crawlo 1.3.1__py3-none-any.whl → 1.3.3__py3-none-any.whl

crawlo/queue/queue_manager.py

@@ -1,379 +1,379 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-统一的队列管理器
-提供简洁、一致的队列接口，自动处理不同队列类型的差异
-"""
-import asyncio
-import traceback
-from enum import Enum
-from typing import Optional, Dict, Any, Union
-
-from crawlo import Request
-from crawlo.queue.pqueue import SpiderPriorityQueue
-from crawlo.utils.error_handler import ErrorHandler
-from crawlo.utils.log import get_logger
-from crawlo.utils.request_serializer import RequestSerializer
-
-try:
-    from crawlo.queue.redis_priority_queue import RedisPriorityQueue
-
-    REDIS_AVAILABLE = True
-except ImportError:
-    RedisPriorityQueue = None
-    REDIS_AVAILABLE = False
-
-
-class QueueType(Enum):
-    """Queue type enumeration"""
-    MEMORY = "memory"
-    REDIS = "redis"
-    AUTO = "auto"  # 自动选择
-
-
-class QueueConfig:
-    """Queue configuration class"""
-
-    def __init__(
-            self,
-            queue_type: Union[QueueType, str] = QueueType.AUTO,
-            redis_url: Optional[str] = None,
-            redis_host: str = "127.0.0.1",
-            redis_port: int = 6379,
-            redis_password: Optional[str] = None,
-            redis_db: int = 0,
-            queue_name: str = "crawlo:requests",
-            max_queue_size: int = 1000,
-            max_retries: int = 3,
-            timeout: int = 300,
-            **kwargs
-    ):
-        self.queue_type = QueueType(queue_type) if isinstance(queue_type, str) else queue_type
-
-        # Redis 配置
-        if redis_url:
-            self.redis_url = redis_url
-        else:
-            if redis_password:
-                self.redis_url = f"redis://:{redis_password}@{redis_host}:{redis_port}/{redis_db}"
-            else:
-                self.redis_url = f"redis://{redis_host}:{redis_port}/{redis_db}"
-
-        self.queue_name = queue_name
-        self.max_queue_size = max_queue_size
-        self.max_retries = max_retries
-        self.timeout = timeout
-        self.extra_config = kwargs
-
-    @classmethod
-    def from_settings(cls, settings) -> 'QueueConfig':
-        """Create configuration from settings"""
-        return cls(
-            queue_type=settings.get('QUEUE_TYPE', QueueType.AUTO),
-            redis_url=settings.get('REDIS_URL'),
-            redis_host=settings.get('REDIS_HOST', '127.0.0.1'),
-            redis_port=settings.get_int('REDIS_PORT', 6379),
-            redis_password=settings.get('REDIS_PASSWORD'),
-            redis_db=settings.get_int('REDIS_DB', 0),
-            queue_name=settings.get('SCHEDULER_QUEUE_NAME', 'crawlo:requests'),
-            max_queue_size=settings.get_int('SCHEDULER_MAX_QUEUE_SIZE', 1000),
-            max_retries=settings.get_int('QUEUE_MAX_RETRIES', 3),
-            timeout=settings.get_int('QUEUE_TIMEOUT', 300)
-        )
-
-
-class QueueManager:
-    """Unified queue manager"""
-
-    def __init__(self, config: QueueConfig):
-        self.config = config
-        self.logger = get_logger(self.__class__.__name__)
-        self.error_handler = ErrorHandler(self.__class__.__name__)
-        self.request_serializer = RequestSerializer()
-        self._queue = None
-        self._queue_semaphore = None
-        self._queue_type = None
-        self._health_status = "unknown"
-
-    async def initialize(self) -> bool:
-        """初始化队列"""
-        try:
-            queue_type = await self._determine_queue_type()
-            self._queue = await self._create_queue(queue_type)
-            self._queue_type = queue_type
-
-            # 测试队列健康状态
-            health_check_result = await self._health_check()
-
-            self.logger.info(f"Queue initialized successfully Type: {queue_type.value}")
-            # 只在调试模式下输出详细配置信息
-            self.logger.debug(f"Queue configuration: {self._get_queue_info()}")
-
-            # 如果健康检查返回True，表示队列类型发生了切换，需要更新配置
-            if health_check_result:
-                return True
-
-            # 如果队列类型是Redis，检查是否需要更新配置
-            if queue_type == QueueType.REDIS:
-                # 这个检查需要在调度器中进行，因为队列管理器无法访问crawler.settings
-                # 但我们不需要总是返回True，只有在确实需要更新时才返回True
-                # 调度器会进行更详细的检查
-                pass
-
-            return False  # 默认不需要更新配置
-
-        except Exception as e:
-            # 记录详细的错误信息和堆栈跟踪
