crawlo 1.1.4__py3-none-any.whl → 1.1.6__py3-none-any.whl

tests/test_config_validator.py

@@ -0,0 +1,194 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+配置验证器测试脚本
+用于验证配置验证器的功能
+"""
+import sys
+import os
+import unittest
+
+# 添加项目根目录到路径
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
+
+from crawlo.config_validator import ConfigValidator, validate_config
+
+
+class TestConfigValidator(unittest.TestCase):
+    """配置验证器测试类"""
+    
+    def setUp(self):
+        """测试前准备"""
+        self.validator = ConfigValidator()
+    
+    def test_valid_standalone_config(self):
+        """测试有效的单机配置"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'memory',
+            'CONCURRENCY': 8,
+            'DOWNLOAD_DELAY': 1.0,
+            'DOWNLOAD_TIMEOUT': 30,
+            'CONNECTION_POOL_LIMIT': 50,
+            'SCHEDULER_MAX_QUEUE_SIZE': 2000,
+            'LOG_LEVEL': 'INFO',
+            'MIDDLEWARES': [
+                'crawlo.middleware.request_ignore.RequestIgnoreMiddleware',
+                'crawlo.middleware.download_delay.DownloadDelayMiddleware'
+            ],
+            'PIPELINES': [
+                'crawlo.pipelines.console_pipeline.ConsolePipeline'
+            ]
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertTrue(is_valid)
+        self.assertEqual(len(errors), 0)
+    
+    def test_valid_distributed_config(self):
+        """测试有效的分布式配置"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'redis',
+            'CONCURRENCY': 16,
+            'DOWNLOAD_DELAY': 1.0,
+            'DOWNLOAD_TIMEOUT': 30,
+            'CONNECTION_POOL_LIMIT': 50,
+            'SCHEDULER_MAX_QUEUE_SIZE': 2000,
+            'SCHEDULER_QUEUE_NAME': 'crawlo:test_project:queue:requests',
+            'REDIS_HOST': '127.0.0.1',
+            'REDIS_PORT': 6379,
+            'REDIS_URL': 'redis://127.0.0.1:6379/0',
+            'LOG_LEVEL': 'INFO',
+            'MIDDLEWARES': [
+                'crawlo.middleware.request_ignore.RequestIgnoreMiddleware',
+                'crawlo.middleware.download_delay.DownloadDelayMiddleware'
+            ],
+            'PIPELINES': [
+                'crawlo.pipelines.console_pipeline.ConsolePipeline'
+            ]
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertTrue(is_valid)
+        self.assertEqual(len(errors), 0)
+    
+    def test_invalid_project_name(self):
+        """测试无效的项目名称"""
+        config = {
+            'PROJECT_NAME': '',  # 空字符串
+            'QUEUE_TYPE': 'memory',
+            'CONCURRENCY': 8
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertFalse(is_valid)
+        self.assertIn("PROJECT_NAME 必须是非空字符串", errors)
+    
+    def test_invalid_concurrency(self):
+        """测试无效的并发数"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'memory',
+            'CONCURRENCY': -1  # 负数
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertFalse(is_valid)
+        self.assertIn("CONCURRENCY 必须是正整数", errors)
+    
+    def test_invalid_queue_type(self):
+        """测试无效的队列类型"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'invalid_type',  # 无效类型
+            'CONCURRENCY': 8
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertFalse(is_valid)
+        self.assertIn("QUEUE_TYPE 必须是以下值之一: ['memory', 'redis', 'auto']", errors)
+    
+    def test_invalid_redis_queue_name(self):
+        """测试无效的Redis队列名称"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'redis',
+            'CONCURRENCY': 8,
+            'SCHEDULER_QUEUE_NAME': 'invalid_queue_name'  # 不符合命名规范
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertTrue(is_valid)  # 队列名称错误是警告，不是错误
+        self.assertGreater(len(warnings), 0)
+        self.assertTrue(any("Redis队列名称" in warning for warning in warnings))
+    
+    def test_missing_redis_queue_name(self):
+        """测试缺少Redis队列名称"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'redis',
+            'CONCURRENCY': 8
+            # 缺少 SCHEDULER_QUEUE_NAME
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertFalse(is_valid)
+        self.assertIn("使用Redis队列时，SCHEDULER_QUEUE_NAME 不能为空", errors)
+    
+    def test_invalid_redis_port(self):
+        """测试无效的Redis端口"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'redis',
+            'CONCURRENCY': 8,
+            'SCHEDULER_QUEUE_NAME': 'crawlo:test_project:queue:requests',
+            'REDIS_HOST': '127.0.0.1',
+            'REDIS_PORT': 99999  # 无效端口
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertFalse(is_valid)
+        self.assertIn("REDIS_PORT 必须是1-65535之间的整数", errors)
+    
+    def test_invalid_log_level(self):
+        """测试无效的日志级别"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'memory',
+            'CONCURRENCY': 8,
+            'LOG_LEVEL': 'INVALID_LEVEL'  # 无效日志级别
+        }
+        
+        is_valid, errors, warnings = self.validator.validate(config)
+        self.assertFalse(is_valid)
+        self.assertIn("LOG_LEVEL 必须是以下值之一: ['DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL']", errors)
+    
+    def test_convenience_function(self):
+        """测试便利函数"""
+        config = {
+            'PROJECT_NAME': 'test_project',
+            'QUEUE_TYPE': 'memory',
+            'CONCURRENCY': 8,
+            'LOG_LEVEL': 'INFO'
+        }
+        
+        is_valid, errors, warnings = validate_config(config)
+        self.assertTrue(is_valid)
+        self.assertEqual(len(errors), 0)
+
+
+def main():
+    """主测试函数"""
+    print("🚀 开始配置验证器测试...")
+    print("=" * 50)
+    
+    # 运行测试
+    unittest.main(argv=['first-arg-is-ignored'], exit=False, verbosity=2)
+    
+    print("=" * 50)
+    print("✅ 配置验证器测试完成")
+
+
+if __name__ == "__main__":
+    main()

