chore: audit RF4 media quality

rf4_research/media_assets.py

@@ -251,6 +251,45 @@ def media_coverage(root: Path) -> dict:
     return {"baseline_date": baseline["verified_at"], "entities": result}
 
 
+def media_quality_report(root: Path, *, minimum_dimension: int = 256, display_dimension: int = 180) -> dict:
+    """Report published raster quality and known higher-quality source alternatives offline."""
+    manifest = json.loads((root / "manifest.json").read_text(encoding="utf-8"))
+    assets = manifest.get("assets", [])
+    approved = [item for item in assets if item.get("status") == "approved" and item.get("entity_type") == "fish"]
+    low = [item for item in approved if min(int(item.get("width") or 0), int(item.get("height") or 0)) < minimum_dimension]
+    upsampled = [item for item in approved if min(int(item.get("width") or 0), int(item.get("height") or 0)) < display_dimension]
+    low_urls = {item.get("asset_url") for item in low}
+    alternatives = [
+        item for item in assets
+        if item.get("entity_type") == "fish"
+        and item.get("status") in {"duplicate", "queued"}
+        and (item.get("duplicate_of") in low_urls or item.get("status") == "queued")
+    ]
+
+    by_source: dict[str, dict[str, int]] = {}
+    for item in approved:
+        host = urlsplit(str(item.get("asset_url") or "")).hostname or "unknown"
+        summary = by_source.setdefault(host, {"published": 0, "below_minimum": 0, "upsampled_in_cards": 0})
+        summary["published"] += 1
+        summary["below_minimum"] += item in low
+        summary["upsampled_in_cards"] += item in upsampled
+
+    return {
+        "entity_type": "fish",
+        "minimum_dimension": minimum_dimension,
+        "display_dimension": display_dimension,
+        "published": len(approved),
+        "below_minimum": len(low),
+        "upsampled_in_cards": len(upsampled),
+        "known_alternative_urls": len(alternatives),
+        "by_source": by_source,
+        "examples": [
+            {"label": item.get("label"), "width": item.get("width"), "height": item.get("height"), "asset_url": item.get("asset_url")}
+            for item in sorted(low, key=lambda row: str(row.get("label") or ""))[:20]
+        ],
+    }
+
+
 def inspect_image(body: bytes) -> tuple[int, int, str]:
     try:
         with Image.open(io.BytesIO(body)) as image: