|
|
|
|
|
import csv
|
|
|
|
|
|
|
|
|
|
|
|
from selenium import webdriver
|
|
|
|
|
|
from selenium.webdriver.common.by import By
|
|
|
|
|
|
from time import sleep
|
|
|
|
|
|
|
|
|
|
|
|
driver = webdriver.Chrome()
|
|
|
|
|
|
driver.get('http://www.jd.com/')
|
|
|
|
|
|
|
|
|
|
|
|
#通过id找到搜索框，输入内容
|
|
|
|
|
|
driver.find_element(By.ID,'key').send_keys("口红")
|
|
|
|
|
|
#通过类名找到button提交，并点击
|
|
|
|
|
|
driver.find_element(By.CLASS_NAME, 'button').click()
|
|
|
|
|
|
# 扫码完之后随便输入一个内容
|
|
|
|
|
|
input()
|
|
|
|
|
|
print("登录成功")
|
|
|
|
|
|
# 登录完成后，跳转到指定链接
|
|
|
|
|
|
driver.get('http://www.jd.com/')
|
|
|
|
|
|
|
|
|
|
|
|
#通过id找到搜索框，输入内容
|
|
|
|
|
|
driver.find_element(By.ID,'key').send_keys("口红")
|
|
|
|
|
|
#通过类名找到button提交，并点击
|
|
|
|
|
|
driver.find_element(By.CLASS_NAME, 'button').click()
|
|
|
|
|
|
driver.implicitly_wait(10)
|
|
|
|
|
|
|
|
|
|
|
|
with open('JD.csv',mode='w',encoding='UTF-8',newline='') as file:
|
|
|
|
|
|
csv.writer(file).writerow(['商品', '价格','店铺','评论数'])
|
|
|
|
|
|
for page in range(0,50):
|
|
|
|
|
|
sleep(5)
|
|
|
|
|
|
# 下滑页面
|
|
|
|
|
|
js_down = "window.scrollTo(0,8000)"
|
|
|
|
|
|
driver.execute_script(js_down)
|
|
|
|
|
|
sleep(5)
|
|
|
|
|
|
goods = driver.find_elements(By.CLASS_NAME,'gl-i-wrap')
|
|
|
|
|
|
for good in goods:
|
|
|
|
|
|
title = good.find_element(By.CSS_SELECTOR,'.p-name em').text.strip()
|
|
|
|
|
|
price = good.find_element(By.CSS_SELECTOR,'.p-price strong').text.strip()
|
|
|
|
|
|
shop = good.find_element(By.CSS_SELECTOR,'.p-shop span a').text.strip()
|
|
|
|
|
|
comment = good.find_element(By.CSS_SELECTOR,'.p-commit strong a').text.strip()
|
|
|
|
|
|
print('title: ' + title)
|
|
|
|
|
|
print('price: ' + price)
|
|
|
|
|
|
print('shop: ' + shop)
|
|
|
|
|
|
print('comment: ' + comment)
|
|
|
|
|
|
# 用a+模式创建csv文件并写入
|
|
|
|
|
|
f = open('JD.csv', 'a+', encoding='utf-8')
|
|
|
|
|
|
# 基于文件对象构建csv写入
|
|
|
|
|
|
csv_a = csv.writer(f)
|
|
|
|
|
|
# 将数据写入
|
|
|
|
|
|
csv_a.writerow([title, price, shop,comment])
|
|
|
|
|
|
# 关闭文件
|
|
|
|
|
|
f.close()
|
|
|
|
|
|
driver.find_element(By.CLASS_NAME,'pn-next').click()
|
|
|
|
|
|
print('下一页')
|
|
|
|
|
|
|
|
|
|
|
|
input()
|
