crawlo 1.4.1__py3-none-any.whl → 1.4.3__py3-none-any.whl

tests/test_template_redis_key.py

@@ -1,135 +1,135 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-模板项目Redis Key测试脚本
-用于验证通过模板生成的项目是否符合新的Redis key命名规范
-"""
-import sys
-import os
-import tempfile
-import shutil
-import subprocess
-from pathlib import Path
-
-# 添加项目根目录到路径
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
-
-
-def test_template_project_redis_key():
-    """测试模板项目Redis key命名规范"""
-    print("🔍 测试模板项目Redis key命名规范...")
-    
-    # 创建临时目录
-    with tempfile.TemporaryDirectory() as temp_dir:
-        try:
-            # 在原始工作目录中创建项目，然后移动到临时目录
-            original_cwd = os.getcwd()
-            
-            # 创建测试项目（在原始工作目录中）
-            print("   1. 创建测试项目...")
-            cmd_path = os.path.join(original_cwd, "crawlo", "commands", "startproject.py")
-            result = subprocess.run([
-                sys.executable, cmd_path, "test_project"
-            ], cwd=original_cwd, capture_output=True, text=True)
-            
-            if result.returncode != 0:
-                print(f"创建项目失败: {result.stderr}")
-                return False
-            
-            print("      项目创建成功")
-            
-            # 检查生成的文件
-            project_dir = Path(original_cwd) / "test_project"
-            if not project_dir.exists():
-                print("项目目录未创建")
-                return False
-                
-            # 移动项目到临时目录
-            target_dir = Path(temp_dir) / "test_project"
-            shutil.move(str(project_dir), str(target_dir))
-            project_dir = target_dir
-            
-            settings_file = project_dir / "test_project" / "settings.py"
-            if not settings_file.exists():
-                print("settings.py文件未创建")
-                return False
-            
-            # 读取settings.py内容
-            with open(settings_file, 'r', encoding='utf-8') as f:
-                settings_content = f.read()
-            
-            # 检查是否移除了旧的REDIS_KEY配置
-            if "REDIS_KEY = f'{{project_name}}:fingerprint'" in settings_content:
-                print("仍然存在旧的REDIS_KEY配置")
-                return False
-                
-            # 检查是否添加了新的注释
-            if "# crawlo:{project_name}:filter:fingerprint (请求去重)" not in settings_content:
-                print("缺少新的Redis key命名规范注释")
-                return False
-                
-            if "# crawlo:{project_name}:item:fingerprint (数据项去重)" not in settings_content:
-                print("缺少数据项去重的Redis key命名规范注释")
-                return False
-            
-            print("      settings.py符合新的Redis key命名规范")
-            
-            # 检查crawlo.cfg
-            cfg_file = project_dir / "crawlo.cfg"
-            if not cfg_file.exists():
-                print("crawlo.cfg文件未创建")
-                return False
-                
-            with open(cfg_file, 'r', encoding='utf-8') as f:
-                cfg_content = f.read()
-                
-            if "default = test_project.settings" not in cfg_content:
-                print("crawlo.cfg配置不正确")
-                return False
-                
-            print("      crawlo.cfg配置正确")
-            
-            print("模板项目Redis key命名规范测试通过！")
-            return True
-            
-        except Exception as e:
-            print(f"测试过程中发生错误: {e}")
-            import traceback
-            traceback.print_exc()
-            return False
-        finally:
-            # 清理创建的项目目录
-            project_dir = Path(original_cwd) / "test_project"
-            if project_dir.exists():
-                shutil.rmtree(str(project_dir), ignore_errors=True)
-            
-            # 恢复原始工作目录
-            os.chdir(original_cwd)
-
-
-def main():
-    """主测试函数"""
-    print("开始模板项目Redis key命名规范测试...")
