思路

這里抓取的是中國天氣網(wǎng)的數(shù)據(jù)，它是個靜態(tài)頁面，請求的URL是：

http://d1.weather.com.cn/sk_2d/101190101.html?_=1527152861827

兩個變量，一個是html前的一串?dāng)?shù)字，這個數(shù)字代表一個城市，因此我們需要城市名和數(shù)字編號的轉(zhuǎn)換關(guān)系。后面一串?dāng)?shù)字很明顯是當(dāng)前時間。

該請求會返回一個字典，每個key的命名還是很容易猜到其意義的，就不多說了。

代碼

weather-result.jpg

weather.py

import re
import Sqlite3api as sqlite3
import Configure as Configs
import requests
from random import choice
from bs4 import BeautifulSoup
import time
import ast

def import_data():
    with open('cityWeather18-03-4.sql', 'r', encoding='utf-8') as file:
        content = file.read()

    pattern = re.compile('(\d+),\'(.*?)\',\'(\d+)\'',re.S)
    data = pattern.findall(content)
    print ("Retrieve {0:d} weather codes.".format(len(data)))

    conn = sqlite3.sqlite3_init()

    for d in data:
        print (d)
        sql = "INSERT INTO weather VALUES ({0:s},'{1:s}',{2:s})".format(d[0], d[1].strip(),d[2])
        sqlite3.sqlite3_execute(conn, sql)

    sqlite3.sqlite3_close(conn)

def get_citycode_by_cityname(cityname):
    conn = sqlite3.sqlite3_init()

    ret = sqlite3.sqlite3_execute(conn, "select cityname, citycode from weather WHERE cityname LIKE '%{0:s}%'".format(cityname))

    sqlite3.sqlite3_close(conn)

    return ret

def get_weather_by_citycode(citycode):
    url = "http://d1.weather.com.cn/sk_2d/{0:d}.html?_={1:d}".format(citycode, int(time.time()*1000))

    header = {}
    header['user-agent'] = choice(Configs.FakeUserAgents)
    header['Referer'] = "http://www.weather.com.cn/weather1d/101190101.shtml"

    try:
        response = requests.get(url, headers=header)
        content = ''
        if response.status_code == requests.codes.ok:
            response.encoding = 'utf-8'
            content = response.text
    except Exception as e:
        print (e)

    pattern = re.compile('{(.*?)}', re.S)
    data = pattern.findall(content)[0]
    data = ast.literal_eval("{"+data+"}")

    print ("城市:", data.get('cityname'))
    print ("日期:", data.get('date'))
    print ("時間:", data.get('time'))
    print ("攝氏溫度:", data.get('temp'))
    print ("華氏溫度:", data.get('tempf'))
    print ("天氣:", data.get('weather'))
    print ("濕度:", data.get('SD'))
    print ("風(fēng)向:", data.get('WD'))
    print ("風(fēng)級:", data.get('WS'))
    print ("空氣質(zhì)量:", data.get('aqi'))
    print ("空氣質(zhì)量PM2.5:", data.get('aqi_pm25'))



def main():
    while True:
        print ("-"*80)
        cityname = input("請輸入要查詢的城市名: ")

        ret = get_citycode_by_cityname(cityname)

        if not ret:
            print ("沒有找到該城市，請重新輸入。")
            continue
        
        if 1 == len(ret):
            get_weather_by_citycode(int(ret[0][1]))
            continue

        cnt = 1
        table = PrettyTable()
        table.field_names = ["編號", "城市名字"]
        table.sortby = "編號"

        for r in ret:
            table.add_row([cnt, r[0]])
            cnt += 1

        print (table)

        code = int(input("找到不只一個城市，請輸入其編號: "))
        get_weather_by_citycode(int(ret[code-1][1]))
    


if __name__ == '__main__':
    #import_data()
    main()