데이터 추출 및 분류 완료

@@ -6,15 +6,36 @@ const cheerio = require('cheerio')
 const sanitizeHtml =require('sanitize-html')
 const sanitizeHtml =require('sanitize-html')
 
 
 const splitData = (string)=>{
 const splitData = (string)=>{
-    return string.replace(/(<([^>]+)>)*(\\t)?/gi, "").split("\t").filter((ele)=> ele != '')
+    const temp = /(<a([^>]+)>)/g.exec(string)
+    const temp2 = temp[0].split('"');
+    url = "https://recruit.navercorp.com" + temp2[1];
+    result = string.replace(/(<([^>]+)>)*(\\t)?/gi, "").split("\t").filter((ele)=> ele != '')
+    result.pop()
+    result.push(url)
+    return result;
 }
 }
 
 
 const makeObject = (array)=>{
 const makeObject = (array)=>{
-    const result = [];
+    const result = []
+    let tempData = null
     for(let i of array){
     for(let i of array){
-        result.push(splitData(i))
+        tempData = splitData(i)
+        if(tempData.length == 5){
+            result.push({
+                title : tempData[0],
+                term : tempData[2],
+                tags : tempData[3].split('#').filter((ele)=>ele != ''),
+                url : tempData[4]
+            })
+        }
+        else if (tempData.length == 4){
+            result.push({
+                title : tempData[0],
+                term : tempData[2],
+                url : tempData[3]
+            })
+        }
     }
     }
-
     return result
     return result
 }
 }
 
 

+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