Update app.py

app.py

@@ -21,7 +21,7 @@ scheduler = CommitScheduler(
     repo_id="stabletoolbench/StableToolBench_data",  # 替换为你的用户名和数据集仓库名
     repo_type="dataset",
     folder_path=dataset_dir,
-    path_in_repo="leaderboard_data.json",
+    path_in_repo="data",
 )
 
 # 根据排名生成表格数据
@@ -51,40 +51,39 @@ protected_methods = ["GPT-4-Turbo-Preview (DFS)", "GPT-3.5-Turbo-1106 (DFS)", "G
 def merge_data(uploaded_data_json):
     # No need to call json.loads here because uploaded_data is already a Python dict
     new_data = uploaded_data_json
-
-    # Define a helper function to merge scores for an entry
-    def merge_scores(existing_scores, new_scores):
-        for key, value in new_scores.items():
-            existing_scores[key] = value
-
-    # Merge 'SolvablePassRateScores'
-    # 更新 'SolvablePassRateScores'
-    for new_entry in new_data["SolvablePassRateScores"]:
-        if new_entry["Method"] not in protected_methods:
-            existing_entry = next(
-                (item for item in existing_data["SolvablePassRateScores"] if item["Method"] == new_entry["Method"]),
-                None)
-            if existing_entry:
-                # 这是一个非保护的方法，更新它
-                merge_scores(existing_entry["Scores"], new_entry["Scores"])
-            else:
-                # 这是一个新的方法，添加到列表中
-                existing_data["SolvablePassRateScores"].append(new_entry)
-
-    # Merge 'SolvableWinRateScores'
-    for new_entry in new_data["SolvableWinRateScores"]:
-        if new_entry["Method"] not in protected_methods:
-            existing_entry = next(
-                (item for item in existing_data["SolvableWinRateScores"] if item["Method"] == new_entry["Method"]),
-                None)
-            if existing_entry:
-                merge_scores(existing_entry["Scores"], new_entry["Scores"])
-            else:
-                existing_data["SolvableWinRateScores"].append(new_entry)
-
-         # After merging the data, write the updated data back to the file
-        with open(data_file_path, 'w') as file:
-            json.dump(existing_data, file, indent=4)
+    with scheduler.lock:  # 确保文件操作的线程安全
+        # Define a helper function to merge scores for an entry
+        def merge_scores(existing_scores, new_scores):
+            for key, value in new_scores.items():
+                existing_scores[key] = value
+
+        # Merge 'SolvablePassRateScores'
+        for new_entry in new_data["SolvablePassRateScores"]:
+            if new_entry["Method"] not in protected_methods:
+                existing_entry = next(
+                    (item for item in existing_data["SolvablePassRateScores"] if item["Method"] == new_entry["Method"]),
+                    None)
+                if existing_entry:
+                    # 这是一个非保护的方法，更新它
+                    merge_scores(existing_entry["Scores"], new_entry["Scores"])
+                else:
+                    # 这是一个新的方法，添加到列表中
+                    existing_data["SolvablePassRateScores"].append(new_entry)
+
+        # Merge 'SolvableWinRateScores'
+        for new_entry in new_data["SolvableWinRateScores"]:
+            if new_entry["Method"] not in protected_methods:
+                existing_entry = next(
+                    (item for item in existing_data["SolvableWinRateScores"] if item["Method"] == new_entry["Method"]),
+                    None)
+                if existing_entry:
+                    merge_scores(existing_entry["Scores"], new_entry["Scores"])
+                else:
+                    existing_data["SolvableWinRateScores"].append(new_entry)
+
+            data_file_path = dataset_dir / "leaderboard_data.json"
+            with open(data_file_path, 'w') as file:
+                json.dump(existing_data, file, indent=4)
 
     # No need to sort here since we're doing it in the generate_table function
     return existing_data
@@ -97,7 +96,7 @@ def process_file(file_info):
             uploaded_data_json = json.loads(data_content)
         # Merge the uploaded data
         merge_data(uploaded_data_json)
-        # No need to call generate_table here since we are not returning it
+        scheduler.commit()  # Trigger manual commit after processing
     # If we don't return anything here, gradio will not update the table
     # We need to return the new data for the table if it's interactive
     pass_rate_table = generate_table(existing_data["SolvablePassRateScores"])[1]
@@ -245,3 +244,4 @@ with gr.Blocks() as app:
 
 if __name__ == "__main__":
     app.launch()
+    scheduler.commit()  # Ensure all changes are committed on exit