企业动态内容小于300过滤掉

comData/Tyc/tyc_qydt_add.py

@@ -158,6 +158,8 @@ def beinWork(tyc_code, social_code,start_time):
                 contentText = smart.extract_by_url(link).text
                 #不带标签正文
                 content = smart.extract_by_url(link).cleaned_text
+                if len(content) < 300:
+                    continue
                 # time.sleep(3)
             except Exception as e:
                 contentText = ''

comData/weixin_solo/wxList.py

@@ -49,6 +49,7 @@ def getSourceInfo(infoSourceCode):
     sql = f"SELECT site_uri,id,site_name,info_source_code from info_source where info_source_code = '{infoSourceCode}' "
     cursor.execute(sql)
     row = cursor.fetchone()
+    cnx.commit()
     dic_url = {
         'url_': row[0],
         'sid': row[1],
@@ -143,6 +144,7 @@ def updateCookieToken(token,cookies):
 def getToken():
     cursor_.execute(f"select token,cookies from weixin_tokenCookies  where fenghao_time < DATE_SUB(NOW(), INTERVAL 2 HOUR) order by update_time asc limit 1")
     row = cursor_.fetchall()
+    cnx_.commit()
     if row:
         pass
     else:
@@ -159,7 +161,7 @@ def getPageData(dic_url,page):
     info_source_code = dic_url['info_source_code']
     biz = dic_url['biz']
     fakeid = biz + '=='
-    tokenAndCookie =  getToken()
+    tokenAndCookie = getToken()
     if tokenAndCookie:
         pass
     else: