修改aiqicha login逻辑， cookie文件位置

.gitignore

@@ -0,0 +1,25 @@
+
+.gitignore
+.git/
+.idea/
+.vscode/
+.DS_Store
+.env
+.env.local
+.env.development.local
+.env.test.local
+.env.production.local
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+pnpm-debug.log*
+node_modules/
+dist/
+build/
+.next/
+.out/
+.serverless/
+cdk.out/
+.vscode-test/
+cookies/
+data/

cookies/.gitignore

@@ -0,0 +1 @@
+*

crawler_bing_main.py

@@ -346,7 +346,7 @@ if __name__ == '__main__':
                 filter_list_with_not_match = filter_aiqicha_qcc(data_list, company_name, with_not_match=True)
                 # 创建CSV工具实例
                 csv_tool = CSVTool(
-                    csv_file_name='company_search_filter_is_none_data.csv',
+                    csv_file_name='data/company_search_filter_is_none_data.csv',
                     headers=['company_name','title', 'web_site_type','url',  'request_url',  'create_time']
                 )
 

data/.gitignore

@@ -0,0 +1 @@
+*

tool/aiqicha_detail.py

@@ -7,7 +7,7 @@ from tool.aiqicha_login import AiqichaLoginManager  # 导入登录管理器
 from tool.aiqicha_detail_parser import AiqichaDetailParser  # 导入解析器
 
 class AiqichaDetailCrawler:
-    def __init__(self, cookie_path="aiqicha_cookies.json"):
+    def __init__(self, cookie_path="cookies/aiqicha_cookies.json"):
         self.browser = WebBrowser(cookie_path)
         self.browser_started = False
         self.login_manager = None  # 添加登录管理器实例
@@ -28,11 +28,11 @@ class AiqichaDetailCrawler:
                     print("已加载Cookie")
 
                 # 使用登录管理器检测登录状态
-                # logined = self.login_manager.check_and_login()
-                # if logined:
-                #     print("登录成功")
-                # else:
-                #     print("登录失败")
+                logined = self.login_manager.check_and_login()
+                if logined:
+                    print("登录成功")
+                else:
+                    print("登录失败")
                 self.browser_started = True
             except Exception as e:
                 print(f"启动浏览器失败: {e}")

tool/aiqicha_login.py

@@ -16,6 +16,21 @@ class AiqichaLoginManager:
         else:
             print("已加载Cookie，验证登录状态...")
             # 加载cookie后访问页面验证是否真正登录
+            # 判断当前的url地址 如果是 https://aiqicha.baidu.com/company_detail_* 地址
+            if "company_detail" in self.browser.page.url:
+                # 等待页面加载完成
+                self.browser.page.wait_for_load_state("networkidle")
+
+                # 验证登录状态
+                if not self.check_login_status():
+                    print("Cookie已过期或无效，重新登录...")
+                    return self.login()
+                else:
+                    print("Cookie有效，已登录")
+                    return True
+
+            else:
+                print("未登录或已过期，开始登录流程...")
                 self.browser.page.goto("https://aiqicha.baidu.com")
                 # 等待页面加载完成
                 self.browser.page.wait_for_load_state("networkidle")

tool/bing_search.py

@@ -6,7 +6,7 @@ from tool.web_browser import WebBrowser
 
 
 class BingSearcher:
-    def __init__(self, cookie_path="bing_cookies.json"):
+    def __init__(self, cookie_path="cookies/bing_cookies.json"):
         self.browser = WebBrowser(cookie_path)
         self.browser_started = False