import json from pathlib import Path from rf4_research import media_cli from rf4_research.media_cli import _download_batch, media_queue_plan def test_queue_plan_prioritizes_entity_gaps_and_respects_domain_cooldown(tmp_path: Path) -> None: root = tmp_path / "media" root.mkdir() (root / "catalog-baseline.json").write_text(json.dumps({ "verified_at": "2026-09-12", "entities": { "fish": {"count": 2}, "waterbody": {"count": 1}, "tackle": {"count": None}, }, }), encoding="utf-8") (root / "manifest.json").write_text(json.dumps({"version": 1, "assets": [ {"status": "queued", "asset_url": "https://rf4map.ru/tackle.png", "source_page": "https://rf4map.ru/fishes", "entity_type": "tackle", "label": "Катушка", "external_id": "3"}, {"status": "queued", "asset_url": "https://rf4map.ru/fish.png", "source_page": "https://rf4map.ru/fishes", "entity_type": "fish", "label": "Щука", "external_id": "2"}, {"status": "queued", "asset_url": "https://rf4map.ru/water.png", "source_page": "https://rf4map.ru/maps", "entity_type": "waterbody", "label": "Ладога", "external_id": "1"}, {"status": "queued", "asset_url": "https://rf4-posts.com/fish.png", "source_page": "https://rf4-posts.com/fishes", "entity_type": "fish", "label": "Ёрш", "external_id": "4"}, {"status": "approved", "asset_url": "https://rf4map.ru/approved.png", "source_page": "https://rf4map.ru/fishes", "entity_type": "fish", "label": "Окунь", "entity_key": "perch"}, ]}), encoding="utf-8") state = tmp_path / "state.json" state.write_text(json.dumps({"rf4map.ru": 1000}), encoding="utf-8") plan = media_queue_plan(root, state, now=2000) assert plan["network_requests"] == 0 assert plan["batch_limit_per_domain"] == 40 assert plan["queued_total"] == 4 assert plan["domains"]["rf4map.ru"]["ready"] is False assert plan["domains"]["rf4map.ru"]["retry_in_seconds"] == 800 assert plan["domains"]["rf4map.ru"]["queued_by_type"] == {"fish": 1, "tackle": 1, "waterbody": 1} assert plan["domains"]["rf4map.ru"]["next_asset"]["entity_type"] == "waterbody" assert plan["domains"]["rf4-posts.com"]["ready"] is True assert plan["exact_catalog_gaps_known"] is False def test_batch_reserves_once_per_domain_and_caps_each_window(tmp_path: Path, monkeypatch) -> None: root = tmp_path / "media" root.mkdir() assets = [ {"status": "queued", "asset_url": f"https://rf4map.ru/fish-{index}.png", "source_page": "https://rf4map.ru/fishes", "entity_type": "fish", "label": f"Рыба {index}"} for index in range(45) ] + [ {"status": "queued", "asset_url": f"https://rf4-posts.com/fish-{index}.png", "source_page": "https://rf4-posts.com/fishes", "entity_type": "fish", "label": f"Рыба {index}"} for index in range(2) ] (root / "manifest.json").write_text(json.dumps({"version": 1, "assets": assets}), encoding="utf-8") state = tmp_path / "state.json" monkeypatch.setattr(media_cli, "_attempt_asset", lambda *_: ("stored", None)) report = _download_batch(root, state, limit=40) assert report["attempted_total"] == 42 assert report["domains"]["rf4map.ru"]["attempted"] == 40 assert report["domains"]["rf4-posts.com"]["attempted"] == 2 assert set(json.loads(state.read_text())) == {"rf4map.ru", "rf4-posts.com"} cooling = _download_batch(root, state, limit=40) assert cooling["attempted_total"] == 0 assert all("cooldown is active" in item["skipped"] for item in cooling["domains"].values()) def test_batch_stops_domain_after_three_consecutive_invalid_assets(tmp_path: Path, monkeypatch) -> None: root = tmp_path / "media" root.mkdir() assets = [ {"status": "queued", "asset_url": f"https://rf4map.ru/bad-{index}.png", "source_page": "https://rf4map.ru/fishes", "entity_type": "fish", "label": f"Рыба {index}"} for index in range(10) ] (root / "manifest.json").write_text(json.dumps({"version": 1, "assets": assets}), encoding="utf-8") def invalid(_root, _path, _manifest, item): item["status"] = "invalid" return "invalid", ValueError("not an image") monkeypatch.setattr(media_cli, "_attempt_asset", invalid) report = _download_batch(root, tmp_path / "state.json", limit=10) assert report["attempted_total"] == 3 assert report["failed_total"] == 3 assert "3 consecutive failures" in report["domains"]["rf4map.ru"]["stopped_reason"]