kakao 채용정보를 가져와서 분류하기 구현

@@ -2,7 +2,7 @@ const express = require('express')
 const puppeteer = require('puppeteer')
 const puppeteer = require('puppeteer')
 
 
 const naverRouter = require('./router/naver')
 const naverRouter = require('./router/naver')
-
+const kakaoRouter = require('./router/kakao')
 
 
 
 
 const app = express();
 const app = express();
@@ -13,6 +13,7 @@ const server = app.listen(3000,()=>{
 })
 })
 
 
 app.use('/naver', naverRouter)
 app.use('/naver', naverRouter)
+app.use('/kakao', kakaoRouter)
 
 
 app.get('/', (req, res)=>{
 app.get('/', (req, res)=>{
     res.send("this is home!")
     res.send("this is home!")

 const express = require('express')
 const express = require('express')
 const router = express.Router()
 const router = express.Router()
 
 
-const puppeteer = require('puppeteer')
-const cheerio = require('cheerio')
-const sanitizeHtml =require('sanitize-html')
+const kakaoFunction = require('./kakaoFunction')
 
 
 router.get('/', (req, res, error)=>{
 router.get('/', (req, res, error)=>{
     res.send('hello this is kakao')
     res.send('hello this is kakao')
 })
 })
 
 
 router.get('/recruitment', async (req, res, error)=>{
 router.get('/recruitment', async (req, res, error)=>{
-    res.send('kakao recruitment')
-})
\ No newline at end of file
+    res.send(await kakaoFunction.getData())
+})
+
+module.exports = router
\ No newline at end of file

+const puppeteer = require('puppeteer')
+const cheerio = require('cheerio')
+const sanitizeHtml =require('sanitize-html')
+
+const splitData = (string)=>{
+    const temp = /(<a([^>]+)>)/g.exec(string)
+    const temp2 = temp[0].split('"');
+    url = "https://careers.kakao.com" + temp2[1];
+    result = string.replace(/(<([^>]+)>)*(\\t)?/gi, "").replace(/ /g, "").split("\n").filter((ele)=> ele != '')
+    result.pop()
+    result.push(url)
+    return result;
+}
+
+const makeObject = (array)=>{
+    const result = []
+    let tempData = null
+    for(let i of array){
+        tempData = splitData(i)
+        result.push({
+            title : tempData[0],
+            term : tempData[2],
+            tags : [],
+            url : tempData[tempData.length - 1]
+        })
+        for(let j of tempData){
+            if(j[0] == '#'){
+                result[result.length - 1].tags.push(j)
+            }
+        }
+    }
+    return result
+}
+
+const moveNextPage = async (page)=>{
+
+    await page.click('#mArticle > div > div.paging_list > span > a:nth-child(11) > span > span').catch((error)=>{
+    })
+    await page.waitForTimeout(300)
+    return await page.content()
+}
+
+const getData = async ()=>{
+    const browser = await puppeteer.launch();
+    const page = await browser.newPage();
+
+    let result = []
+    let temp = ""
+
+    await page.goto('https://careers.kakao.com/jobs')
+    let content = await page.content()
+    while(true){
+        if(temp == content){
+            console.log("finish", result.length)
+            break;
+        }
+        let $ = cheerio.load(content, {decodeEntities: true})
+    
+        const selector = '#mArticle > div > ul.list_jobs'
+    
+        let resultString = sanitizeHtml($(selector), {
+            parser : {
+                decodeEntities: true
+            }
+        })
+        resArr = resultString.split('</li>')
+        resArr.pop()
+        result = result.concat(await makeObject(resArr))
+        temp = content
+        content = await moveNextPage(page)
+    }
+    console.log(result)
+    return result
+}
+
+module.exports = {
+    getData : getData
+}
+

-const splitData = (string)=>{
-    const temp = /(<a([^>]+)>)/g.exec(string)
-    console.log(temp)
-    const temp2 = temp[0].split('"');
-    console.log(temp2)
-    url = "https://recruit.navercorp.com" + temp2[1];
-    console.log(url)
-    result = string.replace(/(<([^>]+)>)*(\\t)?/gi, "").split("\t").filter((ele)=> ele != '')
-    result.pop()
-    result.push(url)
-    return result;
-}
-
-const string = `\t<a href="/naver/job/detail/developer?annoId=20002849&amp;classId=&amp;jobId=&amp;entTypeCd=&amp;searchTxt=&amp;searchSysComCd=">\t\t<span>\t\t\t<strong>[Platform Engineering] Server Engineer</strong>\t\t\t<span>NEW</span>\t\t\t<em>2021.04.26~2021.05.10</em>\t\t\t<span></span>\t\t</span>\t</a>\t<span><a href="/naver/job/list/developer?entTypeCd=&amp;searchSysComCd=&amp;searchTxt=Platform%20Engineering">#Platform Engineering</a><a href="/naver/job/list/developer?entTypeCd=&amp;searchSysComCd=&amp;searchTxt=Software%20Development">#Software Development</a><a href="/naver/job/list/developer?entTypeCd=&amp;searchSysComCd=&amp;searchTxt=Server">#Server</a><a href="/naver/job/list/developer?entTypeCd=&amp;searchSysComCd=&amp;searchTxt=%EA%B2%BD%EB%A0%A5">#경력</a><a href="/naver/job/list/developer?entTypeCd=&amp;searchSysComCd=&amp;searchTxt=Backend">#Backend</a>\t </span>`
-
-console.log(splitData(string))
\ No newline at end of file