Create scorer.py

scorer.py

@@ -0,0 +1,52 @@
+from dataclasses import dataclass
+from typing import Dict, Any, List
+import re
+
+REQ = [
+    "failure_surface_type",
+    "subgroup_affected",
+    "intervention_route",
+    "recalibration_needed",
+    "protocol_reset_plan",
+    "escalation_priority",
+    "confidence_score",
+]
+
+@dataclass
+class ScoreResult:
+    score: float
+    details: Dict[str, Any]
+
+def _yesno(text: str, key: str) -> bool:
+    return bool(re.search(rf"{key}\s*[:=]\s*(yes|no)\b", text))
+
+def _float01(text: str, key: str) -> bool:
+    return bool(re.search(rf"{key}\s*[:=]\s*(0\.\d+|1\.0)\b", text))
+
+def score(sample: Dict[str, Any], prediction: str) -> ScoreResult:
+    p = (prediction or "").lower()
+    words_ok = len(p.split()) <= 1200
+
+    hits = sum(1 for k in REQ if k in p)
+
+    recal_ok = int(_yesno(p, "recalibration_needed"))
+    conf_ok = int(_float01(p, "confidence_score"))
+
+    priority_ok = int("escalation_priority" in p and any(x in p for x in [
+        "low", "medium", "high", "critical"
+    ]))
+
+    raw = (
+        0.20 * int(words_ok) +
+        0.55 * (hits / len(REQ)) +
+        0.10 * recal_ok +
+        0.10 * conf_ok +
+        0.05 * priority_ok
+    )
+
+    return ScoreResult(score=min(1.0, raw), details={"id": sample.get("id"), "hits": hits})
+
+def aggregate(results: List[ScoreResult]) -> Dict[str, Any]:
+    if not results:
+        return {"mean": 0.0, "n": 0}
+    return {"mean": sum(r.score for r in results)/len(results), "n": len(results)}