import requests
from lxml import etree
import pymongo
class article(object):
def __init__(self):
# 创建连接,ip地址和端口号
self.client = pymongo.MongoClient("iP地址", 端口号)
# 连接的数据库
db = self.client.数据库的名字
# 账号密码的认证
db.authenticate("账号", "密码")
# 连接的表
self.conn_table = db.表名
def write_to_mongodb(self):
url = 'http://caipiao.163.com/help/14/0805/21/A2TOI4UF00754IHE_3.html'
response = requests.get(url)
content = response.content.decode('gb2312')
tree = etree.HTML(content)
# 转为字符串
p_list = tree.xpath('//div[@class="help_t_int"]/p')
text = ''
for num in range(len(p_list)):
conten = etree.tostring(p_list[num], encoding="utf-8",pretty_print=True,
method="xml").decode('utf-8')
text += conten
data = {
'article_name': '易乐11选5奖项规则',
'article_sub_name': '',
'lottery_name': '易乐11选5',
'article_tag': '',
'article_class_name': '奖项规则',
'article_recommend': '',
'article_from': '',
'article_from_url': url,
'article_author': '',
'article_content': text,
'article_abstract': '',
'article_img': '',
'is_delete': 0
}
self.conn_table.insert(data)
article().write_to_mongodb()