-    print("=" * 50)
-    
-    try:
-        success = test_template_project_redis_key()
-        
-        print("=" * 50)
-        if success:
-            print("所有测试通过！模板项目符合新的Redis key命名规范")
-        else:
-            print("测试失败，请检查模板文件")
-            return 1
-            
-    except Exception as e:
-        print("=" * 50)
-        print(f"测试过程中发生异常: {e}")
-        return 1
-    
-    return 0
-
-
-if __name__ == "__main__":
-    exit_code = main()
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+模板项目Redis Key测试脚本
+用于验证通过模板生成的项目是否符合新的Redis key命名规范
+"""
+import sys
+import os
+import tempfile
+import shutil
+import subprocess
+from pathlib import Path
+
+# 添加项目根目录到路径
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
+
+
+def test_template_project_redis_key():
+    """测试模板项目Redis key命名规范"""
+    print("🔍 测试模板项目Redis key命名规范...")
+    
+    # 创建临时目录
+    with tempfile.TemporaryDirectory() as temp_dir:
+        try:
+            # 在原始工作目录中创建项目，然后移动到临时目录
+            original_cwd = os.getcwd()
+            
+            # 创建测试项目（在原始工作目录中）
+            print("   1. 创建测试项目...")
+            cmd_path = os.path.join(original_cwd, "crawlo", "commands", "startproject.py")
+            result = subprocess.run([
+                sys.executable, cmd_path, "test_project"
+            ], cwd=original_cwd, capture_output=True, text=True)
+            
+            if result.returncode != 0:
+                print(f"创建项目失败: {result.stderr}")
+                return False
+            
+            print("      项目创建成功")
+            
+            # 检查生成的文件
+            project_dir = Path(original_cwd) / "test_project"
+            if not project_dir.exists():
+                print("项目目录未创建")
+                return False
+                
+            # 移动项目到临时目录
+            target_dir = Path(temp_dir) / "test_project"
+            shutil.move(str(project_dir), str(target_dir))
+            project_dir = target_dir
+            
+            settings_file = project_dir / "test_project" / "settings.py"
+            if not settings_file.exists():
+                print("settings.py文件未创建")
+                return False
+            
+            # 读取settings.py内容
+            with open(settings_file, 'r', encoding='utf-8') as f:
+                settings_content = f.read()
+            
+            # 检查是否移除了旧的REDIS_KEY配置
+            if "REDIS_KEY = f'{{project_name}}:fingerprint'" in settings_content:
+                print("仍然存在旧的REDIS_KEY配置")
+                return False
+                
+            # 检查是否添加了新的注释
+            if "# crawlo:{project_name}:filter:fingerprint (请求去重)" not in settings_content:
+                print("缺少新的Redis key命名规范注释")
+                return False
+                
+            if "# crawlo:{project_name}:item:fingerprint (数据项去重)" not in settings_content:
+                print("缺少数据项去重的Redis key命名规范注释")
+                return False
+            
+            print("      settings.py符合新的Redis key命名规范")
+            
+            # 检查crawlo.cfg
+            cfg_file = project_dir / "crawlo.cfg"
+            if not cfg_file.exists():
+                print("crawlo.cfg文件未创建")
+                return False
+                
+            with open(cfg_file, 'r', encoding='utf-8') as f:
+                cfg_content = f.read()
+                
+            if "default = test_project.settings" not in cfg_content:
+                print("crawlo.cfg配置不正确")
+                return False
+                
+            print("      crawlo.cfg配置正确")
+            
+            print("模板项目Redis key命名规范测试通过！")
+            return True
+            
+        except Exception as e:
+            print(f"测试过程中发生错误: {e}")
+            import traceback
+            traceback.print_exc()
+            return False
+        finally:
+            # 清理创建的项目目录
+            project_dir = Path(original_cwd) / "test_project"
+            if project_dir.exists():
+                shutil.rmtree(str(project_dir), ignore_errors=True)
+            
+            # 恢复原始工作目录
+            os.chdir(original_cwd)
+
+
+def main():
+    """主测试函数"""
+    print("开始模板项目Redis key命名规范测试...")
