Enhance manual crawling functionality in GXGPMonitorApp to skip database saving for both announcements and filtered results. Update logging to reflect manual crawl mode actions, ensuring clarity in the process. Adjusted comments for better understanding of the manual crawling logic.

2026-01-08 19:12:37 +08:00
parent 04d254cfef
commit 85cd7e9d3e
10 changed files with 1115 additions and 6 deletions
@@ -0,0 +1,97 @@
+#!/usr/bin/env python3
+"""
+测试定时爬取脚本配置
+"""
+
+import sys
+import os
+
+# 添加项目路径
+sys.path.insert(0, os.path.dirname(__file__))
+
+try:
+    from gx_gp_monitor.core.config_manager import load_config
+    from gx_gp_monitor.core.logger import init_logger
+    from gx_gp_monitor.storage.postgresql import init_storage
+
+    def test_config():
+        """测试配置加载"""
+        print("🔍 测试配置加载...")
+
+        config = load_config()
+        if not config:
+            print("❌ 配置加载失败")
+            return False
+
+        print("✅ 配置加载成功")
+        print(f"   关键词: {config.crawler.keyword}")
+        print(f"   企业微信启用: {config.wechat_app.enabled}")
+        print(f"   数据库启用: {config.database.enabled}")
+
+        return True
+
+    def test_logger():
+        """测试日志系统"""
+        print("\n🔍 测试日志系统...")
+
+        try:
+            init_logger()
+            print("✅ 日志系统初始化成功")
+            return True
+        except Exception as e:
+            print(f"❌ 日志系统初始化失败: {e}")
+            return False
+
+    def test_database():
+        """测试数据库连接"""
+        print("\n🔍 测试数据库连接...")
+
+        try:
+            init_storage()
+            print("✅ 数据库连接成功")
+            return True
+        except Exception as e:
+            print(f"❌ 数据库连接失败: {e}")
+            return False
+
+    def main():
+        """主测试函数"""
+        print("🚀 开始测试定时爬取脚本配置\n")
+
+        results = []
+        results.append(("配置加载", test_config()))
+        results.append(("日志系统", test_logger()))
+        results.append(("数据库连接", test_database()))
+
+        print("\n" + "="*50)
+        print("📊 测试结果:")
+
+        all_passed = True
+        for test_name, passed in results:
+            status = "✅ 通过" if passed else "❌ 失败"
+            print(f"   {test_name}: {status}")
+            if not passed:
+                all_passed = False
+
+        print("\n" + "="*50)
+        if all_passed:
+            print("🎉 所有测试通过！可以安全使用定时爬取脚本。")
+            print("\n💡 使用方法:")
+            print("   直接运行: python cron_crawl.py")
+            print("   定时运行: 参考 CRON_README.md 或 cron_example.txt")
+        else:
+            print("⚠️  部分测试失败，请检查配置后再使用脚本。")
+
+        return all_passed
+
+    if __name__ == "__main__":
+        success = main()
+        sys.exit(0 if success else 1)
+
+except ImportError as e:
+    print(f"❌ 导入失败: {e}", file=sys.stderr)
+    print("请确保已安装所有依赖: pip install -r gx_gp_monitor/requirements.txt", file=sys.stderr)
+    sys.exit(1)
+except Exception as e:
+    print(f"❌ 测试失败: {e}", file=sys.stderr)
+    sys.exit(1)