-            self.logger.error(f"Queue initialization failed: {e}")
-            self.logger.debug(f"详细错误信息:\n{traceback.format_exc()}")
-            self._health_status = "error"
-            return False
-
-    async def put(self, request: Request, priority: int = 0) -> bool:
-        """Unified enqueue interface"""
-        if not self._queue:
-            raise RuntimeError("队列未初始化")
-
-        try:
-            # 序列化处理（仅对 Redis 队列）
-            if self._queue_type == QueueType.REDIS:
-                request = self.request_serializer.prepare_for_serialization(request)
-
-            # 背压控制（仅对内存队列）
-            if self._queue_semaphore:
-                # 对于大量请求，使用非阻塞式检查
-                if not self._queue_semaphore.locked():
-                    await self._queue_semaphore.acquire()
-                else:
-                    # 如果队列已满，返回 False 而不是阻塞
-                    self.logger.warning("Queue is full, skipping current request")
-                    return False
-
-            # 统一的入队操作
-            if hasattr(self._queue, 'put'):
-                if self._queue_type == QueueType.REDIS:
-                    success = await self._queue.put(request, priority)
-                else:
-                    await self._queue.put(request)
-                    success = True
-            else:
-                raise RuntimeError(f"队列类型 {self._queue_type} 不支持 put 操作")
-
-            if success:
-                self.logger.debug(f"Request enqueued successfully: {request.url}")
-
-            return success
-
-        except Exception as e:
-            self.logger.error(f"Failed to enqueue request: {e}")
-            if self._queue_semaphore:
-                self._queue_semaphore.release()
-            return False
-
-    async def get(self, timeout: float = 5.0) -> Optional[Request]:
-        """Unified dequeue interface"""
-        if not self._queue:
-            raise RuntimeError("队列未初始化")
-
-        try:
-            request = await self._queue.get(timeout=timeout)
-
-            # 释放信号量（仅对内存队列）
-            if self._queue_semaphore and request:
-                self._queue_semaphore.release()
-
-            # 反序列化处理（仅对 Redis 队列）
-            if request and self._queue_type == QueueType.REDIS:
-                # 这里需要 spider 实例，暂时返回原始请求
-                # 实际的 callback 恢复在 scheduler 中处理
-                pass
-
-            return request
-
-        except Exception as e:
-            self.logger.error(f"Failed to dequeue request: {e}")
-            return None
-
-    async def size(self) -> int:
-        """Get queue size"""
-        if not self._queue:
-            return 0
-
-        try:
-            if hasattr(self._queue, 'qsize'):
-                if asyncio.iscoroutinefunction(self._queue.qsize):
-                    return await self._queue.qsize()
-                else:
-                    return self._queue.qsize()
-            return 0
-        except Exception as e:
-            self.logger.warning(f"Failed to get queue size: {e}")
-            return 0
-
-    def empty(self) -> bool:
-        """Check if queue is empty (synchronous version, for compatibility)"""
-        try:
-            # 对于内存队列，可以同步检查
-            if self._queue_type == QueueType.MEMORY:
-                return self._queue.qsize() == 0
-            # 对于 Redis 队列，由于需要异步操作，这里返回近似值
-            # 为了确保程序能正常退出，我们返回True，让上层通过更精确的异步检查来判断
-            return True
-        except Exception:
-            return True
-
-    async def async_empty(self) -> bool:
-        """Check if queue is empty (asynchronous version, more accurate)"""
-        try:
-            # 对于内存队列
-            if self._queue_type == QueueType.MEMORY:
-                return self._queue.qsize() == 0
-            # 对于 Redis 队列，使用异步检查
-            elif self._queue_type == QueueType.REDIS:
-                size = await self.size()
-                return size == 0
-            return True
-        except Exception:
-            return True
-
-    async def close(self) -> None:
-        """Close queue"""
-        if self._queue and hasattr(self._queue, 'close'):
-            try:
-                await self._queue.close()
-                # Change INFO level log to DEBUG level to avoid redundant output
-                self.logger.debug("Queue closed")
-            except Exception as e:
-                self.logger.warning(f"Error closing queue: {e}")
