crawlo 1.2.0__py3-none-any.whl → 1.2.1__py3-none-any.whl

tests/test_double_crawlo_fix_simple.py

@@ -1,125 +1,125 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-双重 crawlo 前缀问题修复简单测试脚本
-用于验证 Redis 队列名称中双重 crawlo 前缀问题的修复，不依赖于实际的 Redis 连接
-"""
-import sys
-import os
-import asyncio
-import traceback
-
-# 添加项目根目录到路径
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
-
-# 导入相关模块
-from crawlo.queue.redis_priority_queue import RedisPriorityQueue
-
-
-def test_redis_queue_naming():
-    """测试 Redis 队列命名修复"""
-    print("🚀 开始测试 Redis 队列命名修复...")
-    print("=" * 50)
-    
-    test_cases = [
-        {
-            "name": "正常命名",
-            "queue_name": "crawlo:test_project:queue:requests",
-            "expected_queue": "crawlo:test_project:queue:requests",
-            "expected_processing": "crawlo:test_project:queue:processing",
-            "expected_failed": "crawlo:test_project:queue:failed"
-        },
-        {
-            "name": "双重 crawlo 前缀",
-            "queue_name": "crawlo:crawlo:queue:requests",
-            "expected_queue": "crawlo:crawlo:queue:requests",  # 保持原始名称不变
-            "expected_processing": "crawlo:crawlo:queue:processing",  # 保持一致的双重前缀
-            "expected_failed": "crawlo:crawlo:queue:failed"
-        },
-        {
-            "name": "三重 crawlo 前缀",
-            "queue_name": "crawlo:crawlo:crawlo:queue:requests",
-            "expected_queue": "crawlo:crawlo:crawlo:queue:requests",  # 保持原始名称不变
-            "expected_processing": "crawlo:crawlo:crawlo:queue:processing",  # 保持一致的前缀
-            "expected_failed": "crawlo:crawlo:crawlo:queue:failed"
-        },
-        {
-            "name": "无 crawlo 前缀",
-            "queue_name": "test_project:queue:requests",
-            "expected_queue": "test_project:queue:requests",  # 保持原始名称不变
-            "expected_processing": "test_project:queue:processing",
-            "expected_failed": "test_project:queue:failed"
-        }
-    ]
-    
-    try:
-        for i, test_case in enumerate(test_cases, 1):
-            print(f"测试 {i}: {test_case['name']}")
-            print(f"  输入队列名称: {test_case['queue_name']}")
-            
-            # 测试 RedisPriorityQueue 初始化
-            try:
-                queue = RedisPriorityQueue(
-                    redis_url="redis://127.0.0.1:6379/15",
-                    queue_name=test_case['queue_name'],
-                    module_name="test_project"
-                )
-                
-                print(f"  修复后队列名称: {queue.queue_name}")
-                print(f"  修复后处理队列: {queue.processing_queue}")
-                print(f"  修复后失败队列: {queue.failed_queue}")
-                
-                # 验证结果
-                assert queue.queue_name == test_case['expected_queue'], \
-                    f"队列名称不匹配: {queue.queue_name} != {test_case['expected_queue']}"
-                assert queue.processing_queue == test_case['expected_processing'], \
-                    f"处理队列名称不匹配: {queue.processing_queue} != {test_case['expected_processing']}"
-                assert queue.failed_queue == test_case['expected_failed'], \
-                    f"失败队列名称不匹配: {queue.failed_queue} != {test_case['expected_failed']}"
-                
-                print("  ✅ 测试通过")
-            except Exception as e:
-                print(f"  ❌ 测试失败: {e}")
-                traceback.print_exc()
-                return False
-            
-            print()
-        
-        print("✅ Redis 队列命名修复测试通过！")
-        return True
-        
-    except Exception as e:
-        print(f"❌ Redis 队列命名修复测试失败: {e}")
-        traceback.print_exc()
-        return False
-
-
-def main():
-    """主测试函数"""
-    print("🚀 开始双重 crawlo 前缀问题修复测试...")
