crawlo 1.4.6__py3-none-any.whl → 1.4.8__py3-none-any.whl

tests/test_dynamic_proxy_config.py

@@ -1,147 +0,0 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-测试动态下载器（Selenium和Playwright）的代理配置逻辑
-"""
-
-from crawlo.tools import AuthenticatedProxy
-
-
-def test_selenium_proxy_logic():
-    """测试Selenium下载器的代理配置逻辑"""
-    print("=== 测试Selenium下载器的代理配置逻辑 ===")
-    
-    # 代理配置
-    proxy_config = {
-        "http": "http://dwe20241014:Dwe0101014@182.201.243.186:58111",
-        "https": "http://dwe20241014:Dwe0101014@182.201.243.186:58111"
-    }
-    
-    # 创建代理对象
-    proxy_url = proxy_config["http"]
-    proxy = AuthenticatedProxy(proxy_url)
-    
-    print(f"原始代理URL: {proxy_url}")
-    print(f"清洁URL: {proxy.clean_url}")
-    print(f"认证信息: {proxy.get_auth_credentials()}")
-    
-    # 模拟Selenium的代理配置逻辑
-    print(f"\nSelenium代理配置逻辑:")
-    print(f"  1. 在爬虫设置中配置:")
-    print(f"     settings = {{")
-    print(f"         'SELENIUM_PROXY': '{proxy.clean_url}',")
-    print(f"     }}")
-    
-    # 对于带认证的代理，需要特殊处理
-    if proxy.username and proxy.password:
-        print(f"\n  2. 带认证代理的处理逻辑:")
-        print(f"     - 用户名: {proxy.username}")
-        print(f"     - 密码: {proxy.password}")
-        print(f"     - 认证头: {proxy.get_auth_header()}")
-        print(f"     - 处理方式: 通过浏览器扩展或手动输入认证信息")
-    
-    print("\nSelenium配置逻辑测试完成!")
-
-
-def test_playwright_proxy_logic():
-    """测试Playwright下载器的代理配置逻辑"""
-    print("\n=== 测试Playwright下载器的代理配置逻辑 ===")
-    
-    # 代理配置
-    proxy_config = {
-        "http": "http://dwe20241014:Dwe0101014@182.201.243.186:58111",
-        "https": "http://dwe20241014:Dwe0101014@182.201.243.186:58111"
-    }
-    
-    # 创建代理对象
-    proxy_url = proxy_config["http"]
-    proxy = AuthenticatedProxy(proxy_url)
-    
-    print(f"原始代理URL: {proxy_url}")
-    print(f"清洁URL: {proxy.clean_url}")
-    print(f"认证信息: {proxy.get_auth_credentials()}")
-    
-    # 模拟Playwright的代理配置逻辑
-    print(f"\nPlaywright代理配置逻辑:")
-    print(f"  1. 简单代理配置:")
-    print(f"     settings = {{")
-    print(f"         'PLAYWRIGHT_PROXY': '{proxy.clean_url}',")
-    print(f"     }}")
-    
-    # 对于带认证的代理，Playwright可以直接在代理配置中包含认证信息
-    if proxy.username and proxy.password:
-        print(f"\n  2. 带认证的代理配置逻辑:")
-        print(f"     settings = {{")
-        print(f"         'PLAYWRIGHT_PROXY': {{")
-        print(f"             'server': '{proxy.clean_url}',")
-        print(f"             'username': '{proxy.username}',")
-        print(f"             'password': '{proxy.password}'")
-        print(f"         }}")
-        print(f"     }}")
-        print(f"     实现方式: 在启动浏览器时传递proxy参数")
-    
-    print("\nPlaywright配置逻辑测试完成!")
-
-
-def show_proxy_configuration_examples():
-    """显示代理配置示例"""
-    print("\n=== 代理配置示例 ===")
-    
-    # 不同类型的代理配置示例
-    examples = [
-        {
-            "name": "带认证HTTP代理",
-            "url": "http://username:password@proxy.example.com:8080",
-            "selenium_config": "SELENIUM_PROXY = 'http://proxy.example.com:8080'",
-            "playwright_config": "PLAYWRIGHT_PROXY = {'server': 'http://proxy.example.com:8080', 'username': 'username', 'password': 'password'}"
-        },
-        {
-            "name": "带认证HTTPS代理",
-            "url": "https://user:pass@secure-proxy.example.com:443",
-            "selenium_config": "SELENIUM_PROXY = 'https://secure-proxy.example.com:443'",
-            "playwright_config": "PLAYWRIGHT_PROXY = {'server': 'https://secure-proxy.example.com:443', 'username': 'user', 'password': 'pass'}"
-        },
-        {
-            "name": "SOCKS5代理",
-            "url": "socks5://username:password@socks-proxy.example.com:1080",
-            "selenium_config": "SELENIUM_PROXY = 'socks5://socks-proxy.example.com:1080'",
-            "playwright_config": "PLAYWRIGHT_PROXY = {'server': 'socks5://socks-proxy.example.com:1080', 'username': 'username', 'password': 'password'}"
-        },
-        {
-            "name": "不带认证代理",
-            "url": "http://proxy.example.com:8080",
-            "selenium_config": "SELENIUM_PROXY = 'http://proxy.example.com:8080'",
-            "playwright_config": "PLAYWRIGHT_PROXY = 'http://proxy.example.com:8080'"
-        }
-    ]
-    
-    for i, example in enumerate(examples, 1):
-        print(f"\n示例 {i}: {example['name']}")
-        print(f"  代理URL: {example['url']}")
-        proxy = AuthenticatedProxy(example['url'])
-        print(f"  清洁URL: {proxy.clean_url}")
-        print(f"  用户名: {proxy.username or '无'}")
-        print(f"  密码: {proxy.password or '无'}")
-        print(f"  Selenium配置: {example['selenium_config']}")
-        print(f"  Playwright配置: {example['playwright_config']}")
-
-
-def main():
-    """主函数"""
-    print("开始测试动态下载器的代理配置逻辑...\n")
-    
-    # 测试各个下载器的配置逻辑
-    test_selenium_proxy_logic()
-    test_playwright_proxy_logic()
-    show_proxy_configuration_examples()
-    
-    print("\n所有配置逻辑测试完成!")
-    print("\n总结:")
-    print("1. Selenium和Playwright都支持代理配置")
-    print("2. 带认证的代理需要特殊处理")
-    print("3. Playwright对带认证代理的支持更直接")
-    print("4. Selenium需要通过扩展或其他方式处理认证")
-
-
-if __name__ == "__main__":
-    main()

tests/test_dynamic_proxy_real.py

@@ -1,110 +0,0 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-实际测试动态下载器（Playwright）通过代理访问网站
-"""
-
-import asyncio
-from crawlo.spider import Spider
-from crawlo.network.request import Request
-from crawlo.tools import AuthenticatedProxy
-
-
-class ProxyTestSpider(Spider):
-    """代理测试爬虫"""
-    name = "proxy_test_spider"  # 添加name属性
-    
-    # 自定义配置
-    custom_settings = {
-        "DOWNLOADER_TYPE": "playwright",
-        "PLAYWRIGHT_BROWSER_TYPE": "chromium",
-        "PLAYWRIGHT_HEADLESS": True,
-        # 配置带认证的代理
-        "PLAYWRIGHT_PROXY": {
-            "server": "http://182.201.243.186:58111",
-            "username": "dwe20241014",
-            "password": "Dwe0101014"
-        }
-    }
-    
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-        print("代理测试爬虫初始化完成")
-        print(f"代理配置: {self.custom_settings.get('PLAYWRIGHT_PROXY')}")
-
-    def start_requests(self):
-        """开始请求"""
-        urls = [
-            "https://httpbin.org/ip",  # 查看IP地址
-            "https://httpbin.org/headers",  # 查看请求头
-        ]
-        
-        for url in urls:
-            request = Request(url, callback=self.parse)
-            yield request
-
-    def parse(self, response):
-        """解析响应"""
-        print(f"\n=== 响应信息 ===")
-        print(f"URL: {response.url}")
-        print(f"状态码: {response.status_code}")
-        print(f"响应内容: {response.text[:500]}")
-        
-        # 保存响应内容
-        filename = response.url.split("/")[-1].replace("?", "_").replace("&", "_")
-        with open(f"proxy_test_{filename}.html", "w", encoding="utf-8") as f:
-            f.write(response.text)
-        print(f"响应已保存到 proxy_test_{filename}.html")
-        
-        # 返回一个简单的item
-        yield {"url": response.url, "status": response.status_code}
-
-
-# 异步运行函数
-async def run_spider():
-    """运行爬虫"""
-    print("开始测试动态下载器通过代理访问网站...")
