make leaderboards go vrooom (#1120)

* make leaderboards go vrooom

* goog

.gitignore

@@ -56,3 +56,6 @@ pg-dump*
 # Generated by `rails docs:generate_llms`
 /public/llms.txt
 /public/llms-full.txt
+
+rust-stats-server/target/
+scratch/

app/jobs/leaderboard_update_job.rb

@@ -29,20 +29,18 @@ class LeaderboardUpdateJob < ApplicationJob
     Rails.logger.info "Building leaderboard for #{period} on #{date}"
 
     range = LeaderboardDateRange.calculate(date, period)
+    timestamp = Time.current
+    eligible_users = User.where.not(github_uid: nil)
+                         .where.not(trust_level: User.trust_levels[:red])
 
     ActiveRecord::Base.transaction do
-      # Build the base heartbeat query
+      heartbeat_query = Heartbeat.where(user_id: eligible_users.select(:id), time: range)
-      heartbeat_query = Heartbeat.where(time: range)
+                                 .leaderboard_eligible
-                                 .with_valid_timestamps
-                                 .joins(:user)
-                                 .coding_only
-                                 .where.not(users: { github_uid: nil })
-                                 .where.not(users: { trust_level: User.trust_levels[:red] })
 
       data = heartbeat_query.group(:user_id).duration_seconds
                             .filter { |_, seconds| seconds > 60 }
 
-      streaks = Heartbeat.daily_streaks_for_users(data.keys)
+      streaks = Heartbeat.daily_streaks_for_users(data.keys, exclude_browser_time: true)
 
       entries = data.map do |user_id, seconds|
         {
@@ -50,8 +48,8 @@ class LeaderboardUpdateJob < ApplicationJob
           user_id: user_id,
           total_seconds: seconds,
           streak_count: streaks[user_id] || 0,
-          created_at: Time.current,
+          created_at: timestamp,
-          updated_at: Time.current
+          updated_at: timestamp
         }
       end
 
@@ -63,7 +61,7 @@ class LeaderboardUpdateJob < ApplicationJob
         board.entries.delete_all
       end
 
-      board.update!(finished_generating_at: Time.current)
+      board.update!(finished_generating_at: timestamp)
     end
 
     # Cache the board

app/models/concerns/heartbeatable.rb

@@ -1,9 +1,31 @@
 module Heartbeatable
   extend ActiveSupport::Concern
 
+  BROWSER_EDITORS = %w[
+    arc
+    brave
+    chrome
+    chromium
+    edge
+    firefox
+    floorp
+    librewolf
+    microsoft-edge
+    opera
+    opera-gx
+    safari
+    vivaldi
+    waterfox
+    zen
+  ].freeze
+
   included do
     # Filter heartbeats to only include those with category equal to "coding"
     scope :coding_only, -> { where(category: "coding") }
+    scope :excluding_browser_time, -> {
+      where("editor IS NULL OR LOWER(editor) NOT IN (?)", BROWSER_EDITORS)
+    }
+    scope :leaderboard_eligible, -> { coding_only.excluding_browser_time.with_valid_timestamps }
 
     # This is to prevent PG timestamp overflow errors if someones gives us a
     # heartbeat with a time that is enormously far in the future.
@@ -97,19 +119,27 @@ module Heartbeatable
       end
     end
 
-    def daily_streaks_for_users(user_ids, start_date: 31.days.ago)
+    def daily_streaks_for_users(user_ids, start_date: 31.days.ago, exclude_browser_time: false)
       return {} if user_ids.empty?
       start_date = [ start_date, 30.days.ago ].max
-      keys = user_ids.map { |id| "user_streak_#{id}" }
+      cache_prefix = exclude_browser_time ? "user_streak_without_browser" : "user_streak"
+      keys = user_ids.map { |id| "#{cache_prefix}_#{id}" }
       streak_cache = Rails.cache.read_multi(*keys)
 
