Merge pull request #616 from ahuangg/alan-branch

update: improved retrieval of cookies workflow

Author: dataabc

weibo_spider/config_util.py

@@ -174,29 +174,44 @@ def add_user_uri_list(user_config_file_path, user_uri_list):
         user_uri_list[0] = '\n' + user_uri_list[0]
     with codecs.open(user_config_file_path, 'a', encoding='utf-8') as f:
         f.write('\n'.join(user_uri_list))
-        
+      
 def get_cookie():
     """Get weibo.cn cookie from Chrome browser"""
     try:
         chrome_cookies = browser_cookie3.chrome(domain_name='weibo.cn')
         cookies_dict = {cookie.name: cookie.value for cookie in chrome_cookies}
-        cookie_string = '; '.join(f'{name}={value}' for name, value in cookies_dict.items())
-        return cookie_string
+        return cookies_dict
     except Exception as e:
         logger.error(u'Failed to obtain weibo.cn cookie from Chrome browser: %s', str(e))
         raise 
 
-def update_cookie_config(user_config_file_path):
-    "Update cookie in config.json"
+def update_cookie_config(cookie, user_config_file_path):
+    """Update cookie in config.json"""
     if not user_config_file_path:
         user_config_file_path = os.getcwd() + os.sep + 'config.json' 
     try:
-        cookie = get_cookie()
         with codecs.open(user_config_file_path, 'r', encoding='utf-8') as f:
             config = json.load(f)
-        config['cookie'] = cookie
-        with codecs.open(user_config_file_path, 'w', encoding='utf-8') as f:
-            json.dump(config, f, indent=4, ensure_ascii=False)
+            
+        cookie_string = '; '.join(f'{name}={value}' for name, value in cookie.items())
+        
+        if config['cookie'] != cookie_string:
+            config['cookie'] = cookie_string
+            with codecs.open(user_config_file_path, 'w', encoding='utf-8') as f:
+                json.dump(config, f, indent=4, ensure_ascii=False)
     except Exception as e:
         logger.error(u'Failed to update cookie in config file: %s', str(e))
         raise 
+    
+def check_cookie(user_config_file_path): 
+    """Checks if user is logged in"""
+    try:
+        cookie = get_cookie()
+        if cookie["MLOGIN"] == '0':
+            logger.warning("使用 Chrome 在此登录 %s", "https://passport.weibo.com/sso/signin?entry=wapsso&source=wapssowb&url=https://m.weibo.cn/")
+            sys.exit()
+        else:
+            update_cookie_config(cookie, user_config_file_path)
+    except Exception as e:
+        logger.error(u'Check for cookie failed: %s', str(e))
+        raise 

weibo_spider/spider.py

@@ -366,14 +366,14 @@ def _get_config():
         config_path = FLAGS.config_path
     elif not os.path.isfile(config_path):
         shutil.copy(src, config_path)
-        config_util.update_cookie_config(config_path)
         logger.info(u'请先配置当前目录(%s)下的config.json文件，'
                     u'如果想了解config.json参数的具体意义及配置方法，请访问\n'
                     u'https://github.com/dataabc/weiboSpider#2程序设置' %
                     os.getcwd())
         sys.exit()
     try:
         with open(config_path) as f:
+            config_util.check_cookie(config_path)
             config = json.loads(f.read())
             return config
     except ValueError: