Add sitemap generator and serve sitemap from S3

Gemfile

@@ -42,6 +42,9 @@ gem 'sentry-rails'
 gem 'sentry-ruby'
 gem 'stackprof'
 
+# Sitemap generation
+gem 'sitemap_generator'
+
 # Storext for easier json attributes, custom fork for Better Together
 gem 'storext', github: 'better-together-org/storext'
 

Gemfile.lock

@@ -58,6 +58,7 @@ PATH
       rswag (>= 2.3.1, < 2.17.0)
       ruby-openai
       simple_calendar
+      sitemap_generator
       sprockets-rails
       stackprof
       stimulus-rails (~> 1.3)
@@ -732,6 +733,8 @@ GEM
       logger (>= 1.6.2)
       rack (>= 3.1.0)
       redis-client (>= 0.23.2)
+    sitemap_generator (6.3.0)
+      builder (~> 3.0)
     simple_calendar (3.1.0)
       rails (>= 6.1)
     simplecov (0.22.0)
@@ -866,6 +869,7 @@ DEPENDENCIES
   shoulda-callback-matchers
   shoulda-matchers
   sidekiq (~> 8.0.7)
+  sitemap_generator
   simplecov
   spring
   spring-watcher-listen (~> 2.1.0)

app/controllers/better_together/sitemaps_controller.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module BetterTogether
+  # Serves the generated sitemap stored in Active Storage
+  class SitemapsController < ApplicationController
+    def show
+      sitemap = Sitemap.current(helpers.host_platform)
+      if sitemap.file.attached?
+        redirect_to sitemap.file.url, allow_other_host: true
+      else
+        head :not_found
+      end
+    end
+  end
+end

app/jobs/better_together/sitemap_refresh_job.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+require 'rake'
+
+module BetterTogether
+  # Generates the sitemap in a background job so newly published pages are included
+  class SitemapRefreshJob < ApplicationJob
+    queue_as :default
+
+    def perform
+      Rails.application.load_tasks unless Rake::Task.task_defined?('sitemap:refresh')
+      Rake::Task['sitemap:refresh'].invoke
+      Rake::Task['sitemap:refresh'].reenable
+    end
+  end
+end

app/models/better_together/page.rb

@@ -57,6 +57,8 @@ class Page < ApplicationRecord
     scope :published, -> { where.not(published_at: nil).where('published_at <= ?', Time.zone.now) }
     scope :by_publication_date, -> { order(published_at: :desc) }
 
+    after_commit :refresh_sitemap, on: %i[create update destroy]
+
     def hero_block
       @hero_block ||= blocks.where(type: 'BetterTogether::Content::Hero').with_attached_background_image_file.with_translations.first
     end
@@ -103,6 +105,12 @@ def url
 
     private
 
+    def refresh_sitemap
+      return if Rails.env.test?
+
+      SitemapRefreshJob.perform_later
+    end
+
     def add_creator_as_author
       return unless respond_to?(:creator_id) && creator_id.present?
 

app/models/better_together/platform.rb

@@ -36,6 +36,8 @@ class Platform < ApplicationRecord
     has_one_attached :profile_image
     has_one_attached :cover_image
 
+    has_one :sitemap, class_name: '::BetterTogether::Sitemap', dependent: :destroy
+
     has_many :platform_blocks, dependent: :destroy, class_name: 'BetterTogether::Content::PlatformBlock'
     has_many :blocks, through: :platform_blocks
 

app/models/better_together/sitemap.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module BetterTogether
+  # Stores the generated sitemap in Active Storage for serving via S3
+  class Sitemap < ApplicationRecord
+    belongs_to :platform
+
+    has_one_attached :file
+
+    validates :platform_id, uniqueness: true
+
+    def self.current(platform)
+      find_or_create_by!(platform: platform)
+    end
+  end
+end

app/views/layouts/better_together/application.html.erb

@@ -17,6 +17,7 @@
     <meta name="viewport" content="width=device-width, initial-scale=1.0">
     <%= csrf_meta_tags %>
     <%= csp_meta_tag %>
+    <link rel="sitemap" type="application/xml" href="<%= sitemap_path %>">
 