-
-    def get_status(self) -> Dict[str, Any]:
-        """Get queue status information"""
-        return {
-            "type": self._queue_type.value if self._queue_type else "unknown",
-            "health": self._health_status,
-            "config": self._get_queue_info(),
-            "initialized": self._queue is not None
-        }
-
-    async def _determine_queue_type(self) -> QueueType:
-        """Determine queue type"""
-        if self.config.queue_type == QueueType.AUTO:
-            # 自动选择：优先使用 Redis（如果可用）
-            if REDIS_AVAILABLE and self.config.redis_url:
-                # 测试 Redis 连接
-                try:
-                    test_queue = RedisPriorityQueue(self.config.redis_url)
-                    await test_queue.connect()
-                    await test_queue.close()
-                    # Change INFO level log to DEBUG level to avoid redundant output
-                    self.logger.debug("Auto-detection: Redis available, using distributed queue")
-                    return QueueType.REDIS
-                except Exception as e:
-                    self.logger.debug(f"Auto-detection: Redis unavailable ({e}), using memory queue")
-                    return QueueType.MEMORY
-            else:
-                self.logger.debug("Auto-detection: Redis not configured, using memory queue")
-                return QueueType.MEMORY
-
-        elif self.config.queue_type == QueueType.REDIS:
-            if not REDIS_AVAILABLE:
-                raise RuntimeError("Redis 队列不可用：未安装 redis 依赖")
-            if not self.config.redis_url:
-                raise RuntimeError("Redis 队列不可用：未配置 REDIS_URL")
-            # 测试 Redis 连接
-            try:
-                test_queue = RedisPriorityQueue(self.config.redis_url)
-                await test_queue.connect()
-                await test_queue.close()
-                return QueueType.REDIS
-            except Exception as e:
-                # 如果强制使用Redis但连接失败，则抛出异常
-                raise RuntimeError(f"Redis 队列不可用：无法连接到 Redis ({e})")
-
-        elif self.config.queue_type == QueueType.MEMORY:
-            return QueueType.MEMORY
-
-        else:
-            raise ValueError(f"不支持的队列类型: {self.config.queue_type}")
-
-    async def _create_queue(self, queue_type: QueueType):
-        """Create queue instance"""
-        if queue_type == QueueType.REDIS:
-            # 简化项目名称提取逻辑
-            project_name = "default"
-            if ':' in self.config.queue_name:
-                parts = self.config.queue_name.split(':')
-                # 跳过所有"crawlo"前缀，取第一个非"crawlo"部分作为项目名称
-                for part in parts:
-                    if part != "crawlo":
-                        project_name = part
-                        break
-            else:
-                project_name = self.config.queue_name or "default"
-
-            queue = RedisPriorityQueue(
-                redis_url=self.config.redis_url,
-                queue_name=self.config.queue_name,
-                max_retries=self.config.max_retries,
-                timeout=self.config.timeout,
-                module_name=project_name  # 传递项目名称作为module_name
-            )
-            # 不需要立即连接，使用 lazy connect
-            return queue
-
-        elif queue_type == QueueType.MEMORY:
-            queue = SpiderPriorityQueue()
-            # 为内存队列设置背压控制
-            self._queue_semaphore = asyncio.Semaphore(self.config.max_queue_size)
-            return queue
-
-        else:
-            raise ValueError(f"不支持的队列类型: {queue_type}")
-
-    async def _health_check(self) -> bool:
-        """Health check"""
-        try:
-            if self._queue_type == QueueType.REDIS:
-                # 测试 Redis 连接
-                await self._queue.connect()
-                self._health_status = "healthy"
-            else:
-                # 内存队列总是健康的
-                self._health_status = "healthy"
-                return False  # 内存队列不需要更新配置
-        except Exception as e:
-            self.logger.warning(f"Queue health check failed: {e}")
-            self._health_status = "unhealthy"
-            # 如果是Redis队列且健康检查失败，尝试切换到内存队列
-            if self._queue_type == QueueType.REDIS and self.config.queue_type == QueueType.AUTO:
-                self.logger.info("Redis queue unavailable, attempting to switch to memory queue...")
