feat: reconcile RF4DB media catalog

tests/test_media_assets.py

@@ -1,11 +1,12 @@
 from pathlib import Path
+import json
 
 import io
 
 import pytest
 from PIL import Image
 
-from rf4_research.media_assets import audit_media_catalog, extract_media_candidates, inspect_image, media_coverage, merge_manifest, review_asset, store_asset
+from rf4_research.media_assets import audit_media_catalog, extract_media_candidates, inspect_image, media_coverage, merge_manifest, reconcile_queued_duplicates, review_asset, store_asset
 
 
 def test_extracts_and_classifies_unique_https_media() -> None:
@@ -42,6 +43,14 @@ def test_rf4map_gateway_fish_filename_is_classified_as_fish() -> None:
     assert items[0].entity_type == "fish"
 
 
+def test_rf4db_asset_filename_is_stable_external_id() -> None:
+    item = extract_media_candidates(
+        '<img src="https://oss.rf4db.com/game/fish/a.sleeper.webp" alt="Ротан">',
+        source_page="https://download.rf4db.com/ru/fishes",
+    )[0]
+    assert (item.entity_type, item.external_id) == ("fish", "a.sleeper")
+
+
 def test_tackle_path_wins_over_fish_word_in_product_name() -> None:
     items = extract_media_candidates(
         '<img src="https://cdn.example/bait/crab.png" alt="Краб и рыба 14">',
@@ -140,3 +149,20 @@ def test_media_coverage_deduplicates_candidate_labels(tmp_path: Path) -> None:
     )
     fish = media_coverage(tmp_path)["entities"]["fish"]
     assert (fish["candidate_assets"], fish["unique_candidates"], fish["candidate_gap"]) == (2, 1, 1)
+
+
+def test_reconcile_queue_preserves_source_but_skips_normalized_duplicate(tmp_path: Path) -> None:
+    path = tmp_path / "manifest.json"
+    path.write_text(json.dumps({"assets": [
+        {"entity_type": "fish", "label": "Ерш", "status": "stored", "asset_url": "https://a/fish.png"},
+        {"entity_type": "fish", "label": " ЁРШ ", "status": "queued", "asset_url": "https://b/fish.webp"},
+        {"entity_type": "fish", "label": "Щука", "status": "queued", "asset_url": "https://b/pike.webp"},
+    ]}), encoding="utf-8")
+
+    report = reconcile_queued_duplicates(path)
+    assets = json.loads(path.read_text(encoding="utf-8"))["assets"]
+
+    assert report == {"duplicates": 1, "queued": 1}
+    assert assets[1]["status"] == "duplicate"
+    assert assets[1]["duplicate_of"] == "https://a/fish.png"
+    assert assets[2]["status"] == "queued"