Import LFS Objects as part of GitLab Migration

 - Add LFS Objects to GitLab Project Migration in order
   to have closer feature parity with file-based
   Import/Export

ee/spec/lib/ee/bulk_imports/projects/stage_spec.rb

@@ -26,6 +26,7 @@ RSpec.describe BulkImports::Projects::Stage do
       [4, BulkImports::Projects::Pipelines::ServiceDeskSettingPipeline],
       [5, BulkImports::Common::Pipelines::WikiPipeline],
       [5, BulkImports::Common::Pipelines::UploadsPipeline],
+      [5, BulkImports::Common::Pipelines::LfsObjectsPipeline],
       [5, BulkImports::Projects::Pipelines::AutoDevopsPipeline],
       [5, BulkImports::Projects::Pipelines::PipelineSchedulesPipeline],
       [6, BulkImports::Common::Pipelines::EntityFinisher]

lib/bulk_imports/common/pipelines/lfs_objects_pipeline.rb

+# frozen_string_literal: true
+
+module BulkImports
+  module Common
+    module Pipelines
+      class LfsObjectsPipeline
+        include Pipeline
+
+        def extract(_context)
+          download_service.execute
+          decompression_service.execute
+          extraction_service.execute
+
+          file_paths = Dir.glob(File.join(tmpdir, '*'))
+
+          BulkImports::Pipeline::ExtractedData.new(data: file_paths)
+        end
+
+        # rubocop: disable CodeReuse/ActiveRecord
+        def load(_context, file_path)
+          Gitlab::Utils.check_path_traversal!(file_path)
+          Gitlab::Utils.check_allowed_absolute_path!(file_path, [Dir.tmpdir])
+
+          return if tar_filepath?(file_path)
+          return if lfs_json_filepath?(file_path)
+          return if File.directory?(file_path)
+          return if File.lstat(file_path).symlink?
+
+          size = File.size(file_path)
+          oid = LfsObject.calculate_oid(file_path)
+
+          lfs_object = LfsObject.find_or_initialize_by(oid: oid, size: size)
+          lfs_object.file = File.open(file_path) unless lfs_object.file&.exists?
+          lfs_object.save! if lfs_object.changed?
+
+          repository_types(oid)&.each do |type|
+            create_lfs_objects_project(lfs_object, type)
+          end
+        end
+        # rubocop: enable CodeReuse/ActiveRecord
+
+        def after_run(_)
+          FileUtils.remove_entry(tmpdir) if Dir.exist?(tmpdir)
+        end
+
+        private
+
+        def download_service
+          BulkImports::FileDownloadService.new(
+            configuration: context.configuration,
+            relative_url: context.entity.relation_download_url_path(relation),
+            tmpdir: tmpdir,
+            filename: targz_filename
+          )
+        end
+
+        def decompression_service
+          BulkImports::FileDecompressionService.new(tmpdir: tmpdir, filename: targz_filename)
+        end
+
+        def extraction_service
+          BulkImports::ArchiveExtractionService.new(tmpdir: tmpdir, filename: tar_filename)
+        end
+
+        def lfs_json
+          @lfs_json ||= Gitlab::Json.parse(File.read(lfs_json_filepath))
+        rescue StandardError
+          raise BulkImports::Error, 'LFS Objects JSON read failed'
+        end
+
+        def tmpdir
+          @tmpdir ||= Dir.mktmpdir('bulk_imports')
+        end
+
+        def relation
+          BulkImports::FileTransfer::ProjectConfig::LFS_OBJECTS_RELATION
+        end
+
+        def tar_filename
+          "#{relation}.tar"
+        end
+
+        def targz_filename
+          "#{tar_filename}.gz"
+        end
+
+        def lfs_json_filepath?(file_path)
+          file_path == lfs_json_filepath
+        end
+
+        def tar_filepath?(file_path)
+          File.join(tmpdir, tar_filename) == file_path
+        end
+
+        def lfs_json_filepath
+          File.join(tmpdir, "#{relation}.json")
+        end
+
+        def create_lfs_objects_project(lfs_object, repository_type)
+          return unless allowed_repository_types.include?(repository_type)
+
+          lfs_objects_project = LfsObjectsProject.create(
+            project: portable,
+            lfs_object: lfs_object,
+            repository_type: repository_type
+          )
+
+          return if lfs_objects_project.persisted?
+
+          logger.warn(
+            project_id: portable.id,
+            message: 'Failed to save lfs objects project',
+            errors: lfs_objects_project.errors.full_messages.to_sentence,
+            **Gitlab::ApplicationContext.current
+          )
+        end
+
+        def repository_types(oid)
+          types = lfs_json[oid]
+
+          return [] unless types
+          return [] unless types.is_a?(Array)
+
+          # only return allowed repository types
+          types.uniq & allowed_repository_types
+        end
+
+        def allowed_repository_types
+          @allowed_repository_types ||= LfsObjectsProject.repository_types.values.push(nil)
+        end
+      end
+    end
+  end
+end

