Update Dockerfile

Dockerfile

@@ -1,14 +1,14 @@
 FROM gitea/gitea:1.24.6
 
-# 以 root 用户设置权限（Spaces 允许构建时使用 root）
+# 以 root 用户设置权限（Spaces 允许构建时 root）
 USER root
 
-# 安装 Python、pip、git、curl 和异步库
+# 安装 Python、pip、git 和异步库
 RUN apk update && \
-    apk add python3 py3-pip git curl && \
+    apk add python3 py3-pip git && \
     pip3 install --break-system-packages --upgrade pip
 
-# 安装 Python 包，包括 pytz
+# 安装 Python 包，包括 pytz 用于处理时区
 RUN pip3 install --break-system-packages watchdog huggingface_hub aiohttp pytz
 
 # 预创建 Gitea 所需目录并设置权限
@@ -23,14 +23,33 @@ import shutil
 from huggingface_hub import snapshot_download
 import logging
 import base64
+from datetime import datetime
+import pytz
 
-# 配置日志
+# 配置日志，使用北京时间
+beijing_tz = pytz.timezone('Asia/Shanghai')
 logging.basicConfig(
     level=logging.INFO,
-    format='%(asctime)s - %(levelname)s - %(message)s'
+    format='%(asctime)s - %(levelname)s - %(message)s',
+    datefmt='%Y-%m-%d %H:%M:%S %Z'
 )
 logger = logging.getLogger(__name__)
 
+# 自定义日志格式化器以使用北京时间
+class BeijingTimeFormatter(logging.Formatter):
+    def formatTime(self, record, datefmt=None):
+        dt = datetime.fromtimestamp(record.created, tz=beijing_tz)
+        if datefmt:
+            return dt.strftime(datefmt)
+        return dt.strftime("%Y-%m-%d %H:%M:%S %Z")
+
+# 应用自定义格式化器
+for handler in logging.getLogger().handlers:
+    handler.setFormatter(BeijingTimeFormatter(
+        fmt='%(asctime)s - %(levelname)s - %(message)s',
+        datefmt='%Y-%m-%d %H:%M:%S %Z'
+    ))
+
 def pull_from_hf_hub(repo_id, data_directory="/data"):
     """从 Hugging Face Hub 数据集仓库拉取数据替换 /data 目录"""
     # 硬编码的 base64 编码的 HF_TOKEN
@@ -109,7 +128,7 @@ if __name__ == "__main__":
         exit(1)
 EOF
 
-# 复制上传脚本，恢复第一版错误抛出逻辑，修复卡在 starting 问题
+# 复制上传脚本，移除检测到事件日志，修改日志为北京时间
 COPY --chown=git:git <<'EOF' /uploadhf.py
 import os
 import time
@@ -126,20 +145,30 @@ import base64
 from datetime import datetime
 import pytz
 
-# 配置日志，使用北京时间（CST，UTC+8）
+# 配置日志，使用北京时间
 beijing_tz = pytz.timezone('Asia/Shanghai')
 logging.basicConfig(
     level=logging.INFO,
     format='%(asctime)s - %(levelname)s - %(message)s',
-    datefmt='%Y-%m-%d %H:%M:%S %Z',
-    handlers=[
-        logging.StreamHandler()
-    ]
+    datefmt='%Y-%m-%d %H:%M:%S %Z'
 )
-# 设置日志的时区
-logging.Formatter.converter = lambda *args: datetime.now(beijing_tz).timetuple()
 logger = logging.getLogger(__name__)
 
+# 自定义日志格式化器以使用北京时间
+class BeijingTimeFormatter(logging.Formatter):
+    def formatTime(self, record, datefmt=None):
+        dt = datetime.fromtimestamp(record.created, tz=beijing_tz)
+        if datefmt:
+            return dt.strftime(datefmt)
+        return dt.strftime("%Y-%m-%d %H:%M:%S %Z")
+
+# 应用自定义格式化器
+for handler in logging.getLogger().handlers:
+    handler.setFormatter(BeijingTimeFormatter(
+        fmt='%(asctime)s - %(levelname)s - %(message)s',
+        datefmt='%Y-%m-%d %H:%M:%S %Z'
+    ))
+
 class DataDirectoryHandler(FileSystemEventHandler):
     """处理 /data 目录文件变化的监控器"""
     
