Add annual reports for accounts (#28693)

app/lib/annual_report.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+class AnnualReport
+  include DatabaseHelper
+
+  SOURCES = [
+    AnnualReport::Archetype,
+    AnnualReport::TypeDistribution,
+    AnnualReport::TopStatuses,
+    AnnualReport::MostUsedApps,
+    AnnualReport::CommonlyInteractedWithAccounts,
+    AnnualReport::TimeSeries,
+    AnnualReport::TopHashtags,
+    AnnualReport::MostRebloggedAccounts,
+    AnnualReport::Percentiles,
+  ].freeze
+
+  SCHEMA = 1
+
+  def initialize(account, year)
+    @account = account
+    @year = year
+  end
+
+  def generate
+    return if GeneratedAnnualReport.exists?(account: @account, year: @year)
+
+    GeneratedAnnualReport.create(
+      account: @account,
+      year: @year,
+      schema_version: SCHEMA,
+      data: data
+    )
+  end
+
+  private
+
+  def data
+    with_read_replica do
+      SOURCES.each_with_object({}) { |klass, hsh| hsh.merge!(klass.new(@account, @year).generate) }
+    end
+  end
+end

app/lib/annual_report/archetype.rb

@@ -0,0 +1,49 @@
+# frozen_string_literal: true
+
+class AnnualReport::Archetype < AnnualReport::Source
+  # Average number of posts (including replies and reblogs) made by
+  # each active user in a single year (2023)
+  AVERAGE_PER_YEAR = 113
+
+  def generate
+    {
+      archetype: archetype,
+    }
+  end
+
+  private
+
+  def archetype
+    if (standalone_count + replies_count + reblogs_count) < AVERAGE_PER_YEAR
+      :lurker
+    elsif reblogs_count > (standalone_count * 2)
+      :booster
+    elsif polls_count > (standalone_count * 0.1) # standalone_count includes posts with polls
+      :pollster
+    elsif replies_count > (standalone_count * 2)
+      :replier
+    else
+      :oracle
+    end
+  end
+
+  def polls_count
+    @polls_count ||= base_scope.where.not(poll_id: nil).count
+  end
+
+  def reblogs_count
+    @reblogs_count ||= base_scope.where.not(reblog_of_id: nil).count
+  end
+
+  def replies_count
+    @replies_count ||= base_scope.where.not(in_reply_to_id: nil).where.not(in_reply_to_account_id: @account.id).count
+  end
+
+  def standalone_count
+    @standalone_count ||= base_scope.without_replies.without_reblogs.count
+  end
+
+  def base_scope
+    @account.statuses.where(id: year_as_snowflake_range)
+  end
+end

app/lib/annual_report/commonly_interacted_with_accounts.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+class AnnualReport::CommonlyInteractedWithAccounts < AnnualReport::Source
+  SET_SIZE = 40
+
+  def generate
+    {
+      commonly_interacted_with_accounts: commonly_interacted_with_accounts.map do |(account_id, count)|
+                                           {
+                                             account_id: account_id,
+                                             count: count,
+                                           }
+                                         end,
+    }
+  end
+
+  private
+
+  def commonly_interacted_with_accounts
+    @account.statuses.reorder(nil).where(id: year_as_snowflake_range).where.not(in_reply_to_account_id: @account.id).group(:in_reply_to_account_id).having('count(*) > 1').order(total: :desc).limit(SET_SIZE).pluck(Arel.sql('in_reply_to_account_id, count(*) AS total'))
+  end
+end

app/lib/annual_report/most_reblogged_accounts.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+class AnnualReport::MostRebloggedAccounts < AnnualReport::Source
+  SET_SIZE = 10
+
+  def generate
+    {
+      most_reblogged_accounts: most_reblogged_accounts.map do |(account_id, count)|
+                                 {
+                                   account_id: account_id,
+                                   count: count,
+                                 }
+                               end,
+    }
+  end
+
+  private
+
+  def most_reblogged_accounts
+    @account.statuses.reorder(nil).where(id: year_as_snowflake_range).where.not(reblog_of_id: nil).joins(reblog: :account).group('accounts.id').having('count(*) > 1').order(total: :desc).limit(SET_SIZE).pluck(Arel.sql('accounts.id, count(*) as total'))
+  end
+end