+    print("=" * 50)
+    
+    try:
+        success = test_template_project_redis_key()
+        
+        print("=" * 50)
+        if success:
+            print("所有测试通过！模板项目符合新的Redis key命名规范")
+        else:
+            print("测试失败，请检查模板文件")
+            return 1
+            
+    except Exception as e:
+        print("=" * 50)
+        print(f"测试过程中发生异常: {e}")
+        return 1
+    
+    return 0
+
+
+if __name__ == "__main__":
+    exit_code = main()
     sys.exit(exit_code)

tests/test_tools.py

@@ -1,160 +1,160 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-工具包测试
-"""
-import asyncio
-import unittest
-from crawlo.tools import (
-    # 日期工具
-    parse_time,
-    format_time,
-    time_diff,
-    
-    # 数据清洗工具
-    clean_text,
-    format_currency,
-    extract_emails,
-    
-    # 数据验证工具
-    validate_email,
-    validate_url,
-    validate_phone,
-    validate_chinese_id_card,
-    validate_date,
-    validate_number_range,
-    
-    # 请求处理工具
-    build_url,
-    add_query_params,
-    merge_headers,
-    
-    # 反爬虫应对工具
-    get_random_user_agent,
-    rotate_proxy,
-    
-    # 分布式协调工具
-    generate_task_id,
-    get_cluster_info
-)
-
-
-class TestTools(unittest.TestCase):
-    """工具包测试类"""
-
-    def test_date_tools(self):
-        """测试日期工具"""
-        # 测试时间解析
-        time_str = "2025-09-10 14:30:00"
-        parsed_time = parse_time(time_str)
-        self.assertIsNotNone(parsed_time)
-        
-        # 测试时间格式化
-        formatted_time = format_time(parsed_time, "%Y-%m-%d")
-        self.assertEqual(formatted_time, "2025-09-10")
-        
-        # 测试时间差计算
-        time_str2 = "2025-09-11 14:30:00"
-        parsed_time2 = parse_time(time_str2)
-        diff = time_diff(parsed_time2, parsed_time)
-        self.assertEqual(diff, 86400)  # 24小时 = 86400秒
-
-    def test_data_cleaning_tools(self):
-        """测试数据清洗工具"""
-        # 测试文本清洗
-        dirty_text = "<p>这是一个&nbsp;<b>测试</b>&amp;文本</p>"
-        clean_result = clean_text(dirty_text)
-        self.assertEqual(clean_result, "这是一个 测试&文本")
-        
-        # 测试货币格式化
-        price = 1234.567
-        formatted_price = format_currency(price, "¥", 2)
-        self.assertEqual(formatted_price, "¥1,234.57")
-        
-        # 测试邮箱提取
-        text_with_email = "联系邮箱: test@example.com, support@crawler.com"
-        emails = extract_emails(text_with_email)
-        self.assertIn("test@example.com", emails)
-        self.assertIn("support@crawler.com", emails)
-
-    def test_data_validation_tools(self):
-        """测试数据验证工具"""
-        # 测试邮箱验证
-        self.assertTrue(validate_email("test@example.com"))
-        self.assertFalse(validate_email("invalid-email"))
-        
-        # 测试URL验证
-        self.assertTrue(validate_url("https://example.com"))
-        self.assertFalse(validate_url("invalid-url"))
-        
-        # 测试电话验证
-        self.assertTrue(validate_phone("13812345678"))
-        self.assertFalse(validate_phone("12345"))
-        
-        # 测试身份证验证
-        self.assertTrue(validate_chinese_id_card("110101199001011234"))
-        self.assertFalse(validate_chinese_id_card("invalid-id"))
-        
-        # 测试日期验证
-        self.assertTrue(validate_date("2025-09-10"))
-        self.assertFalse(validate_date("invalid-date"))
-        
-        # 测试数值范围验证
-        self.assertTrue(validate_number_range(50, 1, 100))
-        self.assertFalse(validate_number_range(150, 1, 100))
-
-    def test_request_handling_tools(self):
-        """测试请求处理工具"""