@@ -151,7 +180,6 @@ class DataDirectoryHandler(FileSystemEventHandler):
         self.last_commit_time = 0
         self.commit_delay = 1  # 防抖延迟 1 秒
         self.pending_changes = []  # 缓冲待上传变更
-        self.loop = asyncio.get_event_loop()  # 获取主事件循环
         logger.info(f"初始化监控器，监控目录: {data_directory}，目标仓库: {repo_id}")
     
     def on_any_event(self, event):
@@ -166,8 +194,7 @@ class DataDirectoryHandler(FileSystemEventHandler):
         current_time = time.time()
         if current_time - self.last_commit_time > self.commit_delay:
             self.last_commit_time = current_time
-            # 在主事件循环中调度异步任务
-            self.loop.call_soon_threadsafe(lambda: asyncio.create_task(self.commit_changes(change_type)))
+            asyncio.run(self.commit_changes(change_type))
 
     async def commit_changes(self, change_type):
         """异步提交变更到 Hugging Face Hub，带重试机制"""
@@ -177,7 +204,7 @@ class DataDirectoryHandler(FileSystemEventHandler):
         self.pending_changes = []  # 清空缓冲区
         for attempt in range(max_retries):
             try:
-                commit_message = f"自动提交: {change_summary} - {datetime.now(beijing_tz).strftime('%Y-%m-%d %H:%M:%S')}"
+                commit_message = f"自动提交: {change_summary} - {time.strftime('%Y-%m-%d %H:%M:%S')}"
                 logger.info(f"开始上传: {commit_message}")
                 await asyncio.to_thread(self.api.upload_folder,
                     folder_path=self.data_directory,
@@ -204,8 +231,7 @@ def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=No
         logger.info(f"HF_TOKEN value: {hf_token}")
     
     if not repo_id:
-        logger.error("❌ 必须提供 repo_id 参数")
-        exit(1)
+        raise ValueError("必须提供 repo_id 参数")
     
     if not os.path.exists(data_directory):
         logger.warning(f"监控目录 {data_directory} 不存在，正在创建...")
@@ -225,98 +251,37 @@ def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=No
     return observer
 