app/lib/annual_report/most_used_apps.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+class AnnualReport::MostUsedApps < AnnualReport::Source
+  SET_SIZE = 10
+
+  def generate
+    {
+      most_used_apps: most_used_apps.map do |(name, count)|
+                        {
+                          name: name,
+                          count: count,
+                        }
+                      end,
+    }
+  end
+
+  private
+
+  def most_used_apps
+    @account.statuses.reorder(nil).where(id: year_as_snowflake_range).joins(:application).group('oauth_applications.name').order(total: :desc).limit(SET_SIZE).pluck(Arel.sql('oauth_applications.name, count(*) as total'))
+  end
+end

app/lib/annual_report/percentiles.rb

@@ -0,0 +1,62 @@
+# frozen_string_literal: true
+
+class AnnualReport::Percentiles < AnnualReport::Source
+  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,
+      },
+    }
+  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 ||= @account.statuses.where(id: year_as_snowflake_range).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')
+  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')
+  end
+end

app/lib/annual_report/source.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+class AnnualReport::Source
+  attr_reader :account, :year
+
+  def initialize(account, year)
+    @account = account
+    @year = year
+  end
+
+  protected
+
+  def year_as_snowflake_range
+    (Mastodon::Snowflake.id_at(DateTime.new(year, 1, 1))..Mastodon::Snowflake.id_at(DateTime.new(year, 12, 31)))
+  end
+end

app/lib/annual_report/time_series.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+class AnnualReport::TimeSeries < AnnualReport::Source
+  def generate
+    {
+      time_series: (1..12).map do |month|
+                     {
+                       month: month,
+                       statuses: statuses_per_month[month] || 0,
+                       following: following_per_month[month] || 0,
+                       followers: followers_per_month[month] || 0,
+                     }
+                   end,
+    }
+  end
+
+  private
+
+  def statuses_per_month
+    @statuses_per_month ||= @account.statuses.reorder(nil).where(id: year_as_snowflake_range).group(:period).pluck(Arel.sql("date_part('month', created_at)::int AS period, count(*)")).to_h
+  end
+
+  def following_per_month
+    @following_per_month ||= @account.active_relationships.where("date_part('year', created_at) = ?", @year).group(:period).pluck(Arel.sql("date_part('month', created_at)::int AS period, count(*)")).to_h
+  end
+
+  def followers_per_month
+    @followers_per_month ||= @account.passive_relationships.where("date_part('year', created_at) = ?", @year).group(:period).pluck(Arel.sql("date_part('month', created_at)::int AS period, count(*)")).to_h
+  end
+end

app/lib/annual_report/top_hashtags.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+class AnnualReport::TopHashtags < AnnualReport::Source
+  SET_SIZE = 40
+
+  def generate
+    {
+      top_hashtags: top_hashtags.map do |(name, count)|
+                      {
+                        name: name,
+                        count: count,
+                      }
+                    end,
+    }
+  end
+
+  private
+
+  def top_hashtags
+    Tag.joins(:statuses).where(statuses: { id: @account.statuses.where(id: year_as_snowflake_range).reorder(nil).select(:id) }).group(:id).having('count(*) > 1').order(total: :desc).limit(SET_SIZE).pluck(Arel.sql('COALESCE(tags.display_name, tags.name), count(*) AS total'))
+  end
+end

app/lib/annual_report/top_statuses.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+class AnnualReport::TopStatuses < AnnualReport::Source
+  def generate
+    top_reblogs = base_scope.order(reblogs_count: :desc).first&.id
+    top_favourites = base_scope.where.not(id: top_reblogs).order(favourites_count: :desc).first&.id
+    top_replies = base_scope.where.not(id: [top_reblogs, top_favourites]).order(replies_count: :desc).first&.id
+
+    {
+      top_statuses: {
+        by_reblogs: top_reblogs,
+        by_favourites: top_favourites,
+        by_replies: top_replies,
+      },
+    }
+  end
+
+  def base_scope
+    @account.statuses.with_public_visibility.joins(:status_stat).where(id: year_as_snowflake_range).reorder(nil)
+  end
+end

app/lib/annual_report/type_distribution.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+class AnnualReport::TypeDistribution < AnnualReport::Source
+  def generate
+    {
+      type_distribution: {
+        total: base_scope.count,
+        reblogs: base_scope.where.not(reblog_of_id: nil).count,
+        replies: base_scope.where.not(in_reply_to_id: nil).where.not(in_reply_to_account_id: @account.id).count,
+        standalone: base_scope.without_replies.without_reblogs.count,
+      },
+    }
+  end
+
+  private
+
+  def base_scope
+    @account.statuses.where(id: year_as_snowflake_range)
+  end
+end