Merge remote-tracking branch 'origin/master'

comData/weixin_solo/oneWeixin2.py

@@ -2,6 +2,7 @@
 '''
 成功100 发送数据失败200  请求失败400  文章内容为空500
 '''
+import re
 
 import requests, time, random, json, pymysql, redis
 
@@ -32,7 +33,7 @@ def updatewxLink(link,info_source_code,state):
 
 def getjsonInfo():
     #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 = "select * from wx_link  where state=100 order by id asc limit 1"
     cursor_.execute(select_sql)
     row = cursor_.fetchone()
     cnx_.commit()
@@ -85,8 +86,10 @@ def get_info(dict_json):
     soup_news = BeautifulSoup(res_news.content, 'html.parser')
 
     news_html = soup_news.find('div', {'id': 'js_content'})
+    news_html['style'] = 'width: 814px ; margin: 0 auto;'
     try:
-        del news_html['style']
+        #del news_html['style']
+        news_html=rm_style_attr(news_html)
         del news_html['id']
         del news_html['class']
     except:
@@ -209,6 +212,27 @@ def get_info(dict_json):
     updatewxLink(url_news,info_source_code,100)
     return True
 
+def rm_style_attr(soup):
+    # 查找所有含有style属性的标签
+    style_tags = soup.find_all(style=True)
+    # 遍历每个style标签
+    for style_tag in style_tags:
+        try:
+            # 使用正则表达式替换
+            styleattr = style_tag['style']
+            styleattr = re.sub(r'visibility:(?s).{1,}?;', '', styleattr)
+            styleattr = re.sub(r'font-family:(?s).{1,}?;', '', styleattr)
+            styleattr = re.sub(r'color:(?s).{1,}?;', '', styleattr)
+            styleattr = re.sub(r'font-size:(?s).{1,}?;', '', styleattr)
+            style_tag['style'] = styleattr
+        except:
+            continue
+
+    first_div = soup.select('div[id="js_content"]')
+    # 设置style属性
+    first_div['style'] = 'width: 814px ; margin: 0 auto;'
+    return soup
+
 if __name__=="__main__":
     num_caiji = 0
     list_all_info = []