     <!-- Default Stylesheets -->
     <link rel="stylesheet" type="text/css" href="https://unpkg.com/[email protected]/dist/trix.css">

better_together.gemspec

@@ -66,6 +66,7 @@ Gem::Specification.new do |spec|
   spec.add_dependency 'rswag', '>= 2.3.1', '< 2.17.0'
   spec.add_dependency 'ruby-openai'
   spec.add_dependency 'simple_calendar'
+  spec.add_dependency 'sitemap_generator'
   spec.add_dependency 'sprockets-rails'
   spec.add_dependency 'stackprof'
   spec.add_dependency 'stimulus-rails', '~> 1.3'

config/routes.rb

@@ -2,7 +2,9 @@
 
 require 'sidekiq/web'
 
-BetterTogether::Engine.routes.draw do # rubocop:todo Metrics/BlockLength
+  BetterTogether::Engine.routes.draw do # rubocop:todo Metrics/BlockLength
+    get '/sitemap.xml.gz', to: 'sitemaps#show', as: :sitemap
+
   scope ':locale', # rubocop:todo Metrics/BlockLength
         locale: /#{I18n.available_locales.join('|')}/ do
     # bt base path

config/sitemap.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+SitemapGenerator::Sitemap.default_host =
+  "#{ENV.fetch('APP_PROTOCOL', 'http')}://#{ENV.fetch('APP_HOST', 'localhost:3000')}"
+
+helpers = BetterTogether::Engine.routes.url_helpers
+
+SitemapGenerator::Sitemap.create do
+  add helpers.home_page_path(locale: I18n.default_locale)
+
+  add helpers.communities_path(locale: I18n.default_locale)
+  BetterTogether::Community.find_each do |community|
+    add helpers.community_path(community, locale: I18n.default_locale), lastmod: community.updated_at
+  end
+
+  add helpers.conversations_path(locale: I18n.default_locale)
+  BetterTogether::Conversation.find_each do |conversation|
+    add helpers.conversation_path(conversation, locale: I18n.default_locale), lastmod: conversation.updated_at
+  end
+
+  add helpers.posts_path(locale: I18n.default_locale)
+  BetterTogether::Post.published.find_each do |post|
+    add helpers.post_path(post, locale: I18n.default_locale), lastmod: post.updated_at
+  end
+
+  add helpers.events_path(locale: I18n.default_locale)
+  BetterTogether::Event.find_each do |event|
+    add helpers.event_path(event, locale: I18n.default_locale), lastmod: event.updated_at
+  end
+
+  BetterTogether::Page.published.privacy_public.find_each do |page|
+    add helpers.render_page_path(path: page.slug, locale: I18n.default_locale), lastmod: page.updated_at
+  end
+end

db/migrate/20250821120000_create_better_together_sitemaps.rb

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+
+class CreateBetterTogetherSitemaps < ActiveRecord::Migration[7.1]
+  def change
+    create_bt_table :sitemaps do |t|
+      t.bt_references :platform,
+                      null: false,
+                      index: { unique: true, name: 'unique_sitemaps_platform' }
+    end
+  end
+end

lib/tasks/sitemap.rake

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+namespace :sitemap do
+  desc 'Generate sitemap and upload to Active Storage'
+  task refresh: :environment do
+    require 'sitemap_generator'
+
+    SitemapGenerator::Sitemap.public_path = Rails.root.join('tmp')
+    SitemapGenerator::Sitemap.sitemaps_path = ''
+
+    load Rails.root.join('config/sitemap.rb')
+
+    file_path = Rails.root.join('tmp', 'sitemap.xml.gz')
+    platform = BetterTogether::Platform.find_by!(host: true)
+    BetterTogether::Sitemap.current(platform).file.attach(
+      io: File.open(file_path),
+      filename: 'sitemap.xml.gz',
+      content_type: 'application/gzip'
+    )
+  end
+end
+
+begin
+  Rake::Task['assets:precompile'].enhance do
+    Rake::Task['sitemap:refresh'].invoke
+  end
+rescue RuntimeError
+  # assets:precompile may not be defined in some environments
+end

spec/dummy/db/schema.rb

@@ -1135,6 +1135,14 @@
     t.index ["reporter_id"], name: "index_better_together_reports_on_reporter_id"
   end
 
