find job
1.0.0
Recently I discovered a very interesting project on GitHub: GitHub Link.
The author cleverly combined GPT and RPA technology to create an automatic resume submission assistant. This is the effect demonstration video shared by the original author: B station video link.
However, the original project had the following problems:
After struggling to no avail, I decided to use Node.js to re-implement the project. It is completely free and can be run with one click without setting up an agent: GitHub project address.
In this cold recruiting season, this script can provide you with some help and bring you some warmth. If you think this project is valuable, I hope you can help by giving it a star. It will be greatly appreciated.
First, we will use selenium-webdriver to simulate user behavior. This library is a powerful automated testing tool. It enables programmatic control of browser interactions and is commonly used for tasks such as automated testing, web scraping, and simulating user interactions.
DOM node of the login button, simulate the user's click to trigger the login, and wait for the user to scan the QR code.GPT , waiting for GPT 's responseGPT responds, click the "Communicate Now" button to enter the communication chat interfaceGPT into the chat box, and trigger the send eventThose who have done GPT development should know that calling the GPT interface requires payment, and the recharge process is extremely cumbersome and requires the use of overseas bank cards.
In order to simplify this process, I found a project on GitCode that provides free API_KEY, which can be easily obtained by logging in with a GitHub account.
This way you can initialize the OpenAI client with the free API_KEY .
// 初始化OpenAI客户端
const openai = new OpenAI ( {
// 代理地址，这样国内用户就可以访问了
baseURL : "https://api.chatanywhere.com.cn" ,
apiKey : "你的apiKey" ,
} ) ;In this step, what we want to achieve is to open the browser and navigate to the specified URL. The specific operation is to call the API of selenium-webdriver and directly enter the code:
const { Builder , By , until } = require ( "selenium-webdriver" ) ;
const chrome = require ( "selenium-webdriver/chrome" ) ;
// 全局 WebDriver 实例
let driver ;
// 使用指定的选项打开浏览器
async function openBrowserWithOptions ( url , browser ) {
const options = new chrome . Options ( ) ;
options . addArguments ( "--detach" ) ;
if ( browser === "chrome" ) {
// 初始化一个谷歌浏览器客户端
driver = await new Builder ( )
. forBrowser ( "chrome" )
. setChromeOptions ( options )
. build ( ) ;
// 全屏打开浏览器
await driver . manage ( ) . window ( ) . maximize ( ) ;
} else {
throw new Error ( "不支持的浏览器类型" ) ;
}
await driver . get ( url ) ;
// 等待直到页面包含登录按钮dom
const loginDom = By . xpath ( "//*[@id='header']/div[1]/div[3]/div/a" ) ;
await driver . wait ( until . elementLocated ( loginDom ) , 10000 ) ;
}
// 主函数
async function main ( url , browserType ) {
try {
// 打开浏览器
await openBrowserWithOptions ( url , browserType ) ;
} catch ( error ) {
console . error ( `发生错误: ${ error } ` ) ;
}
}
const url =
"https://www.zhipin.com/web/geek/job-recommend?ka=header-job-recommend" ;
const browserType = "chrome" ;
main ( url , browserType ) ;In this step, we need to find the DOM node of the login button , and then simulate clicking to log in.
// 省略上一步的代码
// 点击登录按钮，并等待登录成功
async function logIn ( ) {
// 点击登录
const loginButton = await driver . findElement (
By . xpath ( "//*[@id='header']/div[1]/div[3]/div/a" )
) ;
await loginButton . click ( ) ;
// 等待微信登录按钮出现
const xpathLocatorWechatLogin =
"//*[@id='wrap']/div/div[2]/div[2]/div[2]/div[1]/div[4]/a" ;
await driver . wait (
until . elementLocated ( By . xpath ( xpathLocatorWechatLogin ) ) ,
10000
) ;
const wechatButton = await driver . findElement (
By . xpath ( "//*[@id='wrap']/div/div[2]/div[2]/div[2]/div[1]/div[4]/a" )
) ;
// 选择微信扫码登录
await wechatButton . click ( ) ;
const xpathLocatorWechatLogo =
"//*[@id='wrap']/div/div[2]/div[2]/div[1]/div[2]/div[1]/img" ;
await driver . wait (
until . elementLocated ( By . xpath ( xpathLocatorWechatLogo ) ) ,
10000
) ;
// 等待用户扫码，登录成功
const xpathLocatorLoginSuccess = "//*[@id='header']/div[1]/div[3]/ul/li[2]/a" ;
await driver . wait (
until . elementLocated ( By . xpath ( xpathLocatorLoginSuccess ) ) ,
60000
) ;
}
// 主函数
async function main ( url , browserType ) {
try {
// 打开浏览器
// 点击登录按钮，并等待登录成功
+ await logIn ( ) ;
} catch ( error ) {
console . error ( `发生错误: ${ error } ` ) ;
}
}After successfully logging in, you will enter the recruitment information list page. In this step, we need to traverse the recruitment information and click in order to find the job description information of each recruitment information, as shown in the figure:
// 省略上一步的代码
// 根据索引获取职位描述
async function getJobDescriptionByIndex ( index ) {
try {
const jobSelector = `//*[@id='wrap']/div[2]/div[2]/div/div/div[1]/ul/li[ ${ index } ]` ;
const jobElement = await driver . findElement ( By . xpath ( jobSelector ) ) ;
// 点击招聘信息列表中的项
await jobElement . click ( ) ;
// 找到描述信息节点并获取文字
const descriptionSelector =
"//*[@id='wrap']/div[2]/div[2]/div/div/div[2]/div/div[2]/p" ;
await driver . wait (
until . elementLocated ( By . xpath ( descriptionSelector ) ) ,
10000
) ;
const jobDescriptionElement = await driver . findElement (
By . xpath ( descriptionSelector )
) ;
return jobDescriptionElement . getText ( ) ;
} catch ( error ) {
console . log ( `在索引 ${ index } 处找不到工作。` ) ;
return null ;
}
}
// 主函数
async function main ( url , browserType ) {
try {
// 打开浏览器
// 点击登录按钮，并等待登录成功
// 开始的索引
+ let jobIndex = 1 ;
+ while ( true ) {
+ // 获取对应下标的职位描述
+ const jobDescription = await getJobDescriptionByIndex ( jobIndex ) ;
+ console . log ( `职位描述信息/n： ${ jobDescription } ` ) ;
+ if ( jobDescription ) {
+ //
+ }
+ jobIndex += 1 ;
}
} catch ( error ) {
console . error ( `发生错误: ${ error } ` ) ;
}
} Then combine the uploaded resume information and recruitment information and pass it to GPT , waiting for GPT 's response:
// 省略上一步的代码
// 读取简历信息
const getResumeInfo = ( ) => {
fs . readFile ( "./简历基本信息.txt" , "utf8" , ( err , data ) => {
if ( err ) {
console . error ( "读取文件时出错:" , err ) ;
return ;
}
// 输出文件内容
return data ;
} ) ;
} ;
// 与GPT进行聊天的函数
async function chat ( jobDescription ) {
// 获取简历信息
const resumeInfo = getResumeInfo ( ) ;
const askMessage = `你好，这是我的简历： ${ resumeInfo } ，这是我所应聘公司的要求： ${ jobDescription } 。我希望您能帮我直接给HR写一个礼貌专业的求职新消息，要求能够用专业的语言将简历中的技能结合应聘工作的描述，来阐述自己的优势，尽最大可能打动招聘者。并且请您始终使用中文来进行消息的编写,开头是招聘负责人。这是一封完整的求职信，不要包含求职信内容以外的东西，例如“根据您上传的求职要求和个人简历，我来帮您起草一封求职邮件：”这一类的内容，以便于我直接自动化复制粘贴发送，字数控制在80字左右为宜` ;
try {
const completion = await openai . chat . completions . create ( {
messages : [
{
role : "system" ,
content : askMessage ,
} ,
] ,
model : "gpt-3.5-turbo" ,
} ) ;
// 获取gpt返回的信息
const formattedMessage = completion . choices [ 0 ] . message . content . replace (
/ n / g ,
" "
) ;
return formattedMessage ;
} catch ( error ) {
console . error ( `gpt返回时发生错误: ${ error } ` ) ;
const errorResponse = JSON . stringify ( { error : String ( error ) } ) ;
return errorResponse ;
}
}
// 主函数
async function main ( url , browserType ) {
try {
// 打开浏览器
// 点击登录按钮，并等待登录成功
// 开始的索引
while ( true ) {
// 获取对应下标的职位描述
if ( jobDescription ) {
// 发送描述到聊天并打印响应
+ const response = await chat ( jobDescription ) ;
+ console . log ( "gpt给的回复" , response ) ;
}
jobIndex += 1 ;
}
} catch ( error ) {
console . error ( `发生错误: ${ error } ` ) ;
}
}After the GPT response is completed, find the Communicate Now button and simulate clicking it. At this time, you will enter the communication and chat interface, as shown in the figure:
// 省略上一步的代码
// 主函数
async function main ( url , browserType ) {
try {
// 打开浏览器
// 点击登录按钮，并等待登录成功
// 开始的索引
while ( true ) {
// 获取对应下标的职位描述
if ( jobDescription ) {
// 发送描述到聊天并打印响应
// 点击沟通按钮
+ const contactButton = await driver . findElement (
+ By . xpath (
+ "//*[@id='wrap']/div[2]/div[2]/div/div/div[2]/div/div[1]/div[2]/a[2]"
+ )
+ ) ;
+ await contactButton . click ( ) ;
}
jobIndex += 1 ;
}
} catch ( error ) {
console . error ( `发生错误: ${ error } ` ) ;
}
}At this point, enter the chat interface, fill in the GPT return information into the input box, and trigger the send event.
// 省略上一步的代码
// 发送响应到聊天框
async function sendResponseToChatBox ( driver , response ) {
try {
// 请找到聊天输入框
const chatBox = await driver . findElement ( By . xpath ( "//*[@id='chat-input']" ) ) ;
// 清除输入框中可能存在的任何文本
await chatBox . clear ( ) ;
// 将响应粘贴到输入框
await chatBox . sendKeys ( response ) ;
await sleep ( 1000 ) ;
// 模拟按下回车键来发送消息
await chatBox . sendKeys ( Key . RETURN ) ;
await sleep ( 2000 ) ; // 模拟等待2秒
} catch ( error ) {
console . error ( `发送响应到聊天框时发生错误: ${ error } ` ) ;
}
}
// 主函数
async function main ( url , browserType ) {
try {
// 打开浏览器
// 点击登录按钮，并等待登录成功
// 开始的索引
while ( true ) {
// 获取对应下标的职位描述
if ( jobDescription ) {
// 发送描述到聊天并打印响应
// 点击沟通按钮
// 等待回复框出现
+ const chatBox = await driver . wait (
+ until . elementLocated ( By . xpath ( "//*[@id='chat-input']" ) ) ,
+ 10000
+ ) ;
+ // 调用函数发送响应
+ await sendResponseToChatBox ( driver , response ) ;
+ // 返回到上一个页面
+ await driver . navigate ( ) . back ( ) ;
+ await sleep ( 2000 ) ; // 模拟等待3秒
}
jobIndex += 1 ;
}
} catch ( error ) {
console . error ( `发生错误: ${ error } ` ) ;
}
}After sending, return to the recruitment list page and repeat the process.
This project simply sends the resume information combined with the job information to GPT, and then sends the GPT reply to the recruiter. In fact, it is not difficult and is intended to attract new ideas.
There are actually more elegant ways here, such as passing your resume to GPT and letting GPT extract effective information (this is what the original author did). However, since the GPT-API-free project does not provide assistant services, payment is required to achieve this. Friends who have recharge channels can try it.
In addition, for those who are interested, you can dig deeper, such as:
Finally, I reiterate the original author’s point of view:
I hope no one will use my script to cut leeks. They have already been forced to use this script to submit resumes. There is nothing left to squeeze out. Just be a human being.