-                try:
-                    await self._queue.close()
-                except:
-                    pass
-                self._queue = None
-                # 重新创建内存队列
-                self._queue = await self._create_queue(QueueType.MEMORY)
-                self._queue_type = QueueType.MEMORY
-                self._queue_semaphore = asyncio.Semaphore(self.config.max_queue_size)
-                self._health_status = "healthy"
-                self.logger.info("Switched to memory queue")
-                # 返回一个信号，表示需要更新过滤器和去重管道配置
-                return True
-        return False
-
-    def _get_queue_info(self) -> Dict[str, Any]:
-        """Get queue configuration information"""
-        info = {
-            "queue_name": self.config.queue_name,
-            "max_queue_size": self.config.max_queue_size
-        }
-
-        if self._queue_type == QueueType.REDIS:
-            info.update({
-                "redis_url": self.config.redis_url,
-                "max_retries": self.config.max_retries,
-                "timeout": self.config.timeout
-            })
-
-        return info
+#!/usr/bin/python
+# -*- coding: UTF-8 -*-
+"""
+统一的队列管理器
+提供简洁、一致的队列接口，自动处理不同队列类型的差异
+"""
+import asyncio
+import traceback
+from enum import Enum
+from typing import Optional, Dict, Any, Union
+
+from crawlo import Request
+from crawlo.queue.pqueue import SpiderPriorityQueue
+from crawlo.utils.error_handler import ErrorHandler
+from crawlo.utils.log import get_logger
+from crawlo.utils.request_serializer import RequestSerializer
+
+try:
+    from crawlo.queue.redis_priority_queue import RedisPriorityQueue
+
+    REDIS_AVAILABLE = True
+except ImportError:
+    RedisPriorityQueue = None
+    REDIS_AVAILABLE = False
+
+
+class QueueType(Enum):
+    """Queue type enumeration"""
+    MEMORY = "memory"
+    REDIS = "redis"
+    AUTO = "auto"  # 自动选择
+
+
+class QueueConfig:
+    """Queue configuration class"""
+
+    def __init__(
+            self,
+            queue_type: Union[QueueType, str] = QueueType.AUTO,
+            redis_url: Optional[str] = None,
+            redis_host: str = "127.0.0.1",
+            redis_port: int = 6379,
+            redis_password: Optional[str] = None,
+            redis_db: int = 0,
+            queue_name: str = "crawlo:requests",
+            max_queue_size: int = 1000,
+            max_retries: int = 3,
+            timeout: int = 300,
+            **kwargs
+    ):
+        self.queue_type = QueueType(queue_type) if isinstance(queue_type, str) else queue_type
+
+        # Redis 配置
+        if redis_url:
+            self.redis_url = redis_url
+        else:
+            if redis_password:
+                self.redis_url = f"redis://:{redis_password}@{redis_host}:{redis_port}/{redis_db}"
+            else:
+                self.redis_url = f"redis://{redis_host}:{redis_port}/{redis_db}"
+
+        self.queue_name = queue_name
+        self.max_queue_size = max_queue_size
+        self.max_retries = max_retries
+        self.timeout = timeout
+        self.extra_config = kwargs
+
+    @classmethod
+    def from_settings(cls, settings) -> 'QueueConfig':
+        """Create configuration from settings"""
+        return cls(
+            queue_type=settings.get('QUEUE_TYPE', QueueType.AUTO),
+            redis_url=settings.get('REDIS_URL'),
+            redis_host=settings.get('REDIS_HOST', '127.0.0.1'),
+            redis_port=settings.get_int('REDIS_PORT', 6379),
+            redis_password=settings.get('REDIS_PASSWORD'),
+            redis_db=settings.get_int('REDIS_DB', 0),
+            queue_name=settings.get('SCHEDULER_QUEUE_NAME', 'crawlo:requests'),
+            max_queue_size=settings.get_int('SCHEDULER_MAX_QUEUE_SIZE', 1000),
+            max_retries=settings.get_int('QUEUE_MAX_RETRIES', 3),