tests/test_date_tools.py

@@ -0,0 +1,124 @@
+#!/usr/bin/python
+# -*- coding: UTF-8 -*-
+"""
+日期工具测试
+"""
+import unittest
+from crawlo.tools import (
+    TimeUtils,
+    parse_time,
+    format_time,
+    time_diff,
+    to_timestamp,
+    to_datetime,
+    now,
+    to_timezone,
+    to_utc,
+    to_local
+)
+
+
+class TestDateTools(unittest.TestCase):
+    """日期工具测试类"""
+
+    def test_date_parsing(self):
+        """测试日期解析功能"""
+        # 测试标准格式
+        dt = parse_time("2025-09-10 14:30:00")
+        self.assertIsNotNone(dt)
+        self.assertEqual(dt.year, 2025)
+        self.assertEqual(dt.month, 9)
+        self.assertEqual(dt.day, 10)
+        self.assertEqual(dt.hour, 14)
+        self.assertEqual(dt.minute, 30)
+        self.assertEqual(dt.second, 0)
+        
+        # 测试不同格式
+        dt2 = parse_time("September 10, 2025 2:30 PM")
+        self.assertIsNotNone(dt2)
+        self.assertEqual(dt2.year, 2025)
+        self.assertEqual(dt2.month, 9)
+        self.assertEqual(dt2.day, 10)
+        self.assertEqual(dt2.hour, 14)
+        self.assertEqual(dt2.minute, 30)
+
+    def test_date_formatting(self):
+        """测试日期格式化功能"""
+        dt = parse_time("2025-09-10 14:30:00")
+        
+        # 测试标准格式
+        formatted = format_time(dt, "%Y-%m-%d")
+        self.assertEqual(formatted, "2025-09-10")
+        
+        # 测试中文格式
+        formatted_cn = format_time(dt, "%Y年%m月%d日")
+        self.assertEqual(formatted_cn, "2025年09月10日")
+
+    def test_time_difference(self):
+        """测试时间差计算功能"""
+        start_time = "2025-09-10 10:00:00"
+        end_time = "2025-09-10 14:30:45"
+        
+        # 测试秒级差值
+        diff_seconds = time_diff(start_time, end_time, "seconds")
+        self.assertEqual(diff_seconds, 16245)  # 4小时30分45秒 = 16245秒
+        
+        # 测试分钟级差值
+        diff_minutes = time_diff(start_time, end_time, "minutes")
+        self.assertEqual(diff_minutes, 270)  # 270分钟
+        
+        # 测试小时级差值
+        diff_hours = time_diff(start_time, end_time, "hours")
+        self.assertEqual(diff_hours, 4)  # 4小时
+
+    def test_timestamp_conversion(self):
+        """测试时间戳转换功能"""
+        # 测试转换为时间戳
+        dt = parse_time("2025-09-10 14:30:00")
+        timestamp = to_timestamp(dt)
+        self.assertIsInstance(timestamp, float)
+        
+        # 测试从时间戳转换
+        dt_from_ts = to_datetime(timestamp)
+        self.assertEqual(dt.year, dt_from_ts.year)
+        self.assertEqual(dt.month, dt_from_ts.month)
+        self.assertEqual(dt.day, dt_from_ts.day)
+        self.assertEqual(dt.hour, dt_from_ts.hour)
+        self.assertEqual(dt.minute, dt_from_ts.minute)
+        self.assertEqual(dt.second, dt_from_ts.second)
+
+    def test_timezone_conversion(self):
+        """测试时区转换功能"""
+        dt = parse_time("2025-09-10 14:30:00")
+        
+        # 测试UTC转换
+        utc_time = to_utc(dt)
+        self.assertIsNotNone(utc_time)
+        
+        # 测试本地时区转换
+        local_time = to_local(dt)
+        self.assertIsNotNone(local_time)
+        
+        # 测试指定时区转换
+        ny_time = to_timezone(dt, "America/New_York")
+        self.assertIsNotNone(ny_time)
+
+    def test_time_utils_class(self):
+        """测试TimeUtils类方法"""
+        # 测试日期加减
+        base_date = "2025-09-10"
+        plus_30_days = TimeUtils.add_days(base_date, 30)
+        self.assertEqual(plus_30_days.month, 10)
+        self.assertEqual(plus_30_days.day, 10)
+        
+        # 测试月份加减
+        plus_3_months = TimeUtils.add_months(base_date, 3)
+        self.assertEqual(plus_3_months.month, 12)
+        
+        # 测试闰年判断
+        self.assertTrue(TimeUtils.is_leap_year(2024))
+        self.assertFalse(TimeUtils.is_leap_year(2025))
+
+
+if __name__ == '__main__':
+    unittest.main()

