Merge nucleic/plucky-north-vole-sdna into dev

label_swe_chat_prompts.py

@@ -306,6 +306,16 @@ def _append(path: Path, states: Sequence[dict[str, Any]]) -> None:
         os.fsync(handle.fileno())
 
 
+def record_from_decision(candidate: Candidate, decision: dict[str, Any]) -> dict[str, Any]:
+    """Construct the exact seven-field student record from a teacher decision."""
+    record = {field: decision[field] for field in SOURCE_FIELDS - {"prompt"}}
+    record["prompt"] = candidate.prompt
+    if not decision["recoverableFromFirst"]:
+        record["slice"] = "vague-eval"
+    validate_source_record(record, candidate.id)
+    return record
+
+
 def run(args: argparse.Namespace) -> dict[str, int]:
     lines = base.source_lines(args.input)
     source = candidates(lines)
@@ -335,10 +345,7 @@ def run(args: argparse.Namespace) -> dict[str, int]:
         newly: list[dict[str, Any]] = []
         for candidate, decision in decisions:
             if decision["keep"]:
-                record = {field: decision[field] for field in SOURCE_FIELDS}
-                if not decision["recoverableFromFirst"]:
-                    record["slice"] = "vague-eval"
-                validate_source_record(record, candidate.id)
+                record = record_from_decision(candidate, decision)
                 newly.append(_state(candidate, status="labeled", record=record, reason=None, recoverable=decision["recoverableFromFirst"]))
             else:
                 newly.append(_state(candidate, status="rejected", record=None, reason=f"semantic_junk:{decision['junkReason'].strip()}", recoverable=None))