crawlo 1.2.5__py3-none-any.whl → 1.2.7__py3-none-any.whl

crawlo/middleware/download_delay.py

@@ -1,105 +1,105 @@
-#!/usr/bin/python
-# -*- coding:UTF-8 -*-
-"""
-DownloadDelayMiddleware 中间件
-用于控制请求之间的延迟时间，支持固定延迟和随机延迟
-"""
-
-from asyncio import sleep
-from random import uniform
-from crawlo.utils.log import get_logger
-from crawlo.exceptions import NotConfiguredError
-
-
-class DownloadDelayMiddleware(object):
-    """
-    DownloadDelayMiddleware 中间件
-    用于控制请求之间的延迟时间，支持固定延迟和随机延迟
-    
-    功能特性:
-    - 支持固定延迟时间
-    - 支持随机延迟时间
-    - 提供详细的日志信息
-    - 记录延迟统计信息
-    """
-
-    def __init__(self, settings, log_level, stats=None):
-        """
-        初始化中间件
-        
-        Args:
-            settings: 设置管理器
-            log_level: 日志级别
-            stats: 统计信息收集器（可选）
-        """
-        self.delay = settings.get_float("DOWNLOAD_DELAY")
-        if not self.delay:
-            raise NotConfiguredError("DOWNLOAD_DELAY not set or is zero")
-            
-        self.randomness = settings.get_bool("RANDOMNESS", False)
-        
-        # 安全地获取随机范围配置
-        random_range = settings.get_list("RANDOM_RANGE")
-        if len(random_range) >= 2:
-            try:
-                self.floor = float(random_range[0])
-                self.upper = float(random_range[1])
-            except (ValueError, TypeError):
-                # 如果配置无效，使用默认值
-                self.floor, self.upper = 0.5, 1.5
-        else:
-            # 如果配置不完整，使用默认值
-            self.floor, self.upper = 0.5, 1.5
-            
-        self.logger = get_logger(self.__class__.__name__, log_level)
-        self.stats = stats
-
-    @classmethod
-    def create_instance(cls, crawler):
-        """
-        创建中间件实例
-        
-        Args:
-            crawler: 爬虫实例
-            
-        Returns:
-            DownloadDelayMiddleware: 中间件实例
-        """
-        o = cls(
-            settings=crawler.settings, 
-            log_level=crawler.settings.get('LOG_LEVEL'),
-            stats=getattr(crawler, 'stats', None)
-        )
-        return o
-
-    async def process_request(self, _request, _spider):
-        """
-        处理请求，添加延迟
-        
-        Args:
-            _request: 请求对象
-            _spider: 爬虫实例
-        """
-        if self.randomness:
-            # 计算随机延迟时间
-            delay_time = uniform(self.delay * self.floor, self.delay * self.upper)
-            await sleep(delay_time)
-            
-            # 记录统计信息
-            if self.stats:
-                self.stats.inc_value('download_delay/random_count')
-                self.stats.inc_value('download_delay/random_total_time', delay_time)
-                
-            # 记录日志
-            self.logger.debug(f"应用随机延迟: {delay_time:.2f}秒 (范围: {self.delay * self.floor:.2f} - {self.delay * self.upper:.2f})")
-        else:
-            # 应用固定延迟
-            await sleep(self.delay)
-            
-            # 记录统计信息
-            if self.stats:
-                self.stats.inc_value('download_delay/fixed_count')
-                self.stats.inc_value('download_delay/fixed_total_time', self.delay)
-                
-            # 记录日志
+#!/usr/bin/python
+# -*- coding:UTF-8 -*-
+"""
+DownloadDelayMiddleware 中间件
+用于控制请求之间的延迟时间，支持固定延迟和随机延迟
+"""
+
+from asyncio import sleep
+from random import uniform
+from crawlo.utils.log import get_logger
+from crawlo.exceptions import NotConfiguredError
+
+
+class DownloadDelayMiddleware(object):
+    """
+    DownloadDelayMiddleware 中间件
+    用于控制请求之间的延迟时间，支持固定延迟和随机延迟
+    
+    功能特性:
+    - 支持固定延迟时间
+    - 支持随机延迟时间
+    - 提供详细的日志信息
+    - 记录延迟统计信息
+    """
+
+    def __init__(self, settings, log_level, stats=None):
+        """
+        初始化中间件
+        
+        Args:
+            settings: 设置管理器
+            log_level: 日志级别
+            stats: 统计信息收集器（可选）
+        """
+        self.delay = settings.get_float("DOWNLOAD_DELAY")
+        if not self.delay:
+            raise NotConfiguredError("DOWNLOAD_DELAY not set or is zero")
+            
+        self.randomness = settings.get_bool("RANDOMNESS", False)
+        
+        # 安全地获取随机范围配置
+        random_range = settings.get_list("RANDOM_RANGE")
+        if len(random_range) >= 2:
+            try:
+                self.floor = float(random_range[0])
+                self.upper = float(random_range[1])
+            except (ValueError, TypeError):
+                # 如果配置无效，使用默认值
+                self.floor, self.upper = 0.5, 1.5
+        else:
+            # 如果配置不完整，使用默认值
+            self.floor, self.upper = 0.5, 1.5
+            
+        self.logger = get_logger(self.__class__.__name__, log_level)
+        self.stats = stats
+
+    @classmethod
+    def create_instance(cls, crawler):
+        """
+        创建中间件实例
+        
+        Args:
