从http://www.pm25.com/shenzhen.html抓取北京,深圳,上海,广州,成都的pm2.5指数,并按照空气质量从优到差排序,保存在txt文档里
代码如下:
#coding=utf-8 from selenium import webdriver from time import sleep class PM: def __init__(self): self.dr = webdriver.Chrome() self.pm25_info = self.get_pm25_info() def get_pm25_info(self): city_list = ['beijing','shenzhen','shanghai','guangzhou','chengdu'] pm_info_list = [] i = 0 while i < len(city_list): self.dr.get('http://www.pm25.com/'+city_list[i]+'.html') sleep(3) city_name = self.dr.find_element_by_css_selector('.bi_loaction_city').text #定位城市名字 pm_exp = self.dr.find_element_by_css_selector('.pm25_span').text #定位PM2.5指数 air_num = self.dr.find_element_by_css_selector('.bi_aqiarea_num').text #定位AQI指数 air_quality = self.dr.find_element_by_css_selector('.bi_aqiarea_right span').text #定位空气质量描述 pm_info_list.append((city_name, pm_exp, air_num, air_quality)) i += 1 pm_info_list.sort(key=lambda x:float(x[2])) #按AQI指数排序(AQI指数决定了空气质量的优良) return pm_info_list def get_pm_info_file(self): self.file_title = '北上广深成五地PM2.5数据信息' self.file = open(self.file_title + '.txt', 'wb') self.floor = 1 for item in self.pm25_info: separate_line = '**********' + 'TOP' + str(self.floor) + '**********\n' self.file.write(separate_line.encode('utf-8')) self.file.write(('城市:' + item[0] + '\n' 'PM2.5指数:' + item[1] + ' '+ '微克/立方米' + '\n' 'AQI指数:' + item[2] + '\n' '空气质量:' + item[3]+ '\n').encode('utf-8')) self.floor += 1 self.file.close() def quit(self): self.dr.quit() if __name__ == '__main__': pm_info = PM() pm_info.get_pm_info_file() pm_info.quit()
网页如下:
生成txt如下: