More elegant scrape of lat and long.

app/crawler/specificCrawlers/prostor.js

@@ -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/saljic.js

@@ -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(
-        latAndLongSrc.indexOf("marker=") + 7,
-        latAndLongSrc.indexOf("%2C", latAndLongSrc.indexOf("marker="))
-      );
-      const longText = latAndLongSrc.substring(
-        latAndLongSrc.indexOf("%2C", latAndLongSrc.indexOf("marker=")) + 3,
-        latAndLongSrc.length
-      );
+      const tmpLatLong = latAndLongSrc.split("marker=")[1];
+      const latText = tmpLatLong.split("%2C")[0];
+      const longText = tmpLatLong.split("%2C")[1];
+
       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(
-        0,
-        descriptions.indexOf(".")
-      );
-      let longDescription = descriptions;
+      let shortDescription = descriptions
+        ? descriptions.substring(0, descriptions.indexOf("."))
+        : "";
+      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)
-          .trim();
+          ? mainField
+              .substring(mainField.indexOf("  "), mainField.length)
+              .trim()
+          : "";
 
         switch (mainFieldTitle) {
           case "Površina":