-    print("=" * 50)
-    
-    try:
-        # 测试 Redis 队列命名修复
-        redis_test_success = test_redis_queue_naming()
-        print()
-        
-        print("=" * 50)
-        if redis_test_success:
-            print("🎉 Redis 队列命名修复测试通过！双重 crawlo 前缀问题已修复")
-        else:
-            print("❌ Redis 队列命名修复测试失败，请检查实现")
-            return 1
-            
-    except Exception as e:
-        print("=" * 50)
-        print(f"❌ 测试过程中发生异常: {e}")
-        traceback.print_exc()
-        return 1
-    
-    return 0
-
-
-if __name__ == "__main__":
-    exit_code = main()
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+双重 crawlo 前缀问题修复简单测试脚本
+用于验证 Redis 队列名称中双重 crawlo 前缀问题的修复，不依赖于实际的 Redis 连接
+"""
+import sys
+import os
+import asyncio
+import traceback
+
+# 添加项目根目录到路径
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
+
+# 导入相关模块
+from crawlo.queue.redis_priority_queue import RedisPriorityQueue
+
+
+def test_redis_queue_naming():
+    """测试 Redis 队列命名修复"""
+    print("🚀 开始测试 Redis 队列命名修复...")
+    print("=" * 50)
+    
+    test_cases = [
+        {
+            "name": "正常命名",
+            "queue_name": "crawlo:test_project:queue:requests",
+            "expected_queue": "crawlo:test_project:queue:requests",
+            "expected_processing": "crawlo:test_project:queue:processing",
+            "expected_failed": "crawlo:test_project:queue:failed"
+        },
+        {
+            "name": "双重 crawlo 前缀",
+            "queue_name": "crawlo:crawlo:queue:requests",
+            "expected_queue": "crawlo:crawlo:queue:requests",  # 保持原始名称不变
+            "expected_processing": "crawlo:crawlo:queue:processing",  # 保持一致的双重前缀
+            "expected_failed": "crawlo:crawlo:queue:failed"
+        },
+        {
+            "name": "三重 crawlo 前缀",
+            "queue_name": "crawlo:crawlo:crawlo:queue:requests",
+            "expected_queue": "crawlo:crawlo:crawlo:queue:requests",  # 保持原始名称不变
+            "expected_processing": "crawlo:crawlo:crawlo:queue:processing",  # 保持一致的前缀
+            "expected_failed": "crawlo:crawlo:crawlo:queue:failed"
+        },
+        {
+            "name": "无 crawlo 前缀",
+            "queue_name": "test_project:queue:requests",
+            "expected_queue": "test_project:queue:requests",  # 保持原始名称不变
+            "expected_processing": "test_project:queue:processing",
+            "expected_failed": "test_project:queue:failed"
+        }
+    ]
+    
+    try:
+        for i, test_case in enumerate(test_cases, 1):
+            print(f"测试 {i}: {test_case['name']}")
+            print(f"  输入队列名称: {test_case['queue_name']}")
+            
+            # 测试 RedisPriorityQueue 初始化
+            try:
+                queue = RedisPriorityQueue(
+                    redis_url="redis://127.0.0.1:6379/15",
+                    queue_name=test_case['queue_name'],
+                    module_name="test_project"
+                )
+                
+                print(f"  修复后队列名称: {queue.queue_name}")
+                print(f"  修复后处理队列: {queue.processing_queue}")
+                print(f"  修复后失败队列: {queue.failed_queue}")
+                
+                # 验证结果
+                assert queue.queue_name == test_case['expected_queue'], \
+                    f"队列名称不匹配: {queue.queue_name} != {test_case['expected_queue']}"
+                assert queue.processing_queue == test_case['expected_processing'], \
+                    f"处理队列名称不匹配: {queue.processing_queue} != {test_case['expected_processing']}"
+                assert queue.failed_queue == test_case['expected_failed'], \
+                    f"失败队列名称不匹配: {queue.failed_queue} != {test_case['expected_failed']}"
+                
+                print("  ✅ 测试通过")
+            except Exception as e:
+                print(f"  ❌ 测试失败: {e}")
+                traceback.print_exc()
+                return False
+            
+            print()
+        
+        print("✅ Redis 队列命名修复测试通过！")
+        return True
+        
+    except Exception as e:
+        print(f"❌ Redis 队列命名修复测试失败: {e}")
+        traceback.print_exc()
+        return False
+
+
+def main():
+    """主测试函数"""
+    print("🚀 开始双重 crawlo 前缀问题修复测试...")
+    print("=" * 50)
+    
+    try:
+        # 测试 Redis 队列命名修复
+        redis_test_success = test_redis_queue_naming()
+        print()
+        
+        print("=" * 50)
+        if redis_test_success:
+            print("🎉 Redis 队列命名修复测试通过！双重 crawlo 前缀问题已修复")
+        else:
+            print("❌ Redis 队列命名修复测试失败，请检查实现")
+            return 1
+            
+    except Exception as e:
+        print("=" * 50)
+        print(f"❌ 测试过程中发生异常: {e}")
+        traceback.print_exc()
+        return 1
+    
+    return 0
+
+
+if __name__ == "__main__":
+    exit_code = main()
     sys.exit(exit_code)

tests/test_download_delay_middleware.py

@@ -0,0 +1,222 @@
+#!/usr/bin/python
+# -*- coding:UTF-8 -*-
+"""
+DownloadDelayMiddleware 测试文件
+用于测试下载延迟中间件的功能
+"""
+
+import asyncio
+import unittest
+from unittest.mock import Mock, patch
+
+from crawlo.middleware.download_delay import DownloadDelayMiddleware
+from crawlo.exceptions import NotConfiguredError
+from crawlo.settings.setting_manager import SettingManager
+
+
+class MockLogger:
+    """Mock Logger 类，用于测试日志输出"""
+    def __init__(self, name, level=None):
+        self.name = name
+        self.level = level
+        self.logs = []
+
+    def debug(self, msg):
+        self.logs.append(('debug', msg))
+
+    def info(self, msg):
+        self.logs.append(('info', msg))
+
+    def warning(self, msg):
+        self.logs.append(('warning', msg))
+
+    def error(self, msg):
+        self.logs.append(('error', msg))
+
+
+class MockStats:
+    """Mock Stats 类，用于测试统计信息"""
+    def __init__(self):
+        self.stats = {}
+
+    def inc_value(self, key, value=1):
+        if key in self.stats:
+            self.stats[key] += value
+        else:
+            self.stats[key] = value
+
+
+class TestDownloadDelayMiddleware(unittest.TestCase):
+    """DownloadDelayMiddleware 测试类"""
+
+    def setUp(self):
+        """测试前准备"""
+        # 创建设置管理器
+        self.settings = SettingManager()
+        
+        # 创建爬虫模拟对象
+        self.crawler = Mock()
+        self.crawler.settings = self.settings
+        
+        # 创建请求和爬虫模拟对象
+        self.request = Mock()
+        self.spider = Mock()
+
+    @patch('crawlo.utils.log.get_logger')
+    def test_middleware_initialization_without_delay(self, mock_get_logger):
+        """测试没有设置DOWNLOAD_DELAY时中间件初始化"""
+        # 设置DOWNLOAD_DELAY为0
+        self.settings.set('DOWNLOAD_DELAY', 0)
+        mock_get_logger.return_value = MockLogger('DownloadDelayMiddleware')
+        
+        # 应该抛出NotConfiguredError异常
+        with self.assertRaises(NotConfiguredError) as context:
+            DownloadDelayMiddleware.create_instance(self.crawler)
+        
+        self.assertIn("DOWNLOAD_DELAY not set or is zero", str(context.exception))
+
+    @patch('crawlo.utils.log.get_logger')
+    def test_middleware_initialization_with_delay(self, mock_get_logger):
+        """测试正确设置DOWNLOAD_DELAY时中间件初始化"""
+        # 设置DOWNLOAD_DELAY
+        self.settings.set('DOWNLOAD_DELAY', 2.0)
+        self.settings.set('RANDOMNESS', False)
+        self.settings.set('RANDOM_RANGE', [0.5, 1.5])
+        self.settings.set('LOG_LEVEL', 'INFO')
+        
+        mock_get_logger.return_value = MockLogger('DownloadDelayMiddleware')
+        
+        # 应该正常创建实例
+        middleware = DownloadDelayMiddleware.create_instance(self.crawler)
+        self.assertIsInstance(middleware, DownloadDelayMiddleware)
+        self.assertEqual(middleware.delay, 2.0)
+        self.assertFalse(middleware.randomness)
+
+    @patch('crawlo.utils.log.get_logger')
+    def test_middleware_initialization_with_randomness(self, mock_get_logger):
+        """测试启用随机延迟时中间件初始化"""
+        # 设置DOWNLOAD_DELAY和随机配置
+        self.settings.set('DOWNLOAD_DELAY', 1.0)
+        self.settings.set('RANDOMNESS', True)
+        self.settings.set('RANDOM_RANGE', [0.5, 2.0])
+        self.settings.set('LOG_LEVEL', 'INFO')
+        
+        mock_get_logger.return_value = MockLogger('DownloadDelayMiddleware')
+        
+        # 应该正常创建实例
+        middleware = DownloadDelayMiddleware.create_instance(self.crawler)
+        self.assertIsInstance(middleware, DownloadDelayMiddleware)
+        self.assertEqual(middleware.delay, 1.0)
+        self.assertTrue(middleware.randomness)
+        self.assertEqual(middleware.floor, 0.5)
+        self.assertEqual(middleware.upper, 2.0)
+
+    @patch('crawlo.utils.log.get_logger')
+    def test_middleware_initialization_with_invalid_random_range(self, mock_get_logger):
+        """测试随机范围配置无效时中间件初始化"""
+        # 设置DOWNLOAD_DELAY和无效的随机配置
+        self.settings.set('DOWNLOAD_DELAY', 1.0)
+        self.settings.set('RANDOMNESS', True)
+        self.settings.set('RANDOM_RANGE', ['invalid', 'range'])
+        self.settings.set('LOG_LEVEL', 'INFO')
+        
+        mock_get_logger.return_value = MockLogger('DownloadDelayMiddleware')
+        
+        # 应该正常创建实例，使用默认随机范围
+        middleware = DownloadDelayMiddleware.create_instance(self.crawler)
+        self.assertIsInstance(middleware, DownloadDelayMiddleware)
+        self.assertEqual(middleware.floor, 0.5)
+        self.assertEqual(middleware.upper, 1.5)
+
+    @patch('crawlo.utils.log.get_logger')
+    def test_middleware_initialization_with_incomplete_random_range(self, mock_get_logger):
+        """测试随机范围配置不完整时中间件初始化"""
+        # 设置DOWNLOAD_DELAY和不完整的随机配置
+        self.settings.set('DOWNLOAD_DELAY', 1.0)
+        self.settings.set('RANDOMNESS', True)
+        self.settings.set('RANDOM_RANGE', [0.8])  # 只有一个值
+        self.settings.set('LOG_LEVEL', 'INFO')
+        
+        mock_get_logger.return_value = MockLogger('DownloadDelayMiddleware')
+        
+        # 应该正常创建实例，使用默认随机范围
+        middleware = DownloadDelayMiddleware.create_instance(self.crawler)
+        self.assertIsInstance(middleware, DownloadDelayMiddleware)
+        self.assertEqual(middleware.floor, 0.5)
+        self.assertEqual(middleware.upper, 1.5)
+
+    @patch('crawlo.middleware.download_delay.sleep')
+    @patch('crawlo.utils.log.get_logger')
