Finish loading the vscode examples v1; Improve on the infra: Add accessibility tree into the observation; Add activate window function, etc

desktop_env/evaluators/metrics/__init__.py

@@ -13,4 +13,4 @@ from .vlc import is_vlc_playing, is_vlc_recordings_folder, is_vlc_fullscreen, co
 from .gimp import increase_saturation, decrease_brightness, check_file_exists, compare_triangle_positions
 from .general import check_csv, check_accessibility_tree, check_list, run_sqlite3
 from .thunderbird import check_thunderbird_prefs, check_thunderbird_filter
-from .vscode import compare_text_file, compare_config, compare_answer
+from .vscode import compare_text_file, compare_config, compare_answer, is_extension_installed

desktop_env/evaluators/metrics/vscode.py

@@ -10,6 +10,8 @@ def compare_text_file(actual: str, expected: str, **options) -> float:
     Return:
         float: the score
     """
+    if not actual:
+        return 0.
 
     with open(actual) as f1:
         actual_text = f1.read()
@@ -22,6 +24,9 @@ def compare_text_file(actual: str, expected: str, **options) -> float:
 
 
 def compare_config(actual: str, rules: Dict, **options) -> float:
+    if not actual:
+        return 0.
+
     with open(actual) as f1:
         actual_text = f1.read()
 
@@ -39,9 +44,24 @@ def compare_answer(actual: str, rules: Dict, **options) -> float:
     Return:
         float: the score
     """
+    if not actual:
+        return 0.
 
     if actual == rules['expect']:
         return 1.0
 
     # TODO: can use text embedding to get non-zero return
     return 0.0
+
+
+def is_extension_installed(actual: str, rules: Dict, **options):
+    if rules['type'] == 'contain':
+        if rules['expected'] in actual:
+            return 1.0
+        return 0.0
+    elif rules['type'] == 'not_contain':
+        if rules['expected'] not in actual:
+            return 1.0
+        return 0.0
+    else:
+        raise NotImplementedError