src/lightspeed_evaluation/core/system/validator.py (1)

468-472: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Duplicate canonicalization logic.

The canonical = DEPRECATED_METRIC_NAMES.get(metric, metric) + dual-membership check is repeated verbatim for turn-level and conversation-level validation. Consider extracting a small helper, e.g. _is_metric_known(metric, known_metrics), to avoid drift if the logic changes later.

♻️ Proposed refactor

+    `@staticmethod`
+    def _is_metric_known(metric: str, known_metrics: set[str]) -> bool:
+        """Check if a metric (or its canonical alias) is in the known set."""
+        canonical = DEPRECATED_METRIC_NAMES.get(metric, metric)
+        return metric in known_metrics or canonical in known_metrics
+
     def _validate_metrics_availability(self, data: EvaluationData) -> None:
         ...
         for turn_data in data.turns:
             if turn_data.turn_metrics:
                 for metric in turn_data.turn_metrics:
-                    canonical = DEPRECATED_METRIC_NAMES.get(metric, metric)
-                    if (
-                        metric not in self._turn_level_metrics
-                        and canonical not in self._turn_level_metrics
-                    ):
+                    if not self._is_metric_known(metric, self._turn_level_metrics):
                         turn_data.add_invalid_metric(metric)
                         ...

Also applies to: 482-486

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/lightspeed_evaluation/core/system/validator.py` around lines 468 - 472,
The turn-level and conversation-level metric checks in validator.py duplicate
the same canonicalization and membership logic. Extract that repeated logic into
a small helper such as _is_metric_known(metric, known_metrics) in Validator, and
have both validation branches call it instead of inlining
DEPRECATED_METRIC_NAMES.get(...) and the dual-membership check. Keep the helper
generic so both _turn_level_metrics and the conversation-level known metrics use
the same source of truth and future changes stay consistent.