Merge branch 'main' of https://github.com/xlang-ai/DesktopEnv

desktop_env/evaluators/metrics/docs.py

@@ -13,6 +13,7 @@ from odf.text import P
 from odf.text import Span
 from skimage.color import deltaE_ciede2000
 from skimage.color import rgb2lab
+from fuzzywuzzy import fuzz
 
 logger = logging.getLogger("desktopenv.metric.docs")
 
@@ -57,6 +58,8 @@ def contains_page_break(docx_file):
 
 def compare_docx_files(file1, file2, **options):
     ignore_blanks = options.get('ignore_blanks', True)
+    content_only = options.get('content_only', False)
+
     def get_paragraph_texts_odt(document):
         paragraphs = document.getElementsByType(P)
         paragraph_texts = []
@@ -89,6 +92,13 @@ def compare_docx_files(file1, file2, **options):
         print("Unsupported file types or mismatch between file types.")
         return 0
 
+    if content_only:
+        # Compare the content of the documents
+        text1 = re.sub(r'\s+', ' ', '\n'.join(doc1_paragraphs)).strip()
+        text2 = re.sub(r'\s+', ' ', '\n'.join(doc2_paragraphs)).strip()
+        similarity = fuzz.ratio(text1, text2) / 100.0
+        return similarity
+
     # Process and compare documents
     if ignore_blanks:
         text1 = re.sub(r'\s+', ' ', '\n'.join(doc1_paragraphs)).strip()

desktop_env/evaluators/metrics/vscode.py

@@ -1,5 +1,7 @@
+import copy
+import json
 from typing import Dict
-import json, copy
+
 
 def check_json_keybindings(actual: str, expected: str, **options) -> float:
     """
@@ -10,6 +12,7 @@ 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:
@@ -17,7 +20,7 @@ def check_json_keybindings(actual: str, expected: str, **options) -> float:
             return data
         except:
             return None
-    
+
     def skip_first_line_load_json(fp):
         try:
             with open(fp, 'r') as f:
@@ -54,7 +57,7 @@ def check_json_settings(actual: str, expected: str, **options) -> float:
 
     with open(actual, 'r') as f:
         data = json.load(f)
-    
+
     expect = expected['expected']
     data_copy = copy.deepcopy(data)
     data_copy.update(expect)
@@ -128,3 +131,11 @@ def is_extension_installed(actual: str, rules: Dict, **options):
         return 0.0
     else:
         raise NotImplementedError
+
+
+def check_python_file_by_test_suite(actual_file, test_suites, **options) -> float:
+    pass
+
+
+def check_python_file_by_gold_file(actual_file, gold_file: str, **options) -> float:
+    pass