1/24

comData/BaseInfo_qcc/baseinfo1122.py

@@ -55,7 +55,19 @@ def baseinfo(com_soup):
     data = {}
     for cominfo in cominfo_list:
         # print(cominfo)
+        try:
             value = cominfo.find('span', class_='val').text.replace('复制', '').strip(' ')
+        except:
+            try:
+                value_tags = cominfo.find_all('span')
+                for _ in value_tags:
+                    if len(_.attrs) == 0:
+                        value = _.text.replace('复制', '').strip(' ')
+                        break
+                else:
+                    return data
+            except:
+                return data
         pattern = r'\（\d{4}\s*年\）'
         match = re.search(pattern, value)
         if match:

comData/weixin_solo/get_tokenCookies.py

@@ -56,7 +56,7 @@ if __name__=="__main__":
     url = "https://mp.weixin.qq.com/"
     browser.get(url)
     # 可改动
-    time.sleep(20)
+    time.sleep(70)
 
     s = requests.session()
     #获取到token和cookies

comData/weixin_solo/oneWeixin2.py

@@ -32,8 +32,15 @@ def updatewxLink(link,info_source_code,state):
 
 
 def getjsonInfo():
+    # todo:从redis中获取一条
+    linkid = baseCore.redicPullData('WeiXinGZH:linkid')
+    if linkid:
+        pass
+    else:
+        log.info('-----没有数据了-----')
+        return False
     #从数据库中获取信息 一条
-    select_sql = "select * from wx_link  where state=0 order by id asc limit 1"
+    select_sql = f"select * from wx_link  where state=0 and id= '{linkid}'"
     cursor_.execute(select_sql)
     row = cursor_.fetchone()
     cnx_.commit()

comData/weixin_solo/wxList.py

@@ -4,6 +4,7 @@ import time
 import random
 
 import pymysql
+import redis
 import requests
 import urllib3
 from pymysql.converters import escape_string
@@ -113,7 +114,20 @@ def insertWxList(dic_url,json_search,page):
                 cnx_.commit()
             except Exception as e:
                 log.error(f"保存数据库失败：{e}")
+
+            # 查询放入之后的id
+            selectIdSql = f"select id from wx_link where sid='{dic_url['sid']}' and link='{escape_string(url_news)}'"
+            cursor_.execute(selectIdSql)
+            linkid = cursor_.fetchone()[0]
             # todo: 放入redis
+            try:
+                r.ping()
+            except:
+                r_ = redis.Redis(host="114.115.236.206", port=6379, password='clbzzsn', db=6)
+                r_.lpush('WeiXinGZH:url', linkid)
+                continue
+            r.lpush('WeiXinGZH:linkid',linkid)
+
     log.info(f"---{dic_url['name']}--第{page}页----总数：{listCount}---重复数：{repetCount}---新增数：{insertCount}-------------")
     if listCount==0:
         #列表为空认为结束