From 511b2900961a470c80ba22d64bbab3ec57193ee9 Mon Sep 17 00:00:00 2001
From: Naida Vatric <naida.dozic@gmail.com>
Date: Mon, 13 Jan 2020 12:05:33 +0100
Subject: [PATCH] Login to prostor.ba befoure crawl.

---
 app/crawler/specificCrawlers/prostor.js | 44 ++++++++++++-------------
 1 file changed, 22 insertions(+), 22 deletions(-)

diff --git a/app/crawler/specificCrawlers/prostor.js b/app/crawler/specificCrawlers/prostor.js
index 01e2402..04be5f3 100644
--- a/app/crawler/specificCrawlers/prostor.js
+++ b/app/crawler/specificCrawlers/prostor.js
@@ -62,15 +62,16 @@ class ProstorCrawler {
 
   async crawl() {
     const crawlAdCategories = this.crawlerAdCategories;
+    //We need session cookie to use login privileges
+    const prostorCookie = await this.getCookies();
     //New tag to check if crawler loged in
-    const login = await this.loginForScraping(PROSTOR_LOGIN);
+    const login = await this.loginForScraping(PROSTOR_LOGIN, prostorCookie);
     const newRealEstates = [];
-    //
-    console.log("login before crawl:", login);
+    //Crawl only if login was successful
     if (crawlAdCategories && login) {
       const indexGenerators = [];
       for (const adCategory of crawlAdCategories) {
-        indexGenerators.push(this.categoryIndexer(adCategory));
+        indexGenerators.push(this.categoryIndexer(adCategory, prostorCookie));
       }
 
       let done = false;
@@ -123,13 +124,14 @@ class ProstorCrawler {
     return newRealEstates;
   }
 
-  async *categoryIndexer(adCategory) {
+  async *categoryIndexer(adCategory, prostorCookie) {
     const urlAdTypePart = PROSTOR_ENUMS.PROSTOR_AD_TYPE[this.crawlerAdTypes];
     const urlCategoryPart = PROSTOR_ENUMS.PROSTOR_AD_CATEGORY[adCategory];
     if (urlAdTypePart !== undefined && urlCategoryPart !== undefined) {
       const urlPageToCrawl = `${this.baseUrl}?remove_sold=0${urlAdTypePart}${urlCategoryPart}`;
       const listOfAllRealEstates = await this.extractRealEstates(
-        urlPageToCrawl
+        urlPageToCrawl,
+        prostorCookie
       );
 
       let elementToStartIndexFrom = 0;
@@ -143,7 +145,8 @@ class ProstorCrawler {
           elementToStartIndexFrom += realEstatesForSinglePage.length;
 
           const singlePageResults = await this.indexSinglePage(
-            realEstatesForSinglePage
+            realEstatesForSinglePage,
+            prostorCookie
           );
 
           const filteredSinglePageResults = singlePageResults.filter(
@@ -167,10 +170,10 @@ class ProstorCrawler {
     }
   }
 
-  async indexSinglePage(realEstatesList) {
+  async indexSinglePage(realEstatesList, prostorCookie) {
     const asyncActions = [];
     for (const realEstate of realEstatesList) {
-      asyncActions.push(this.scrapeAd(realEstate));
+      asyncActions.push(this.scrapeAd(realEstate, prostorCookie));
     }
 
     try {
@@ -184,7 +187,7 @@ class ProstorCrawler {
     }
   }
 
-  async scrapeAd(realEstate) {
+  async scrapeAd(realEstate, prostorCookie) {
     const { lat, lng, property_name, price, size, link, status } = realEstate;
 
     //Status information is given already in realestate list
@@ -200,7 +203,9 @@ class ProstorCrawler {
 
     // console.log("[PROSTOR] Scraping : ", url);
     try {
-      const adPageSource = await fetch(url);
+      const adPageSource = await fetch(url, {
+        headers: { Cookie: prostorCookie }
+      });
       const body = await adPageSource.text();
       const $ = cheerio.load(body);
 
@@ -422,13 +427,15 @@ class ProstorCrawler {
     }
   }
 
-  async extractRealEstates(url) {
+  async extractRealEstates(url, prostorCookie) {
     if (PRINT_CRAWLER_DEBUG) {
       console.log("[PROSTOR] Index page : ", url);
     }
 
     try {
-      const res = await fetch(url);
+      const res = await fetch(url, {
+        headers: { Cookie: prostorCookie }
+      });
       const body = await res.text();
       const $ = cheerio.load(body);
 
@@ -585,28 +592,21 @@ class ProstorCrawler {
     return savers[0].save(results);
     //so that we can use some sequelize options and information when data is inserted
   }
-  async loginForScraping(PROSTOR_LOGIN) {
-    console.log("PROSTOR_LOGIN", PROSTOR_LOGIN);
-    const prostorCookie = await this.getCookies();
-    console.log("prostor cookie", prostorCookie);
+  async loginForScraping(PROSTOR_LOGIN, prostorCookie) {
     let formData = new FormData();
     formData.append("email", PROSTOR_LOGIN.EMAIL);
     formData.append("password", PROSTOR_LOGIN.PASSWORD);
-    //When once loged in it stays loged in with same credentials.
-    //Do we need to log out ??
+
     return fetch("https://prostor.ba/moj-prostor/prijava", {
       method: "POST",
       body: formData,
       headers: { Cookie: prostorCookie }
     })
       .then(page => {
-        //
-        console.log("headers: ", page.headers.raw()["set-cookie"]);
         return page.text();
       })
       .then(resp => {
         const $ = cheerio.load(resp);
-        console.log("$ ", $("h1").text());
         if (
           $("h1")
             .text()