture already has the data, so trimming the request would NOT trip the parse half — this half guards the request shape directly). Pure: no network, no DB. """ import json from pathlib import Path from backend.app.services.patreon_client import MediaItem, PatreonClient _FIXTURE = Path(__file__).parent / "fixtures" / "patreon_posts_page1.json" # Pinned media total across the 4 fixture posts: 1001→2 (gallery, image_large # cover deduped), 1002→1 (attachment+postfile collapse), 1003→1 (inline content # img), 1004→1 (video postfile). A parser change that drops or doubles media # trips this. _EXPECTED_MEDIA_TOTAL = 5 _KNOWN_KINDS = {"images", "image_large", "attachments", "postfile", "content"} def test_recorded_fixture_parses_end_to_end(): response = json.loads(_FIXTURE.read_text()) client = PatreonClient(cookies_path=None) client._validate_response(response) # top-level shape must hold index = client._transform(response) total = 0 for post in response["data"]: items = client.extract_media(post, index) # must not raise drift for m in items: assert isinstance(m, MediaItem) assert m.url, f"empty url in post {post['id']}" assert m.filename, f"empty filename in post {post['id']}" assert m.kind in _KNOWN_KINDS, f"unknown kind {m.kind!r}" assert m.post_id == post["id"] total += len(items) assert total == _EXPECTED_MEDIA_TOTAL def test_request_field_set_carries_parser_dependencies(): """The params we SEND must keep requesting the fields the parser resolves against. Trimming `fields[media]=file_name`, an `include` relationship, or a `fields[post]` key would make real responses parse to nothing — caught here, not by the fixture parse above.""" params = PatreonClient(cookies_path=None)._params("5555", None) media_fields = params["fields[media]"] assert "file_name" in media_fields assert "image_urls" in media_fields or "download_url" in media_fields includes = params["include"] for rel in ("images", "attachments_media", "media"): assert rel in includes, f"missing include relationship {rel}" post_fields = params["fields[post]"] # content_json_string is the CURRENT body field (Patreon nulled flat `content`, # #842); content kept for legacy posts. Both must stay requested. for field in ("content", "content_json_string", "post_file", "image"): assert field in post_fields, f"missing post field {field}" assert params["filter[campaign_id]"] == "5555"