FEATURE: Cache embed contents in the database (#25133)

* FEATURE: Cache embed contents in the database

This will be useful for features that rely on the semantic content of topics, like the many AI features



Co-authored-by: Roman Rizzi <rizziromanalejandro@gmail.com>

app/models/topic_embed.rb

@@ -3,10 +3,13 @@
 class TopicEmbed < ActiveRecord::Base
   include Trashable
 
+  EMBED_CONTENT_CACHE_MAX_LENGTH = 32_000
+
   belongs_to :topic
   belongs_to :post
   validates_presence_of :embed_url
   validates_uniqueness_of :embed_url
+  validates :embed_content_cache, length: { maximum: EMBED_CONTENT_CACHE_MAX_LENGTH }
 
   before_validation(on: :create) do
     unless (
@@ -43,6 +46,7 @@ class TopicEmbed < ActiveRecord::Base
   def self.import(user, url, title, contents, category_id: nil, cook_method: nil, tags: nil)
     return unless url =~ %r{\Ahttps?\://}
 
+    original_contents = contents.dup.truncate(EMBED_CONTENT_CACHE_MAX_LENGTH)
     contents = first_paragraph_from(contents) if SiteSetting.embed_truncate && cook_method.nil?
     contents ||= ""
     contents = contents.dup << imported_from_html(url)
@@ -77,6 +81,7 @@ class TopicEmbed < ActiveRecord::Base
         }
 
         post = PostCreator.create(user, create_args)
+        post.topic.topic_embed.update!(embed_content_cache: original_contents)
       end
     else
       absolutize_urls(url, contents)
@@ -101,7 +106,7 @@ class TopicEmbed < ActiveRecord::Base
           changes[:title] = title if title.present?
 
           post.revise(user, changes, skip_validations: true, bypass_rate_limiter: true)
-          embed.update!(content_sha1: content_sha1)
+          embed.update!(content_sha1: content_sha1, embed_content_cache: original_contents)
         end
       end
     end
@@ -296,6 +301,11 @@ class TopicEmbed < ActiveRecord::Base
         response = TopicEmbed.find_remote(url)
 
         body = response.body
+        if post&.topic&.topic_embed && body.present?
+          post.topic.topic_embed.update!(
+            embed_content_cache: body.truncate(EMBED_CONTENT_CACHE_MAX_LENGTH),
+          )
+        end
         body << TopicEmbed.imported_from_html(url)
         body
       end
@@ -306,15 +316,16 @@ end
 #
 # Table name: topic_embeds
 #
-#  id            :integer          not null, primary key
-#  topic_id      :integer          not null
-#  post_id       :integer          not null
-#  embed_url     :string(1000)     not null
-#  content_sha1  :string(40)
-#  created_at    :datetime         not null
-#  updated_at    :datetime         not null
-#  deleted_at    :datetime
-#  deleted_by_id :integer
+#  id                  :integer          not null, primary key
+#  topic_id            :integer          not null
+#  post_id             :integer          not null
+#  embed_url           :string(1000)     not null
+#  content_sha1        :string(40)
+#  created_at          :datetime         not null
+#  updated_at          :datetime         not null
+#  deleted_at          :datetime
+#  deleted_by_id       :integer
+#  embed_content_cache :text
 #
 # Indexes
 #

db/migrate/20240104155715_add_embed_content_cache_to_topic_embed.rb

@@ -0,0 +1,7 @@
+# frozen_string_literal: true
+
+class AddEmbedContentCacheToTopicEmbed < ActiveRecord::Migration[7.0]
+  def change
+    add_column :topic_embeds, :embed_content_cache, :text
+  end
+end

spec/models/topic_embed_spec.rb

@@ -73,6 +73,9 @@ RSpec.describe TopicEmbed do
         expect(post.cooked).to have_tag("a", with: { href: "http://eviltrout.com/hello" })
         expect(post.cooked).to have_tag("img", with: { src: "http://eviltrout.com/images/wat.jpg" })
 
+        # It caches the embed content
+        expect(post.topic.topic_embed.embed_content_cache).to eq(contents)
+
         # It converts relative URLs to absolute when expanded
         stub_request(:get, url).to_return(status: 200, body: contents)
         expect(TopicEmbed.expanded_for(post)).to have_tag(
@@ -107,6 +110,13 @@ RSpec.describe TopicEmbed do
         expect(topic_embed.post.topic.user).to eq(new_user)
       end
 
+      it "Supports updating the embed content cache" do
+        expect do TopicEmbed.import(user, url, title, "new contents") end.to change {
+          topic_embed.reload.embed_content_cache
+        }
+        expect(topic_embed.embed_content_cache).to eq("new contents")
+      end
+
       it "Should leave uppercase Feed Entry URL untouched in content" do
         cased_url = "http://eviltrout.com/ABCD"
         post = TopicEmbed.import(user, cased_url, title, "some random content")
@@ -559,4 +569,30 @@ RSpec.describe TopicEmbed do
       expect(html).to eq(expected_html)
     end
   end
+
+  describe ".expanded_for" do
+    fab!(:user)
+    let(:title) { "How to turn a fish from good to evil in 30 seconds" }
+    let(:url) { "http://eviltrout.com/123" }
+    let(:contents) { "<p>hello world new post :D</p>" }
+    fab!(:embeddable_host)
+    fab!(:category)
+    fab!(:tag)
+
+    it "returns embed content" do
+      stub_request(:get, url).to_return(status: 200, body: contents)
+      post = TopicEmbed.import(user, url, title, contents)
+      expect(TopicEmbed.expanded_for(post)).to include(contents)
+    end
+
+    it "updates the embed content cache" do
+      stub_request(:get, url)
+        .to_return(status: 200, body: contents)
+        .then
+        .to_return(status: 200, body: "contents changed")
+      post = TopicEmbed.import(user, url, title, contents)
+      TopicEmbed.expanded_for(post)
+      expect(post.topic.topic_embed.reload.embed_content_cache).to include("contents changed")
+    end
+  end
 end