fix: count unique media coverage

2026-09-13 08:22:10 +07:00
parent 7e464ef598
commit 0b7df7f721
4 changed files with 27 additions and 5 deletions
@@ -169,13 +169,20 @@ def media_coverage(root: Path) -> dict:
    result: dict[str, dict] = {}
    for entity_type, target in baseline["entities"].items():
        candidates = [item for item in manifest.get("assets", []) if item.get("entity_type") == entity_type]
+        candidate_labels = {
+            " ".join(item["label"].split()).casefold()
+            for item in candidates
+            if isinstance(item.get("label"), str) and item["label"].strip()
+        }
        approved_keys = {item.get("entity_key") for item in candidates if item.get("status") == "approved" and item.get("entity_key")}
        expected = target.get("count")
        result[entity_type] = {
            "expected": expected,
-            "candidates": len(candidates),
+            "candidate_assets": len(candidates),
+            "unique_candidates": len(candidate_labels),
+            "unidentified_assets": sum(1 for item in candidates if not isinstance(item.get("label"), str) or not item["label"].strip()),
            "approved": len(approved_keys),
-            "candidate_gap": max(expected - len(candidates), 0) if isinstance(expected, int) else None,
+            "candidate_gap": max(expected - len(candidate_labels), 0) if isinstance(expected, int) else None,
            "approved_gap": max(expected - len(approved_keys), 0) if isinstance(expected, int) else None,
        }
    return {"baseline_date": baseline["verified_at"], "entities": result}