Bug 2004746 - Add URL normalization step for New Tab frecency based tile sorting. r=home-newtab-reviewers,nbarrett

Differential Revision: https://phabricator.services.mozilla.com/D275524

Author: Irene Ni

browser/extensions/newtab/lib/TopSitesFeed.sys.mjs

@@ -1449,6 +1449,19 @@ export class TopSitesFeed {
     );
   }
 
+  normalizeUrl(url) {
+    let normalized = url;
+    if (normalized.startsWith("https://")) {
+      normalized = normalized.slice(8);
+    } else if (normalized.startsWith("http://")) {
+      normalized = normalized.slice(7);
+    }
+    if (normalized.startsWith("www.")) {
+      normalized = normalized.slice(4);
+    }
+    return normalized;
+  }
+
   /**
    * Build frecency-boosted spocs from a list of sponsor domains by checking Places history.
    * Checks if domains exist in history, dedupes against organic topsites,
@@ -1481,9 +1494,13 @@ export class TopSitesFeed {
 
     const candidates = [];
     frecent.forEach(site => {
+      const normalizedSiteUrl = this.normalizeUrl(site.url);
+
       for (const domainObj of sponsorsToCheck) {
+        const normalizedDomain = this.normalizeUrl(domainObj.domain);
+
         if (
-          !site.url.startsWith(domainObj.domain) ||
+          !normalizedSiteUrl.startsWith(normalizedDomain) ||
           lazy.NewTabUtils.blockedLinks.isBlocked({ url: domainObj.domain })
         ) {
           continue;

browser/extensions/newtab/test/xpcshell/test_TopSitesFeed_frecencyRanking.js

@@ -108,6 +108,7 @@ function getTopSitesFeedForTest(
   return feed;
 }
 
+// eslint-disable-next-line max-statements
 add_task(async function test_frecency_sponsored_topsites() {
   let sandbox = sinon.createSandbox();
   {
@@ -197,7 +198,7 @@ add_task(async function test_frecency_sponsored_topsites() {
 
     sandbox.restore();
   }
-  /*{
+  {
     info(
       "TopSitesFeed.fetchFrecencyBoostedSpocs - " +
         "Should return a single match with a subdomain"
@@ -216,7 +217,7 @@ add_task(async function test_frecency_sponsored_topsites() {
     Assert.equal(frecencyBoostedSpocs[0].hostname, "hostname1");
 
     sandbox.restore();
-  }*/
+  }
   {
     info(
       "TopSitesFeed.fetchFrecencyBoostedSpocs - " +
@@ -264,6 +265,45 @@ add_task(async function test_frecency_sponsored_topsites() {
     Assert.equal(frecencyBoostedSpocs.length, 1);
     Assert.equal(frecencyBoostedSpocs[0].hostname, "hostname4");
 
+    sandbox.restore();
+  }
+  {
+    info(
+      "TopSitesFeed.fetchFrecencyBoostedSpocs - " +
+        "Should not return a match with a different subdomain"
+    );
+    const feed = getTopSitesFeedForTest(sandbox, {
+      frecent: [
+        {
+          url: "https://bus.domain1.com",
+          frecency: 1234,
+        },
+      ],
+    });
+
+    const frecencyBoostedSpocs = await feed.fetchFrecencyBoostedSpocs();
+    Assert.equal(frecencyBoostedSpocs.length, 0);
+
+    sandbox.restore();
+  }
+  {
+    info(
+      "TopSitesFeed.fetchFrecencyBoostedSpocs - " +
+        "Should return a match with the same subdomain"
+    );
+    const feed = getTopSitesFeedForTest(sandbox, {
+      frecent: [
+        {
+          url: "https://sub.domain1.com",
+          frecency: 1234,
+        },
+      ],
+    });
+
+    const frecencyBoostedSpocs = await feed.fetchFrecencyBoostedSpocs();
+    Assert.equal(frecencyBoostedSpocs.length, 1);
+    Assert.equal(frecencyBoostedSpocs[0].hostname, "hostname1sub");
+
     sandbox.restore();
   }
 });