环保部的网站改了，原来的链接没有了，发送请求的方式也稍微改了一些。
1 #!/usr/bin/env python3 2 # -*- coding: utf-8 -*- 3 # 获取国内城市 30天内 每天空气质量指数（AQI） 4 # 数据来源：中华人民共和国环境保护部 5 import requests 6 from bs4 import BeautifulSoup 7 import datetime 8 9 ref_url = 'http://datacenter.mep.gov.cn:8099/ths-report/report!list.action' 10 req_url = 'http://datacenter.mep.gov.cn:8099/ths-report/report!list.action' 11 12 13 def get_html_content(city='西安市', V_DATE='2017-03-01', E_DATE='2017-03-03'): 14 headers = { 15 'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8', 16 'Accept-Encoding': 'gzip, deflate', 17 'Accept-Language': 'zh-CN,zh;q=0.8', 18 'Content-Type': 'application/x-www-form-urlencoded', 19 'Referer': ref_url, 20 'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2924.87 Safari/537.36' 21 } 22 data_dict = { 23 'page.pageNo': '1', 24 'xmlname': '1462259560614', 25 'queryflag': 'close', 26 'CITY': city, 27 'isdesignpatterns': 'false', 28 'V_DATE': V_DATE, 29 'E_DATE': E_DATE, 30 } 31 r = requests.post(req_url, data=data_dict, headers=headers, timeout=60) 32 return r.text 33 34 35 def date_format(): 36 dd = datetime.datetime.now() 37 E_DATE = dd.strftime('%Y-%m-%d') 38 _31DayAgo = (dd - datetime.timedelta(days=31)) 39 V_DATE = _31DayAgo.strftime('%Y-%m-%d') 40 return (V_DATE, E_DATE) 41 42 43 def get_aqi_info(html_content): 44 soup = BeautifulSoup(html_content, 'html.parser') 45 aqi_table = soup.find(id='GridView1') 46 aqi_trs = aqi_table.find_all('tr')[1:] 47 result = {} 48 for aqi_tr in aqi_trs: 49 aqi_tds = aqi_tr.find_all('td') 50 aqi = aqi_tds[3].string 51 dd = aqi_tds[6].string 52 result[dd] = aqi 53 return result 54 55 56 if __name__ == '__main__': 57 city = input('请输入城市名称，如"西安市"： ') 58 (V, E) = date_format() 59 content = get_html_content(city, V, E) 60 result = get_aqi_info(content) 61 print(city, '最近30天空气质量指数如下：\n') 62 for key in sorted(result.keys()): 63 print(key, '\t', result[key])
执行的结果类似这样：
