More elegant scrape of lat and long.

app/config/appConfig.js

@@ -41,13 +41,6 @@ const PROSTOR_LOGIN = {
   PASSWORD: process.env.PROSTOR_LOGIN_PASS
 };
 
-const USER_AGENT =
-  process.env.USER_AGENT ||
-  "Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/65.0.3325.181 Safari/537.36";
-
-const USE_SCRAPER_API = process.env.USE_SCRAPER_API || 1; //Default to use
-const SCRAPER_API_KEY = process.env.SCRAPER_API_KEY || "";
-
 module.exports = {
   APP_PORT,
   APP_URL,
@@ -61,8 +54,5 @@ module.exports = {
   API_MAP_KEY,
   STAGING,
   CHECK_UP_DAYS,
-  PROSTOR_LOGIN,
+  PROSTOR_LOGIN
-  USER_AGENT,
-  USE_SCRAPER_API,
-  SCRAPER_API_KEY
 };

app/crawler/specificCrawlers/aktido.js

@@ -1,6 +1,6 @@
 "use strict";
 
-const fetch = require("../../helpers/fetchWrapper");
+const fetch = require("node-fetch");
 const cheerio = require("cheerio");
 const Promise = require("bluebird");
 const moment = require("moment-timezone");

app/crawler/specificCrawlers/olx.js

@@ -1,6 +1,6 @@
 "use strict";
 
-const fetch = require("../../helpers/fetchWrapper");
+const fetch = require("node-fetch");
 const cheerio = require("cheerio");
 const Promise = require("bluebird");
 const moment = require("moment-timezone");

app/crawler/specificCrawlers/prostor.js

@@ -1,6 +1,6 @@
 "use strict";
 
-const fetch = require("../../helpers/fetchWrapper");
+const fetch = require("node-fetch");
 const cheerio = require("cheerio");
 const moment = require("moment-timezone");
 const FormData = require("form-data");
@@ -191,7 +191,13 @@ class ProstorCrawler {
     const { lat, lng, property_name, price, size, link, status } = realEstate;
 
     //Status information is given already in realestate list
-    const adStatus = ProstorCrawler.getStatusId(status);
+    //For VIP Ads status ='' canot be used, but no VIP ads are crawled
+    //We will make "fake" vip ad for RE that have size=55
+    //It is weird because yesterday it said 'VIP ponuda' ???
+    const adStatus =
+      size === "55"
+        ? ProstorCrawler.getStatusId("VIP ponuda")
+        : ProstorCrawler.getStatusId(status);
 
     const url = `https://prostor.ba${link}`;
 

app/crawler/specificCrawlers/rental.js

@@ -1,6 +1,6 @@
 "use strict";
 
-const fetch = require("../../helpers/fetchWrapper");
+const fetch = require("node-fetch");
 const cheerio = require("cheerio");
 const Promise = require("bluebird");
 const moment = require("moment-timezone");
@@ -399,9 +399,7 @@ class RentalCrawler {
         );
         if (!publishedDateMoment.isValid()) {
           throw {
-            message: `Invalid published date : ${
+            message: `Invalid published date : ${extractedData["re_realEstates_inserted"]}`
-              extractedData["re_realEstates_inserted"]
-            }`
           };
         }
 
@@ -412,9 +410,7 @@ class RentalCrawler {
         );
         if (!renewedDateMoment.isValid()) {
           throw {
-            message: `Invalid renewed date : ${
+            message: `Invalid renewed date : ${extractedData["re_realEstates_edited"]}`
-              extractedData["re_realEstates_edited"]
-            }`
           };
         }
 

app/crawler/specificCrawlers/saljic.js

@@ -1,6 +1,6 @@
 "use strict";
 
-const fetch = require("../../helpers/fetchWrapper");
+const fetch = require("node-fetch");
 const cheerio = require("cheerio");
 const moment = require("moment-timezone");
 
@@ -218,7 +218,7 @@ class SaljicCrawler {
   }
 
   async scrapeAd(url, adType) {
-    // console.log("[SALJIC] Scraping : ", url);
+    console.log("[SALJIC] Scraping : ", url);
     try {
       const adPageSource = await fetch(url);
       const body = await adPageSource.text();
@@ -227,7 +227,9 @@ class SaljicCrawler {
       // No information for status ex. PRODAN
       const status = AD_STATUS.STATUS_NORMAL;
       //Extracting agency ID from url
-      const agencyObjectId = parseInt(url.substring(46, url.length));
+      const agencyObjectId = url
+        ? parseInt(url.substring(46, url.length))
+        : null;
 
