Fix performance of percentile calculation for annual reports (#32765)

app/lib/annual_report.rb

@@ -17,11 +17,21 @@ class AnnualReport
 
   SCHEMA = 1
 
+  def self.table_name_prefix
+    'annual_report_'
+  end
+
   def initialize(account, year)
     @account = account
     @year = year
   end
 
+  def self.prepare(year)
+    SOURCES.each do |klass|
+      klass.prepare(year)
+    end
+  end
+
   def generate
     return if GeneratedAnnualReport.exists?(account: @account, year: @year)
 

app/lib/annual_report/percentiles.rb

@@ -1,62 +1,37 @@
 # frozen_string_literal: true
 
 class AnnualReport::Percentiles < AnnualReport::Source
+  def self.prepare(year)
+    AnnualReport::StatusesPerAccountCount.connection.exec_query(<<~SQL.squish, nil, [year, Mastodon::Snowflake.id_at(DateTime.new(year).beginning_of_year), Mastodon::Snowflake.id_at(DateTime.new(year).end_of_year)])
+      INSERT INTO annual_report_statuses_per_account_counts (year, account_id, statuses_count)
+      SELECT $1, account_id, count(*)
+      FROM statuses
+      WHERE id BETWEEN $2 AND $3
+      AND (local OR uri IS NULL)
+      GROUP BY account_id
+      ON CONFLICT (year, account_id) DO NOTHING
+    SQL
+  end
+
   def generate
     {
       percentiles: {
-        followers: (total_with_fewer_followers / (total_with_any_followers + 1.0)) * 100,
-        statuses: (total_with_fewer_statuses / (total_with_any_statuses + 1.0)) * 100,
+        statuses: 100.0 - ((total_with_fewer_statuses / (total_with_any_statuses + 1.0)) * 100),
       },
     }
   end
 
   private
 
-  def followers_gained
-    @followers_gained ||= @account.passive_relationships.where("date_part('year', follows.created_at) = ?", @year).count
-  end
-
   def statuses_created
     @statuses_created ||= report_statuses.count
   end
 
-  def total_with_fewer_followers
-    @total_with_fewer_followers ||= Follow.find_by_sql([<<~SQL.squish, { year: @year, comparison: followers_gained }]).first.total
-      WITH tmp0 AS (
-        SELECT follows.target_account_id
-        FROM follows
-        INNER JOIN accounts ON accounts.id = follows.target_account_id
-        WHERE date_part('year', follows.created_at) = :year
-          AND accounts.domain IS NULL
-        GROUP BY follows.target_account_id
-        HAVING COUNT(*) < :comparison
-      )
-      SELECT count(*) AS total
-      FROM tmp0
-    SQL
-  end
-
   def total_with_fewer_statuses
-    @total_with_fewer_statuses ||= Status.find_by_sql([<<~SQL.squish, { comparison: statuses_created, min_id: year_as_snowflake_range.first, max_id: year_as_snowflake_range.last }]).first.total
-      WITH tmp0 AS (
-        SELECT statuses.account_id
-        FROM statuses
-        INNER JOIN accounts ON accounts.id = statuses.account_id
-        WHERE statuses.id BETWEEN :min_id AND :max_id
-          AND accounts.domain IS NULL
-        GROUP BY statuses.account_id
-        HAVING count(*) < :comparison
-      )
-      SELECT count(*) AS total
-      FROM tmp0
-    SQL
-  end
-
-  def total_with_any_followers
-    @total_with_any_followers ||= Follow.where("date_part('year', follows.created_at) = ?", @year).joins(:target_account).merge(Account.local).count('distinct follows.target_account_id')
+    @total_with_fewer_statuses ||= AnnualReport::StatusesPerAccountCount.where(year: year).where(statuses_count: ...statuses_created).count
   end
 
   def total_with_any_statuses
-    @total_with_any_statuses ||= Status.where(id: year_as_snowflake_range).joins(:account).merge(Account.local).count('distinct statuses.account_id')
+    @total_with_any_statuses ||= AnnualReport::StatusesPerAccountCount.where(year: year).count
   end
 end

app/lib/annual_report/source.rb

@@ -8,6 +8,14 @@ class AnnualReport::Source
     @year = year
   end
 
+  def self.prepare(_year)
+    # Use this method if any pre-calculations must be made before individual annual reports are generated
+  end
+
+  def generate
+    raise NotImplementedError
+  end
+
   protected
 
   def report_statuses

app/models/annual_report/statuses_per_account_count.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+# == Schema Information
+#
+# Table name: annual_report_statuses_per_account_counts
+#
+#  id             :bigint(8)        not null, primary key
+#  year           :integer          not null
+#  account_id     :bigint(8)        not null
+#  statuses_count :bigint(8)        not null
+#
+
+class AnnualReport::StatusesPerAccountCount < ApplicationRecord
+  # This table facilitates percentile calculations
+end