Fixes for bulk creating job

This commit is contained in:
Eugene Burmakin 2025-07-12 22:04:14 +02:00
parent 2425b2423a
commit 418df71c53
2 changed files with 156 additions and 28 deletions

View file

@ -1,27 +1,48 @@
# frozen_string_literal: true
# This job is being run on daily basis to create tracks for all users
# for the past 24 hours.
# This job is being run on daily basis to create tracks for all users.
# For each user, it starts from the end of their last track (or from their oldest point
# if no tracks exist) and processes points until the specified end_at time.
#
# To manually run for a specific time range:
# Tracks::BulkCreatingJob.perform_later(start_at: 1.week.ago, end_at: Time.current)
#
# To run for specific users only:
# Tracks::BulkCreatingJob.perform_later(user_ids: [1, 2, 3])
#
# To let the job determine start times automatically (recommended):
# Tracks::BulkCreatingJob.perform_later(end_at: Time.current)
class Tracks::BulkCreatingJob < ApplicationJob
queue_as :tracks
sidekiq_options retry: false
def perform(start_at: 1.day.ago.beginning_of_day, end_at: 1.day.ago.end_of_day, user_ids: [])
def perform(start_at: nil, end_at: 1.day.ago.end_of_day, user_ids: [])
users = user_ids.any? ? User.active.where(id: user_ids) : User.active
start_at = start_at.to_datetime
end_at = end_at.to_datetime
users.find_each do |user|
next if user.tracked_points.empty?
next unless user.tracked_points.where(timestamp: start_at.to_i..end_at.to_i).exists?
Tracks::CreateJob.perform_later(user.id, start_at: start_at, end_at: end_at, cleaning_strategy: :daily)
# Start from the end of the last track, or from the beginning if no tracks exist
user_start_at = start_at&.to_datetime || start_time(user)
next unless user.tracked_points.where(timestamp: user_start_at.to_i..end_at.to_i).exists?
Tracks::CreateJob.perform_later(user.id, start_at: user_start_at, end_at: end_at, cleaning_strategy: :daily)
end
end
private
def start_time(user)
# Find the latest track for this user
latest_track = user.tracks.order(end_at: :desc).first
if latest_track
latest_track.end_at
else
oldest_point = user.tracked_points.order(:timestamp).first
oldest_point ? Time.zone.at(oldest_point.timestamp) : 1.day.ago.beginning_of_day
end
end
end

View file

@ -20,26 +20,28 @@ RSpec.describe Tracks::BulkCreatingJob, type: :job do
create(:point, user: inactive_user, timestamp: start_at.to_i + 1.hour.to_i)
end
it 'schedules tracks creation jobs for active users with points in the timeframe' do
expect {
described_class.new.perform(start_at: start_at, end_at: end_at)
}.to have_enqueued_job(Tracks::CreateJob).with(active_user.id, start_at: start_at, end_at: end_at, cleaning_strategy: :daily)
end
context 'when explicit start_at is provided' do
it 'schedules tracks creation jobs for active users with points in the timeframe' do
expect {
described_class.new.perform(start_at: start_at, end_at: end_at)
}.to have_enqueued_job(Tracks::CreateJob).with(active_user.id, start_at: start_at, end_at: end_at, cleaning_strategy: :daily)
end
it 'does not schedule jobs for users without tracked points' do
expect {
described_class.new.perform(start_at: start_at, end_at: end_at)
}.not_to have_enqueued_job(Tracks::CreateJob).with(user_without_points.id, start_at: start_at, end_at: end_at, cleaning_strategy: :daily)
end
it 'does not schedule jobs for users without tracked points' do
expect {
described_class.new.perform(start_at: start_at, end_at: end_at)
}.not_to have_enqueued_job(Tracks::CreateJob).with(user_without_points.id, start_at: start_at, end_at: end_at, cleaning_strategy: :daily)
end
it 'does not schedule jobs for users without points in the specified timeframe' do
# Create a user with points outside the timeframe
user_with_old_points = create(:user)
create(:point, user: user_with_old_points, timestamp: 2.days.ago.to_i)
it 'does not schedule jobs for users without points in the specified timeframe' do
# Create a user with points outside the timeframe
user_with_old_points = create(:user)
create(:point, user: user_with_old_points, timestamp: 2.days.ago.to_i)
expect {
described_class.new.perform(start_at: start_at, end_at: end_at)
}.not_to have_enqueued_job(Tracks::CreateJob).with(user_with_old_points.id, start_at: start_at, end_at: end_at, cleaning_strategy: :daily)
expect {
described_class.new.perform(start_at: start_at, end_at: end_at)
}.not_to have_enqueued_job(Tracks::CreateJob).with(user_with_old_points.id, start_at: start_at, end_at: end_at, cleaning_strategy: :daily)
end
end
context 'when specific user_ids are provided' do
@ -56,17 +58,122 @@ RSpec.describe Tracks::BulkCreatingJob, type: :job do
end
end
context 'with default parameters' do
it 'uses yesterday as the default timeframe' do
context 'with automatic start time determination' do
let(:user_with_tracks) { create(:user) }
let(:user_without_tracks) { create(:user) }
let(:current_time) { Time.current }
before do
# Create some historical points and tracks for user_with_tracks
create(:point, user: user_with_tracks, timestamp: 3.days.ago.to_i)
create(:point, user: user_with_tracks, timestamp: 2.days.ago.to_i)
# Create a track ending 1 day ago
create(:track, user: user_with_tracks, end_at: 1.day.ago)
# Create newer points after the last track
create(:point, user: user_with_tracks, timestamp: 12.hours.ago.to_i)
create(:point, user: user_with_tracks, timestamp: 6.hours.ago.to_i)
# Create points for user without tracks
create(:point, user: user_without_tracks, timestamp: 2.days.ago.to_i)
create(:point, user: user_without_tracks, timestamp: 1.day.ago.to_i)
end
it 'starts from the end of the last track for users with existing tracks' do
track_end_time = user_with_tracks.tracks.order(end_at: :desc).first.end_at
expect {
described_class.new.perform
described_class.new.perform(end_at: current_time, user_ids: [user_with_tracks.id])
}.to have_enqueued_job(Tracks::CreateJob).with(
active_user.id,
start_at: 1.day.ago.beginning_of_day.to_datetime,
user_with_tracks.id,
start_at: track_end_time,
end_at: current_time.to_datetime,
cleaning_strategy: :daily
)
end
it 'starts from the oldest point for users without tracks' do
oldest_point_time = Time.zone.at(user_without_tracks.tracked_points.order(:timestamp).first.timestamp)
expect {
described_class.new.perform(end_at: current_time, user_ids: [user_without_tracks.id])
}.to have_enqueued_job(Tracks::CreateJob).with(
user_without_tracks.id,
start_at: oldest_point_time,
end_at: current_time.to_datetime,
cleaning_strategy: :daily
)
end
it 'falls back to 1 day ago for users with no points' do
expect {
described_class.new.perform(end_at: current_time)
}.not_to have_enqueued_job(Tracks::CreateJob).with(
user_without_points.id,
start_at: anything,
end_at: anything,
cleaning_strategy: :daily
)
end
end
context 'with default parameters' do
let(:user_with_recent_points) { create(:user) }
before do
# Create points within yesterday's timeframe
create(:point, user: user_with_recent_points, timestamp: 1.day.ago.beginning_of_day.to_i + 2.hours.to_i)
create(:point, user: user_with_recent_points, timestamp: 1.day.ago.beginning_of_day.to_i + 6.hours.to_i)
end
it 'uses automatic start time determination with yesterday as end_at' do
oldest_point_time = Time.zone.at(user_with_recent_points.tracked_points.order(:timestamp).first.timestamp)
expect {
described_class.new.perform(user_ids: [user_with_recent_points.id])
}.to have_enqueued_job(Tracks::CreateJob).with(
user_with_recent_points.id,
start_at: oldest_point_time,
end_at: 1.day.ago.end_of_day.to_datetime,
cleaning_strategy: :daily
)
end
end
end
describe '#start_time' do
let(:user) { create(:user) }
let(:job) { described_class.new }
context 'when user has tracks' do
let!(:old_track) { create(:track, user: user, end_at: 3.days.ago) }
let!(:recent_track) { create(:track, user: user, end_at: 1.day.ago) }
it 'returns the end time of the most recent track' do
result = job.send(:start_time, user)
expect(result).to eq(recent_track.end_at)
end
end
context 'when user has no tracks but has points' do
let!(:old_point) { create(:point, user: user, timestamp: 5.days.ago.to_i) }
let!(:recent_point) { create(:point, user: user, timestamp: 2.days.ago.to_i) }
it 'returns the timestamp of the oldest point' do
result = job.send(:start_time, user)
expect(result).to eq(Time.zone.at(old_point.timestamp))
end
end
context 'when user has no tracks and no points' do
it 'returns 1 day ago beginning of day' do
result = job.send(:start_time, user)
expect(result).to eq(1.day.ago.beginning_of_day)
end
end
end
end