Update app.py

app.py

@@ -12,13 +12,25 @@ from typing import Iterable
 from gradio.themes import Soft
 from gradio.themes.utils import colors, fonts, sizes
 
-# 自动安装 Playwright 浏览器（如果不存在）
-try:
-    import playwright
-    # 检查是否已经安装了浏览器，没有则下载
-    subprocess.run(["playwright", "install", "chromium"], check=True)
-except Exception as e:
-    print(f"Playwright setup failed: {e}")
+
+import threading
+import subprocess
+# ==========================================
+# --- 🌐 异步安装 Playwright 浏览器 ---
+# ==========================================
+def setup_playwright():
+    """在后台静默安装 Playwright，防止阻塞 Gradio 启动导致 HF 500 超时"""
+    try:
+        import playwright
+        print("⏳ [System] Downloading Playwright Chromium in background...")
+        # 增加 --with-deps 尝试安装系统级依赖 (虽然在非 root 容器可能失效，但有备无患)
+        subprocess.run(["playwright", "install", "chromium"], check=True)
+        print("✅ [System] Playwright browsers ready.")
+    except Exception as e:
+        print(f"❌ [System] Playwright setup failed: {e}")
+
+# 这一步非常关键：启动一个后台守护线程去下载，主进程直接往下走！
+threading.Thread(target=setup_playwright, daemon=True).start()
 
 # ==========================================
 # --- 📁 全局目录配置 (修改为 Session 基础目录) ---
@@ -38,6 +50,51 @@ def get_user_dirs(session_id):
     os.makedirs(output_dir, exist_ok=True)
     return papers_dir, output_dir, zip_path
 
+
+import time
+
+# ==========================================
+# --- 🧹 垃圾回收 (后台清理过期 Session) ---
+# ==========================================
+# 设定 Session 过期时间（例如：2 小时 = 7200 秒）
+SESSION_MAX_AGE_SECONDS = 2 * 60 * 60
+# 设定清理器执行间隔（例如：每 30 分钟扫描一次 = 1800 秒）
+CLEANUP_INTERVAL_SECONDS = 30 * 60
+
+def cleanup_expired_sessions():
+    """后台运行的垃圾回收任务"""
+    while True:
+        try:
+            if os.path.exists(SESSIONS_BASE_DIR):
+                current_time = time.time()
+                for session_folder in os.listdir(SESSIONS_BASE_DIR):
+                    folder_path = os.path.join(SESSIONS_BASE_DIR, session_folder)
+                    
+                    # 确保只处理目录
+                    if os.path.isdir(folder_path):
+                        # 获取文件夹的最后修改时间
+                        folder_mtime = os.path.getmtime(folder_path)
+                        
+                        # 判断是否超过了最大存活时间
+                        if (current_time - folder_mtime) > SESSION_MAX_AGE_SECONDS:
+                            try:
+                                shutil.rmtree(folder_path)
+                                print(f"🧹 [Garbage Collector] Deleted expired session: {session_folder}")
+                            except Exception as e:
+                                print(f"⚠️ [Garbage Collector] Failed to delete {session_folder}: {e}")
+        except Exception as e:
+            print(f"⚠️ [Garbage Collector] Error during cleanup scan: {e}")
+        
+        # 休眠到下一次扫描时间
+        time.sleep(CLEANUP_INTERVAL_SECONDS)
+
+def start_garbage_collector():
+    """启动后台守护线程"""
+    gc_thread = threading.Thread(target=cleanup_expired_sessions, daemon=True)
+    gc_thread.start()
+    print("🚀 [Garbage Collector] Background cleanup service started.")
+
+
 # ==========================================
 # --- 🎨 Custom Purple Theme Definition ---
 # ==========================================
@@ -681,5 +738,7 @@ with gr.Blocks(theme=purple_theme, css=custom_css) as demo:
     )
 
 if __name__ == "__main__":
+    start_garbage_collector()
     # 并发放宽至 5
-    demo.queue(default_concurrency_limit=5).launch()
+    # demo.queue(default_concurrency_limit=5).launch()
+    demo.launch(server_name="0.0.0.0", server_port=7860)