Update app.py

app.py

@@ -270,6 +270,7 @@ def sample_audio_paths(audio_paths, k=5, max_count=1):
 
     # return selected, count_data
     # Keep count_data atomic
+    
     return selected
 
 # ==============================================================================
@@ -466,7 +467,7 @@ def back_to_welcome():
 # Retry Function Definitions
 # ==============================================================================
 
-# Function for handling connection error
+# Decorator function that allows to use ThreadPoolExecutor to retry a function with timeout
 def retry_with_timeout(max_retries=3, timeout=10, backoff=1):
     def decorator(func):
         @wraps(func)
@@ -495,12 +496,12 @@ def retry_with_timeout(max_retries=3, timeout=10, backoff=1):
         return wrapper
     return decorator
 
-def save_with_retry(all_results, user_data, count_data):
+def save_with_retry(all_results, user_data):
     # 尝试上传到Hugging Face Hub
     try:
         # 使用线程安全的保存方式
         with ThreadPoolExecutor(max_workers=1) as executor:
-            future = executor.submit(save_all_results_to_file, all_results, user_data, count_data)
+            future = executor.submit(save_all_results_to_file, all_results, user_data)
             try:
                 future.result(timeout=30)  # 设置30秒超时
                 return True
@@ -597,6 +598,7 @@ def update_count_with_retry(count_data, question_set, max_retries=3):
             gr.update(), gr.update(),
         ) + (gr.update(),) * MAX_SUB_DIMS + (all_results, result_str)"""
 
+# user_data now no further contain "updated_count_data", which should be read/write with filelock and be directly accessed from working directory
 def submit_question_and_advance(q_idx, d_idx, selections, final_choice, all_results, user_data):
     try:
         # 准备数据
@@ -633,7 +635,8 @@ def submit_question_and_advance(q_idx, d_idx, selections, final_choice, all_resu
             
             # 尝试上传（带重试）
             try:
-                success = save_with_retry(all_results, user_data, user_data.get("updated_count_data"))
+                # success = save_with_retry(all_results, user_data, user_data.get("updated_count_data"))
+                success = save_with_retry(all_results, user_data)
             except Exception as e:
                 print(f"上传过程中发生错误: {e}")
                 success = False
@@ -646,7 +649,7 @@ def submit_question_and_advance(q_idx, d_idx, selections, final_choice, all_resu
                 
                 # 准备数据包
                 user_info_clean = {
-                    k: v for k, v in user_data.items() if k not in ["question_set", "updated_count_data"]
+                    k: v for k, v in user_data.items() if k not in ["question_set"]
                 }
                 final_data_package = {
                     "user_info": user_info_clean,
@@ -663,10 +666,10 @@ def submit_question_and_advance(q_idx, d_idx, selections, final_choice, all_resu
                 
                 # 更新count.json（剔除未完成的题目）
                 try:
-                    count_update_success = update_count_with_retry(
-                        user_data.get("updated_count_data", {}),
-                        user_data["question_set"]
-                    )
+                    with FileLock(COUNT_JSON_PATH + ".lock", timeout=5):
+                        with open(COUNT_JSON_PATH, "r", encoding="utf-8") as f:
+                            count_data = json.load(f, object_pairs_hook=collections.OrderedDict)
+                    count_update_success = update_count_with_retry(count_data, user_data["question_set"])
                 except Exception as e:
                     print(f"更新count.json失败: {e}")
                     count_update_success = False
@@ -743,14 +746,14 @@ def submit_question_and_advance(q_idx, d_idx, selections, final_choice, all_resu
     except Exception as e:
         print(f"上传出错: {e}")"""
 
-def save_all_results_to_file(all_results, user_data, count_data=None):
+def save_all_results_to_file(all_results, user_data):
     repo_id = "intersteller2887/Turing-test-dataset"
     username = user_data.get("username", "user")
     timestamp = pd.Timestamp.now().strftime('%Y%m%d_%H%M%S')
     submission_filename = f"submissions_{username}_{timestamp}.json"
 
     user_info_clean = {
-        k: v for k, v in user_data.items() if k not in ["question_set", "updated_count_data"]
+        k: v for k, v in user_data.items() if k not in ["question_set"]
     }
     
     final_data_package = {
@@ -775,19 +778,21 @@ def save_all_results_to_file(all_results, user_data, count_data=None):
         commit_message=f"Add new submission from {username}"
     )
 
-    if count_data:
+    try:
         with FileLock(COUNT_JSON_PATH + ".lock", timeout=5):
-            with open(COUNT_JSON_PATH, "w", encoding="utf-8") as f:
-                json.dump(count_data, f, indent=4, ensure_ascii=False)
-    
+            with open(COUNT_JSON_PATH, "r", encoding="utf-8") as f:
+                count_data_str = f.read()
+        
         api.upload_file(
-            path_or_fileobj=COUNT_JSON_PATH,
+            path_or_fileobj=bytes(count_data_str, "utf-8"),
             path_in_repo=COUNT_JSON_REPO_PATH,
             repo_id=repo_id,
             repo_type="dataset",
             token=hf_token,
             commit_message=f"Update count.json after submission by {username}"
         )
+    except Exception as e:
+        print(f"上传 count.json 失败: {e}")
 
 # ==============================================================================
 # Gradio 界面定义 (Gradio UI Definition)