基于requests库制作的爬虫。
使用方法:打开终端输入 “python3 weather.py 北京(或你所在的城市)"
程序正常运行需要在同文件夹下加入一个“data.csv”文件,内容请参考链接: 城市代码
运行效果:
源码:
'''
想要学习Python?Python学习交流群:973783996满足你的需求,资料都已经上传群文件,可以自行下载!
'''
import sys
import re
import requests
import webbrowser
from PIL import Image
from requests.exceptions import RequestException
import csv
data={}
with open("data.csv",'r') as f:
rawinfos=list(csv.reader(f))
for i in rawinfos:
data[i[0]]=i[1]
def get_one_page(url,headers):
try:
response=requests.get(url,headers=headers)
if response.status_code==200:
response.encoding='utf-8'
return response.text
return None
except RequestException:
return None
headers={'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_2) AppleWebKit/604.4.7 (KHTML, like Gecko) Version/11.0.2 Safari/604.4.7'}
try:
address=data[sys.argv[1]]
except:
sys.exit("\033[31m无该城市!\033[0m")
html=get_one_page('http://www.weather.com.cn/weather1d/'+address+'.shtml',headers)
if not html:
print("城市代码有误!")
exit(1)
ADDRESS=re.findall('<title>(.*?)</title>',html)
aim=re.findall('<input type="hidden" id="hidden_title" value="(.*?)月(.*?)日(.*?)时(.*?) (.*?) (.*?) (.*?)"',html,re.S)
airdata=re.findall('<li class="li6 hot">\n<i></i>\n<span>(.*?)</span>\n<em>(.*?)</em>\n<p>(.*?)</p>\n</li>',html,re.S)
print(ADDRESS[0][1:5])
print("当前日期:%s月%s日,%s"%(aim[0][0],aim[0][1],aim[0][4]))
print("更新时间:%s:00"%aim[0][2])
print("当前天气:%s"%aim[0][5])
print("今日温度:%s"%aim[0][6])
print("空气质量:"+airdata[0][0]+","+airdata[0][2])
ask_ok=input("是否深入查看(Y/N):")
if ask_ok=='Y' or ask_ok=='y':
lightdata=re.findall('<li class="li1 hot">\n<i></i>\n<span>(.*?)</span>\n<em>(.*?)</em>\n<p>(.*?)</p>\n</li>',html,re.S)
colddata=re.findall('<li class="li2 hot">\n(.*?)</span>\n<em>(.*?)</em>\n<p>(.*?)</p>',html,re.S)
weardata=re.findall('<li class="li3 hot" id="chuanyi">\n(.*?)<span>(.*?)</span>\n<em>(.*?)</em>\n<p>(.*?)</p>',html,re.S)
washdata=re.findall('<li class="li4 hot">\n<i></i>\n<span>(.*?)</span>\n<em>(.*?)</em>\n<p>(.*?)</p>\n</li>',html,re.S)
bloodata=re.findall('<li class="li5 hot">\n<i></i>\n<span>(.*?)</span>\n<em>(.*?)</em>\n<p>(.*?)</p>\n</li>',html,re.S)
detail = re.findall('hour3data={"1d":(.*?),"23d"', html, re.S)
detail = re.findall('"(.*?)"', detail[0], re.S)
print("--"*40)
print('详细数据:')
print("%-10s\t%-10s\t%-10s\t%-10s\t%-10s"%("时间","状态","温度","风向","风力"))
for each in detail:
each=each.split(',')
print("%-10s\t%-10s\t%-10s\t%-10s\t%-10s"%(each[0],each[2],each[3],each[4],each[5]))
print("--"*40)
print("%s:\t%s\t%s"%(lightdata[0][1],lightdata[0][0],lightdata[0][2]))
print("%s:\t%s"%(colddata[0][1],colddata[0][2]))
print("%s:\t%s\t%s"%(washdata[0][1],washdata[0][0],washdata[0][2]))
print("血糖指数:\t%s,%s"%(bloodata[0][0],bloodata[0][2]))
print("%s:\t%s\t%s"%(weardata[0][2],weardata[0][1],weardata[0][3]))
print("--"*40)
flag=input("是否查看详细穿衣建议(Y/N):")
if flag=='Y' or flag=='y':
webbrowser.open("http://www.weather.com.cn/forecast/ct.shtml?areaid="+address)
print("数据来源:中央气象台")