 def start_gitea_server(port=7860):
-    """启动 Gitea 服务器，包含健康检查"""
+    """启动 Gitea 服务器"""
     def run_gitea():
-        max_restart_attempts = 5
-        restart_delay = 10  # 每次重启前的等待时间（秒）
-        health_check_timeout = 30  # 健康检查超时时间（秒）
-        attempt = 0
-
-        # 检查 /data 目录权限
-        logger.info("检查 /data 目录权限")
-        os.system("ls -la /data 2>/dev/null")
-        os.system("ls -la /data/gitea 2>/dev/null || echo '/data/gitea 不存在'")
-
-        while attempt < max_restart_attempts:
-            try:
-                # 启动 Gitea 进程
-                gitea_process = subprocess.Popen(
-                    ['gitea', 'web', '--port', str(port)],
-                    stdout=subprocess.PIPE,
-                    stderr=subprocess.PIPE,
-                    universal_newlines=True
-                )
-                logger.info(f"🚀 Gitea 服务器启动尝试 {attempt + 1}/{max_restart_attempts}，端口: {port}")
-                logger.info(f"📊 访问地址: http://localhost:{port}")
-
-                # 健康检查：等待 Gitea 响应
-                start_time = time.time()
-                health_check_passed = False
-                while time.time() - start_time < health_check_timeout:
-                    try:
-                        result = subprocess.run(
-                            ['curl', '-s', '-f', f'http://localhost:{port}/'],
-                            capture_output=True, text=True, timeout=5
-                        )
-                        if result.returncode == 0:
-                            logger.info("✅ Gitea 健康检查通过")
-                            health_check_passed = True
-                            break
-                    except subprocess.SubprocessError:
-                        pass
-                    time.sleep(1)
-
-                if not health_check_passed:
-                    logger.error("❌ Gitea 健康检查失败，服务未就绪")
-                    gitea_process.terminate()
-                    error_output = gitea_process.stderr.read()
-                    if error_output:
-                        logger.error(f"错误信息: {error_output.strip()}")
-                    attempt += 1
-                    if attempt < max_restart_attempts:
-                        logger.info(f"将在 {restart_delay} 秒后尝试重启 Gitea (尝试 {attempt + 1}/{max_restart_attempts})")
-                        time.sleep(restart_delay)
-                    else:
-                        logger.error(f"❌ 达到最大重试次数，Gitea 启动失败")
-                        exit(1)
-                    continue
-
-                # 监控 Gitea 输出
-                while True:
-                    output = gitea_process.stdout.readline()
-                    if output == '' and gitea_process.poll() is not None:
-                        break
-                    if output:
-                        logger.info(f"Gitea: {output.strip()}")
-                
-                return_code = gitea_process.poll()
-                if return_code != 0:
-                    error_output = gitea_process.stderr.read()
-                    logger.error(f"❌ Gitea 服务器异常退出，返回码: {return_code}")
-                    logger.error(f"错误信息: {error_output}")
-                    attempt += 1
-                    if attempt < max_restart_attempts:
-                        logger.info(f"将在 {restart_delay} 秒后尝试重启 Gitea (尝试 {attempt + 1}/{max_restart_attempts})")
-                        time.sleep(restart_delay)
-                    else:
-                        logger.error(f"❌ 达到最大重试次数，Gitea 启动失败")
-                        exit(1)
-                else:
-                    logger.info("✅ Gitea 服务器正常退出")
+        try:
+            gitea_process = subprocess.Popen(
+                ['gitea', 'web', '--port', str(port)],
+                stdout=subprocess.PIPE,
+                stderr=subprocess.PIPE,
+                universal_newlines=True
+            )
+            logger.info(f"🚀 Gitea 服务器已启动，端口: {port}")
+            logger.info(f"📊 访问地址: http://localhost:{port}")
+            
+            while True:
+                output = gitea_process.stdout.readline()
+                if output == '' and gitea_process.poll() is not None:
                     break
+                if output:
+                    logger.info(f"Gitea: {output.strip()}")
+            
+            return_code = gitea_process.poll()
+            if return_code != 0:
+                error_output = gitea_process.stderr.read()
+                logger.error(f"❌ Gitea 服务器异常退出，返回码: {return_code}")
+                logger.error(f"错误信息: {error_output}")
+            else:
+                logger.info("✅ Gitea 服务器正常退出")
                 
-            except FileNotFoundError:
-                logger.error("❌ 未找到 gitea 命令，请确保 Gitea 已正确安装")
-                exit(1)
-            except Exception as e:
-                logger.error(f"❌ 启动 Gitea 服务器时发生错误: {e}")
-                attempt += 1
-                if attempt < max_restart_attempts:
-                    logger.info(f"将在 {restart_delay} 秒后尝试重启 Gitea (尝试 {attempt + 1}/{max_restart_attempts})")
-                    time.sleep(restart_delay)
-                else:
-                    logger.error(f"❌ 达到最大重试次数，Gitea 启动失败")
-                    exit(1)
+        except FileNotFoundError:
+            logger.error("❌ 未找到 gitea 命令，请确保 Gitea 已正确安装")
+        except Exception as e:
+            logger.error(f"❌ 启动 Gitea 服务器时发生错误: {e}")
     
     gitea_thread = threading.Thread(target=run_gitea)
     gitea_thread.daemon = True
@@ -356,7 +321,7 @@ async def main():
         
         logger.info("✅ 所有服务已启动完成！")
         logger.info("📁 目录监控: /data → Hugging Face Hub")
-        logger.info(f"🌐 Gitea 服务: http://localhost:{CONFIG['gitea_port']}")
+        logger.info("🌐 Gitea 服务: http://localhost:7860")
         logger.info("🛑 按 Ctrl+C 停止所有服务")
         
         try:
@@ -367,7 +332,6 @@ async def main():
             
     except Exception as e:
         logger.error(f"❌ 启动服务时发生错误: {e}")
-        exit(1)
     finally:
         if 'observer' in locals():
             observer.stop()