from lxml import etree
import requests
import csv
import time
def writecsv(item):
with open('qfang.csv','a',encoding='utf-8') as f:
writer = csv.writer(f)
try:
writer.writerow(item)
except:
print('writer error!')
if __name__ =='__main__':
header={'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; WOW64)AppleWebKit/537.36 (KHTML, like Gecko) Chrome/74.0.3729.169Safari/537.36'}
start_url = 'https://xiny.fang.anjuke.com/loupan/all/p'
for x in range (1,7):
url = start_url + str(x)+'/'
html = requests.get(url,headers = header)
time.sleep(1)
selector = etree.HTML(html.text)
xiaoqulist = selector.xpath('//*[@id="container"]/div[2]/div[1]/div[3]/div')
for xiaoqu in xiaoqulist:
mingcheng = xiaoqu.xpath('div/a[1]/span/text()')
dizhi = xiaoqu.xpath('div/a[2]/span/text()')
mianji = xiaoqu.xpath('div/a[3]/span[3]/text()')
junjia = xiaoqu.xpath('a[2]/p/span/text()')
item = [mingcheng,dizhi,mianji,junjia]
writecsv(item)
print('正在爬取:',mingcheng)