FIX: Strip audio/video content from excerpt (#8881)

lib/pretty_text.rb

@@ -340,8 +340,8 @@ module PrettyText
     doc = Nokogiri::HTML.fragment(html)
     DiscourseEvent.trigger(:reduce_excerpt, doc, options)
     strip_image_wrapping(doc)
+    strip_oneboxed_media(doc)
     html = doc.to_html
-
     ExcerptParser.get_excerpt(html, max_length, options)
   end
 
@@ -374,6 +374,11 @@ module PrettyText
     doc.css(".lightbox-wrapper .meta").remove
   end
 
+  def self.strip_oneboxed_media(doc)
+    doc.css("audio").remove
+    doc.css("video").remove
+  end
+
   def self.convert_vimeo_iframes(doc)
     doc.css("iframe[src*='player.vimeo.com']").each do |iframe|
       if iframe["data-original-href"].present?

spec/components/pretty_text_spec.rb

@@ -771,6 +771,29 @@ describe PrettyText do
         )).to eq("boom")
       end
     end
+
+    it 'should strip audio/video' do
+      html = <<~HTML
+        <audio controls>
+          <source src="https://awebsite.com/audio.mp3"><a href="https://awebsite.com/audio.mp3">https://awebsite.com/audio.mp3</a></source>
+        </audio>
+        <p>Listen to this!</p>
+      HTML
+
+      expect(PrettyText.excerpt(html, 100)).to eq("Listen to this!")
+
+      html = <<~HTML
+        <div class="onebox video-onebox">
+          <video controlslist="nodownload" width="100%" height="100%" controls="">
+            <source src="http://videosource.com/running.mp4">
+            <a href="http://videosource.com/running.mp4">http://videosource.com/running.mp4</a>
+          </video>
+        </div>
+        <p>Watch this, but not in the excerpt.</p>
+      HTML
+
+      expect(PrettyText.excerpt(html, 100)).to eq("Watch this, but not in the excerpt.")
+    end
   end
 
   describe "strip links" do