爬取城市历史天气数据并保存在Excel中

2962615177

输入城市名称和年份，获取2011~2019年中任意一年的历史天气并保持在excel中
感觉用正则表达式有点繁杂了
城市和年份太多，只测试了几个城市，不保证每个城市每个年份都能查到

1. import requests
   
2. from bs4 import BeautifulSoup
   
3. import openpyxl
   
4. import re
   
5. 
   
6. #查找日期正则表达式
   
7. find_date = re.compile(r'<div><a href="//lishi.tianqi.com/.*?/(.*?).html"',re.S)
   
8. #查找最高温度正则表达式
   
9. find_htemp = re.compile(r'<div style="width: 100px">(\d{1,2}|-\d{1,2})</div>',re.S)
   
10. #查找最低温度正则表达式
    
11. find_dtemp = re.compile(r'<div>(\d{1,2}|-\d{1,2})</div>',re.S)
    
12. #查找天气情况正则表达式
    
13. find_tianqi = re.compile(r'<div>([^0-9]*?)</div>',re.S)
    
14. #查找风力正则表达式
    
15. find_fengli = re.compile(r'<div style="width:200px;">(.*?)</div>',re.S)
    
16. 
    
17. #生成查询网址的函数
    
18. def genHTML(cityname,year):
    
19.     months = ["%d%02d" % (year, month + 1) for month in range(12)]
    
20.     todo_urls = [f"http://lishi.tianqi.com/{cityname}/{month}.html" for month in months]
    
21.     return todo_urls
    
22. 
    
23. 
    
24. #抓取网页信息
    
25. def askURL(url):
    
26.     headers = {
    
27.         "User-Agent": "Mozilla / 5.0(Windows NT 10.0; WOW64) AppleWebKit / 537.36(KHTML, like Gecko) Chrome/78.0.3904.108 Safari / 537.36"
    
28.     }
    
29.     res = requests.get(url, headers=headers)
    
30.     res = res.content.decode("utf-8")
    
31.     return res
    
32. 
    
33. #解析网页信息
    
34. def getdata(res):
    
35.     datalist = [ ]
    
36.     soup = BeautifulSoup(res, "html.parser")
    
37.     data = soup.find_all('ul',class_="lishitable_content clearfix")
    
38.     data = str(data)
    
39. 
    
40.     dates = re.findall(find_date,data)
    
41.     dtemp = re.findall(find_dtemp,data)
    
42.     htemp = re.findall(find_htemp,data)
    
43.     tianqi = re.findall(find_tianqi,data)
    
44.     fengli = re.findall(find_fengli,data)
    
45. 
    
46.     for i in range(len(dates)):
    
47.         datalist.append([dates[i],dtemp[i],htemp[i],tianqi[i],fengli[i]])
    
48. 
    
49.     return datalist
    
50. 
    
51. #保存信息至文件
    
52. def saveData(datalist,cityname,year):
    
53.     wb = openpyxl.Workbook()
    
54.     ws = wb.active
    
55. 
    
56.     ws['A1'] = "日期"
    
57.     ws['B1'] = "最低温度"
    
58.     ws['C1'] = "最高温度"
    
59.     ws['D1'] = "天气"
    
60.     ws['E1'] = "风力"
    
61. 
    
62.     #datalist为双层列表
    
63.     for i in range(12):
    
64.         for each in datalist[i]:
    
65.             ws.append(each)
    
66. 
    
67.     wb.save("%d年%s历史天气.xlsx"%(year,cityname))
    
68. 
    
69. def main():
    
70.     print("-------历史天气查询系统-------")
    
71.     cityname = input("请输入城市名称（小写全拼）：")
    
72.     year = int(input("请输入查询年份："))
    
73.     urls = genHTML(cityname,year)
    
74.     result = []
    
75.     for i in range(12):
    
76.         res = askURL(urls[i])
    
77.         result.append(getdata(res))
    
78. 
    
79.     saveData(result,cityname,year)
    
80. 
    
81. 
    
82. if __name__ == '__main__':
    
83.     main()

复制代码

Mike_python小

沙发