-    
-    # 创建爬虫实例
-    spider = ProxyTestSpider()
-    
-    # 创建一个简单的crawler模拟器
-    class MockCrawler:
-        def __init__(self):
-            from crawlo.settings.setting_manager import SettingManager
-            self.settings = SettingManager()
-            # 应用爬虫的自定义设置
-            if hasattr(spider, 'custom_settings'):
-                for key, value in spider.custom_settings.items():
-                    self.settings.set(key, value)
-    
-    crawler = MockCrawler()
-    
-    # 创建爬虫实例并绑定crawler
-    spider_instance = spider.create_instance(crawler)
-    
-    # 执行初始请求
-    requests = list(spider_instance.start_requests())
-    print(f"生成了 {len(requests)} 个请求")
-    
-    # 使用Playwright下载器处理请求
-    try:
-        from crawlo.downloader import PlaywrightDownloader
-        downloader = PlaywrightDownloader(crawler)
-        await downloader.download(requests[0])  # 测试第一个请求
-        print("Playwright下载器测试成功!")
-    except Exception as e:
-        print(f"Playwright下载器测试失败: {e}")
-        import traceback
-        traceback.print_exc()
-    
-    print("\n测试完成!")
-
-
-async def main():
-    """主函数"""
-    await run_spider()
-
-
-if __name__ == "__main__":
-    asyncio.run(main())

tests/test_env_config.py

@@ -1,122 +0,0 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-环境变量配置工具测试
-"""
-import os
-import sys
-import unittest
-from unittest.mock import patch
-
-# 添加项目根目录到Python路径
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
-
-from crawlo.utils.env_config import EnvConfigManager, get_env_var, get_redis_config, get_runtime_config
-
-
-class TestEnvConfigManager(unittest.TestCase):
-    """环境变量配置管理器测试"""
-
-    def test_get_env_var_str(self):
-        """测试字符串环境变量获取"""
-        with patch.dict(os.environ, {'TEST_VAR': 'test_value'}):
-            result = EnvConfigManager.get_env_var('TEST_VAR', 'default', str)
-            self.assertEqual(result, 'test_value')
-        
-        # 测试默认值
-        result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 'default', str)
-        self.assertEqual(result, 'default')
-
-    def test_get_env_var_int(self):
-        """测试整数环境变量获取"""
-        with patch.dict(os.environ, {'TEST_INT': '42'}):
-            result = EnvConfigManager.get_env_var('TEST_INT', 0, int)
-            self.assertEqual(result, 42)
-        
-        # 测试默认值
-        result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 10, int)
-        self.assertEqual(result, 10)
-        
-        # 测试无效值
-        with patch.dict(os.environ, {'INVALID_INT': 'not_a_number'}):
-            result = EnvConfigManager.get_env_var('INVALID_INT', 5, int)
-            self.assertEqual(result, 5)
-
-    def test_get_env_var_float(self):
-        """测试浮点数环境变量获取"""
-        with patch.dict(os.environ, {'TEST_FLOAT': '3.14'}):
-            result = EnvConfigManager.get_env_var('TEST_FLOAT', 0.0, float)
-            self.assertEqual(result, 3.14)
-        
-        # 测试默认值
-        result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', 2.5, float)
-        self.assertEqual(result, 2.5)
-
-    def test_get_env_var_bool(self):
-        """测试布尔环境变量获取"""
-        # 测试 True 值
-        for true_val in ['1', 'true', 'True', 'TRUE', 'yes', 'on']:
-            with patch.dict(os.environ, {'TEST_BOOL': true_val}):
-                result = EnvConfigManager.get_env_var('TEST_BOOL', False, bool)
-                self.assertTrue(result)
