#print(i.split(',')[4], end=' ')
#print(i.split(',')[5])
data3 = [
i.split(',')[0],
i.split(',')[2],
i.split(',')[3],
i.split(',')[4],
i.split(',')[5]
]
print(data3)
data4.append(data3)
weather_data = {'city': city, '24hour_weather': data4}
db_utils.save_one_to_mongo('24hour_weather', weather_data)
print('save to mongo')
if __name__ == "__main__":
db_util = DB_Utils()
db_util.drop_collection('24hour_weather')
web_data_list = []
for key in cities:
url = get_city_weather_url(key)
web_data = get_weather_html(url)
get_1day_weather_data(web_data)
#web_data_list.append(web_data)
#get_weather_data(web_data)
#pool = Pool(1)
#pool.map(get_1day_weather_data,web_data_list)
#pool.close()
f.write('}')
f.close()
#使用BeautifulSoup解析列表xml并提取城市信息存储到MongoDB
def get_citylist_to_mongo(citylist_xml):
save_to_db = DB_Utils()
soup = BeautifulSoup(citylist_xml, 'lxml')
#print(soup.prettify())
city_info = soup.select('d')
for i in city_info:
data = {
'province': i.get('d4'),
'city_name': i.get('d2'),
'city_code': i.get('d1'),
'city_piny': i.get('d3')
}
if i.get('d4') == '韩国':
break
save_to_db.save_one_to_mongo('citylist', data)
print('正在将城市列表存储到mongodb: #', i.get('d4'), i.get('d2'))
if __name__ == "__main__":
#url = CITYLIST_URL
db_utils = DB_Utils()
db_utils.drop_collection('citylist')
url = 'http://mobile.weather.com.cn/js/citylist.xml'
citylist_xml = get_citylist_xml(url)
get_citylist_to_mongo(citylist_xml)
weather_data_of_1day)
print('save to mongo')
except:
print(province, city_name, city_code)
db_utils1 = DB_Utils()
error_city = {
'province': province,
'city_name': city_name,
'city_code': city_code,
}
db_utils1.save_one_to_mongo('error', error_city)
print('save error to mongo')
pass
if __name__ == "__main__":
db_util = DB_Utils()
db_util.drop_collection('1day_weather_data')
data = db_util.query_all('citylist')
for i in data:
url = city_code_to_url(i.get('city_code'))
province = i.get('province')
city_name = i.get('city_name')
city_code = i.get('city_code')
city_piny = i.get('city_piny')
info = [city_name, city_code, city_piny]
print(info)
web_data = get_weather_html(url)
get_1day_weather_data(province, city_name, city_code, city_piny,
web_data)