FEATURE: Extend summary backfill to also generate gists

Updates default batch size to 0 and max to 10000

Author: romanrizzi

app/jobs/scheduled/summaries_backfill.rb

@@ -14,10 +14,13 @@ def execute(_args)
       limit_per_job = [SiteSetting.ai_summary_backfill_maximum_topics_per_hour, 12].max / 12
       budget = [current_budget, limit_per_job].min
 
+      system_user = Discourse.system_user
+
       backfill_candidates
         .limit(budget)
         .each do |topic|
-          DiscourseAi::Summarization.topic_summary(topic).force_summarize(Discourse.system_user)
+          DiscourseAi::Summarization.topic_summary(topic).force_summarize(system_user)
+          DiscourseAi::Summarization.topic_gist(topic).force_summarize(system_user)
         end
     end
 
@@ -27,10 +30,12 @@ def backfill_candidates
         .joins(
           "LEFT OUTER JOIN ai_summaries ais ON topics.id = ais.target_id AND ais.target_type = 'Topic'",
         )
-        .where(
-          "ais.id IS NULL OR UPPER(ais.content_range) < topics.highest_post_number + 1",
-        ) # (1..1) gets stored ad (1..2).
-        .order("ais.created_at DESC NULLS FIRST, topics.last_posted_at DESC")
+        .group("topics.id")
+        .having(<<~SQL)
+          BOOL_OR(ais.id IS NULL) OR 
+          BOOL_OR(ais.created_at IS NOT NULL AND UPPER(ais.content_range) < topics.highest_post_number + 1)
+        SQL
+        .order("MAX(ais.created_at) DESC NULLS FIRST, topics.last_posted_at DESC")
     end
 
     def current_budget

config/settings.yml

@@ -376,9 +376,9 @@ discourse_ai:
     type: list
     list_type: compact
   ai_summary_backfill_maximum_topics_per_hour:
-    default: 10
+    default: 0
     min: 0
-    max: 1000
+    max: 10000
   ai_summary_backfill_minimum_word_count:
     default: 200
     hidden: true

spec/jobs/scheduled/summaries_backfill_spec.rb

@@ -90,14 +90,18 @@
       Fabricate(:ai_summary, target: topic, created_at: 3.hours.ago, content_range: (1..1))
 
       summary_1 = "Summary of topic_2"
+      gist_1 = "Gist of topic_2"
       summary_2 = "Summary of topic"
+      gist_2 = "Gist of topic"
 
-      DiscourseAi::Completions::Llm.with_prepared_responses([summary_1, summary_2]) do
-        subject.execute({})
-      end
+      DiscourseAi::Completions::Llm.with_prepared_responses(
+        [summary_1, gist_1, summary_2, gist_2],
+      ) { subject.execute({}) }
 
-      expect(AiSummary.find_by(target: topic_2).summarized_text).to eq(summary_1)
-      expect(AiSummary.find_by(target: topic).summarized_text).to eq(summary_2)
+      expect(AiSummary.complete.find_by(target: topic_2).summarized_text).to eq(summary_1)
+      expect(AiSummary.gist.find_by(target: topic_2).summarized_text).to eq(gist_1)
+      expect(AiSummary.complete.find_by(target: topic).summarized_text).to eq(summary_2)
+      expect(AiSummary.gist.find_by(target: topic).summarized_text).to eq(gist_2)
     end
   end
 end