       //Extracting main properties
       const propertySelectors = {
@@ -272,14 +274,10 @@ class SaljicCrawler {
         .trim();
 
       const latAndLongSrc = $(propertySelectors.latAndLong).attr("src");
-      const latText = latAndLongSrc.substring(
+      const tmpLatLong = latAndLongSrc.split("marker=")[1];
-        latAndLongSrc.indexOf("marker=") + 7,
+      const latText = tmpLatLong.split("%2C")[0];
-        latAndLongSrc.indexOf("%2C", latAndLongSrc.indexOf("marker="))
+      const longText = tmpLatLong.split("%2C")[1];
-      );
+
-      const longText = latAndLongSrc.substring(
-        latAndLongSrc.indexOf("%2C", latAndLongSrc.indexOf("marker=")) + 3,
-        latAndLongSrc.length
-      );
       const locationLat = parseFloat(latText) || null;
       const locationLong = parseFloat(longText) || null;
 
@@ -328,11 +326,10 @@ class SaljicCrawler {
       let numberOfViewsKivi = null;
       let streetNumber = 0;
       let adStatus = status;
-      let shortDescription = descriptions.substring(
+      let shortDescription = descriptions
-        0,
+        ? descriptions.substring(0, descriptions.indexOf("."))
-        descriptions.indexOf(".")
+        : "";
-      );
+      let longDescription = descriptions || "";
-      let longDescription = descriptions;
       //Extracting data - Glavne karakteristike
       let mainFieldIndex = 1;
       do {
@@ -343,10 +340,14 @@ class SaljicCrawler {
           .replace(/[\n\r\t]/gm, "")
           .trim();
 
-        const mainFieldTitle = mainField.substring(0, mainField.indexOf("  "));
+        const mainFieldTitle = mainField
+          ? mainField.substring(0, mainField.indexOf("  "))
+          : "";
         const mainFieldValue = mainField
-          .substring(mainField.indexOf("  "), mainField.length)
+          ? mainField
-          .trim();
+              .substring(mainField.indexOf("  "), mainField.length)
+              .trim()
+          : "";
 
         switch (mainFieldTitle) {
           case "Površina":

app/helpers/db/realEstate.js

@@ -332,14 +332,10 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
     };
   }
 
-  //When includeIncompleteAds are not defined - null it will consider it true
   const order = [["updatedAt", "desc"]];
 
   return db.RealEstate.findAll({
-    where:
+    where: includeIncompleteAds ? queryIncludeIncomplete : query,
-      includeIncompleteAds || includeIncompleteAds == null
-        ? queryIncludeIncomplete
-        : query,
     limit: maxResults,
     order
   });

app/helpers/db/searchRequest.js

@@ -157,7 +157,7 @@ const findSearchRequestsForRealEstate = async realEstate => {
     } else {
       // If real estate dont have defined number of rooms ex. null
       //It returns requests that didn't choose number of rooms - also null
-      //Or ones that picked some values but also picked to includeIncomplete ads (or default)
+      //Or ones that picked some values but also picked to includeIncomplete ads
       numberOfRoomsQuery = {
         [Op.or]: [
           {
@@ -176,10 +176,7 @@ const findSearchRequestsForRealEstate = async realEstate => {
           },
           {
             includeIncompleteAds: {
-              [Op.or]: {
+              [Op.eq]: true
-                [Op.eq]: true,
-                [Op.is]: null
-              }
             }
           }
         ]
@@ -229,10 +226,7 @@ const findSearchRequestsForRealEstate = async realEstate => {
           },
           {
             includeIncompleteAds: {
-              [Op.or]: {
+              [Op.eq]: true
-                [Op.eq]: true,
-                [Op.is]: null
-              }
             }
           }
         ]
@@ -281,10 +275,7 @@ const findSearchRequestsForRealEstate = async realEstate => {
           },
           {
             includeIncompleteAds: {
-              [Op.or]: {
+              [Op.eq]: true
-                [Op.eq]: true,
-                [Op.is]: null
-              }
             }
           }
         ]
@@ -322,10 +313,7 @@ const findSearchRequestsForRealEstate = async realEstate => {
               },
               {
                 includeIncompleteAds: {
-                  [Op.or]: {
+                  [Op.eq]: true
-                    [Op.eq]: true,
-                    [Op.is]: null
-                  }
                 }
               }
             ]
@@ -359,10 +347,7 @@ const findSearchRequestsForRealEstate = async realEstate => {
               },
               {
                 includeIncompleteAds: {
-                  [Op.or]: {
+                  [Op.eq]: true
-                    [Op.eq]: true,
-                    [Op.is]: null
-                  }
                 }
               }
             ]
@@ -396,10 +381,7 @@ const findSearchRequestsForRealEstate = async realEstate => {
               },
               {
                 includeIncompleteAds: {
-                  [Op.or]: {
+                  [Op.eq]: true
-                    [Op.eq]: true,
-                    [Op.is]: null
-                  }
                 }
               }
             ]
@@ -441,13 +423,10 @@ const findSearchRequestsForRealEstate = async realEstate => {
       [Op.eq]: "ANY"
     };
   }
-  //Tag to check if incomplete ads are accepted in query which is default
+  //Tag to check if incomplete ads are accepted in query
   if (checkForIncompleteWanted) {
     query.includeIncompleteAds = {
-      [Op.or]: {
+      [Op.eq]: true
-        [Op.eq]: true,
-        [Op.is]: null
-      }
     };
   }
 

