feat: expand RF4 media inventory

2026-09-13 08:06:48 +07:00
parent 4042c80f00
commit 674bc5627f
9 changed files with 5054 additions and 19 deletions
@@ -34,6 +34,14 @@ def test_gallery_page_provenance_classifies_generic_asset_urls() -> None:
    assert chrome[0].entity_type == "reference"


+def test_rf4map_gateway_fish_filename_is_classified_as_fish() -> None:
+    items = extract_media_candidates(
+        '<img src="https://gw.rf4map.ru/public/images/fish_123.webp" alt="Ерш">',
+        source_page="https://rf4map.ru/fishes",
+    )
+    assert items[0].entity_type == "fish"
+
+
 def test_manifest_merges_and_binary_store_is_content_addressed(tmp_path: Path) -> None:
    item = extract_media_candidates('<img src="/bait/worm.png" alt="Bait worm">', source_page="https://example.test")[0]
    first = merge_manifest(tmp_path / "manifest.json", [item])