Adjust hot topics gist strategy and add a job to generate gists

Author: romanrizzi

app/jobs/regular/hot_topics_gist_batch.rb

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module ::Jobs
+  class HotTopicsGistBatch < ::Jobs::Base
+    def execute(args)
+      return if !SiteSetting.discourse_ai_enabled
+      return if !SiteSetting.ai_summarization_enabled
+      return if !SiteSetting.ai_summarize_hot_topics_list
+
+      Topic
+        .joins("JOIN topic_hot_scores on topics.id = topic_hot_scores.topic_id")
+        .order("topic_hot_scores.score DESC")
+        .limit(100)
+        .each do |topic|
+          summarizer = DiscourseAi::Summarization.topic_gist(topic)
+          gist = summarizer.existing_summary
+
+          summarizer.delete_cached_summaries! if gist && gist.outdated
+
+          summarizer.summarize(Discourse.system_user)
+        end
+    end
+  end
+end

lib/summarization.rb

@@ -17,7 +17,7 @@ def self.topic_gist(topic)
       if SiteSetting.ai_summarization_model.present? && SiteSetting.ai_summarization_enabled
         DiscourseAi::Summarization::FoldContent.new(
           DiscourseAi::Completions::Llm.proxy(SiteSetting.ai_summarization_model),
-          DiscourseAi::Summarization::Strategies::TopicGist.new(topic),
+          DiscourseAi::Summarization::Strategies::HotTopicGists.new(topic),
         )
       else
         nil

lib/summarization/entry_point.rb

@@ -36,6 +36,10 @@ def inject_into(plugin)
             SiteSetting.ai_summarization_enabled && SiteSetting.ai_summarize_hot_topics_list
           end,
         ) { object.ai_summaries.to_a.first&.summarized_text }
+
+        # To make sure hot topic gists are inmediately up to date, we rely on this event
+        # instead of using a scheduled job.
+        plugin.on(:topic_hot_scores_updated) { Jobs.enqueue(:summarize_hot_topics_batch) }
       end
     end
   end

lib/summarization/strategies/topic_gist.rb renamed to lib/summarization/strategies/hot_topic_gists.rb

@@ -3,7 +3,7 @@
 module DiscourseAi
   module Summarization
     module Strategies
-      class TopicGist < Base
+      class HotTopicGists < Base
         def type
           AiSummary.summary_types[:gist]
         end
@@ -13,20 +13,21 @@ def targets_data
 
           op_post_number = 1
 
-          last_twenty_posts =
+          hot_topics_recent_cutoff = Time.zone.now - SiteSetting.hot_topics_recent_days.days
+
+          recent_hot_posts =
             Post
               .where(topic_id: target.id)
               .where("post_type = ?", Post.types[:regular])
               .where("NOT hidden")
-              .order("post_number DESC")
-              .limit(20)
+              .where("created_at >= ?", hot_topics_recent_cutoff)
               .pluck(:post_number)
 
           posts_data =
             Post
               .where(topic_id: target.id)
               .joins(:user)
-              .where("post_number IN (?)", last_twenty_posts << op_post_number)
+              .where("post_number IN (?)", recent_hot_posts << op_post_number)
               .order(:post_number)
               .pluck(:post_number, :raw, :username)
 

