feat: download media in bounded domain batches

tests/test_media_cli.py

@@ -1,7 +1,8 @@
 import json
 from pathlib import Path
 
-from rf4_research.media_cli import media_queue_plan
+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:
@@ -26,6 +27,7 @@ def test_queue_plan_prioritizes_entity_gaps_and_respects_domain_cooldown(tmp_pat
     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
@@ -33,3 +35,50 @@ def test_queue_plan_prioritizes_entity_gaps_and_respects_domain_cooldown(tmp_pat
     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"]