-        # 测试URL构建
-        base_url = "https://api.example.com"
-        path = "/v1/users"
-        query_params = {"page": 1, "limit": 10}
-        full_url = build_url(base_url, path, query_params)
-        self.assertIn("https://api.example.com/v1/users", full_url)
-        self.assertIn("page=1", full_url)
-        self.assertIn("limit=10", full_url)
-        
-        # 测试添加查询参数
-        existing_url = "https://api.example.com/v1/users?page=1"
-        new_params = {"sort": "name"}
-        updated_url = add_query_params(existing_url, new_params)
-        self.assertIn("sort=name", updated_url)
-        
-        # 测试合并请求头
-        base_headers = {"Content-Type": "application/json"}
-        additional_headers = {"Authorization": "Bearer token123"}
-        merged_headers = merge_headers(base_headers, additional_headers)
-        self.assertEqual(merged_headers["Content-Type"], "application/json")
-        self.assertEqual(merged_headers["Authorization"], "Bearer token123")
-
-    def test_anti_crawler_tools(self):
-        """测试反爬虫应对工具"""
-        # 测试随机User-Agent
-        user_agent = get_random_user_agent()
-        self.assertIsInstance(user_agent, str)
-        self.assertGreater(len(user_agent), 0)
-        
-        # 测试代理轮换
-        proxy = rotate_proxy()
-        self.assertIsInstance(proxy, dict)
-
-    def test_distributed_coordinator_tools(self):
-        """测试分布式协调工具"""
-        # 测试任务ID生成
-        task_id = generate_task_id("https://example.com", "test_spider")
-        self.assertIsInstance(task_id, str)
-        self.assertEqual(len(task_id), 32)  # MD5 hash长度
-        
-        # 测试集群信息获取（异步函数需要特殊处理）
-        async def test_cluster_info():
-            cluster_info = await get_cluster_info()
-            self.assertIsInstance(cluster_info, dict)
-            return cluster_info
-            
-        # 运行异步测试
-        cluster_info = asyncio.run(test_cluster_info())
-        self.assertIsInstance(cluster_info, dict)
-
-
-if __name__ == '__main__':
+#!/usr/bin/python
+# -*- coding: UTF-8 -*-
+"""
+工具包测试
+"""
+import asyncio
+import unittest
+from crawlo.tools import (
+    # 日期工具
+    parse_time,
+    format_time,
+    time_diff,
+    
+    # 数据清洗工具
+    clean_text,
+    format_currency,
+    extract_emails,
+    
+    # 数据验证工具
+    validate_email,
+    validate_url,
+    validate_phone,
+    validate_chinese_id_card,
+    validate_date,
+    validate_number_range,
+    
+    # 请求处理工具
+    build_url,
+    add_query_params,
+    merge_headers,
+    
+    # 反爬虫应对工具
+    get_random_user_agent,
+    rotate_proxy,
+    
+    # 分布式协调工具
+    generate_task_id,
+    get_cluster_info
+)
+
+
+class TestTools(unittest.TestCase):
+    """工具包测试类"""
+
+    def test_date_tools(self):
+        """测试日期工具"""
+        # 测试时间解析
+        time_str = "2025-09-10 14:30:00"
+        parsed_time = parse_time(time_str)
+        self.assertIsNotNone(parsed_time)
+        
+        # 测试时间格式化
+        formatted_time = format_time(parsed_time, "%Y-%m-%d")
+        self.assertEqual(formatted_time, "2025-09-10")
+        
+        # 测试时间差计算
+        time_str2 = "2025-09-11 14:30:00"
+        parsed_time2 = parse_time(time_str2)
+        diff = time_diff(parsed_time2, parsed_time)
+        self.assertEqual(diff, 86400)  # 24小时 = 86400秒
+
+    def test_data_cleaning_tools(self):
+        """测试数据清洗工具"""
+        # 测试文本清洗
+        dirty_text = "<p>这是一个&nbsp;<b>测试</b>&amp;文本</p>"
+        clean_result = clean_text(dirty_text)
