feat(highlights) - Decrease score for consecutive highlights with sam…

…e image

* Score lower for items with same preview
* Fixes #1274

common/recommender/Baseline.js

@@ -275,25 +275,46 @@ class Baseline {
   }
 
   /**
-   *  Decrease the score for consecutive items from the same domain.
-   *  Combined with sorting by score the result is we don't see consecutive
-   *  items from the same domain.
+   * Determine if two entries are similar. Used to lower the score for similar consecutive items.
+   *
+   * @param {Object} prev
+   * @param {Object} curr
+   * @returns {boolean}
+   * @private
+   */
+  _similarItems(prev, curr) {
+    const imgPrev = getBestImage(prev.images);
+    const imgCurr = getBestImage(curr.images);
+    const hasImage = imgPrev && imgCurr;
+    return prev.host === curr.host || (hasImage && imgPrev.url === imgCurr.url);
+  }
+
+  /**
+   *  Decrease the score for consecutive items which are similar (see `_similarItems`).
+   *  Combined with sorting by score the result is we don't see similar consecutive
+   *  items.
    *
    *  @param {Array} entries - scored and sorted highlight items.
    */
   dedupe(entries) {
-    let penalty = 0.6;
+    let penalty = 0.8;
 
-    return entries.map((entry, i, arr) => {
-      if (i > 0 && arr[i - 1].host === entry.host) {
-        let score = entry.score * penalty;
+    if (entries.length < 2) {
+      return entries;
+    }
+
+    entries.reduce((prev, curr) => {
+      if (this._similarItems(prev, curr)) {
+        curr.score *= penalty;
         penalty -= 0.2;
-        return Object.assign({}, entry, {score});
+      } else {
+        penalty = 0.8;
       }
 
-      penalty = 0.8;
-      return entry;
+      return curr;
     });
+
+    return entries;
   }
 }
 

content-test/common/Baseline.test.js

@@ -159,6 +159,44 @@ describe("Baseline", () => {
     assert.ok(items[1].score > items[2].score);
   });
 
+  it("should decrease score for consecutive items with same image", () => {
+    let fakeUrlsWithScore = [
+      {host: "foo.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.sameimage.jpg"}], score: 1},
+      {host: "bar.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.sameimage.jpg"}], score: 1},
+      {host: "bar.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.sameimage.jpg"}], score: 1},
+      {host: "diff.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.diffimage.jpg"}], score: 1}
+    ];
+    let items = baseline.dedupe(fakeUrlsWithScore);
+    assert.ok(items[0].score > items[1].score);
+    assert.ok(items[1].score > items[2].score);
+    assert.equal(items[1].score, 0.8);
+    assert.equal(items[3].score, 1);
+  });
+
+  it("should decrease score for consecutive items with same image (1 entry)", () => {
+    let fakeUrlsWithScore = [
+      {host: "foo.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.sameimage.jpg"}], score: 1}
+    ];
+    let items = baseline.dedupe(fakeUrlsWithScore);
+    assert.equal(items[0].score, 1);
+  });
+
+  it("should decrease score for consecutive items with same image (no entries)", () => {
+    let items = baseline.dedupe([]);
+    assert.equal(items.length, 0);
+  });
+
+  it("should reset penalty coefficient after 2 consecutive & different items in the list", () => {
+    let fakeUrlsWithScore = [
+      {host: "foo.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.sameimage.jpg"}], score: 1},
+      {host: "bar.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.sameimage.jpg"}], score: 1},
+      {host: "diff.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.diffimage.jpg"}], score: 1},
+      {host: "diff.com", images: [{size: 1000, width: 300, height: 300, url: "http://www.diffimage.jpg"}], score: 1}
+    ];
+    let items = baseline.dedupe(fakeUrlsWithScore);
+    assert.equal(items[3].score, 0.8);
+  });
+
   it("should decrease by the right amount", () => {
     let fakeUrlsWithScore = fakeUrls.map(url => Object.assign({}, url, {score: 1}));
     let items = baseline.dedupe(fakeUrlsWithScore);