lib/bulk_imports/projects/stage.rb

@@ -87,6 +87,10 @@ module BulkImports
             pipeline: BulkImports::Common::Pipelines::UploadsPipeline,
             stage: 5
           },
+          lfs_objects: {
+            pipeline: BulkImports::Common::Pipelines::LfsObjectsPipeline,
+            stage: 5
+          },
           auto_devops: {
             pipeline: BulkImports::Projects::Pipelines::AutoDevopsPipeline,
             stage: 5

spec/lib/bulk_imports/common/pipelines/lfs_objects_pipeline_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe BulkImports::Common::Pipelines::LfsObjectsPipeline do
+  let_it_be(:portable) { create(:project) }
+  let_it_be(:oid) { 'e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855' }
+
+  let(:tmpdir) { Dir.mktmpdir }
+  let(:entity) { create(:bulk_import_entity, :project_entity, project: portable, source_full_path: 'test') }
+  let(:tracker) { create(:bulk_import_tracker, entity: entity) }
+  let(:context) { BulkImports::Pipeline::Context.new(tracker) }
+  let(:lfs_dir_path) { tmpdir }
+  let(:lfs_json_file_path) { File.join(lfs_dir_path, 'lfs_objects.json')}
+  let(:lfs_file_path) { File.join(lfs_dir_path, oid)}
+
+  subject(:pipeline) { described_class.new(context) }
+
+  before do
+    FileUtils.mkdir_p(lfs_dir_path)
+    FileUtils.touch(lfs_json_file_path)
+    FileUtils.touch(lfs_file_path)
+    File.write(lfs_json_file_path, { oid => [0, 1, 2, nil] }.to_json )
+
+    allow(Dir).to receive(:mktmpdir).with('bulk_imports').and_return(tmpdir)
+  end
+
+  after do
+    FileUtils.remove_entry(tmpdir) if Dir.exist?(tmpdir)
+  end
+
+  describe '#run' do
+    it 'imports lfs objects into destination project and removes tmpdir' do
+      allow(pipeline)
+        .to receive(:extract)
+        .and_return(BulkImports::Pipeline::ExtractedData.new(data: [lfs_json_file_path, lfs_file_path]))
+
+      pipeline.run
+
+      expect(portable.lfs_objects.count).to eq(1)
+      expect(portable.lfs_objects_projects.count).to eq(4)
+      expect(Dir.exist?(tmpdir)).to eq(false)
+    end
+  end
+
+  describe '#extract' do
+    it 'downloads & extracts lfs objects filepaths' do
+      download_service = instance_double("BulkImports::FileDownloadService")
+      decompression_service = instance_double("BulkImports::FileDecompressionService")
+      extraction_service = instance_double("BulkImports::ArchiveExtractionService")
+
+      expect(BulkImports::FileDownloadService)
+        .to receive(:new)
+        .with(
+          configuration: context.configuration,
+          relative_url: "/#{entity.pluralized_name}/test/export_relations/download?relation=lfs_objects",
+          tmpdir: tmpdir,
+          filename: 'lfs_objects.tar.gz')
+        .and_return(download_service)
+      expect(BulkImports::FileDecompressionService).to receive(:new).with(tmpdir: tmpdir, filename: 'lfs_objects.tar.gz').and_return(decompression_service)
+      expect(BulkImports::ArchiveExtractionService).to receive(:new).with(tmpdir: tmpdir, filename: 'lfs_objects.tar').and_return(extraction_service)
+
+      expect(download_service).to receive(:execute)
+      expect(decompression_service).to receive(:execute)
+      expect(extraction_service).to receive(:execute)
+
+      extracted_data = pipeline.extract(context)
+
+      expect(extracted_data.data).to contain_exactly(lfs_json_file_path, lfs_file_path)
+    end
+  end
+
+  describe '#load' do
+    before do
+      allow(pipeline)
+        .to receive(:extract)
+        .and_return(BulkImports::Pipeline::ExtractedData.new(data: [lfs_json_file_path, lfs_file_path]))
+    end
+
+    context 'when file path is lfs json' do
+      it 'returns' do
+        filepath = File.join(tmpdir, 'lfs_objects.json')
+
+        allow(Gitlab::Json).to receive(:parse).with(filepath).and_return({})
+
+        expect { pipeline.load(context, filepath) }.not_to change { portable.lfs_objects.count }
+      end
+    end
+
+    context 'when file path is tar file' do
+      it 'returns' do
+        filepath = File.join(tmpdir, 'lfs_objects.tar')
+
+        expect { pipeline.load(context, filepath) }.not_to change { portable.lfs_objects.count }
+      end
+    end
+
+    context 'when lfs json read failed' do
+      it 'raises an error' do
+        File.write(lfs_json_file_path, 'invalid json')
+