spec/jobs/regular/hot_topics_gist_batch.rb

@@ -0,0 +1,121 @@
+# frozen_string_literal: true
+
+RSpec.describe Jobs::HotTopicsGistBatch do
+  fab!(:topic_1) { Fabricate(:topic) }
+  fab!(:post_1) { Fabricate(:post, topic: topic_1, post_number: 1) }
+  fab!(:post_2) { Fabricate(:post, topic: topic_1, post_number: 2) }
+
+  before do
+    assign_fake_provider_to(:ai_summarization_model)
+    SiteSetting.ai_summarization_enabled = true
+    SiteSetting.ai_summarize_hot_topics_list = true
+  end
+
+  describe "#execute" do
+    context "When there is a topic with a hot score" do
+      before { TopicHotScore.create!(topic_id: topic_1.id, score: 0.1) }
+
+      it "does nothing if the plugin is disabled" do
+        SiteSetting.discourse_ai_enabled = false
+
+        subject.execute({})
+
+        gist = AiSummary.gist.find_by(target: topic_1)
+        expect(gist).to be_nil
+      end
+
+      it "does nothing if the summarization module is disabled" do
+        SiteSetting.ai_summarization_enabled = false
+
+        subject.execute({})
+
+        gist = AiSummary.gist.find_by(target: topic_1)
+        expect(gist).to be_nil
+      end
+
+      it "does nothing if hot topics summarization is disabled" do
+        SiteSetting.ai_summarize_hot_topics_list = false
+
+        subject.execute({})
+
+        gist = AiSummary.gist.find_by(target: topic_1)
+        expect(gist).to be_nil
+      end
+
+      it "creates a gist" do
+        gist_result = "I'm a gist"
+
+        DiscourseAi::Completions::Llm.with_prepared_responses([gist_result]) { subject.execute({}) }
+
+        gist = AiSummary.gist.find_by(target: topic_1)
+        expect(gist.summarized_text).to eq("I'm a gist")
+      end
+
+      context "and we already generated a gist of it" do
+        fab!(:ai_gist) do
+          Fabricate(
+            :topic_ai_gist,
+            target: topic_1,
+            original_content_sha: AiSummary.build_sha("12"),
+          )
+        end
+
+        it "does nothing if the gist is up to date" do
+          subject.execute({})
+
+          gist = AiSummary.gist.find_by(target: topic_1)
+          expect(gist.summarized_text).to eq(ai_gist.summarized_text)
+          expect(gist.original_content_sha).to eq(ai_gist.original_content_sha)
+        end
+
+        it "regenerates it if it's outdated" do
+          Fabricate(:post, topic: topic_1, post_number: 3)
+          gist_result = "They updated me"
+
+          DiscourseAi::Completions::Llm.with_prepared_responses([gist_result]) do
+            subject.execute({})
+          end
+
+          gist = AiSummary.gist.find_by(target: topic_1)
+          expect(gist.summarized_text).to eq(gist_result)
+          expect(gist.original_content_sha).to eq(AiSummary.build_sha("123"))
+        end
+      end
+    end
+
+    context "when there is a topic but it doesn't have a hot score" do
+      it "does nothing" do
+        subject.execute({})
+
+        gist = AiSummary.gist.find_by(target: topic_1)
+        expect(gist).to be_nil
+      end
+    end
+
+    context "when there are multiple hot topics" do
+      fab!(:topic_2) { Fabricate(:topic) }
+      fab!(:post_2_1) { Fabricate(:post, topic: topic_2, post_number: 1) }
+      fab!(:post_2_2) { Fabricate(:post, topic: topic_2, post_number: 2) }
+
+      before do
+        TopicHotScore.create!(topic_id: topic_1.id, score: 0.2)
+        TopicHotScore.create!(topic_id: topic_2.id, score: 0.4)
+      end
+
+      it "processes them by score order" do
+        topic_1_gist = "I'm gist of topic 1"
+        topic_2_gist = "I'm gist of topic 2"
+
+        DiscourseAi::Completions::Llm.with_prepared_responses([topic_2_gist, topic_1_gist]) do
+          subject.execute({})
+        end
+
+        gist = AiSummary.gist.find_by(target: topic_1)
+        expect(gist.summarized_text).to eq(topic_1_gist)
+
+        gist_2 = AiSummary.gist.find_by(target: topic_2)
+        expect(gist_2.summarized_text).to eq(topic_2_gist)
+      end
+    end
+  end
+end

spec/lib/modules/summarization/strategies/topic_gist_spec.rb renamed to spec/lib/modules/summarization/strategies/hot_topic_gists_spec.rb

@@ -1,29 +1,20 @@
 # frozen_string_literal: true
 
-RSpec.describe DiscourseAi::Summarization::Strategies::TopicGist do
+RSpec.describe DiscourseAi::Summarization::Strategies::HotTopicGists do
   subject(:gist) { described_class.new(topic) }
 
   fab!(:topic) { Fabricate(:topic, highest_post_number: 25) }
   fab!(:post_1) { Fabricate(:post, topic: topic, post_number: 1) }
   fab!(:post_2) { Fabricate(:post, topic: topic, post_number: 2) }
 
   describe "#targets_data" do
-    context "when the topic has more than 20 posts" do
-      before do
-        offset = 3 # Already created posts 1 and 2
-        (topic.highest_post_number - 2).times do |i|
-          Fabricate(:post, topic: topic, post_number: i + offset)
-        end
-      end
-
-      it "includes the OP and the last 20 posts" do
-        content = gist.targets_data
-        post_numbers = content[:contents].map { |c| c[:id] }
+    it "respects the `hot_topics_recent_days` setting" do
+      post_2.update(created_at: (SiteSetting.hot_topics_recent_days + 1).days.ago)
+      Fabricate(:post, topic: topic, post_number: 3)
 
-        expected = (6..25).to_a << 1
+      post_numbers = gist.targets_data[:contents].map { |c| c[:id] }
 
-        expect(post_numbers).to contain_exactly(*expected)
-      end
+      expect(post_numbers).to contain_exactly(1, 3)
     end
 
     it "only includes visible posts" do