+        self.assertEqual(clean_result, "这是一个 测试&文本")
+        
+        # 测试货币格式化
+        price = 1234.567
+        formatted_price = format_currency(price, "¥", 2)
+        self.assertEqual(formatted_price, "¥1,234.57")
+        
+        # 测试邮箱提取
+        text_with_email = "联系邮箱: test@example.com, support@crawler.com"
+        emails = extract_emails(text_with_email)
+        self.assertIn("test@example.com", emails)
+        self.assertIn("support@crawler.com", emails)
+
+    def test_data_validation_tools(self):
+        """测试数据验证工具"""
+        # 测试邮箱验证
+        self.assertTrue(validate_email("test@example.com"))
+        self.assertFalse(validate_email("invalid-email"))
+        
+        # 测试URL验证
+        self.assertTrue(validate_url("https://example.com"))
+        self.assertFalse(validate_url("invalid-url"))
+        
+        # 测试电话验证
+        self.assertTrue(validate_phone("13812345678"))
+        self.assertFalse(validate_phone("12345"))
+        
+        # 测试身份证验证
+        self.assertTrue(validate_chinese_id_card("110101199001011234"))
+        self.assertFalse(validate_chinese_id_card("invalid-id"))
+        
+        # 测试日期验证
+        self.assertTrue(validate_date("2025-09-10"))
+        self.assertFalse(validate_date("invalid-date"))
+        
+        # 测试数值范围验证
+        self.assertTrue(validate_number_range(50, 1, 100))
+        self.assertFalse(validate_number_range(150, 1, 100))
+
+    def test_request_handling_tools(self):
+        """测试请求处理工具"""
+        # 测试URL构建
+        base_url = "https://api.example.com"
+        path = "/v1/users"
+        query_params = {"page": 1, "limit": 10}
+        full_url = build_url(base_url, path, query_params)
+        self.assertIn("https://api.example.com/v1/users", full_url)
+        self.assertIn("page=1", full_url)
+        self.assertIn("limit=10", full_url)
+        
+        # 测试添加查询参数
+        existing_url = "https://api.example.com/v1/users?page=1"
+        new_params = {"sort": "name"}
+        updated_url = add_query_params(existing_url, new_params)
+        self.assertIn("sort=name", updated_url)
+        
+        # 测试合并请求头
+        base_headers = {"Content-Type": "application/json"}
+        additional_headers = {"Authorization": "Bearer token123"}
+        merged_headers = merge_headers(base_headers, additional_headers)
+        self.assertEqual(merged_headers["Content-Type"], "application/json")
+        self.assertEqual(merged_headers["Authorization"], "Bearer token123")
+
+    def test_anti_crawler_tools(self):
+        """测试反爬虫应对工具"""
+        # 测试随机User-Agent
+        user_agent = get_random_user_agent()
+        self.assertIsInstance(user_agent, str)
+        self.assertGreater(len(user_agent), 0)
+        
+        # 测试代理轮换
+        proxy = rotate_proxy()
+        self.assertIsInstance(proxy, dict)
+
+    def test_distributed_coordinator_tools(self):
+        """测试分布式协调工具"""
+        # 测试任务ID生成
+        task_id = generate_task_id("https://example.com", "test_spider")
+        self.assertIsInstance(task_id, str)
+        self.assertEqual(len(task_id), 32)  # MD5 hash长度
+        
+        # 测试集群信息获取（异步函数需要特殊处理）
+        async def test_cluster_info():
+            cluster_info = await get_cluster_info()
+            self.assertIsInstance(cluster_info, dict)
+            return cluster_info
+            
+        # 运行异步测试
+        cluster_info = asyncio.run(test_cluster_info())
+        self.assertIsInstance(cluster_info, dict)
+
+
+if __name__ == '__main__':
     unittest.main()