+            crawler: 爬虫实例
+            
+        Returns:
+            DownloadDelayMiddleware: 中间件实例
+        """
+        o = cls(
+            settings=crawler.settings, 
+            log_level=crawler.settings.get('LOG_LEVEL'),
+            stats=getattr(crawler, 'stats', None)
+        )
+        return o
+
+    async def process_request(self, _request, _spider):
+        """
+        处理请求，添加延迟
+        
+        Args:
+            _request: 请求对象
+            _spider: 爬虫实例
+        """
+        if self.randomness:
+            # 计算随机延迟时间
+            delay_time = uniform(self.delay * self.floor, self.delay * self.upper)
+            await sleep(delay_time)
+            
+            # 记录统计信息
+            if self.stats:
+                self.stats.inc_value('download_delay/random_count')
+                self.stats.inc_value('download_delay/random_total_time', delay_time)
+                
+            # 记录日志
+            self.logger.debug(f"应用随机延迟: {delay_time:.2f}秒 (范围: {self.delay * self.floor:.2f} - {self.delay * self.upper:.2f})")
+        else:
+            # 应用固定延迟
+            await sleep(self.delay)
+            
+            # 记录统计信息
+            if self.stats:
+                self.stats.inc_value('download_delay/fixed_count')
+                self.stats.inc_value('download_delay/fixed_total_time', self.delay)
+                
+            # 记录日志
             self.logger.debug(f"应用固定延迟: {self.delay:.2f}秒")

crawlo/middleware/middleware_manager.py

@@ -1,135 +1,136 @@
-#!/usr/bin/python
-# -*- coding:UTF-8 -*-
-from pprint import pformat
-from types import MethodType
-from asyncio import create_task
-from collections import defaultdict
-from typing import List, Dict, Callable, Optional
-
-from crawlo import Request, Response
-from crawlo.utils.log import get_logger
-from crawlo.project import load_class
-from crawlo.middleware import BaseMiddleware
-from crawlo.project import common_call
-from crawlo.event import ignore_request, response_received
-from crawlo.exceptions import MiddlewareInitError, InvalidOutputError, RequestMethodError, IgnoreRequestError, \
-    NotConfiguredError
-
-
-class MiddlewareManager:
-
-    def __init__(self, crawler):
-        self.crawler = crawler
-        self.logger = get_logger(self.__class__.__name__, crawler.settings.get('LOG_LEVEL'))
-        self.middlewares: List = []
-        self.methods: Dict[str, List[MethodType]] = defaultdict(list)
-        middlewares = self.crawler.settings.get_list('MIDDLEWARES')
-        self._add_middleware(middlewares)
-        self._add_method()
-
-        self.download_method: Callable = crawler.engine.downloader.download
-        self._stats = crawler.stats
-
-    async def _process_request(self, request: Request):
-        for method in self.methods['process_request']:
-            result = await common_call(method, request, self.crawler.spider)
-            if result is None:
-                continue
-            if isinstance(result, (Request, Response)):
-                return result
-            raise InvalidOutputError(
-                f"{method.__self__.__class__.__name__}. must return None or Request or Response, got {type(result).__name__}"
-            )
-        return await self.download_method(request)
-
-    async def _process_response(self, request: Request, response: Response):
-        for method in reversed(self.methods['process_response']):
-            try:
-                response = await common_call(method, request, response, self.crawler.spider)
-            except IgnoreRequestError as exp:
-                create_task(self.crawler.subscriber.notify(ignore_request, exp, request, self.crawler.spider))
-            if isinstance(response, Request):
-                return response
-            if isinstance(response, Response):
-                continue
-            raise InvalidOutputError(
-                f"{method.__self__.__class__.__name__}. must return Request or Response, got {type(response).__name__}"
-            )
-        return response
-
-    async def _process_exception(self, request: Request, exp: Exception):
-        for method in self.methods['process_exception']:
-            response = await common_call(method, request, exp, self.crawler.spider)
-            if response is None:
-                continue
-            if isinstance(response, (Request, Response)):
-                return response
-            if response:
-                break
-            raise InvalidOutputError(
-                f"{method.__self__.__class__.__name__}. must return None or Request or Response, got {type(response).__name__}"
-            )
-        else:
-            raise exp
-
-    async def download(self, request) -> Optional[Response]:
