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
@@ -107,5 +107,20 @@ def test_media_coverage_keeps_unknown_tackle_total_honest(tmp_path: Path) -> Non
        encoding="utf-8",
    )
    report = media_coverage(tmp_path)["entities"]
-    assert report["fish"] == {"expected": 2, "candidates": 2, "approved": 1, "candidate_gap": 0, "approved_gap": 1}
+    assert report["fish"] == {
+        "expected": 2, "candidate_assets": 2, "unique_candidates": 0,
+        "unidentified_assets": 2, "approved": 1, "candidate_gap": 2, "approved_gap": 1,
+    }
    assert report["tackle"]["candidate_gap"] is None
+
+
+def test_media_coverage_deduplicates_candidate_labels(tmp_path: Path) -> None:
+    (tmp_path / "manifest.json").write_text(
+        '{"assets":[{"entity_type":"fish","status":"queued","label":" Ерш "},'
+        '{"entity_type":"fish","status":"queued","label":"ерш"}]}', encoding="utf-8",
+    )
+    (tmp_path / "catalog-baseline.json").write_text(
+        '{"verified_at":"2026-09-12","entities":{"fish":{"count":2}}}', encoding="utf-8",
+    )
+    fish = media_coverage(tmp_path)["entities"]["fish"]
+    assert (fish["candidate_assets"], fish["unique_candidates"], fish["candidate_gap"]) == (2, 1, 1)