+        expect { pipeline.load(context, lfs_file_path) }.to raise_error(BulkImports::Error, 'LFS Objects JSON read failed')
+      end
+    end
+
+    context 'when file path is being traversed' do
+      it 'raises an error' do
+        expect { pipeline.load(context, File.join(tmpdir, '..')) }.to raise_error(Gitlab::Utils::PathTraversalAttackError, 'Invalid path')
+      end
+    end
+
+    context 'when file path is not under tmpdir' do
+      it 'returns' do
+        expect { pipeline.load(context, '/home/test.txt') }.to raise_error(StandardError, 'path /home/test.txt is not allowed')
+      end
+    end
+
+    context 'when file path is symlink' do
+      it 'returns' do
+        symlink = File.join(tmpdir, 'symlink')
+
+        FileUtils.ln_s(File.join(tmpdir, lfs_file_path), symlink)
+
+        expect { pipeline.load(context, symlink) }.not_to change { portable.lfs_objects.count }
+      end
+    end
+
+    context 'when path is a directory' do
+      it 'returns' do
+        expect { pipeline.load(context, Dir.tmpdir) }.not_to change { portable.lfs_objects.count }
+      end
+    end
+
+    context 'lfs objects project' do
+      context 'when lfs objects json is invalid' do
+        context 'when oid value is not Array' do
+          it 'does not create lfs objects project' do
+            File.write(lfs_json_file_path, { oid => 'test' }.to_json )
+
+            expect { pipeline.load(context, lfs_file_path) }.not_to change { portable.lfs_objects_projects.count }
+          end
+        end
+
+        context 'when oid value is nil' do
+          it 'does not create lfs objects project' do
+            File.write(lfs_json_file_path, { oid => nil }.to_json )
+
+            expect { pipeline.load(context, lfs_file_path) }.not_to change { portable.lfs_objects_projects.count }
+          end
+        end
+
+        context 'when oid value is not allowed' do
+          it 'does not create lfs objects project' do
+            File.write(lfs_json_file_path, { oid => ['invalid'] }.to_json )
+
+            expect { pipeline.load(context, lfs_file_path) }.not_to change { portable.lfs_objects_projects.count }
+          end
+        end
+
+        context 'when repository type is duplicated' do
+          it 'creates only one lfs objects project' do
+            File.write(lfs_json_file_path, { oid => [0, 0, 1, 1, 2, 2] }.to_json )
+
+            expect { pipeline.load(context, lfs_file_path) }.to change { portable.lfs_objects_projects.count }.by(3)
+          end
+        end
+      end
+
+      context 'when lfs objects project fails to be created' do
+        it 'logs the failure' do
+          allow_next_instance_of(LfsObjectsProject) do |object|
+            allow(object).to receive(:persisted?).and_return(false)
+          end
+
+          expect_next_instance_of(Gitlab::Import::Logger) do |logger|
+            expect(logger)
+              .to receive(:warn)
+              .with(project_id: portable.id,
+                    message: 'Failed to save lfs objects project',
+                    errors: '', **Gitlab::ApplicationContext.current)
+              .exactly(4).times
+          end
+
+          pipeline.load(context, lfs_file_path)
+        end
+      end
+    end
+  end
+
+  describe '#after_run' do
+    it 'removes tmpdir' do
+      allow(FileUtils).to receive(:remove_entry).and_call_original
+      expect(FileUtils).to receive(:remove_entry).with(tmpdir).and_call_original
+
+      pipeline.after_run(nil)
+
+      expect(Dir.exist?(tmpdir)).to eq(false)
+    end
+
+    context 'when tmpdir does not exist' do
+      it 'does not attempt to remove tmpdir' do
+        FileUtils.remove_entry(tmpdir)
+
+        expect(FileUtils).not_to receive(:remove_entry).with(tmpdir)
+
+        pipeline.after_run(nil)
+      end
+    end
+  end
+end

spec/lib/bulk_imports/projects/stage_spec.rb

@@ -26,6 +26,7 @@ RSpec.describe BulkImports::Projects::Stage do
       [4, BulkImports::Projects::Pipelines::ServiceDeskSettingPipeline],
       [5, BulkImports::Common::Pipelines::WikiPipeline],
       [5, BulkImports::Common::Pipelines::UploadsPipeline],
+      [5, BulkImports::Common::Pipelines::LfsObjectsPipeline],
       [5, BulkImports::Projects::Pipelines::AutoDevopsPipeline],
       [5, BulkImports::Projects::Pipelines::PipelineSchedulesPipeline],
       [6, BulkImports::Common::Pipelines::EntityFinisher]