python爬虫实践——爬取京东商品信息

 1 ‘‘‘
 2 爬取京东商品信息:
 3     请求url:
 4         https://www.jd.com/
 5     提取商品信息:
 6         1.商品详情页
 7         2.商品名称
 8         3.商品价格
 9         4.评价人数
10         5.商品商家
11 ‘‘‘
12 from selenium import webdriver
13 from selenium.webdriver.common.keys import Keys
14 import time
15 
16 
17 def get_good(driver):
18     try:
19 
20         # 通过JS控制滚轮滑动获取所有商品信息
21         js_code = ‘‘‘
22             window.scrollTo(0,5000);
23         ‘‘‘
24         driver.execute_script(js_code)  # 执行js代码
25 
26         # 等待数据加载
27         time.sleep(2)
28 
29         # 3、查找所有商品div
30         # good_div = driver.find_element_by_id(‘J_goodsList‘)
31         good_list = driver.find_elements_by_class_name(‘gl-item‘)
32         n = 1
33         for good in good_list:
34             # 根据属性选择器查找
35             # 商品链接
36             good_url = good.find_element_by_css_selector(
37                 ‘.p-img a‘).get_attribute(‘href‘)
38 
39             # 商品名称
40             good_name = good.find_element_by_css_selector(
41                 ‘.p-name em‘).text.replace("\n", "--")
42 
43             # 商品价格
44             good_price = good.find_element_by_class_name(
45                 ‘p-price‘).text.replace("\n", ":")
46 
47             # 评价人数
48             good_commit = good.find_element_by_class_name(
49                 ‘p-commit‘).text.replace("\n", " ")
50 
51             good_content = f‘‘‘
52                         商品链接: {good_url}
53                         商品名称: {good_name}
54                         商品价格: {good_price}
55                         评价人数: {good_commit}
56                         \n
57                         ‘‘‘
58             print(good_content)
59             with open(‘jd.txt‘, ‘a‘, encoding=‘utf-8‘) as f:
60                 f.write(good_content)
61 
62         next_tag = driver.find_element_by_class_name(‘pn-next‘)
63         next_tag.click()
64 
65         time.sleep(2)
66 
67         # 递归调用函数
68         get_good(driver)
69 
70         time.sleep(10)
71 
72     finally:
73         driver.close()
74 
75 
76 if __name__ == ‘__main__‘:
77 
78     good_name = input(‘请输入爬取商品信息:‘).strip()
79 
80     driver = webdriver.Chrome()
81     driver.implicitly_wait(10)
82     # 1、往京东主页发送请求
83     driver.get(‘https://www.jd.com/‘)
84 
85     # 2、输入商品名称，并回车搜索
86     input_tag = driver.find_element_by_id(‘key‘)
87     input_tag.send_keys(good_name)
88     input_tag.send_keys(Keys.ENTER)
89     time.sleep(2)
90 
91     get_good(driver)