app/helpers/fetchWrapper.js

@@ -1,21 +0,0 @@
-const nodeFetch = require("node-fetch");
-const {
-  USER_AGENT,
-  USE_SCRAPER_API,
-  SCRAPER_API_KEY
-} = require("../config/appConfig");
-
-const fetch = async (url, options = {}) => {
-  const newOptions = Object.assign({}, options);
-  if (!newOptions["headers"]) {
-    newOptions["headers"] = {};
-  }
-  newOptions["headers"]["User-Agent"] = USER_AGENT;
-  const urlAdaptedForScraping = USE_SCRAPER_API
-    ? `http://api.scraperapi.com/?api_key=${SCRAPER_API_KEY}&url=${url}`
-    : url;
-
-  return nodeFetch(urlAdaptedForScraping, newOptions);
-};
-
-module.exports = fetch;

app/models/searchRequest.js

@@ -15,7 +15,15 @@ module.exports = (sequelize, DataTypes) => {
       allowNull: false,
       defaultValue: {
         type: "Polygon",
-        coordinates: [[[0, 0], [0, 0], [0, 0], [0, 0], [0, 0]]],
+        coordinates: [
+          [
+            [0, 0],
+            [0, 0],
+            [0, 0],
+            [0, 0],
+            [0, 0]
+          ]
+        ],
         crs: { type: "name", properties: { name: "EPSG:4326" } }
       }
     },

app/services/notificationService.js

@@ -131,7 +131,7 @@ const notifyRequestsWithDailyOption = async () => {
 };
 
 const checkUpNotify = async () => {
-  /* const searchRequestsForCheckUp = await findAllRequestsForCheckUp();
+ /* const searchRequestsForCheckUp = await findAllRequestsForCheckUp();
 
   const asyncSendEmailActions = [];
 
@@ -144,7 +144,7 @@ const checkUpNotify = async () => {
     asyncSendEmailActions.push(sendEmailPromise);
     sendEmailPromise.catch(err => console.log("[Email Sending Failed]", err));
   }
-  await Promise.all(asyncSendEmailActions); */
+  await Promise.all(asyncSendEmailActions); */ 
 };
 
 module.exports = {

app/views/advancedFilters.ejs

@@ -61,8 +61,9 @@
 <p class="distinguished">
   <label class="checkbox-label">
     <input type="checkbox" class="filled-in" name="includeIncompleteAds"
+      <% if (includeIncompleteAds) { %>
            checked
-      >
+      <% } %>>
     <span>Uključi i oglase bez potpunih informacija</span>
   </label>
 </p>

development.env

@@ -10,7 +10,6 @@ APP_BASE_URL=base url for the app
 
 ENVIRONMENT=Variable to denote development, staging and production
 
-USER_AGENT=User agent header to send in fetch requests
 
 MAX_REAL_ESTATES_IN_EMAIL=Max number of real estates that will be shown in email, others will be truncated and URL with full list will be shwon
 MAX_REAL_ESTATES_IN_FIRST_EMAIL=Max number of real estates that will be shown in first (welcome) email
@@ -22,10 +21,6 @@ GA_ID=Google Analytics ID
 #=============== GOOGLE MAPS =============#
 API_MAP_KEY=(your-key-here)
 
-#=============== SCRAPER API SUPORT =============#
-USE_SCRAPER_API= To turn it on (1) or off (0)
-SCRAPER_API_KEY= Key for Scraper api 
-
 #=============== AWS SDK EMAIL SETTINGS =======#
 AWS_KEY_ID=(your-key-here)
 AWS_SECRET_ACCESS_KEY=(your-key-here)
@@ -74,4 +69,4 @@ AKTIDO_FORCE_CRAWL=Non-zero value will force crawler to crawl all pages without
 SALJIC_MAX_RESULTS_PER_PAGE=For Saljic crawler, this represents how many ads are crawled at once
 SALJIC_CRAWLER_AD_TYPE=enum name of what type of ads should be crawled, check common/enums.js file for valid values
 SALJIC_CRAWLER_AD_CATEGORIES=comma separated list of enum names of categories to be included, check common/enums.js file for valid values
-SALJIC_FORCE_CRAWL=Non-zero value will force crawler to crawl all pages without stopping when known real estate is found
+SALJIC_FORCE_CRAWL=Non-zero value will force crawler to crawl all pages without stopping when known real estate is found