+            timeout=settings.get_int('QUEUE_TIMEOUT', 300)
+        )
+
+
+class QueueManager:
+    """Unified queue manager"""
+
+    def __init__(self, config: QueueConfig):
+        self.config = config
+        self.logger = get_logger(self.__class__.__name__)
+        self.error_handler = ErrorHandler(self.__class__.__name__)
+        self.request_serializer = RequestSerializer()
+        self._queue = None
+        self._queue_semaphore = None
+        self._queue_type = None
+        self._health_status = "unknown"
+
+    async def initialize(self) -> bool:
+        """初始化队列"""
+        try:
+            queue_type = await self._determine_queue_type()
+            self._queue = await self._create_queue(queue_type)
+            self._queue_type = queue_type
+
+            # 测试队列健康状态
+            health_check_result = await self._health_check()
+
+            self.logger.info(f"Queue initialized successfully Type: {queue_type.value}")
+            # 只在调试模式下输出详细配置信息
+            self.logger.debug(f"Queue configuration: {self._get_queue_info()}")
+
+            # 如果健康检查返回True，表示队列类型发生了切换，需要更新配置
+            if health_check_result:
+                return True
+
+            # 如果队列类型是Redis，检查是否需要更新配置
+            if queue_type == QueueType.REDIS:
+                # 这个检查需要在调度器中进行，因为队列管理器无法访问crawler.settings
+                # 但我们不需要总是返回True，只有在确实需要更新时才返回True
+                # 调度器会进行更详细的检查
+                pass
+
+            return False  # 默认不需要更新配置
+
+        except Exception as e:
+            # 记录详细的错误信息和堆栈跟踪
+            self.logger.error(f"Queue initialization failed: {e}")
+            self.logger.debug(f"详细错误信息:\n{traceback.format_exc()}")
+            self._health_status = "error"
+            return False
+
+    async def put(self, request: Request, priority: int = 0) -> bool:
+        """Unified enqueue interface"""
+        if not self._queue:
+            raise RuntimeError("队列未初始化")
+
+        try:
+            # 序列化处理（仅对 Redis 队列）
+            if self._queue_type == QueueType.REDIS:
+                request = self.request_serializer.prepare_for_serialization(request)
+
+            # 背压控制（仅对内存队列）
+            if self._queue_semaphore:
+                # 对于大量请求，使用非阻塞式检查
+                if not self._queue_semaphore.locked():
+                    await self._queue_semaphore.acquire()
+                else:
+                    # 如果队列已满，返回 False 而不是阻塞
+                    self.logger.warning("Queue is full, skipping current request")
+                    return False
+
+            # 统一的入队操作
+            if hasattr(self._queue, 'put'):
+                if self._queue_type == QueueType.REDIS:
+                    success = await self._queue.put(request, priority)
+                else:
+                    await self._queue.put(request)
+                    success = True
+            else:
+                raise RuntimeError(f"队列类型 {self._queue_type} 不支持 put 操作")
+
+            if success:
+                self.logger.debug(f"Request enqueued successfully: {request.url}")
+
+            return success
+
+        except Exception as e:
+            self.logger.error(f"Failed to enqueue request: {e}")
+            if self._queue_semaphore:
+                self._queue_semaphore.release()
+            return False
+
+    async def get(self, timeout: float = 5.0) -> Optional[Request]:
+        """Unified dequeue interface"""
+        if not self._queue:
+            raise RuntimeError("队列未初始化")
+
+        try:
+            request = await self._queue.get(timeout=timeout)
+
+            # 释放信号量（仅对内存队列）
+            if self._queue_semaphore and request:
+                self._queue_semaphore.release()
+
+            # 反序列化处理（仅对 Redis 队列）
+            if request and self._queue_type == QueueType.REDIS:
