python通过高德对外接口爬取小区信息

FW_G8Z

已于 2023-12-31 20:15:57 修改

阅读量749

点赞数 10

文章标签： python 开发语言爬虫

于 2023-12-31 20:15:14 首次发布

本文链接：https://blog.csdn.net/weixin_38148680/article/details/135319161

版权

import json,time
from urllib import request
import requests

# key ='xxxxxxxxxxxx'
#polygon='121.953679,30.181944|122.322236,29.940442'
# types=120304
# size=25

url_base='http://restapi.amap.com/v3/place/polygon?key=xxxxxxxxxxxxxxxx&polygon=121.956409,30.178268|122.328571,29.93579|122.03606,29.982192|122.291492,30.093944&types=080100|080101|080102|080103|80104|080105|080106|080107|080108|080110|080111|080112|080113|080114|080115|080116|080117|080118|080119|080200|080201|080202&page_size=25&page_num={0}'
num=1
out=[]
while num<=20:
    url=url_base.format(num)
    print(url)
    html=request.urlopen(url).read()
    t=json.loads(html)
    # print(t)
    p=t['pois']
    # print(p)
   
    f=['address', 'pname', 'cityname', 'type', 'typecode', 'adname', 'name', 'location']

    for poi in p:
            line=[]
            for i in f:
                    line.append(poi[i])
            line.append(poi['location'].split(',')[0])
            line.append(poi['location'].split(',')[1])
            out.append(line)
            print(line)
    print(num)
    num+=1 
   

timemark=str(time.strftime('%Y%m%d%H%M%s'))    

output = open('poi_{0}.txt'.format(timemark),'wb')
title ='\t'.join(f)+'\tx\ty\n'
output.write(title.encode('utf8'))
# print(title.encode('utf8'))

for l in out :
        output.write(('\t'.join(l)+'\n').encode('utf8'))
        # print(('\t'.join(l)+'\n').encode('utf8'))

output.close()