Python编程学习之利用selenium分辨出可访问的网页并获取网页内容

Python版本：3.7.1

selenium模块

pymysql模块

from selenium import webdriver
from selenium.webdriver.chrome.options import Options
import pymysql

# 获取存活的域名
def run(cursor):
    # 获取域名
    domains = get_domains(cursor)

    # Chrome的参数选项
    chrome_options = Options()
    # 无头操作
    chrome_options.add_argument(‘--headless‘)
    # 利用这个路径的Chromium来进行操作
    chrome_options.binary_location = r‘%s‘%"/Applications/Chromium.app/Contents/MacOS/Chromium"

    # 创建Chrome实例
    driver = webdriver.Chrome(executable_path=(r‘/Users/hello/Desktop/chromedriver/chromedriver‘), options=chrome_options)
    # 设置20秒的超时时间
    driver.set_page_load_timeout(20)
    success_list = []
    for i in domains:
        try:
            # 请求网站
            driver.get(‘https://‘+i[0])
            #获取网站的信息
            http_length = len(driver.page_source)
            http_status = ‘响应成功‘
            img_path = "/Users/hello/Desktop/py test/%s.png"%i[0]
            screenshot = driver.get_screenshot_as_file(img_path)
            if driver.title:
                title = driver.title
            else:
                title = ‘‘
            success_list.append([i[0], title, http_length, img_path, http_status])
        except :
            print(‘%s 响应失败‘%i[0])
    return success_list

# 去数据库查询域名
def get_domains(cursor):
    sql = "SELECT hostname FROM 数据库"
    cursor.execute(sql)
    domain_lists = cursor.fetchall()
    return domain_lists

# 把可访问的域名插入数据库
def insert(cursor, list, db):
    for i in list:
        select_sql = "SELECT id FROM 数据库 WHERE hostname = ‘%s‘"%i[0]
        cursor.execute(select_sql)
        result = cursor.fetchone()
        update_sql = "UPDATE 数据库 SET page_title = ‘%s‘, http_length = %d, page_jietu_path = ‘%s‘, http_status= ‘%s‘ WHERE id = %s" %(i[1], i[2], i[3], i[4], result[0])
        cursor.execute(update_sql)
        db.commit()

if __name__ == "__main__":
    db = pymysql.connect(‘localhost‘, ‘账户‘, ‘密码‘, ‘test‘)
    cursor = db.cursor()

    list = run(cursor)
    insert(cursor, list, db)
    db.close()