-        
-        # 测试 False 值
-        for false_val in ['0', 'false', 'False', 'FALSE', 'no', 'off', '']:
-            with patch.dict(os.environ, {'TEST_BOOL': false_val}):
-                result = EnvConfigManager.get_env_var('TEST_BOOL', True, bool)
-                self.assertFalse(result)
-        
-        # 测试默认值
-        result = EnvConfigManager.get_env_var('NON_EXISTENT_VAR', True, bool)
-        self.assertTrue(result)
-
-    def test_get_redis_config(self):
-        """测试 Redis 配置获取"""
-        with patch.dict(os.environ, {
-            'REDIS_HOST': 'localhost',
-            'REDIS_PORT': '6380',
-            'REDIS_PASSWORD': 'secret',
-            'REDIS_DB': '1'
-        }):
-            config = get_redis_config()
-            self.assertEqual(config['REDIS_HOST'], 'localhost')
-            self.assertEqual(config['REDIS_PORT'], 6380)
-            self.assertEqual(config['REDIS_PASSWORD'], 'secret')
-            self.assertEqual(config['REDIS_DB'], 1)
-        
-        # 测试默认值
-        with patch.dict(os.environ, {}):
-            config = get_redis_config()
-            self.assertEqual(config['REDIS_HOST'], '127.0.0.1')
-            self.assertEqual(config['REDIS_PORT'], 6379)
-            self.assertEqual(config['REDIS_PASSWORD'], '')
-            self.assertEqual(config['REDIS_DB'], 0)
-
-    def test_get_runtime_config(self):
-        """测试运行时配置获取"""
-        with patch.dict(os.environ, {
-            'CRAWLO_MODE': 'distributed',
-            'PROJECT_NAME': 'test_project',
-            'CONCURRENCY': '20'
-        }):
-            config = get_runtime_config()
-            self.assertEqual(config['CRAWLO_MODE'], 'distributed')
-            self.assertEqual(config['PROJECT_NAME'], 'test_project')
-            self.assertEqual(config['CONCURRENCY'], 20)
-        
-        # 测试默认值
-        with patch.dict(os.environ, {}):
-            config = get_runtime_config()
-            self.assertEqual(config['CRAWLO_MODE'], 'standalone')
-            self.assertEqual(config['PROJECT_NAME'], 'crawlo')
-            self.assertEqual(config['CONCURRENCY'], 8)
-
-    def test_convenience_functions(self):
-        """测试便捷函数"""
-        with patch.dict(os.environ, {'TEST_CONVENIENCE': 'convenience_value'}):
-            result = get_env_var('TEST_CONVENIENCE', 'default')
-            self.assertEqual(result, 'convenience_value')
-
-
-if __name__ == '__main__':
-    unittest.main()

tests/test_framework_env_usage.py

@@ -1,104 +0,0 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-框架环境变量使用测试
-验证整个框架中环境变量的正确使用
-"""
-import sys
-import os
-import unittest
-from unittest.mock import patch, MagicMock
-
-# 添加项目根目录到Python路径
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
-
-from crawlo.utils.env_config import get_env_var, get_redis_config, get_runtime_config
-from crawlo.settings.setting_manager import SettingManager
-from crawlo.settings import default_settings
-from crawlo.mode_manager import from_env
-
-
-class TestFrameworkEnvUsage(unittest.TestCase):
-    """框架环境变量使用测试"""
-
-    def test_default_settings_env_usage(self):
-        """测试 default_settings.py 中的环境变量使用"""
-        # 验证 default_settings.py 不直接使用 os.getenv
-        import inspect
-        import crawlo.settings.default_settings as default_settings_module
-        
-        source_code = inspect.getsource(default_settings_module)
-        # 检查是否还有直接使用 os.getenv 的地方
-        self.assertNotIn('os.getenv', source_code, 
-                         "default_settings.py 不应该直接使用 os.getenv")
-        
-        # 但应该使用 env_config 工具
-        self.assertIn('get_redis_config', source_code,
-                      "default_settings.py 应该使用 get_redis_config")
-        self.assertIn('get_runtime_config', source_code,
-                      "default_settings.py 应该使用 get_runtime_config")
-
-    def test_env_config_tool(self):
-        """测试环境变量配置工具"""
-        # 测试获取Redis配置
-        with patch.dict(os.environ, {
-            'REDIS_HOST': 'test.redis.com',
-            'REDIS_PORT': '6380',
-            'REDIS_PASSWORD': 'test_pass',
-            'REDIS_DB': '2'
-        }):
-            redis_config = get_redis_config()
-            self.assertEqual(redis_config['REDIS_HOST'], 'test.redis.com')
-            self.assertEqual(redis_config['REDIS_PORT'], 6380)
-            self.assertEqual(redis_config['REDIS_PASSWORD'], 'test_pass')
-            self.assertEqual(redis_config['REDIS_DB'], 2)
-        
-        # 测试获取运行时配置
-        with patch.dict(os.environ, {
-            'PROJECT_NAME': 'test_project',
-            'CRAWLO_MODE': 'distributed',
-            'CONCURRENCY': '16'
-        }):
-            runtime_config = get_runtime_config()
-            self.assertEqual(runtime_config['PROJECT_NAME'], 'test_project')
-            self.assertEqual(runtime_config['CRAWLO_MODE'], 'distributed')
-            self.assertEqual(runtime_config['CONCURRENCY'], 16)
-
-    def test_settings_manager_with_env(self):
-        """测试设置管理器与环境变量的集成"""
-        # 设置环境变量
-        env_vars = {
-            'PROJECT_NAME': 'env_test_project',
-            'CONCURRENCY': '12',
-            'REDIS_HOST': 'env.redis.test',
-            'REDIS_PORT': '6381'
-        }
-        
-        with patch.dict(os.environ, env_vars):
-            # 重新导入 default_settings 模块以获取最新的环境变量
-            import importlib
-            import crawlo.settings.default_settings
-            importlib.reload(crawlo.settings.default_settings)
-            
-            # 创建设置管理器
-            settings = SettingManager()
-            settings.set_settings(crawlo.settings.default_settings)
-            
-            # 验证环境变量被正确使用
-            redis_config = get_redis_config()
-            self.assertEqual(settings.get('REDIS_HOST'), redis_config['REDIS_HOST'])
-            
-            runtime_config = get_runtime_config()
-            self.assertEqual(settings.get('PROJECT_NAME'), runtime_config['PROJECT_NAME'])
-
-    def test_mode_manager_env_usage(self):
-        """测试 mode_manager.py 中的环境变量使用"""
-        # 验证 from_env 函数现在会抛出异常
-        with self.assertRaises(RuntimeError) as context:
-            from_env()
-        
-        self.assertIn("环境变量配置已移除", str(context.exception))
-
-
-if __name__ == '__main__':
-    unittest.main()

tests/test_large_scale_config.py

@@ -1,113 +0,0 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-大规模配置工具测试
-测试 LargeScaleConfig, apply_large_scale_config
-"""
-import sys
-import os
-import unittest
-from unittest.mock import Mock, patch, MagicMock
-
-# 添加项目根目录到 Python 路径
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
-
-from crawlo.utils.large_scale_config import LargeScaleConfig, apply_large_scale_config
-
-
-class TestLargeScaleConfig(unittest.TestCase):
-    """大规模配置工具测试"""
-
-    def test_conservative_config(self):
-        """测试保守配置"""
-        config = LargeScaleConfig.conservative_config(concurrency=8)
-        
-        # 验证基本配置
-        self.assertEqual(config['CONCURRENCY'], 8)
-        self.assertEqual(config['SCHEDULER_MAX_QUEUE_SIZE'], 80)  # 8 * 10
-        self.assertEqual(config['DOWNLOAD_DELAY'], 0.2)
-        self.assertEqual(config['MAX_RUNNING_SPIDERS'], 1)
-        
-        # 验证连接池配置
-        self.assertEqual(config['CONNECTION_POOL_LIMIT'], 16)  # 8 * 2
-        
-        # 验证重试配置
-        self.assertEqual(config['MAX_RETRY_TIMES'], 2)
-        
-    def test_balanced_config(self):
-        """测试平衡配置"""
-        config = LargeScaleConfig.balanced_config(concurrency=16)
-        
-        # 验证基本配置
-        self.assertEqual(config['CONCURRENCY'], 16)
-        self.assertEqual(config['SCHEDULER_MAX_QUEUE_SIZE'], 240)  # 16 * 15
-        self.assertEqual(config['DOWNLOAD_DELAY'], 0.1)
-        self.assertEqual(config['MAX_RUNNING_SPIDERS'], 2)
-        
-        # 验证连接池配置
-        self.assertEqual(config['CONNECTION_POOL_LIMIT'], 48)  # 16 * 3
-        
-        # 验证重试配置
-        self.assertEqual(config['MAX_RETRY_TIMES'], 3)
-        
-    def test_aggressive_config(self):
-        """测试激进配置"""
-        config = LargeScaleConfig.aggressive_config(concurrency=32)
-        
-        # 验证基本配置
-        self.assertEqual(config['CONCURRENCY'], 32)
-        self.assertEqual(config['SCHEDULER_MAX_QUEUE_SIZE'], 640)  # 32 * 20
-        self.assertEqual(config['DOWNLOAD_DELAY'], 0.05)
-        self.assertEqual(config['MAX_RUNNING_SPIDERS'], 3)
-        
-        # 验证连接池配置
-        self.assertEqual(config['CONNECTION_POOL_LIMIT'], 128)  # 32 * 4
-        
-        # 验证重试配置
-        self.assertEqual(config['MAX_RETRY_TIMES'], 5)
-        
-    def test_memory_optimized_config(self):
-        """测试内存优化配置"""
-        config = LargeScaleConfig.memory_optimized_config(concurrency=12)
-        
-        # 验证基本配置
-        self.assertEqual(config['CONCURRENCY'], 12)
-        self.assertEqual(config['SCHEDULER_MAX_QUEUE_SIZE'], 60)  # 12 * 5
-        self.assertEqual(config['DOWNLOAD_DELAY'], 0.1)
-        self.assertEqual(config['MAX_RUNNING_SPIDERS'], 1)
-        
-        # 验证连接池配置
-        self.assertEqual(config['CONNECTION_POOL_LIMIT'], 12)  # 12 * 1
-        
-        # 验证内存限制配置
-        self.assertEqual(config['DOWNLOAD_MAXSIZE'], 2 * 1024 * 1024)  # 2MB
-        self.assertEqual(config['DOWNLOAD_WARN_SIZE'], 512 * 1024)  # 512KB
-        
-        # 验证重试配置
-        self.assertEqual(config['MAX_RETRY_TIMES'], 2)
-        
-    def test_apply_large_scale_config(self):
-        """测试应用大规模配置"""
-        settings_dict = {}
-        
-        # 应用平衡配置
-        apply_large_scale_config(settings_dict, "balanced", 16)
-        
-        # 验证配置已应用
-        self.assertEqual(settings_dict['CONCURRENCY'], 16)
-        self.assertEqual(settings_dict['SCHEDULER_MAX_QUEUE_SIZE'], 240)
-        self.assertEqual(settings_dict['DOWNLOAD_DELAY'], 0.1)
-        
-    def test_apply_large_scale_config_invalid_type(self):
-        """测试应用无效的大规模配置类型"""
-        settings_dict = {}
-        
-        # 应用无效配置类型应该抛出异常
-        with self.assertRaises(ValueError) as context:
-            apply_large_scale_config(settings_dict, "invalid_type", 16)
-            
-        self.assertIn("不支持的配置类型", str(context.exception))
-
-
-if __name__ == '__main__':
-    unittest.main()