Update app.py

app.py

@@ -90,7 +90,7 @@ DIMENSIONS_DATA = [
 
 DIMENSION_TITLES = [d["title"] for d in DIMENSIONS_DATA]
 
-def load_or_initialize_count_json(audio_paths):
+"""def load_or_initialize_count_json(audio_paths):
     if os.path.exists(COUNT_JSON_PATH):
         with open(COUNT_JSON_PATH, "r", encoding="utf-8") as f:
             # 使用 object_pairs_hook 保持原始顺序
@@ -118,12 +118,41 @@ def load_or_initialize_count_json(audio_paths):
             # 确保写入时也保持顺序
             json.dump(count_data, f, indent=4, ensure_ascii=False)
     
+    return count_data"""
+
+from filelock import FileLock
+
+def load_or_initialize_count_json(audio_paths):
+    lock_path = COUNT_JSON_PATH + ".lock"
+    with FileLock(lock_path, timeout=5):  # 最多等 10 秒
+        if os.path.exists(COUNT_JSON_PATH):
+            with open(COUNT_JSON_PATH, "r", encoding="utf-8") as f:
+                count_data = json.load(f, object_pairs_hook=collections.OrderedDict)
+        else:
+            count_data = collections.OrderedDict()
+
+        updated = False
+        sample_audio_files = {os.path.basename(d["audio"]) for d in DIMENSIONS_DATA}
+
+        for path in audio_paths:
+            filename = os.path.basename(path)
+            if filename not in count_data:
+                if filename in sample_audio_files:
+                    count_data[filename] = 999
+                else:
+                    count_data[filename] = 0
+                updated = True
+
+        if updated or not os.path.exists(COUNT_JSON_PATH):
+            with open(COUNT_JSON_PATH, "w", encoding="utf-8") as f:
+                json.dump(count_data, f, indent=4, ensure_ascii=False)
+
     return count_data
 
 def append_cache_buster(audio_path):
     return f"{audio_path}?t={int(time.time() * 1000)}"
 
-def sample_audio_paths(audio_paths, count_data, k=5, max_count=1):
+"""def sample_audio_paths(audio_paths, count_data, k=5, max_count=1):
     eligible_paths = [p for p in audio_paths if count_data.get(os.path.basename(p), 0) < max_count]
     
     if len(eligible_paths) < k:
@@ -142,6 +171,25 @@ def sample_audio_paths(audio_paths, count_data, k=5, max_count=1):
     with open(COUNT_JSON_PATH, "w", encoding="utf-8") as f:
         json.dump(count_data, f, indent=4, ensure_ascii=False)
 
+    return selected, count_data"""
+
+def sample_audio_paths(audio_paths, count_data, k=5, max_count=1):
+    eligible_paths = [p for p in audio_paths if count_data.get(os.path.basename(p), 0) < max_count]
+
+    if len(eligible_paths) < k:
+        raise ValueError(f"可用音频数量不足（只剩 {len(eligible_paths)} 条 count<{max_count} 的音频），无法抽取 {k} 条")
+
+    selected = random.sample(eligible_paths, k)
+
+    for path in selected:
+        filename = os.path.basename(path)
+        count_data[filename] = count_data.get(filename, 0) + 1
+
+    lock_path = COUNT_JSON_PATH + ".lock"
+    with FileLock(lock_path, timeout=10):
+        with open(COUNT_JSON_PATH, "w", encoding="utf-8") as f:
+            json.dump(count_data, f, indent=4, ensure_ascii=False)
+
     return selected, count_data
 
 count_data = load_or_initialize_count_json(all_data_audio_paths)
@@ -410,7 +458,7 @@ def save_all_results_to_file(all_results, user_data, count_data=None):
         print(f"上传成功: {submission_filename}")
 
         # 上传 count.json（如果提供）
-        if count_data:
+        """if count_data:
             with open(COUNT_JSON_PATH, "w", encoding="utf-8") as f:
                 json.dump(count_data, f, indent=4, ensure_ascii=False)
 
@@ -422,7 +470,21 @@ def save_all_results_to_file(all_results, user_data, count_data=None):
                 token=hf_token,
                 commit_message=f"Update count.json after submission by {username}"
             )
-            print("count.json 上传成功")
+            print("count.json 上传成功")"""
+
+        if count_data:
+            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)
+    
+            api.upload_file(
+                path_or_fileobj=COUNT_JSON_PATH,
+                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"上传出错: {e}")