+                # 这里需要 spider 实例，暂时返回原始请求
+                # 实际的 callback 恢复在 scheduler 中处理
+                pass
+
+            return request
+
+        except Exception as e:
+            self.logger.error(f"Failed to dequeue request: {e}")
+            return None
+
+    async def size(self) -> int:
+        """Get queue size"""
+        if not self._queue:
+            return 0
+
+        try:
+            if hasattr(self._queue, 'qsize'):
+                if asyncio.iscoroutinefunction(self._queue.qsize):
+                    return await self._queue.qsize()
+                else:
+                    return self._queue.qsize()
+            return 0
+        except Exception as e:
+            self.logger.warning(f"Failed to get queue size: {e}")
+            return 0
+
+    def empty(self) -> bool:
+        """Check if queue is empty (synchronous version, for compatibility)"""
+        try:
+            # 对于内存队列，可以同步检查
+            if self._queue_type == QueueType.MEMORY:
+                return self._queue.qsize() == 0
+            # 对于 Redis 队列，由于需要异步操作，这里返回近似值
+            # 为了确保程序能正常退出，我们返回True，让上层通过更精确的异步检查来判断
+            return True
+        except Exception:
+            return True
+
+    async def async_empty(self) -> bool:
+        """Check if queue is empty (asynchronous version, more accurate)"""
+        try:
+            # 对于内存队列
+            if self._queue_type == QueueType.MEMORY:
+                return self._queue.qsize() == 0
+            # 对于 Redis 队列，使用异步检查
+            elif self._queue_type == QueueType.REDIS:
+                size = await self.size()
+                return size == 0
+            return True
+        except Exception:
+            return True
+
+    async def close(self) -> None:
+        """Close queue"""
+        if self._queue and hasattr(self._queue, 'close'):
+            try:
+                await self._queue.close()
+                # Change INFO level log to DEBUG level to avoid redundant output
+                self.logger.debug("Queue closed")
+            except Exception as e:
+                self.logger.warning(f"Error closing queue: {e}")
+
+    def get_status(self) -> Dict[str, Any]:
+        """Get queue status information"""
+        return {
+            "type": self._queue_type.value if self._queue_type else "unknown",
+            "health": self._health_status,
+            "config": self._get_queue_info(),
+            "initialized": self._queue is not None
+        }
+
+    async def _determine_queue_type(self) -> QueueType:
+        """Determine queue type"""
+        if self.config.queue_type == QueueType.AUTO:
+            # 自动选择：优先使用 Redis（如果可用）
+            if REDIS_AVAILABLE and self.config.redis_url:
+                # 测试 Redis 连接
+                try:
+                    test_queue = RedisPriorityQueue(self.config.redis_url)
+                    await test_queue.connect()
+                    await test_queue.close()
+                    # Change INFO level log to DEBUG level to avoid redundant output
+                    self.logger.debug("Auto-detection: Redis available, using distributed queue")
+                    return QueueType.REDIS
+                except Exception as e:
+                    self.logger.debug(f"Auto-detection: Redis unavailable ({e}), using memory queue")
+                    return QueueType.MEMORY
+            else:
+                self.logger.debug("Auto-detection: Redis not configured, using memory queue")
+                return QueueType.MEMORY
+
+        elif self.config.queue_type == QueueType.REDIS:
+            if not REDIS_AVAILABLE:
+                raise RuntimeError("Redis 队列不可用：未安装 redis 依赖")
+            if not self.config.redis_url:
+                raise RuntimeError("Redis 队列不可用：未配置 REDIS_URL")
