agent-apprenticeship 0.1.9 → 0.2.0

package/src/agent_apprenticeship_trace/trace_normalizer.py

@@ -109,6 +109,30 @@ def _input_source(value: Any, meta: dict[str, Any]) -> dict[str, Any] | None:
         meta["original_input_source"] = value
     return None
 
+def _actor_to_string(value: Any, action: str, meta: dict[str, Any]) -> str:
+    if action == "user_message":
+        return "user"
+    if isinstance(value, str) and value.strip():
+        return value.strip()
+    if isinstance(value, dict):
+        meta["original_actor"] = value
+        actor_kind = str(value.get("actor_kind") or value.get("kind") or "").strip()
+        role = str(value.get("role") or value.get("actor_role") or value.get("name") or "").strip()
+        agent_id = str(value.get("agent_id") or value.get("id") or "").strip()
+        if actor_kind and role:
+            return f"{actor_kind}:{role}"
+        if actor_kind and agent_id:
+            return f"{actor_kind}:{agent_id}"
+        if actor_kind:
+            return actor_kind
+        if role:
+            return role
+        if agent_id:
+            return f"agent:{agent_id}"
+    elif value is not None:
+        meta["original_actor"] = value
+    return "agent:worker"
+
 def _infer_environment_domain(tool: Any, operation: Any = None, input_value: Any = None, output_value: Any = None) -> str:
     text = " ".join(str(x or "") for x in [tool, operation, input_value, output_value]).lower()
     if any(x in text for x in ["mcp__", "mcp ", "model context protocol"]):
@@ -714,8 +738,7 @@ def _normalize_step(raw_step: Any, ordinal: int, used_steps: set[int], report: T
     if warnings:
         meta['normalization_warnings']=warnings; report.warnings.extend(warnings)
 
-    actor=raw.get('actor') or ('user' if action == 'user_message' else 'agent:worker')
-    if action == 'user_message': actor='user'
+    actor=_actor_to_string(raw.get('actor'), action, meta)
     raw_execution_mode = raw.get('execution_mode') if raw.get('execution_mode') in {'serial','parallel'} else None
     execution_mode = None if action == 'user_message' else (raw_execution_mode or 'serial')
     parallel_group = _string_or_none(raw.get('parallel_group'), meta, 'parallel_group') if execution_mode == 'parallel' else None