feat: variable printing tool of data_science coder testing (#658)

* update debug_info printing

* refine infomation printing

* update ensemble evaluator prompt

* substitute repr with reprlib.repr

* strip comments

rdagent/components/coder/data_science/ensemble/eval_tests/ensemble_test.txt

@@ -70,8 +70,31 @@ for key in val_preds_dict.keys():
 
 print(f"val_y.shape: {val_y.shape}" if not isinstance(val_y, list) else f"val_y(list)'s length: {len(val_y)}")
 
+import sys
+import reprlib
+def debug_info_print(func):
+    aRepr = reprlib.Repr()
+    aRepr.maxother=300
+    def wrapper(*args, **kwargs):
+        def local_trace(frame, event, arg):
+            if event == "return" and frame.f_code == func.__code__:
+                print("\n" + "="*20 + "Running ensemble code, local variable values:" + "="*20)
+                for k, v in frame.f_locals.items():
+                    printed = aRepr.repr(v)
+                    print(f"{k}:\n {printed}")
+                print("="*20 + "Local variable values end" + "="*20)
+            return local_trace
+        
+        sys.settrace(local_trace)
+        try:
+            return func(*args, **kwargs)
+        finally:
+            sys.settrace(None)
+    return wrapper
+
+
 # Run ensemble
-final_pred = ensemble_workflow(test_preds_dict, val_preds_dict, val_y)
+final_pred = debug_info_print(ensemble_workflow)(test_preds_dict, val_preds_dict, val_y)
 
 print_preds_info("ensemble", "test", final_pred)
 

rdagent/components/coder/data_science/ensemble/prompts.yaml

@@ -95,6 +95,7 @@ ensemble_eval:
     ## Evaluation Criteria
     - You will be given the standard output (`stdout`) from the ensemble test and, if applicable, the workflow test.
     - Code should have no try-except blocks because they can hide errors.
+    - The stdout includes the local variable values from the ensemble code execution. Check whether the validation score is calculated correctly.
     
     Please respond with your feedback in the following JSON format and order
     ```json

rdagent/components/coder/data_science/feature/eval_tests/feature_test.txt

@@ -23,7 +23,29 @@ print(f"test_ids length: {len(test_ids)}")
 X_loaded = deepcopy(X)
 y_loaded = deepcopy(y)
 X_test_loaded = deepcopy(X_test)
-X, y, X_test = feat_eng(X, y, X_test)
+
+import sys
+import reprlib
+def debug_info_print(func):
+    aRepr = reprlib.Repr()
+    aRepr.maxother=300
+    def wrapper(*args, **kwargs):
+        def local_trace(frame, event, arg):
+            if event == "return" and frame.f_code == func.__code__:
+                print("\n" + "="*20 + "Running feat_eng code, local variable values:" + "="*20)
+                for k, v in frame.f_locals.items():
+                    printed = aRepr.repr(v)
+                    print(f"{k}:\n {printed}")
+                print("="*20 + "Local variable values end" + "="*20)
+            return local_trace
+        
+        sys.settrace(local_trace)
+        try:
+            return func(*args, **kwargs)
+        finally:
+            sys.settrace(None)
+    return wrapper
+X, y, X_test = debug_info_print(feat_eng)(X, y, X_test)
 
 
 def get_length(data):

rdagent/components/coder/data_science/model/eval_tests/model_test.txt

@@ -25,10 +25,32 @@ print(f"train_y.shape: {train_y.shape}" if not isinstance(train_y, list) else f"
 print(f"val_X.shape: {val_X.shape}")
 print(f"val_y.shape: {val_y.shape}" if not isinstance(val_y, list) else f"val_y(list)'s length: {len(val_y)}")
 
+import sys
+import reprlib
+def debug_info_print(func):
+    aRepr = reprlib.Repr()
+    aRepr.maxother=300
+    def wrapper(*args, **kwargs):
+        def local_trace(frame, event, arg):
+            if event == "return" and frame.f_code == func.__code__:
+                print("\n" + "="*20 + "Running model training code, local variable values:" + "="*20)
+                for k, v in frame.f_locals.items():
+                    printed = aRepr.repr(v)
+                    print(f"{k}:\n {printed}")
+                print("="*20 + "Local variable values end" + "="*20)
+            return local_trace
+        
+        sys.settrace(local_trace)
+        try:
+            return func(*args, **kwargs)
+        finally:
+            sys.settrace(None)
+    return wrapper
+
 # First execution
 print("The first execution begins.\n")
 start_time = time.time()
-val_pred, test_pred, hypers = model_workflow(
+val_pred, test_pred, hypers = debug_info_print(model_workflow)(
     X=train_X,
     y=train_y,
     val_X=val_X,
@@ -40,7 +62,7 @@ log_execution_results(start_time, val_pred, test_pred, hypers, "The first execut
 # Second execution
 print("The second execution begins.\n")
 start_time = time.time()
-val_pred, test_pred, final_hypers = model_workflow(
+val_pred, test_pred, final_hypers = debug_info_print(model_workflow)(
     X=train_X,
     y=train_y,
     val_X=None,

rdagent/components/coder/data_science/raw_data_loader/eval_tests/data_loader_test.txt

@@ -12,7 +12,29 @@ import pickle
 import pandas as pd
 from load_data import load_data
 
-X, y, X_test, test_ids = load_data()
+import sys
+import reprlib
+def debug_info_print(func):
+    aRepr = reprlib.Repr()
+    aRepr.maxother=300
+    def wrapper(*args, **kwargs):
+        def local_trace(frame, event, arg):
+            if event == "return" and frame.f_code == func.__code__:
+                print("\n" + "="*20 + "Running data_load code, local variable values:" + "="*20)
+                for k, v in frame.f_locals.items():
+                    printed = aRepr.repr(v)
+                    print(f"{k}:\n {printed}")
+                print("="*20 + "Local variable values end" + "="*20)
+            return local_trace
+        
+        sys.settrace(local_trace)
+        try:
+            return func(*args, **kwargs)
+        finally:
+            sys.settrace(None)
+    return wrapper
+
+X, y, X_test, test_ids = debug_info_print(load_data)()
 
 
 def get_length(data):