链家房价python

# -*- coding: utf-8 -*-
# @Time : 2022/12/23 20:46

import re # 正则表达式
import json
from numpy import * # 创建给定类型的矩阵，并初始化为0
import urllib.request,urllib.error

location = re.compile(r'data-el="region">(.*?)</a>', re.S)
houseInfo = re.compile(r'class="houseIcon"></span>(.*?)</div>', re.S)
years = re.compile(r'')
priceall = re.compile(r'<div class="priceInfo">(.*?)</span>', re.S)
priceaverage = re.compile(r'data-price="(.*?)"',re.S)
four_location_C = ['金茂府']
four_location_P = ['jinmaofu']
def get_data():
'''
通过循环获取租房信息
:return:
'''
four_lo = {}
for j in range(1):
five_page = {} #爬取5页
k = 1
for i in range(1, 2):
print('正在爬取第%d页' % (i))
# 爬取成都链家网租房信息
baseurl = 'https://jn.lianjia.com/ershoufang/pg'+str(i)+'rs'+four_location_P[j]+'/'
url = baseurl
header = {
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'}
request = urllib.request.Request(url, headers=header)
html = ""
response = urllib.request.urlopen(request)
html = response.read().decode("utf-8")

bian1 = re.compile(r'\u2022')
bian2 = re.compile(r'\xa9')
con = re.sub(bian1, '', html)
con = re.sub(bian2, '', con)

locat = re.findall(location, con)
house = re.findall(houseInfo, con)
price1 = re.findall(priceall, con)
price2 = re.findall(priceaverage, con)

#one_page = []
for i in range(len(locat)):
one = {}
one['楼盘'] = locat[i]
one['总价'] = re.findall("[0-9]+", price1[i])[1] + '万'
one['面积'] = house[i].split('|')[1]
print(house[i])
if '年建' in house[i]:
t = house[i].index('年建')
one['建造时间'] = house[i][t - 4:t]
else:
one['建造时间'] = 0
one['单价']=price2[i] + '元/平'
five_page[k]=one
k = k+1

four_lo[four_location_C[j]]=five_page
with open("北京二手房.json", "w", encoding='utf-8') as f: ## 设置'utf-8'编码
f.write(json.dumps(four_lo, ensure_ascii=False,indent=4 ))


get_data()