fix some vscode eval problems

desktop_env/evaluators/metrics/vscode.py

@@ -1,5 +1,5 @@
 from typing import Dict
-import json
+import json, copy
 
 def check_json_keybindings(actual: str, expected: str, **options) -> float:
     """
@@ -10,36 +10,58 @@ def check_json_keybindings(actual: str, expected: str, **options) -> float:
     Return:
         float: the score
     """
+    def direct_load_json(fp):
+        try:
+            with open(fp, 'r') as f:
+                data = json.load(f)
+            return data
+        except:
+            return None
+    
+    def skip_first_line_load_json(fp):
+        try:
+            with open(fp, 'r') as f:
+                f.readline()
+                data = json.load(f)
+            return data
+        except:
+            return None
 
-    with open(actual) as f:
-        data = json.load(f)
-
+    for func in [direct_load_json, skip_first_line_load_json]:
+        data = func(actual)
+        if data is not None and type(data) == list:
+            break
+    else:
+        return 0.0
+    expected = expected['expect']
     if expected in data:
         return 1.0
     else:
         return 0.0
 
+
 def check_json_settings(actual: str, expected: str, **options) -> float:
     """
     Args:
         actual (str): path to result text file
-        expected (str): expected dict{}
+        expected (dict): expected dict{}, containing key "expect"
 
     Return:
         float: the score
     """
 
-    with open(actual) as f:
+    with open(actual, 'r') as f:
         data = json.load(f)
-        
-    expect = set(expected.items())
-    json = set(data.items())
-
-    if expect.issubset(json):
+    
+    expect = expected['expect']
+    data_copy = copy.deepcopy(data)
+    data_copy.update(expect)
+    if data == data_copy:
         return 1.0
     else:
         return 0.0
 
+
 def compare_text_file(actual: str, expected: str, **options) -> float:
     """
     Args: