如何拯救疫情下的中小企业——临时用RPA替代人工实践篇（二）

公司背景

深圳某人力资源公司，通过RPA自动实现刷简历，根据设定条件筛选候选人，打招呼、简单沟通、索要简历等

screencap.gif

开发技术

Python+Selenium

开发步骤

1. 录入职位

age=27
salary=20
company=平安,腾讯,华为,阿里巴巴...
work_time=3
degree=本科,硕士,博士
keyword=JavaScript,HTML5,CSS,Vue,Angular,React,Hybrid,Web,Android,iOS,App
post=qd,资深前端工程师
day_greet_limit=50,100
day_scan_limit=100,200

将这份职位及筛选条件保存为txt文件，供程序读取，包含内容有年龄、薪资、公司、工作年限、学历、关键词等

2. 创建状态机

class StateType(Enum):
    null = 0
    idle = 1
    start = 2
    login = 3
    waiting = 4
    init = 5
    select_account = 6
    check_chat = 7
    collect_chat = 8
    select_post = 9
    collect_info = 10
    parse_online_resume = 11
    parse_chat = 12
    parse_attach_resume = 13
    save_data = 14
    traverse_chat = 15
    monitor_message = 16
    finish = 99

用状态机的好处是既能保持各个状态独立运行，又能数据共享，如果某个状态出错，可以重新尝试，如果修改逻辑还能热加载无须已执行过程重复执行，状态之间可以任意切换

3. 创建浏览器控制驱动

class Chrome(Browser):
    """Chrome Browser"""
    def __init__(self, name=None, home_page=None, wap=True):
        super(Chrome, self).__init__(name, home_page, wap)

    def init_play(self):
        """init virtual display"""
        super(Chrome, self).init_play()

    def init_profile(self):
        """init the Chrome profile object"""
        super(Chrome, self).init_profile()
        self.profile = webdriver.ChromeOptions()
        self.profile.add_argument('lang=zh_CN.UTF-8')
        self.profile.add_argument('disable-infobars')
        # prefs = {"profile.managed_default_content_settings.images":2}
        # self.profile.add_experimental_option("prefs", prefs)

    def init_agent(self):
        super(Chrome, self).init_agent()
        self.profile.add_argument('--ignore-certificate-errors')
        self.profile.add_argument('--ignore-ssl-errors')
        user_agent = self.get_user_agent()
        self.profile.add_argument('user-agent="%s"'%user_agent)
        try:
            ip_proxy = self.get_proxy_ip()
            if ip_proxy != None:
                # self.profile.add_extension(self.get_chrome_proxy_extension(proxy=ip_proxy))
                self.profile.add_argument("--proxy-server=%s"%ip_proxy)
        except Exception as e:
            logger.error(e)
        
    def init_driver(self):
        """init web driver"""
        super(Chrome, self).init_driver()
        self.driver = webdriver.Chrome(chrome_options = self.profile)
        self.driver.set_page_load_timeout(300)
        self.driver.set_script_timeout(60)

做一些浏览器的初始化工作，设置选项参数，比如是否加载图片等

4. 注册心跳

    def tick(self, dt):
        self.last_state_time += dt

        if StateType.null.value == self.curr_state.value:
            self.change_state(StateType.idle)
        elif StateType.idle.value == self.curr_state.value:
            self.change_state(StateType.start)
        elif StateType.start.value == self.curr_state.value:
            self.on_start()
        elif StateType.login.value == self.curr_state.value:
            self.on_login()
        elif StateType.waiting.value == self.curr_state.value:
            self.on_waiting()
        elif StateType.init.value == self.curr_state.value:
            self.on_init()
        elif StateType.select_account.value == self.curr_state.value:
            self.on_select_account()
        elif StateType.check_chat.value == self.curr_state.value:
            self.on_check_chat()
        elif StateType.collect_chat.value == self.curr_state.value:
            self.on_collect_chat()
        elif StateType.parse_online_resume.value == self.curr_state.value:
            self.on_parse_online_resume()
        elif StateType.parse_chat.value == self.curr_state.value:
            self.on_parse_chat()
        elif StateType.parse_attach_resume.value == self.curr_state.value:
            self.on_parse_attach_resume()
        elif StateType.save_data.value == self.curr_state.value:
            self.on_save_data()
        elif StateType.select_post.value == self.curr_state.value:
            self.on_select_post()
        elif StateType.collect_info.value == self.curr_state.value:
            self.on_collect_info()
        elif StateType.traverse_chat.value == self.curr_state.value:
            self.on_traverse_chat()
        elif StateType.monitor_message.value == self.curr_state.value:
            self.on_monitor_message()
        elif StateType.finish.value == self.curr_state.value:
            self.on_finish()

心跳就如同发动机的传输带，让不同的齿轮带动不同的轮子转动

5. 实现具体逻辑
   以某直聘网站为例，创建逻辑类继承状态机基类，实现每个状态的具体逻辑即可，最终效果如开头图片所示，不过只是其中一个逻辑类。

最后，面对疫情，从容应对，做好自己该做的事情，就是对社会的最大支持！