-        """ called in the download method. """
-        try:
-            response = await self._process_request(request)
-        except KeyError:
-            raise RequestMethodError(f"{request.method.lower()} is not supported")
-        except IgnoreRequestError as exp:
-            create_task(self.crawler.subscriber.notify(ignore_request, exp, request, self.crawler.spider))
-            response = await self._process_exception(request, exp)
-        except Exception as exp:
-            self._stats.inc_value(f'download_error/{exp.__class__.__name__}')
-            response = await self._process_exception(request, exp)
-        else:
-            create_task(self.crawler.subscriber.notify(response_received, response, self.crawler.spider))
-            # self.crawler.stats.inc_value('response_received_count')
-        if isinstance(response, Response):
-            response = await self._process_response(request, response)
-        if isinstance(response, Request):
-            await self.crawler.engine.enqueue_request(request)
-            return None
-        return response
-
-    @classmethod
-    def create_instance(cls, *args, **kwargs):
-        return cls(*args, **kwargs)
-
-    def _add_middleware(self, middlewares):
-        enabled_middlewares = [m for m in middlewares if self._validate_middleware(m)]
-        if enabled_middlewares:
-            self.logger.info(f'enabled middleware：\n {pformat(enabled_middlewares)}')
-
-    def _validate_middleware(self, middleware):
-        middleware_cls = load_class(middleware)
-        if not hasattr(middleware_cls, 'create_instance'):
-            raise MiddlewareInitError(
-                f"Middleware init failed, must inherit from `BaseMiddleware` or have a `create_instance` method"
-            )
-        try:
-            instance = middleware_cls.create_instance(self.crawler)
-            self.middlewares.append(instance)
-            return True
-        except NotConfiguredError:
-            return False
-
-    def _add_method(self):
-        for middleware in self.middlewares:
-            if hasattr(middleware, 'process_request'):
-                if self._validate_middleware_method(method_name='process_request', middleware=middleware):
-                    self.methods['process_request'].append(middleware.process_request)
-            if hasattr(middleware, 'process_response'):
-                if self._validate_middleware_method(method_name='process_response', middleware=middleware):
-                    self.methods['process_response'].append(middleware.process_response)
-            if hasattr(middleware, 'process_exception'):
-                if self._validate_middleware_method(method_name='process_exception', middleware=middleware):
-                    self.methods['process_exception'].append(middleware.process_exception)
-
-    @staticmethod
-    def _validate_middleware_method(method_name, middleware) -> bool:
-        method = getattr(type(middleware), method_name)
-        base_method = getattr(BaseMiddleware, method_name)
-        return False if method == base_method else True
+#!/usr/bin/python
+# -*- coding:UTF-8 -*-
+from pprint import pformat
+from types import MethodType
+from asyncio import create_task
+from collections import defaultdict
+from typing import List, Dict, Callable, Optional
+
+from crawlo import Request, Response
+from crawlo.utils.log import get_logger
+from crawlo.project import load_class
+from crawlo.middleware import BaseMiddleware
+from crawlo.project import common_call
+from crawlo.event import ignore_request, response_received
+from crawlo.exceptions import MiddlewareInitError, InvalidOutputError, RequestMethodError, IgnoreRequestError, \
+    NotConfiguredError
+
+
+class MiddlewareManager:
+
+    def __init__(self, crawler):
+        self.crawler = crawler
+        self.logger = get_logger(self.__class__.__name__, crawler.settings.get('LOG_LEVEL'))
+        self.middlewares: List = []
+        self.methods: Dict[str, List[MethodType]] = defaultdict(list)
+        middlewares = self.crawler.settings.get_list('MIDDLEWARES')
+        self._add_middleware(middlewares)
+        self._add_method()
+
+        self.download_method: Callable = crawler.engine.downloader.download
+        self._stats = crawler.stats
+
+    async def _process_request(self, request: Request):
+        for method in self.methods['process_request']:
+            result = await common_call(method, request, self.crawler.spider)
+            if result is None:
+                continue
+            if isinstance(result, (Request, Response)):
+                return result
+            raise InvalidOutputError(
+                f"{method.__self__.__class__.__name__}. must return None or Request or Response, got {type(result).__name__}"
