Merge branch 'rental-crawler-fix' into 'master'

Rental crawler fix

See merge request saburly/marketalarm/web!80

.gitignore

@@ -2,4 +2,4 @@ node_modules/
 .env
 .idea/
 .eslintrc
-.vscode/
+.vscode/

app/crawler/specificCrawlers/rental.js

@@ -312,7 +312,7 @@ class RentalCrawler {
 
         let numberOfRooms =
             parseInt(extractedData["re_realEstates_roomsNO"]) +
-              parseInt(extractedData["re_realEstates_bedroomNO"]) || null,
+              parseInt(extractedData["re_realEstates_bedNO"]) || null,
           numberOfFloors =
             parseInt(extractedData["re_realEstates_floorsNO"]) ||
             this.getNumberOfFloorsFromFloorId(extractedData["re_floorNO_id"]),
@@ -352,7 +352,9 @@ class RentalCrawler {
             realEstatePropertiesFromInfrastructure.phoneConnection,
           cableTV = realEstatePropertiesFromInfrastructure.cableTV,
           internet = realEstatePropertiesFromInfrastructure.internet,
-          basementAttic = realEstatePropertiesFromSpaces.basementAttic,
+          basementAttic =
+            realEstatePropertiesFromSpaces.basementAttic ||
+            this.checkBasemAtticFromFloors(extractedData["re_floorNO_id"]),
           storeRoom = realEstatePropertiesFromSpaces.storeRoom,
           videoSurveillance =
             realEstatePropertiesFromDescriptions.videoSurveillance ||
@@ -397,9 +399,7 @@ class RentalCrawler {
         );
         if (!publishedDateMoment.isValid()) {
           throw {
-            message: `Invalid published date : ${
-              extractedData["re_realEstates_inserted"]
-            }`
+            message: `Invalid published date : ${extractedData["re_realEstates_inserted"]}`
           };
         }
 
@@ -410,9 +410,7 @@ class RentalCrawler {
         );
         if (!renewedDateMoment.isValid()) {
           throw {
-            message: `Invalid renewed date : ${
-              extractedData["re_realEstates_edited"]
-            }`
+            message: `Invalid renewed date : ${extractedData["re_realEstates_edited"]}`
           };
         }
 
@@ -782,8 +780,42 @@ class RentalCrawler {
     if (floorIds.length === 0) {
       return null;
     }
+    let noOfFloors = floorIds.length;
+    // Floors of 'suteren', 'podrum', 'tavan' and 'potkrovlje' are not counted
+    floorIds.forEach(id => {
+      if (
+        parseInt(id) === 1 ||
+        parseInt(id) === 2 ||
+        parseInt(id) === 12 ||
+        parseInt(id) === 14
+      ) {
+        noOfFloors--;
+      }
+    });
+    return noOfFloors;
+  }
 
-    return floorIds.length;
+  checkBasemAtticFromFloors(floorsIdText) {
+    // floorIdText can be array of numbers, separated by comma or number
+    const floorIds = floorsIdText.split(",");
+
+    let check = false;
+
+    if (floorIds.length === 0) {
+      check = false;
+    }
+    //If floors 'suteren', 'podrum', 'tavan' and 'potkrovlje' exists then tag for basement-attic is true
+    floorIds.forEach(id => {
+      if (
+        parseInt(id) === 1 ||
+        parseInt(id) === 2 ||
+        parseInt(id) === 12 ||
+        parseInt(id) === 14
+      ) {
+        check = true;
+      }
+    });
+    return check;
   }
 
   async sleep(ms) {

test/rentalScrapeTest.js

@@ -13,5 +13,5 @@ if (urlToScrape) {
   })();
 } else {
   console.log("No URL to scrape. Use like this : ");
-  console.log("npm run test-olx-scraper -- URL_TO_SCRAPE");
+  console.log("npm run test-rental-scraper -- URL_TO_SCRAPE");
 }