FIX: Improve top links section from user summary (#15675)

* Do not extract links for hotlinked images
* Include only links that have been clicked at least once in user
summary

app/models/user_summary.rb

@@ -40,6 +40,7 @@ class UserSummary
       .merge(Topic.listable_topics.visible.secured(@guardian))
       .where(user: @user)
       .where(internal: false, reflection: false, quote: false)
+      .where('clicks > 0')
       .order('clicks DESC, topic_links.created_at DESC')
       .limit(MAX_SUMMARY_RESULTS)
   end

lib/pretty_text.rb

@@ -371,6 +371,9 @@ module PrettyText
     # remove href inside quotes & oneboxes & elided part
     doc.css("aside.quote a, aside.onebox a, .elided a").remove
 
+    # remove hotlinked images
+    doc.css("a.onebox > img").each { |img| img.parent.remove }
+
     # extract all links
     doc.css("a").each do |a|
       if a["href"].present? && a["href"][0] != "#"

spec/components/pretty_text_spec.rb

@@ -836,6 +836,20 @@ describe PrettyText do
       expect(extract_urls("<aside class=\"quote\" data-topic=\"321\">aside</aside>")).to eq(["/t/321"])
     end
 
+    it "does not extract links from hotlinked images" do
+      html = <<~HTML
+        <p>
+        <a href="https://example.com">example</a>
+
+        <a href="https://images.pexels.com/photos/1525041/pexels-photo-1525041.jpeg?auto=compress&amp;cs=tinysrgb&amp;w=1260&amp;h=750&amp;dpr=2" target="_blank" rel="noopener" class="onebox">
+        <img src="https://images.pexels.com/photos/1525041/pexels-photo-1525041.jpeg?auto=compress&amp;cs=tinysrgb&amp;w=1260&amp;h=750&amp;dpr=2" width="690" height="459">
+        </a>
+        </p>
+      HTML
+
+      expect(extract_urls(html)).to eq(["https://example.com"])
+    end
+
     it "should lazyYT videos" do
       expect(extract_urls("<div class=\"lazyYT\" data-youtube-id=\"yXEuEUQIP3Q\" data-youtube-title=\"Mister Rogers defending PBS to the US Senate\" data-width=\"480\" data-height=\"270\" data-parameters=\"feature=oembed&amp;wmode=opaque\"></div>")).to eq(["https://www.youtube.com/watch?v=yXEuEUQIP3Q"])
     end

spec/models/user_summary_spec.rb

@@ -93,4 +93,11 @@ describe UserSummary do
     expect(summary.top_categories.first[:topic_count]).to eq(1)
     expect(summary.top_categories.first[:post_count]).to eq(1)
   end
+
+  it "does not include summaries with no clicks" do
+    post = Fabricate(:post, raw: "[example](https://example.com)")
+    TopicLink.extract_from(post)
+    summary = UserSummary.new(post.user, Guardian.new)
+    expect(summary.links.length).to eq(0)
+  end
 end