|
|
|
|
|
from info import username,password,account2,password2
|
|
|
|
|
|
|
|
|
|
|
|
from selenium.webdriver.common.by import By
|
|
|
|
|
|
from selenium.webdriver import Chrome
|
|
|
|
|
|
from selenium.webdriver.chrome.options import Options
|
|
|
|
|
|
from bs4 import BeautifulSoup
|
|
|
|
|
|
import time
|
|
|
|
|
|
|
|
|
|
|
|
import smtplib
|
|
|
|
|
|
from email.mime.text import MIMEText
|
|
|
|
|
|
from email.header import Header
|
|
|
|
|
|
def crawl():
|
|
|
|
|
|
chrome_options = Options()
|
|
|
|
|
|
chrome_options.add_experimental_option("debuggerAddress", "localhost:9222") #此处端口保持和命令行启动的端口一致
|
|
|
|
|
|
#使用ChromeDriver创建一个浏览器实例。
|
|
|
|
|
|
driver = Chrome(options=chrome_options)
|
|
|
|
|
|
driver.get('https://login.taobao.com')
|
|
|
|
|
|
# 等待页面加载完成
|
|
|
|
|
|
time.sleep(2)
|
|
|
|
|
|
# # 初次登录
|
|
|
|
|
|
# # 输入用户名和密码
|
|
|
|
|
|
# driver.find_element(By.ID,'fm-login-id').send_keys(username)
|
|
|
|
|
|
# driver.find_element(By.ID,'fm-login-password').send_keys(password)
|
|
|
|
|
|
#
|
|
|
|
|
|
# time.sleep(4)
|
|
|
|
|
|
# # 点击登录按钮
|
|
|
|
|
|
# driver.find_element(By.CLASS_NAME,'fm-button.fm-submit.password-login').click()
|
|
|
|
|
|
# #多次登陆
|
|
|
|
|
|
# driver.find_element(By.ID,'fm-login-password').send_keys(password)
|
|
|
|
|
|
# time.sleep(3)
|
|
|
|
|
|
# driver.find_element(By.CLASS_NAME,'fm-button.fm-submit.password-login').click()
|
|
|
|
|
|
#快速登陆
|
|
|
|
|
|
driver.find_element(By.CLASS_NAME,'fm-btn').click()
|
|
|
|
|
|
#等待登录完成
|
|
|
|
|
|
time.sleep(10)
|
|
|
|
|
|
# 搜索指定商品
|
|
|
|
|
|
driver.find_element(By.CLASS_NAME,'search-combobox-input').send_keys('西装')
|
|
|
|
|
|
time.sleep(2)
|
|
|
|
|
|
driver.find_element(By.CLASS_NAME,'btn-search').click()
|
|
|
|
|
|
#切换到新窗口
|
|
|
|
|
|
handles=driver.window_handles
|
|
|
|
|
|
driver.switch_to.window(handles[-1])
|
|
|
|
|
|
time.sleep(3)
|
|
|
|
|
|
#数据提取，打印
|
|
|
|
|
|
soup=BeautifulSoup(driver.page_source,'html.parser')
|
|
|
|
|
|
title=[]
|
|
|
|
|
|
price=[]
|
|
|
|
|
|
titles=soup.find_all('div',class_='Title--title--jCOPvpf')
|
|
|
|
|
|
for t in titles:
|
|
|
|
|
|
title.append(t.text)
|
|
|
|
|
|
|
|
|
|
|
|
prices=soup.find_all('div',style="margin-right: 8px;")
|
|
|
|
|
|
for p in prices:
|
|
|
|
|
|
price.append(p.text)
|
|
|
|
|
|
print('{:^40}\t\t\t{:^2}'.format('商品名称','价格',chr(12288)))
|
|
|
|
|
|
for i in range(len(price)):
|
|
|
|
|
|
print('{:<20}\t{:<5}'.format(title[i],price[i]))
|
|
|
|
|
|
return title,price
|
|
|
|
|
|
|
|
|
|
|
|
def send_email(content):
|
|
|
|
|
|
account = account2 #发件人邮箱地址
|
|
|
|
|
|
password = password2 #邮箱授权码
|
|
|
|
|
|
receiver = account2 #收件人地址
|
|
|
|
|
|
|
|
|
|
|
|
message = MIMEText(content,'plain', 'utf-8')
|
|
|
|
|
|
# 实例化一个MIMEText邮件对象，该对象需要写进三个参数，分别是邮件正文，文本格式和编码
|
|
|
|
|
|
subject = ('淘宝数据')
|
|
|
|
|
|
message['Subject'] = Header(subject) #实例化了一个Header邮件头对象
|
|
|
|
|
|
message['From'] = Header(account) # 设置发件人地址
|
|
|
|
|
|
message['To'] =Header(receiver) # 设置收件人地址
|
|
|
|
|
|
|
|
|
|
|
|
try:
|
|
|
|
|
|
server = smtplib.SMTP_SSL('smtp.qq.com', 465)
|
|
|
|
|
|
# 连接服务器，第一个参数是服务器地址，第二个参数是SMTP端口号。
|
|
|
|
|
|
server.login(account, password)
|
|
|
|
|
|
server.sendmail(account, [receiver], message.as_string())
|
|
|
|
|
|
print('邮件发送成功')
|
|
|
|
|
|
server.quit()
|
|
|
|
|
|
|
|
|
|
|
|
except Exception as e:
|
|
|
|
|
|
print ('邮件发送失败')
|
|
|
|
|
|
print(e)
|
|
|
|
|
|
|
|
|
|
|
|
def main():
|
|
|
|
|
|
title,prices= crawl()
|
|
|
|
|
|
temp=''
|
|
|
|
|
|
for i in range(len(prices)):
|
|
|
|
|
|
temp=temp+title[i]+'\t'+str(prices[i])+'\n'
|
|
|
|
|
|
send_email(temp)
|
|
|
|
|
|
|
|
|
|
|
|
main()
|
