Optimize archive export service and export zip files instead of gzipped tar files (#23360)

Gemfile

@@ -162,3 +162,4 @@ gem 'xorcist', '~> 1.1'
 gem 'cocoon', '~> 1.2'
 
 gem 'net-http', '~> 0.3.2'
+gem 'rubyzip', '~> 2.3'

Gemfile.lock

@@ -636,6 +636,7 @@ GEM
       nokogiri (>= 1.10.5)
       rexml
     ruby2_keywords (0.0.5)
+    rubyzip (2.3.2)
     rufus-scheduler (3.8.2)
       fugit (~> 1.1, >= 1.1.6)
     safety_net_attestation (0.4.0)
@@ -876,6 +877,7 @@ DEPENDENCIES
   rubocop-rails
   rubocop-rspec
   ruby-progressbar (~> 1.13)
+  rubyzip (~> 2.3)
   sanitize (~> 6.0)
   scenic (~> 1.7)
   sidekiq (~> 6.5)

app/services/backup_service.rb

@@ -1,59 +1,67 @@
 # frozen_string_literal: true
 
-require 'rubygems/package'
+require 'zip'
 
 class BackupService < BaseService
   include Payloadable
+  include ContextHelper
 
-  attr_reader :account, :backup, :collection
+  attr_reader :account, :backup
 
   def call(backup)
     @backup  = backup
     @account = backup.user.account
 
-    build_json!
     build_archive!
   end
 
   private
 
-  def build_json!
+  def build_outbox_json!(file)
-    @collection = serialize(collection_presenter, ActivityPub::CollectionSerializer)
+    skeleton = serialize(collection_presenter, ActivityPub::CollectionSerializer)
+    skeleton[:@context] = full_context
+    skeleton[:orderedItems] = ['!PLACEHOLDER!']
+    skeleton = Oj.dump(skeleton)
+    prepend, append = skeleton.split('"!PLACEHOLDER!"')
+    add_comma = false
+
+    file.write(prepend)
 
     account.statuses.with_includes.reorder(nil).find_in_batches do |statuses|
-      statuses.each do |status|
+      file.write(',') if add_comma
-        item = serialize_payload(ActivityPub::ActivityPresenter.from_status(status), ActivityPub::ActivitySerializer, signer: @account)
+      add_comma = true
-        item.delete(:@context)
+
+      file.write(statuses.map do |status|
+        item = serialize_payload(ActivityPub::ActivityPresenter.from_status(status), ActivityPub::ActivitySerializer)
+        item.delete('@context')
 
         unless item[:type] == 'Announce' || item[:object][:attachment].blank?
           item[:object][:attachment].each do |attachment|
-            attachment[:url] = Addressable::URI.parse(attachment[:url]).path.gsub(/\A\/system\//, '')
+            attachment[:url] = Addressable::URI.parse(attachment[:url]).path.delete_prefix('/system/')
           end
         end
 
-        @collection[:orderedItems] << item
+        Oj.dump(item)
-      end
+      end.join(','))
 
       GC.start
     end
+
+    file.write(append)
   end
 
   def build_archive!
-    tmp_file = Tempfile.new(%w(archive .tar.gz))
+    tmp_file = Tempfile.new(%w(archive .zip))
 
-    File.open(tmp_file, 'wb') do |file|
+    Zip::File.open(tmp_file, create: true) do |zipfile|
-      Zlib::GzipWriter.wrap(file) do |gz|
+      dump_outbox!(zipfile)
-        Gem::Package::TarWriter.new(gz) do |tar|
+      dump_media_attachments!(zipfile)
-          dump_media_attachments!(tar)
+      dump_likes!(zipfile)
-          dump_outbox!(tar)
+      dump_bookmarks!(zipfile)
-          dump_likes!(tar)
+      dump_actor!(zipfile)
-          dump_bookmarks!(tar)
-          dump_actor!(tar)
-        end
-      end
     end
 
-    archive_filename = "#{['archive', Time.now.utc.strftime('%Y%m%d%H%M%S'), SecureRandom.hex(16)].join('-')}.tar.gz"
+    archive_filename = "#{['archive', Time.now.utc.strftime('%Y%m%d%H%M%S'), SecureRandom.hex(16)].join('-')}.zip"
 
     @backup.dump      = ActionDispatch::Http::UploadedFile.new(tempfile: tmp_file, filename: archive_filename)
     @backup.processed = true
@@ -63,27 +71,28 @@ class BackupService < BaseService
     tmp_file.unlink
   end
 
-  def dump_media_attachments!(tar)
+  def dump_media_attachments!(zipfile)
     MediaAttachment.attached.where(account: account).reorder(nil).find_in_batches do |media_attachments|
       media_attachments.each do |m|
-        next unless m.file&.path
+        path = m.file&.path
+        next unless path
 
-        download_to_tar(tar, m.file, m.file.path)
+        path = path.gsub(/\A.*\/system\//, '')
+        path = path.gsub(/\A\/+/, '')
+        download_to_zip(zipfile, m.file, path)
       end
 
       GC.start
     end
   end
 
-  def dump_outbox!(tar)
+  def dump_outbox!(zipfile)
-    json = Oj.dump(collection)
+    zipfile.get_output_stream('outbox.json') do |io|
-
+      build_outbox_json!(io)
-    tar.add_file_simple('outbox.json', 0o444, json.bytesize) do |io|
-      io.write(json)
     end
   end
 
-  def dump_actor!(tar)
+  def dump_actor!(zipfile)
     actor = serialize(account, ActivityPub::ActorSerializer)
 
     actor[:icon][:url]  = "avatar#{File.extname(actor[:icon][:url])}"  if actor[:icon]
@@ -92,51 +101,66 @@ class BackupService < BaseService
     actor[:likes]       = 'likes.json'
     actor[:bookmarks]   = 'bookmarks.json'
 
-    download_to_tar(tar, account.avatar, "avatar#{File.extname(account.avatar.path)}") if account.avatar.exists?
+    download_to_zip(tar, account.avatar, "avatar#{File.extname(account.avatar.path)}") if account.avatar.exists?
-    download_to_tar(tar, account.header, "header#{File.extname(account.header.path)}") if account.header.exists?
+    download_to_zip(tar, account.header, "header#{File.extname(account.header.path)}") if account.header.exists?
 
     json = Oj.dump(actor)
 
-    tar.add_file_simple('actor.json', 0o444, json.bytesize) do |io|
+    zipfile.get_output_stream('actor.json') do |io|
       io.write(json)
     end
   end
 
-  def dump_likes!(tar)
+  def dump_likes!(zipfile)
-    collection = serialize(ActivityPub::CollectionPresenter.new(id: 'likes.json', type: :ordered, size: 0, items: []), ActivityPub::CollectionSerializer)
+    skeleton = serialize(ActivityPub::CollectionPresenter.new(id: 'likes.json', type: :ordered, size: 0, items: []), ActivityPub::CollectionSerializer)
+    skeleton.delete(:totalItems)
+    skeleton[:orderedItems] = ['!PLACEHOLDER!']
+    skeleton = Oj.dump(skeleton)
+    prepend, append = skeleton.split('"!PLACEHOLDER!"')
 
-    Status.reorder(nil).joins(:favourites).includes(:account).merge(account.favourites).find_in_batches do |statuses|
+    zipfile.get_output_stream('likes.json') do |io|
-      statuses.each do |status|
+      io.write(prepend)
-        collection[:totalItems] += 1
+
-        collection[:orderedItems] << ActivityPub::TagManager.instance.uri_for(status)
+      add_comma = false
+
+      Status.reorder(nil).joins(:favourites).includes(:account).merge(account.favourites).find_in_batches do |statuses|
+        io.write(',') if add_comma
+        add_comma = true
+
+        io.write(statuses.map do |status|
+          Oj.dump(ActivityPub::TagManager.instance.uri_for(status))
+        end.join(','))
+
+        GC.start
       end
 
-      GC.start
+      io.write(append)
-    end
-
-    json = Oj.dump(collection)
-
-    tar.add_file_simple('likes.json', 0o444, json.bytesize) do |io|
-      io.write(json)
     end
   end
 
-  def dump_bookmarks!(tar)
+  def dump_bookmarks!(zipfile)
-    collection = serialize(ActivityPub::CollectionPresenter.new(id: 'bookmarks.json', type: :ordered, size: 0, items: []), ActivityPub::CollectionSerializer)
+    skeleton = serialize(ActivityPub::CollectionPresenter.new(id: 'bookmarks.json', type: :ordered, size: 0, items: []), ActivityPub::CollectionSerializer)
+    skeleton.delete(:totalItems)
+    skeleton[:orderedItems] = ['!PLACEHOLDER!']
+    skeleton = Oj.dump(skeleton)
+    prepend, append = skeleton.split('"!PLACEHOLDER!"')
 
-    Status.reorder(nil).joins(:bookmarks).includes(:account).merge(account.bookmarks).find_in_batches do |statuses|
+    zipfile.get_output_stream('bookmarks.json') do |io|
-      statuses.each do |status|
+      io.write(prepend)
-        collection[:totalItems] += 1
+
-        collection[:orderedItems] << ActivityPub::TagManager.instance.uri_for(status)
+      add_comma = false
+      Status.reorder(nil).joins(:bookmarks).includes(:account).merge(account.bookmarks).find_in_batches do |statuses|
+        io.write(',') if add_comma
+        add_comma = true
+
+        io.write(statuses.map do |status|
+          Oj.dump(ActivityPub::TagManager.instance.uri_for(status))
+        end.join(','))
+
+        GC.start
       end
 
-      GC.start
+      io.write(append)
-    end
-
-    json = Oj.dump(collection)
-
-    tar.add_file_simple('bookmarks.json', 0o444, json.bytesize) do |io|
-      io.write(json)
     end
   end
 
@@ -159,10 +183,10 @@ class BackupService < BaseService
 
   CHUNK_SIZE = 1.megabyte
 
-  def download_to_tar(tar, attachment, filename)
+  def download_to_zip(zipfile, attachment, filename)
     adapter = Paperclip.io_adapters.for(attachment)
 
-    tar.add_file_simple(filename, 0o444, adapter.size) do |io|
+    zipfile.get_output_stream(filename) do |io|
       while (buffer = adapter.read(CHUNK_SIZE))
         io.write(buffer)
       end

spec/services/backup_service_spec.rb

@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+RSpec.describe BackupService, type: :service do
+  subject(:service_call) { described_class.new.call(backup) }
+
+  let!(:user)           { Fabricate(:user) }
+  let!(:attachment)     { Fabricate(:media_attachment, account: user.account) }
+  let!(:status)         { Fabricate(:status, account: user.account, text: 'Hello', visibility: :public, media_attachments: [attachment]) }
+  let!(:private_status) { Fabricate(:status, account: user.account, text: 'secret', visibility: :private) }
+  let!(:favourite)      { Fabricate(:favourite, account: user.account) }
+  let!(:bookmark)       { Fabricate(:bookmark, account: user.account) }
+  let!(:backup)         { Fabricate(:backup, user: user) }
+
+  def read_zip_file(backup, filename)
+    file = Paperclip.io_adapters.for(backup.dump)
+    Zip::File.open(file) do |zipfile|
+      entry = zipfile.glob(filename).first
+      return entry.get_input_stream.read
+    end
+  end
+
+  it 'marks the backup as processed' do
+    expect { service_call }.to change(backup, :processed).from(false).to(true)
+  end
+
+  it 'exports outbox.json as expected' do
+    service_call
+
+    json = Oj.load(read_zip_file(backup, 'outbox.json'))
+    expect(json['@context']).to_not be_nil
+    expect(json['type']).to eq 'OrderedCollection'
+    expect(json['totalItems']).to eq 2
+    expect(json['orderedItems'][0]['@context']).to be_nil
+    expect(json['orderedItems'][0]).to include({
+      'type' => 'Create',
+      'object' => include({
+        'id' => ActivityPub::TagManager.instance.uri_for(status),
+        'content' => '<p>Hello</p>',
+      }),
+    })
+    expect(json['orderedItems'][1]).to include({
+      'type' => 'Create',
+      'object' => include({
+        'id' => ActivityPub::TagManager.instance.uri_for(private_status),
+        'content' => '<p>secret</p>',
+      }),
+    })
+  end
+
+  it 'exports likes.json as expected' do
+    service_call
+
+    json = Oj.load(read_zip_file(backup, 'likes.json'))
+    expect(json['type']).to eq 'OrderedCollection'
+    expect(json['orderedItems']).to eq [ActivityPub::TagManager.instance.uri_for(favourite.status)]
+  end
+
+  it 'exports bookmarks.json as expected' do
+    service_call
+
+    json = Oj.load(read_zip_file(backup, 'bookmarks.json'))
+    expect(json['type']).to eq 'OrderedCollection'
+    expect(json['orderedItems']).to eq [ActivityPub::TagManager.instance.uri_for(bookmark.status)]
+  end
+end