tests/test_double_crawlo_fix.py

@@ -0,0 +1,208 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+双重 crawlo 前缀问题修复测试脚本
+用于验证 Redis 队列名称中双重 crawlo 前缀问题的修复
+"""
+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.queue_manager import QueueManager, QueueConfig, QueueType
+from crawlo.queue.redis_priority_queue import RedisPriorityQueue
+
+
+async 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_module": "test_project",
+            "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_module": "crawlo",
+            "expected_queue": "crawlo:crawlo:queue:requests",
+            "expected_processing": "crawlo:crawlo:queue:processing",
+            "expected_failed": "crawlo:crawlo:queue:failed"
+        },
+        {
+            "name": "三重 crawlo 前缀",
+            "queue_name": "crawlo:crawlo:crawlo:queue:requests",
+            "expected_module": "crawlo",
+            "expected_queue": "crawlo:crawlo:queue:requests",
+            "expected_processing": "crawlo:crawlo:queue:processing",
+            "expected_failed": "crawlo:crawlo:queue:failed"
+        },
+        {
+            "name": "无 crawlo 前缀",
+            "queue_name": "test_project:queue:requests",
+            "expected_module": "test_project",
+            "expected_queue": "crawlo:test_project:queue:requests",
+            "expected_processing": "crawlo:test_project:queue:processing",
+            "expected_failed": "crawlo: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 直接初始化
+            try:
+                queue = RedisPriorityQueue(
+                    redis_url="redis://127.0.0.1:6379/15",
+                    queue_name=test_case['queue_name'],
+                    module_name=test_case['expected_module']
+                )
+                
+                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
+
+
+async def test_queue_manager_naming():
+    """测试队列管理器中的项目名称提取"""
+    print("🚀 开始测试队列管理器项目名称提取...")
+    print("=" * 50)
+    
+    test_cases = [
+        {
+            "name": "正常命名",
+            "queue_name": "crawlo:test_project:queue:requests",
+            "expected_module": "test_project"
+        },
+        {
+            "name": "双重 crawlo 前缀",
+            "queue_name": "crawlo:crawlo:queue:requests",
+            "expected_module": "crawlo"
+        },
+        {
+            "name": "三重 crawlo 前缀",
+            "queue_name": "crawlo:crawlo:crawlo:queue:requests",
+            "expected_module": "crawlo"
+        },
+        {
+            "name": "无 crawlo 前缀",
+            "queue_name": "test_project:queue:requests",
+            "expected_module": "test_project"
+        }
+    ]
+    
+    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']}")
+            
+            # 测试队列管理器中的项目名称提取逻辑
+            if ':' in test_case['queue_name']:
+                parts = test_case['queue_name'].split(':')
+                if len(parts) >= 2:
+                    # 处理可能的双重 crawlo 前缀
+                    if parts[0] == "crawlo" and parts[1] == "crawlo":
+                        # 双重 crawlo 前缀，取第三个部分作为项目名称
+                        if len(parts) >= 3:
+                            project_name = parts[2]
+                        else:
+                            project_name = "default"
+                    elif parts[0] == "crawlo":
+                        # 正常的 crawlo 前缀，取第二个部分作为项目名称
+                        project_name = parts[1]
+                    else:
+                        # 没有 crawlo 前缀，使用第一个部分作为项目名称
+                        project_name = parts[0]
+                else:
+                    project_name = test_case['queue_name'] or "default"
+            else:
+                project_name = test_case['queue_name'] or "default"
+            
+            print(f"  提取的项目名称: {project_name}")
+            print(f"  期望的项目名称: {test_case['expected_module']}")
+            
+            # 验证结果
+            assert project_name == test_case['expected_module'], \
+                f"项目名称不匹配: {project_name} != {test_case['expected_module']}"
+            
+            print("  ✅ 测试通过")
+            print()
+        
+        print("✅ 队列管理器项目名称提取测试通过！")
+        return True
+        
+    except Exception as e:
+        print(f"❌ 队列管理器项目名称提取测试失败: {e}")
+        traceback.print_exc()
+        return False
+
+
+async 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 = await test_redis_queue_naming()
+        print()
+        
+        # 测试队列管理器项目名称提取
+        manager_test_success = await test_queue_manager_naming()
+        print()
+        
+        print("=" * 50)
+        if redis_test_success and manager_test_success:
+            print("🎉 所有测试通过！双重 crawlo 前缀问题已修复")
+        else:
+            print("❌ 部分测试失败，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 = asyncio.run(main())
+    sys.exit(exit_code)

tests/test_double_crawlo_fix_simple.py

@@ -0,0 +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()
+    sys.exit(exit_code)