+    def test_process_request_without_randomness(self, mock_get_logger, mock_sleep):
+        """测试不启用随机延迟时的请求处理"""
+        # 设置DOWNLOAD_DELAY
+        self.settings.set('DOWNLOAD_DELAY', 1.5)
+        self.settings.set('RANDOMNESS', False)
+        self.settings.set('LOG_LEVEL', 'DEBUG')  # 使用DEBUG级别以启用日志
+        
+        mock_logger = MockLogger('DownloadDelayMiddleware')
+        mock_get_logger.return_value = mock_logger
+        
+        middleware = DownloadDelayMiddleware.create_instance(self.crawler)
+        
+        # 执行请求处理
+        asyncio.run(middleware.process_request(self.request, self.spider))
+        
+        # 验证sleep被调用且参数正确
+        mock_sleep.assert_called_once_with(1.5)
+
+    @patch('crawlo.middleware.download_delay.sleep')
+    @patch('crawlo.middleware.download_delay.uniform')
+    @patch('crawlo.utils.log.get_logger')
+    def test_process_request_with_randomness(self, mock_get_logger, mock_uniform, mock_sleep):
+        """测试启用随机延迟时的请求处理"""
+        # 设置DOWNLOAD_DELAY和随机配置
+        self.settings.set('DOWNLOAD_DELAY', 2.0)
+        self.settings.set('RANDOMNESS', True)
+        self.settings.set('RANDOM_RANGE', [0.5, 1.5])
+        self.settings.set('LOG_LEVEL', 'DEBUG')  # 使用DEBUG级别以启用日志
+        
+        mock_logger = MockLogger('DownloadDelayMiddleware')
+        mock_get_logger.return_value = mock_logger
+        mock_uniform.return_value = 2.5  # 模拟随机数返回2.5
+        
+        middleware = DownloadDelayMiddleware.create_instance(self.crawler)
+        
+        # 执行请求处理
+        asyncio.run(middleware.process_request(self.request, self.spider))
+        
+        # 验证uniform被调用且参数正确
+        mock_uniform.assert_called_once_with(1.0, 3.0)  # 2.0*0.5=1.0, 2.0*1.5=3.0
+        # 验证sleep被调用且参数正确
+        mock_sleep.assert_called_once_with(2.5)
+
+    @patch('crawlo.middleware.download_delay.sleep')
+    @patch('crawlo.utils.log.get_logger')
+    def test_process_request_with_stats(self, mock_get_logger, mock_sleep):
+        """测试带统计信息的请求处理"""
+        # 设置DOWNLOAD_DELAY
+        self.settings.set('DOWNLOAD_DELAY', 1.0)
+        self.settings.set('RANDOMNESS', False)
+        self.settings.set('LOG_LEVEL', 'INFO')
+        
+        # 添加统计收集器到爬虫
+        mock_stats = MockStats()
+        self.crawler.stats = mock_stats
+        
+        mock_logger = MockLogger('DownloadDelayMiddleware')
+        mock_get_logger.return_value = mock_logger
+        
+        middleware = DownloadDelayMiddleware.create_instance(self.crawler)
+        
+        # 执行请求处理
+        asyncio.run(middleware.process_request(self.request, self.spider))
+        
+        # 验证统计信息
+        self.assertIn('download_delay/fixed_count', mock_stats.stats)
+        self.assertEqual(mock_stats.stats['download_delay/fixed_count'], 1)
+        self.assertIn('download_delay/fixed_total_time', mock_stats.stats)
+        self.assertEqual(mock_stats.stats['download_delay/fixed_total_time'], 1.0)
+
+
+if __name__ == '__main__':
+    unittest.main()