阿里云函数计算使用selenium进行网页截图完成代码示例

随便、先森。

1. # -*- coding: utf-8 -*-
   
2. # tools.bugscaner.com
   
3. import logging
   
4. import hashlib
   
5. import json
   
6. import time
   
7. import oss2
   
8. from selenium import webdriver
   
9. from cgi import parse_qs, escape
   
10. 
    
11. 
    
12. def handler(environ, start_response):
    
13.     context = environ['fc.context']
    
14.     request_uri = environ['fc.request_uri']
    
15.     for k, v in environ.items():
    
16.       if k.startswith('HTTP_'):
    
17.         # process custom request headers
    
18.         pass
    
19.     # do something here
    
20.    
    
21.     #获取post参数
    
22.     try:
    
23.         request_body_size = int(environ.get('CONTENT_LENGTH', 0))
    
24.     except:
    
25.         request_body_size = 0
    
26.     request_body = environ['wsgi.input'].read(request_body_size)
    
27.     input_urls = parse_qs(request_body)
    
28.     logging.warning(input_urls)
    
29.     input_url = input_urls.get("url",[''])[0]
    
30.     logging.warning(input_url)
    
31.     #谷歌浏览器设置
    
32.     option = webdriver.ChromeOptions()
    
33.     option.add_argument('user-agent="Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.17 (KHTML, like Gecko) Chrome/24.0.1312.57 Safari/537.17"')
    
34.     option.add_argument("--headless")
    
35.     option.add_argument("--no-sandbox")
    
36.     option.add_argument("--disable-gpu")
    
37.     option.binary_location = "./headless-chromium"
    
38.     driver = webdriver.Chrome(executable_path="./chromedriver",chrome_options=option)
    
39.     #设置页面访问超时时间
    
40.     driver.set_page_load_timeout(5)
    
41.     driver.set_script_timeout(5)#这两种设置都进行才有效
    
42.     #driver = webdriver.PhantomJS(service_log_path="/tmp/logssssss.log")
    
43.     try:
    
44.         driver.get(input_url)
    
45.     except:
    
46.         return [json.dumps({"status":404})]
    
47.     scroll_width = driver.execute_script('return document.body.parentNode.scrollWidth')
    
48.     scroll_height = driver.execute_script('return document.body.parentNode.scrollHeight')
    
49.     driver.set_window_size(scroll_width, scroll_height)
    
50.     #time.sleep(2)
    
51.     images = driver.get_screenshot_as_png()
    
52.     auth = oss2.Auth("LT33HkjkEz", "33Ns4AjdvG1Ow2sxvtzmjcGbvKtECX")
    
53.     bucket = oss2.Bucket(auth, 'oss-cn-hangzhou.aliyuncs.com', 'pic')
    
54.     filenamemd5 = hashlib.md5(input_url)
    
55.     filenamemd5 = str(filenamemd5.hexdigest())
    
56.     ifover = False
    
57.     for x in range(2):
    
58.         try:
    
59.             bucket.put_object(filenamemd5+".png", images)
    
60.             ifover = True
    
61.             break
    
62.         except:
    
63.             pass
    
64.     if ifover:
    
65.         status = '200 OK'
    
66.         response_headers = [('Content-type', 'text/plain')]
    
67.         start_response(status, response_headers)
    
68.         return [json.dumps({"status":200,"picurl":filenamemd5})]   
    
69.     else:
    
70.         status = '200 OK'
    
71.         response_headers = [('Content-type', 'text/plain')]
    
72.         start_response(status, response_headers)
    
73.         return [json.dumps({"status":403})]   

复制代码

来自：http://www.bugscaner.com/jiaobenyuyan/python/227.html