Exclude audio/video URLs from search result blurbs

Displays translatable "[audio]" or "[video]" placeholders instead of ugly (and often long) URLs.

config/locales/server.en.yml

@@ -2165,6 +2165,8 @@ en:
       topic: "Results"
       user: "Users"
     results_page: "Search results for '%{term}'"
+    audio: "[audio]"
+    video: "[video]"
 
   sso:
     login_error: "Login Error"

lib/search/grouped_search_results.rb

@@ -76,6 +76,14 @@ class Search
       blurb = nil
       cooked = SearchIndexer.scrub_html_for_search(cooked)
 
+      urls = Set.new
+      cooked.scan(URI.regexp(%w{http https})) { urls << $& }
+
+      urls.each do |url|
+        cooked.gsub!(url, I18n.t("search.video")) if url.match(/.(mov|mp4|webm|ogv)/)
+        cooked.gsub!(url, I18n.t("search.audio")) if url.match(/.(mp3|ogg|wav|m4a)/)
+      end
+
       if term
         terms = term.split(/\s+/)
         phrase = terms.first

spec/lib/search_spec.rb

@@ -15,4 +15,32 @@ describe Search do
     end
   end
 
+  context "#GroupedSearchResults.blurb_for" do
+    it "strips audio and video URLs from search blurb" do
+      cooked = <<~RAW
+        link to an external page: https://google.com/?u=bar
+
+        link to an audio file: https://somesite.com/content/file123.m4a
+
+        link to a video file: https://somesite.com/content/somethingelse.mov
+      RAW
+      result = Search::GroupedSearchResults.blurb_for(cooked)
+      expect(result).to eq("link to an external page: https://google.com/?u=bar link to an audio file: #{I18n.t("search.audio")} link to a video file: #{I18n.t("search.video")}")
+    end
+
+    it "strips URLs correctly when blurb is" do
+      cooked = <<~RAW
+        Here goes a test cooked with enough characters to hit the blurb limit.
+
+        Something is very interesting about this audio file.
+
+        http://localhost/uploads/default/original/1X/90adc0092b30c04b761541bc0322d0dce3d896e7.m4a
+      RAW
+
+      result = Search::GroupedSearchResults.blurb_for(cooked)
+      expect(result).to eq("Here goes a test cooked with enough characters to hit the blurb limit. Something is very interesting about this audio file. #{I18n.t("search.audio")}")
+    end
+
+  end
+
 end