-      uncached_users = user_ids.select { |id| streak_cache["user_streak_#{id}"].nil? }
+      uncached_users = user_ids.select { |id| streak_cache["#{cache_prefix}_#{id}"].nil? }
 
       if uncached_users.empty?
-        return user_ids.index_with { |id| streak_cache["user_streak_#{id}"] || 0 }
+        return user_ids.index_with { |id| streak_cache["#{cache_prefix}_#{id}"] || 0 }
       end
 
       timeout = heartbeat_timeout_duration.to_i
+      day_group_sql = "DATE_TRUNC('day', to_timestamp(time) AT TIME ZONE users.timezone)"
+      streak_diff_sql = <<~SQL.squish
+        LEAST(
+          time - LAG(time) OVER (PARTITION BY user_id, #{day_group_sql} ORDER BY time),
+          #{timeout}
+        ) as diff
+      SQL
       raw_durations = joins(:user)
         .where(user_id: uncached_users)
         .coding_only
@@ -118,9 +148,10 @@ module Heartbeatable
         .select(
           :user_id,
           "users.timezone as user_timezone",
-          Arel.sql("DATE_TRUNC('day', to_timestamp(time) AT TIME ZONE users.timezone) as day_group"),
+          Arel.sql("#{day_group_sql} as day_group"),
-          Arel.sql("LEAST(time - LAG(time) OVER (PARTITION BY user_id, DATE_TRUNC('day', to_timestamp(time) AT TIME ZONE users.timezone) ORDER BY time), #{timeout}) as diff")
+          Arel.sql(streak_diff_sql)
         )
+      raw_durations = raw_durations.excluding_browser_time if exclude_browser_time
 
       # Then aggregate the results
       daily_durations = connection.select_all(
@@ -152,7 +183,7 @@ module Heartbeatable
          }
        end
 
-      result = user_ids.index_with { |id| streak_cache["user_streak_#{id}"] || 0 }
+      result = user_ids.index_with { |id| streak_cache["#{cache_prefix}_#{id}"] || 0 }
 
       # Then calculate streaks for each user
       daily_durations.each do |user_id, data|
@@ -183,7 +214,7 @@ module Heartbeatable
         result[user_id] = streak
 
         # Cache the streak for 1 hour
-        Rails.cache.write("user_streak_#{user_id}", streak, expires_in: 1.hour)
+        Rails.cache.write("#{cache_prefix}_#{user_id}", streak, expires_in: 1.hour)
       end
 
       result

app/services/leaderboard_builder.rb

@@ -1,49 +0,0 @@
-module LeaderboardBuilder
-  module_function
-
-  def build_for_users(users, date, scope, period)
-    date = Date.current if date.blank?
-
-    board = ::Leaderboard.new(
-      start_date: date,
-      period_type: period,
-      finished_generating_at: Time.current
-    )
-
-    ids = users.pluck(:id)
-    return board if ids.empty?
-
-    users_map = users.index_by(&:id)
-
-    range = LeaderboardDateRange.calculate(date, period)
-
-    beats = Heartbeat.where(user_id: ids, time: range)
-                    .coding_only
-                    .with_valid_timestamps
-                    .joins(:user)
-                    .where.not(users: { github_uid: nil })
-
-    totals = beats.group(:user_id).duration_seconds
-    totals = totals.filter { |_, seconds| seconds > 60 }
-
-    streak_ids = totals.keys
-    streaks = streak_ids.any? ? Heartbeat.daily_streaks_for_users(streak_ids, start_date: 30.days.ago) : {}
-
-    entries = totals.map do |user_id, seconds|
-      entry = LeaderboardEntry.new(
-        leaderboard: board,
-        user_id: user_id,
-        total_seconds: seconds,
-        streak_count: streaks[user_id] || 0
-      )
-
-      entry.user = users_map[user_id]
-      entry
-    end.sort_by(&:total_seconds).reverse
-
-    board.define_singleton_method(:entries) { entries }
-    board.define_singleton_method(:scope_name) { scope }
-
-    board
-  end
-end

config/ci.rb

@@ -1,6 +1,5 @@
 CI.run do
   step "Setup: Rails", "bin/setup --skip-server"
-  step "Setup: Frontend", "bun install --frozen-lockfile"
   step "Style: Ruby", "bin/rubocop"
 
   step "Zeitwerk", "bin/rails zeitwerk:check"
@@ -19,7 +18,7 @@ CI.run do
   step "Frontend: Lint", "bun run format:svelte:check"
 
   if success?
-    step "Signoff: All systems go. Ready for merge and deploy."
+    heading "Signoff: All systems go. Ready for merge and deploy.", type: :success
   else
     failure "Signoff: CI failed. Do not merge or deploy.", "Fix the issues and try again."
   end

test/jobs/leaderboard_update_job_test.rb

@@ -0,0 +1,66 @@
+require "test_helper"
+
+class LeaderboardUpdateJobTest < ActiveJob::TestCase
+  setup do
+    Rails.cache.clear
+  end
+
+  teardown do
+    Rails.cache.clear
+  end
+
+  test "perform excludes browser editor heartbeats from persisted leaderboard entries" do
+    coded_user = create_user(username: "lb_job_coded", github_uid: "GH_LEADERBOARD_JOB_CODED")
+    browser_only_user = create_user(username: "lb_job_browser", github_uid: "GH_LEADERBOARD_JOB_BROWSER")
+
+    create_heartbeat_pair(user: coded_user, started_at: today_at(9, 0), editor: "vscode")
+    create_heartbeat_pair(user: coded_user, started_at: today_at(11, 0), editor: "firefox")
+    create_heartbeat_pair(user: browser_only_user, started_at: today_at(13, 0), editor: "firefox")
+
+    LeaderboardUpdateJob.perform_now(:daily, Date.current, force_update: true)
+
+    board = Leaderboard.find_by!(
+      start_date: Date.current,
+      period_type: :daily,
+      timezone_utc_offset: nil
+    )
+
+    assert_equal [ coded_user.id ], board.entries.order(:user_id).pluck(:user_id)
+    assert_equal 120, board.entries.find_by!(user_id: coded_user.id).total_seconds
+  end
+
+  private
+
+  def create_user(username:, github_uid:)
+    User.create!(
+      username: username,
+      github_uid: github_uid,
+      timezone: "UTC"
+    )
+  end
+
+  def create_heartbeat_pair(user:, started_at:, editor:)
+    user.heartbeats.create!(
+      entity: "src/#{editor}.rb",
+      type: "file",
+      category: "coding",
+      editor: editor,
+      time: started_at.to_f,
+      project: "leaderboard-job-test",
+      source_type: :test_entry
+    )
+    user.heartbeats.create!(
+      entity: "src/#{editor}.rb",
+      type: "file",
+      category: "coding",
+      editor: editor,
+      time: (started_at + 2.minutes).to_f,
+      project: "leaderboard-job-test",
+      source_type: :test_entry
+    )
+  end
+
+  def today_at(hour, minute)
+    Time.current.change(hour: hour, min: minute, sec: 0)
+  end
+end

test/models/heartbeat_test.rb

@@ -1,8 +1,16 @@
 require "test_helper"
 
 class HeartbeatTest < ActiveSupport::TestCase
+  setup do
+    Rails.cache.clear
+  end
+
+  teardown do
+    Rails.cache.clear
+  end
+
   test "soft delete hides record from default scope and restore brings it back" do
-    user = User.create!
+    user = User.create!(timezone: "UTC")
     heartbeat = user.heartbeats.create!(
       entity: "src/main.rb",
       type: "file",
@@ -23,4 +31,28 @@ class HeartbeatTest < ActiveSupport::TestCase
 
     assert_includes Heartbeat.all, heartbeat
   end
+
+  test "daily streak cache is separated for browser-filtered leaderboard streaks" do
+    user = User.create!(timezone: "UTC", username: "hb_streak_cache")
+    create_heartbeat_sequence(user: user, started_at: Time.current.beginning_of_day + 9.hours, editor: "firefox")
+
+    assert_equal 1, Heartbeat.daily_streaks_for_users([ user.id ])[user.id]
+    assert_equal 0, Heartbeat.daily_streaks_for_users([ user.id ], exclude_browser_time: true)[user.id]
+  end
+
+  private
+
+  def create_heartbeat_sequence(user:, started_at:, editor:, count: 9)
+    count.times do |offset|
+      user.heartbeats.create!(
+        entity: "src/#{editor}.rb",
+        type: "file",
+        category: "coding",
+        editor: editor,
+        time: (started_at + (offset * 2).minutes).to_f,
+        project: "heartbeat-test",
+        source_type: :test_entry
+      )
+    end
+  end
 end