爬取天气信息（我的第一个爬虫）增加简单的tkinter

Cool_Breeze

tkinter

1. #coding=utf-8
   
2. 
   
3. import tkinter as tk
   
4. import time
   
5. from tianqi import main
   
6. def home_mian():
   
7.     font = 'Consolas', 11
   
8.     home = tk.Tk()
   
9.     home.title('天气查询 by cool_breeze')
   
10.     home.geometry('800x600')
    
11.    
    
12.     var = tk.StringVar()
    
13.     varL = []
    
14.     for n in range(1,18):
    
15.         varL.append(tk.StringVar()) #将对象放入列表
    
16. 
    
17.     Llist = []
    
18.     for n in range(1,18):
    
19.         Llist.append(tk.Label(home, textvariable=varL[n-1], bg ='honeydew', font=font))
    
20.     L0 = tk.Label(home, textvariable=var, bg ='honeydew', font=font)
    
21. 
    
22.     for n in Llist:
    
23.         n.pack(fill=tk.X, side=tk.TOP)
    
24.     L0.pack(fill=tk.X, side=tk.TOP)
    
25. 
    
26.     def tishi():
    
27.         var.set('正在获取网页信息...')
    
28.         home.update() #更新窗口
    
29.     def zigong(n):
    
30.         tishi()
    
31.         res = main(n)
    
32.         for i,j in enumerate(varL):
    
33.             j.set(res[i])
    
34.         var.set('')
    
35.         
    
36.     def exit_():
    
37.         exit()
    
38.    
    
39.     zg_bt = tk.Button(home,text='自流井',font=font,bg = 'ivory',width=10,height=1,command=lambda:zigong(1)).pack(side='left')
    
40.     cd_bt = tk.Button(home,text='成都',font=font,bg = 'ivory',width=10,height=1,command=lambda:zigong(2)).pack(side='left')
    
41.     dg_bt = tk.Button(home,text='东莞',font=font,bg = 'ivory',width=10,height=1,command=lambda:zigong(3)).pack(side='left')
    
42.     ex_bt = tk.Button(home,text='退出',font=font,bg = 'ivory',width=10,height=1,command=exit_).pack(side='left')
    
43. 
    
44.     home.mainloop()
    
45. if __name__ == '__main__':
    
46.     home_mian()

复制代码

爬虫

1. #coding=utf-8
   
2. 
   
3. import urllib.request,urllib.error
   
4. from bs4 import BeautifulSoup as bfs
   
5. import os
   
6. import re
   
7. 
   
8. def main(number):
   
9.     city_list = [
   
10.                 'positionCityID=59289; positionCityPinyin=zigong; lastCountyId=71992; lastCountyPinyin=ziliujing',
    
11.                 'lastCountyId=56294; lastCountyPinyin=chengdu; lastCountyTime=1591617161',
    
12.                 'lastCountyId=59289; lastCountyPinyin=dongguan; lastCountyTime=1591612281',
    
13.                 ]
    
14.    
    
15.     url = 'http://tianqi.2345.com/'
    
16.    
    
17.     html = askurl(url, city_list[number-1])
    
18.     data = getdata(html)
    
19.     return data
    
20. 
    
21. def format(string):
    
22.     str_len = len(string)
    
23.     max_len = 13
    
24.     if str_len >= 1 and str_len <= max_len:
    
25.         return string.center(max_len - str_len)
    
26.     else:
    
27.         return string
    
28. 
    
29. def askurl(url,cookkey):
    
30.     head = {
    
31.     'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8',
    
32.     'Accept-Language': 'zh-CN,zh;q=0.9',
    
33.     'Cookie': 'qd_dz_ct=59289; sts=1; theme=-1; wc=59289; lc=59289; lc2=59289; wc_n=%25u4E1C%25u839E; ' + cookkey,
    
34.     'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.25 Safari/537.36 Core/1.70.3641.400 QQBrowser/10.4.3284.400'
    
35.     }
    
36.    
    
37.     req = urllib.request.Request(url=url, headers=head)
    
38.     try:
    
39.         response = urllib.request.urlopen(req)
    
40.         if response.chunked:
    
41.             print('访问：{} 失败！\n请检查网络是否可以正确的访问Internet！'.format(url))
    
42.             exit()
    
43.     except urllib.error.URLError as err:
    
44.         print('\n网络连接失败！\n请检查网络连接！', err)
    
45.         input('按任意键退出！')
    
46.         exit()
    
47.     html = bfs(response,'html.parser') #解析html
    
48.     # print(html)
    
49.    
    
50.     return html
    
51. def getdata(html):
    
52.     datalist = []
    
53.     datalist.append(html.find('em', class_="location-city location-font1").string)
    
54.     date_temp = html.find('p', class_="date").contents
    
55.     datalist.append(date_temp[0].string + ' ' + date_temp[2].string)
    
56.     datalist.append(date_temp[1].string)
    
57.     datalist.append(date_temp[3].string)
    
58.     #天气情况
    
59.     weather = html.find('a', class_="wea-info-index")
    
60.     datalist.append(weather.span.string + weather.b.string)
    
61.     datalist.append(html.find('a', class_="wea-other-a-we").string)
    
62.     datalist.append('空气质量：' + html.find('a', class_="wea-aqi-tip-index").em.string)
    
63.     weather = html.find('ul', class_="wea-info-tip").find_all('li')
    
64.     datalist.append(weather[0].span.string + ' : ' + weather[0].em.string)
    
65.     datalist.append(weather[1].span.string + ' : ' + weather[1].em.string)
    
66.     datalist.append(weather[2].span.string + ' : ' + weather[2].em.string)
    
67.     # print(datalist)
    
68.    
    
69.     #获取未来六天的天气数据
    
70.     tomorrow = [[],[],[],[],[],[]]
    
71.    
    
72.     def get_tomorrw(htmlobj, index): #相应数据
    
73.         temp = htmlobj.contents
    
74.         tomorrow[index].append(format(temp[1].text + ' ' + temp[3].text))
    
75.         tomorrow[index].append(format(temp[7].text))
    
76.         tomorrow[index].append(format(temp[9].text + ' ' + temp[11].text))
    
77.         tomorrow[index].append(format('空气质量：' + temp[13].text))
    
78. 
    
79.     info_tomorrow = html.find('ul', class_="weaday7 wea-white-icon")
    
80.     a_list = info_tomorrow.find_all('a')
    
81.     for day, index in zip(range(2,14,2),range(6)):
    
82.         get_tomorrw(a_list[day], index)
    
83.         
    
84.     #温度
    
85.     script = html.findAll('script')[-1]
    
86.     H = re.compile('var day7DataHight = \[(.*)\]')
    
87.     L = re.compile('var day7DataLow = \[(.*)\]')
    
88.     H_list = re.findall(H,str(script))[0].split(',')
    
89.     L_list = re.findall(L,str(script))[0].split(',')
    
90.     n = 0
    
91.     for i,j in zip(L_list, H_list):
    
92.         if not n:
    
93.             n+=1;continue
    
94.         tomorrow[n-1].insert(3, format(i + ' ~ ' + j))
    
95.         n+=1
    
96.     # print(datalist + tomorrow)   
    
97.     return datalist + tomorrow
    
98. if __name__ == '__main__':
    
99.     print(main(3))

复制代码

myhic

能改变城市吗？没有我所在的城市啊

Cool_Breeze

myhic 发表于 2020-6-9 00:09
能改变城市吗？没有我所在的城市啊

可以自己添加城市的，将Cookie添加到列表就好了！