+            )
+        return await self.download_method(request)
+
+    async def _process_response(self, request: Request, response: Response):
+        for method in reversed(self.methods['process_response']):
+            try:
+                response = await common_call(method, request, response, self.crawler.spider)
+            except IgnoreRequestError as exp:
+                create_task(self.crawler.subscriber.notify(ignore_request, exp, request, self.crawler.spider))
+            if isinstance(response, Request):
+                return response
+            if isinstance(response, Response):
+                continue
+            raise InvalidOutputError(
+                f"{method.__self__.__class__.__name__}. must return Request or Response, got {type(response).__name__}"
+            )
+        return response
+
+    async def _process_exception(self, request: Request, exp: Exception):
+        for method in self.methods['process_exception']:
+            response = await common_call(method, request, exp, self.crawler.spider)
+            if response is None:
+                continue
+            if isinstance(response, (Request, Response)):
+                return response
+            if response:
+                break
+            raise InvalidOutputError(
+                f"{method.__self__.__class__.__name__}. must return None or Request or Response, got {type(response).__name__}"
+            )
+        else:
+            raise exp
+
+    async def download(self, request) -> Optional[Response]:
+        """ called in the download method. """
+        try:
+            response = await self._process_request(request)
+        except KeyError:
+            raise RequestMethodError(f"{request.method.lower()} is not supported")
+        except IgnoreRequestError as exp:
+            create_task(self.crawler.subscriber.notify(ignore_request, exp, request, self.crawler.spider))
+            response = await self._process_exception(request, exp)
+        except Exception as exp:
+            self._stats.inc_value(f'download_error/{exp.__class__.__name__}')
+            response = await self._process_exception(request, exp)
+        else:
+            create_task(self.crawler.subscriber.notify(response_received, response, self.crawler.spider))
+            # self.crawler.stats.inc_value('response_received_count')
+        if isinstance(response, Response):
+            response = await self._process_response(request, response)
+        if isinstance(response, Request):
+            await self.crawler.engine.enqueue_request(request)
+            return None
+        return response
+
+    @classmethod
+    def create_instance(cls, *args, **kwargs):
+        return cls(*args, **kwargs)
+
+    def _add_middleware(self, middlewares):
+        enabled_middlewares = [m for m in middlewares if self._validate_middleware(m)]
+        if enabled_middlewares:
+            # 恢复INFO级别日志，保留关键的启用信息
+            self.logger.info(f'enabled middleware：\n {pformat(enabled_middlewares)}')
+
+    def _validate_middleware(self, middleware):
+        middleware_cls = load_class(middleware)
+        if not hasattr(middleware_cls, 'create_instance'):
+            raise MiddlewareInitError(
+                f"Middleware init failed, must inherit from `BaseMiddleware` or have a `create_instance` method"
+            )
+        try:
+            instance = middleware_cls.create_instance(self.crawler)
+            self.middlewares.append(instance)
+            return True
+        except NotConfiguredError:
+            return False
+
+    def _add_method(self):
+        for middleware in self.middlewares:
+            if hasattr(middleware, 'process_request'):
+                if self._validate_middleware_method(method_name='process_request', middleware=middleware):
+                    self.methods['process_request'].append(middleware.process_request)
+            if hasattr(middleware, 'process_response'):
+                if self._validate_middleware_method(method_name='process_response', middleware=middleware):
+                    self.methods['process_response'].append(middleware.process_response)
+            if hasattr(middleware, 'process_exception'):
+                if self._validate_middleware_method(method_name='process_exception', middleware=middleware):
+                    self.methods['process_exception'].append(middleware.process_exception)
+
+    @staticmethod
+    def _validate_middleware_method(method_name, middleware) -> bool:
+        method = getattr(type(middleware), method_name)
+        base_method = getattr(BaseMiddleware, method_name)
+        return False if method == base_method else True