链家 xpath

 调用封装mysql  保存链家前三页

import requests,re
from lxml import etree
import mysqlhelp

base_url = 'https://bj.lianjia.com/zufang/pg%srp1/'

myhelp = mysqlhelp.mysql_conn()

sql = 'INSERT INTO lianjia (title, region, zone, meters, direction, price) VALUES' \
      ' (%s, %s, %s, %s, %s, %s)'


for i in range(1,4):
    url = base_url % i
    response = requests.get(url)
    html_ele = etree.HTML(response.text)
    # ul 的所有信息
    li_list = html_ele.xpath('.//ul[@id="house-lst"]/li')
    # print(li_list)
    for li_ele in li_list:
        title = li_ele.xpath('./div[2]/h2/a')[0].text
        print(title)
        region = li_ele.xpath('./div[2]/div[1]/div[1]/a/span')[0].text
        print(region)
        zone = li_ele.xpath('./div[2]/div[1]/div[1]/span[1]/span')[0].text
        print(zone)
        meters = li_ele.xpath('.//div[@class="info-panel"]/div[1]/div[1]/span[2]')[0].text
        print(meters)
        direction = li_ele.xpath('.//div[@class="info-panel"]/div[1]/div[1]/span[3]')[0].text
        print(direction)
        price = li_ele.xpath('.//div[@class="price"]/span[1]')[0].text
        print(price)

        res_match = re.match('\d+',meters)
        meters = res_match.group(0)
        data = (title, region, zone, meters, direction, price)
        myhelp.execute_modify_mysql(sql,data)

mysql封装

import pymysql

class mysql_conn(object):

    def __init__(self):
        self.db = pymysql.connect(host='127.0.0.1', user='root', password='123456', port=3306, database='py10',charset='utf8')
        self.cursor = self.db.cursor()

    def execute_modify_mysql(self, sql, data):
        self.cursor.execute(sql,data)
        self.db.commit()

    def __del__(self):
        self.cursor.close()
        self.db.close()

if __name__=='__main__':
    sql = 'insert into xueqiu_test values (3)'
    mc = mysql_conn()
    mc.execute_modify_mysql(sql)       # 执行sql语句

#测试
# if __name__=='__main__':
#     sql = 'insert into xueqiu test values(1)'
#     cursor.execute(sql)
#     db.commit()
#     cursor.close()