crawlo 1.2.8__py3-none-any.whl → 1.3.0__py3-none-any.whl

tests/test_crawlo_proxy_integration.py

@@ -1,173 +1,109 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-Crawlo框架代理集成测试
-====================
-展示如何在Crawlo框架中集成和使用指定的代理API
-"""
-
-import asyncio
-import sys
-import os
-
-# 添加项目根目录到Python路径
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
-
-from crawlo import Spider, Request
-from crawlo.middleware.proxy import ProxyMiddleware
-from crawlo.settings.setting_manager import SettingManager
-
-
-class TestProxySpider(Spider):
-    """测试代理的爬虫示例"""
-    name = 'test_proxy_spider'
-    
-    def __init__(self):
-        super().__init__()
-        self.test_urls = [
-            'https://httpbin.org/ip',  # 查看IP地址
-            'https://httpbin.org/headers',  # 查看请求头
-            'https://stock.10jqka.com.cn/20240315/c655957791.shtml',  # 测试目标链接
-        ]
-    
-    def start_requests(self):
-        """生成初始请求"""
-        for url in self.test_urls:
-            request = Request(url=url, callback=self.parse)
-            yield request
-    
-    def parse(self, response):
-        """解析响应"""
-        print(f"\n=== 响应详情 ===")
-        print(f"URL: {response.url}")
-        print(f"状态码: {response.status_code}")
-        print(f"响应头: {dict(response.headers)}")
-        
-        # 对于httpbin.org/ip，显示IP信息
-        if 'httpbin.org/ip' in response.url:
-            print(f"IP信息: {response.text[:200]}")
-            
-        # 对于httpbin.org/headers，显示请求头信息
-        elif 'httpbin.org/headers' in response.url:
-            print(f"请求头信息: {response.text[:200]}")
-            
-        # 对于目标链接，显示部分内容
-        else:
-            # 只显示前200个字符
-            content_preview = response.text[:200] if response.text else ""
-            print(f"内容预览: {content_preview}")
-            
-        # 返回一个简单的item
-        return {
-            'url': response.url,
-            'status_code': response.status_code,
-            'title': response.css('title::text').get() if response.text else None
-        }
-
-
-def create_proxy_settings():
-    """创建代理配置"""
-    settings = SettingManager()
-    
-    # 基础配置
-    settings.set("LOG_LEVEL", "INFO")
-    settings.set("CONCURRENCY", 1)  # 为了测试，设置并发数为1
-    
-    # 代理配置
-    settings.set("PROXY_ENABLED", True)
-    settings.set("PROXY_API_URL", "http://test.proxy.api:8080/proxy/getitem/")
-    settings.set("PROXY_EXTRACTOR", "proxy")  # 根据API响应结构调整
-    settings.set("PROXY_REFRESH_INTERVAL", 30)  # 30秒刷新一次
-    settings.set("PROXY_API_TIMEOUT", 10)  # 10秒超时
-    settings.set("PROXY_POOL_SIZE", 3)  # 代理池大小
-    settings.set("PROXY_HEALTH_CHECK_THRESHOLD", 0.5)  # 健康检查阈值
-    
-    return settings
-
-
-async def test_proxy_middleware_integration():
-    """测试代理中间件集成"""
-    print("=== 测试Crawlo代理中间件集成 ===")
-    
-    # 创建配置
-    settings = create_proxy_settings()
-    
-    # 创建代理中间件实例
-    proxy_middleware = ProxyMiddleware(settings, "INFO")
-    
-    # 测试代理API连接
-    print(f"代理API URL: {proxy_middleware.api_url}")
-    print(f"代理刷新间隔: {proxy_middleware.refresh_interval}秒")
-    print(f"代理池大小: {proxy_middleware.proxy_pool_size}")
-    
-    # 测试获取代理
-    print("\n--- 测试获取代理 ---")
-    try:
-        # 这里我们直接测试API连接，而不是完整的代理池更新
-        proxy_data = await proxy_middleware._get_proxy_from_api()
-        if proxy_data:
-            print(f"✅ 成功从API获取代理信息: {proxy_data}")
-        else:
-            print("❌ 无法从API获取代理信息")
-    except Exception as e:
-        print(f"❌ 获取代理时出错: {e}")
-    
-    print("\n=== 代理中间件集成测试完成 ===")
-
-
-def show_proxy_configuration_example():
-    """显示代理配置示例"""
-    print("\n=== 代理配置示例 ===")
-    print("""
-在Crawlo项目中配置代理的方法：
-
-1. 在settings.py中添加以下配置：
-
-```python
-# 代理配置
-PROXY_ENABLED = True
-PROXY_API_URL = 'http://test.proxy.api:8080/proxy/getitem/'
-PROXY_EXTRACTOR = 'proxy'
-PROXY_REFRESH_INTERVAL = 30
-PROXY_API_TIMEOUT = 10
-PROXY_POOL_SIZE = 5
-PROXY_HEALTH_CHECK_THRESHOLD = 0.5
-```
-
-2. 确保代理中间件在MIDDLEWARES列表中：
-
-```python
-MIDDLEWARES = [
-    'crawlo.middleware.request_ignore.RequestIgnoreMiddleware',
-    'crawlo.middleware.download_delay.DownloadDelayMiddleware',
-    'crawlo.middleware.default_header.DefaultHeaderMiddleware',
-    'crawlo.middleware.proxy.ProxyMiddleware',  # 代理中间件
-    'crawlo.middleware.retry.RetryMiddleware',
-    'crawlo.middleware.response_code.ResponseCodeMiddleware',
-    'crawlo.middleware.response_filter.ResponseFilterMiddleware',
-]
-```
-
-3. 启动爬虫后，代理中间件会自动：
-   - 定期从API获取代理
-   - 维护代理池
-   - 自动为请求分配代理
-   - 监控代理健康状态
-""")
-
-
-async def main():
-    """主函数"""
-    print("开始Crawlo代理集成测试...\n")
-    
-    # 1. 测试代理中间件集成
-    await test_proxy_middleware_integration()
-    
-    # 2. 显示配置示例
-    show_proxy_configuration_example()
-    
-    print("\n所有测试完成！")
-
-
-if __name__ == "__main__":
-    asyncio.run(main())
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Crawlo框架代理集成测试
+测试代理中间件与框架的集成
+"""
+
+import asyncio
+import sys
+import os
+from unittest.mock import Mock, patch
+
+# 添加项目根目录到路径
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
+
+from crawlo.config import CrawloConfig
+from crawlo.crawler import CrawlerProcess
+from crawlo import Spider, Request, Item
+from crawlo.middleware.proxy import ProxyMiddleware
+
+
+class TestItem(Item):
+    """测试结果项"""
+    url = ''
+    status = 0
+    proxy = ''
+
+
+class ProxyTestSpider(Spider):
+    """代理测试爬虫"""
+    name = 'proxy_test_spider'
+    
+    async def start_requests(self):
+        """发起测试请求"""
+        yield Request('https://httpbin.org/ip', callback=self.parse)
+    
+    async def parse(self, response):
+        """解析响应"""
+        import json
+        try:
+            data = json.loads(response.text)
+            ip_info = data.get('origin', 'Unknown')
+        except:
+            ip_info = 'Parse error'
+        
+        item = TestItem(
+            url=response.url,
+            status=response.status,
+            proxy=str(response.meta.get('proxy', 'No proxy'))
+        )
+        
+        self.logger.info(f"Proxy test result: IP={ip_info}, Proxy={item.proxy}")
+        yield item
+
+
+async def test_proxy_integration():
+    """测试代理集成"""
+    print("🔍 测试代理集成...")
+    
+    # 创建配置
+    config = CrawloConfig.standalone(
+        concurrency=1,
+        download_delay=0.1,
+        PROXY_ENABLED=True,
+        PROXY_API_URL="https://api.proxyprovider.com/get",  # 模拟代理API
+        LOG_LEVEL='WARNING'  # 减少日志输出
+    )
+    
+    # 添加代理中间件
+    config.set('CUSTOM_MIDDLEWARES', [
+        'crawlo.middleware.proxy.ProxyMiddleware',
+    ])
+    
+    # 创建爬虫进程
+    process = CrawlerProcess(settings=config.to_dict())
+    
+    # 添加爬虫
+    process.crawl(ProxyTestSpider)
+    
+    # 运行测试
+    await process.start()
+    
+    print("   代理集成测试完成")
+
+
+async def main():
+    """主测试函数"""
+    print("开始Crawlo框架代理集成测试...")
+    print("=" * 50)
+    
+    try:
+        await test_proxy_integration()
+        
+        print("=" * 50)
+        print("所有代理集成测试通过！")
+        
+    except Exception as e:
+        print("=" * 50)
+        print(f"测试失败: {e}")
+        import traceback
+        traceback.print_exc()
+        return 1
+    
+    return 0
+
+
+if __name__ == "__main__":
+    exit_code = asyncio.run(main())
+    sys.exit(exit_code)

tests/test_date_tools.py

@@ -1,124 +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__':
+#!/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()