fix: try with one single payload

evaluation_function/evaluation.py

@@ -1,69 +1,53 @@
 from typing import Any
-from lf_toolkit.evaluation import Result as LFResult, Params
-
-# note: this file is a temperary workaround, if the frontend -> backend communication succeed, fix this file
-
-from .schemas import FSA#, FSAFrontend
+from lf_toolkit.evaluation import Result as LFResult
+from .schemas import FSA
 from .schemas.result import Result
 from .correction import analyze_fsa_correction
 
-# def evaluation_function(
-#     payload: Any
-# ) -> LFResult:
-#   return LFResult(
-#     is_correct=False,
-#         feedback_items=[("error", f"{payload}")]
-#     )
-
 def validate_fsa(value: str | dict) -> FSA:
     if isinstance(value, str):
         return FSA.model_validate_json(value)
     return FSA.model_validate(value)
 
-def evaluation_function(
-    response: Any,
-    answer: Any,
-    params: Params,
-) -> LFResult:
+def evaluation_function(payload: Any) -> LFResult:
     """
     Evaluate a student's FSA response against the expected answer.
 
     Args:
-        response: Student's FSA (dict with states, alphabet, transitions, etc.), since frontend constriants, this is FSAFrontend
-        answer: Expected FSA still, FSAFrontend for the same reason
-        params: Extra parameters (e.g., require_minimal)
+        payload: dict with keys 'response', 'answer', 'params' (front-end may wrap everything)
 
     Returns:
-        LFResult with is_correct and feedback
+        LFResult
     """
     try:
-        # Parse FSAs from input
-        # student_fsa_ = FSAFrontend.model_validate(response)
-        # expected_fsa_ = FSAFrontend.model_validate(answer)
+        # Extract response/answer from the payload
+        raw_response = payload.get("response") or payload.get("params", {}).get("response")
+        raw_answer = payload.get("answer") or payload.get("params", {}).get("answer")
+        params = payload.get("params", {})
 
-        # student_fsa = student_fsa_.from_flattened()
-        # expected_fsa = expected_fsa_.from_flattened()
+        if raw_response is None or raw_answer is None:
+            raise ValueError("Missing response or answer in payload")
 
-        # as a temporary workaround we assume the response and answer are all valid json strings
-        student_fsa = validate_fsa(response)
-        expected_fsa = validate_fsa(answer)
+        # Parse FSAs
+        student_fsa = validate_fsa(raw_response)
+        expected_fsa = validate_fsa(raw_answer)
 
+        require_minimal = params.get("require_minimal", False)
 
-
-        # Get require_minimal from params if present
-        require_minimal = params.get("require_minimal", False) if hasattr(params, "get") else False
-
-        # Run correction pipeline
+        # Run correction
         result: Result = analyze_fsa_correction(student_fsa, expected_fsa, require_minimal)
-        
-        # Convert to lf_toolkit Result
+
+        # Convert to LFResult
         return LFResult(
             is_correct=result.is_correct,
             feedback_items=[("feedback", result.feedback)]
         )
-        
+
     except Exception as e:
         return LFResult(
             is_correct=False,
-            feedback_items=[("error", f"Invalid FSA format: {str(e)}, received: \n\nresponse: {response}\n\n answer: {answer}, \n\nparams: {params}")]
+            feedback_items=[(
+                "error",
+                f"Invalid FSA format: {str(e)}\n\npayload received:\n{payload}"
+            )]
         )