+  create_table "better_together_sitemaps", id: :uuid, default: -> { "gen_random_uuid()" }, force: :cascade do |t|
+    t.integer "lock_version", default: 0, null: false
+    t.datetime "created_at", null: false
+    t.datetime "updated_at", null: false
+    t.uuid "platform_id", null: false
+    t.index ["platform_id"], name: "unique_sitemaps_platform", unique: true
+  end
+
   create_table "better_together_resource_permissions", id: :uuid, default: -> { "gen_random_uuid()" }, force: :cascade do |t|
     t.integer "lock_version", default: 0, null: false
     t.datetime "created_at", null: false
@@ -1452,6 +1460,7 @@
   add_foreign_key "better_together_platforms", "better_together_communities", column: "community_id"
   add_foreign_key "better_together_posts", "better_together_people", column: "creator_id"
   add_foreign_key "better_together_reports", "better_together_people", column: "reporter_id"
+  add_foreign_key "better_together_sitemaps", "better_together_platforms", column: "platform_id"
   add_foreign_key "better_together_role_resource_permissions", "better_together_resource_permissions", column: "resource_permission_id"
   add_foreign_key "better_together_role_resource_permissions", "better_together_roles", column: "role_id"
   add_foreign_key "better_together_social_media_accounts", "better_together_contact_details", column: "contact_detail_id"

spec/jobs/better_together/sitemap_refresh_job_spec.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+require 'zlib'
+
+RSpec.describe BetterTogether::SitemapRefreshJob, type: :job do
+  it 'generates and attaches a sitemap' do
+    host_platform = create(:platform, :host)
+    BetterTogether::Sitemap.destroy_all
+
+    described_class.new.perform
+
+    expect(BetterTogether::Sitemap.current(host_platform).file).to be_attached
+  end
+
+  it 'includes only public pages in the sitemap' do
+    host_platform = create(:platform, :host)
+    public_page = create(:page, privacy: 'public', slug: 'public-page')
+    private_page = create(:page, privacy: 'private', slug: 'private-page')
+    BetterTogether::Sitemap.destroy_all
+
+    described_class.perform_now
+
+    data = BetterTogether::Sitemap.current(host_platform).file.download
+    xml = Zlib::GzipReader.new(StringIO.new(data)).read
+
+    expect(xml).to include(public_page.slug)
+    expect(xml).not_to include(private_page.slug)
+  end
+end

spec/requests/better_together/sitemaps_spec.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+RSpec.describe 'Sitemap', type: :request do
+  include BetterTogether::Engine.routes.url_helpers
+  include BetterTogether::DeviseSessionHelpers
+
+  let!(:host_platform) { configure_host_platform }
+
+  before do
+    host! 'www.example.com'
+    Rails.application.routes.default_url_options[:host] = 'www.example.com'
+  end
+
+  describe 'GET /sitemap.xml.gz' do
+    context 'when a sitemap is attached' do
+      it 'redirects to the file' do
+        sitemap = BetterTogether::Sitemap.current(host_platform)
+        sitemap.file.attach(io: StringIO.new('test'), filename: 'sitemap.xml.gz', content_type: 'application/gzip')
+
+        get sitemap_path
+
+        expect(response).to redirect_to(sitemap.file.url)
+      end
+    end
+
+    context 'when no sitemap exists' do
+      it 'returns not found' do
+        BetterTogether::Sitemap.current(host_platform).file.detach
+
+        get sitemap_path
+
+        expect(response).to have_http_status(:not_found)
+      end
+    end
+  end
+end