+            # 测试 Redis 连接
+            try:
+                test_queue = RedisPriorityQueue(self.config.redis_url)
+                await test_queue.connect()
+                await test_queue.close()
+                return QueueType.REDIS
+            except Exception as e:
+                # 如果强制使用Redis但连接失败，则抛出异常
+                raise RuntimeError(f"Redis 队列不可用：无法连接到 Redis ({e})")
+
+        elif self.config.queue_type == QueueType.MEMORY:
+            return QueueType.MEMORY
+
+        else:
+            raise ValueError(f"不支持的队列类型: {self.config.queue_type}")
+
+    async def _create_queue(self, queue_type: QueueType):
+        """Create queue instance"""
+        if queue_type == QueueType.REDIS:
+            # 简化项目名称提取逻辑
+            project_name = "default"
+            if ':' in self.config.queue_name:
+                parts = self.config.queue_name.split(':')
+                # 跳过所有"crawlo"前缀，取第一个非"crawlo"部分作为项目名称
+                for part in parts:
+                    if part != "crawlo":
+                        project_name = part
+                        break
+            else:
+                project_name = self.config.queue_name or "default"
+
+            queue = RedisPriorityQueue(
+                redis_url=self.config.redis_url,
+                queue_name=self.config.queue_name,
+                max_retries=self.config.max_retries,
+                timeout=self.config.timeout,
+                module_name=project_name  # 传递项目名称作为module_name
+            )
+            # 不需要立即连接，使用 lazy connect
+            return queue
+
+        elif queue_type == QueueType.MEMORY:
+            queue = SpiderPriorityQueue()
+            # 为内存队列设置背压控制
+            self._queue_semaphore = asyncio.Semaphore(self.config.max_queue_size)
+            return queue
+
+        else:
+            raise ValueError(f"不支持的队列类型: {queue_type}")
+
+    async def _health_check(self) -> bool:
+        """Health check"""
+        try:
+            if self._queue_type == QueueType.REDIS:
+                # 测试 Redis 连接
+                await self._queue.connect()
+                self._health_status = "healthy"
+            else:
+                # 内存队列总是健康的
+                self._health_status = "healthy"
+                return False  # 内存队列不需要更新配置
+        except Exception as e:
+            self.logger.warning(f"Queue health check failed: {e}")
+            self._health_status = "unhealthy"
+            # 如果是Redis队列且健康检查失败，尝试切换到内存队列
+            if self._queue_type == QueueType.REDIS and self.config.queue_type == QueueType.AUTO:
+                self.logger.info("Redis queue unavailable, attempting to switch to memory queue...")
+                try:
+                    await self._queue.close()
+                except:
+                    pass
+                self._queue = None
+                # 重新创建内存队列
+                self._queue = await self._create_queue(QueueType.MEMORY)
+                self._queue_type = QueueType.MEMORY
+                self._queue_semaphore = asyncio.Semaphore(self.config.max_queue_size)
+                self._health_status = "healthy"
+                self.logger.info("Switched to memory queue")
+                # 返回一个信号，表示需要更新过滤器和去重管道配置
+                return True
+        return False
+
+    def _get_queue_info(self) -> Dict[str, Any]:
+        """Get queue configuration information"""
+        info = {
+            "queue_name": self.config.queue_name,
+            "max_queue_size": self.config.max_queue_size
+        }
+
+        if self._queue_type == QueueType.REDIS:
+            info.update({
+                "redis_url": self.config.redis_url,
+                "max_retries": self.config.max_retries,
+                "timeout": self.config.timeout
+            })
+
+        return info