Add Project Uploads Migration to BulkImports

  - Add Project Uploads Migration to GitLab Migration
    feature in order to export/import all project uploads
    from source to destination GitLab instance

app/models/bulk_imports/entity.rb

@@ -112,6 +112,10 @@ class BulkImports::Entity < ApplicationRecord
     @export_relations_url_path ||= EXPORT_RELATIONS_URL % { resource: pluralized_name, full_path: encoded_source_full_path }
   end
 
+  def relation_download_url_path(relation)
+    "#{export_relations_url_path}/download?relation=#{relation}"
+  end
+
   private
 
   def validate_parent_is_a_group

app/services/bulk_imports/file_download_service.rb

@@ -7,8 +7,16 @@ module BulkImports
 
     REMOTE_FILENAME_PATTERN = %r{filename="(?<filename>[^"]+)"}.freeze
     FILENAME_SIZE_LIMIT = 255 # chars before the extension
-
-    def initialize(configuration:, relative_url:, dir:, file_size_limit:, allowed_content_types:, filename: nil)
+    DEFAULT_FILE_SIZE_LIMIT = 5.gigabytes
+    DEFAULT_ALLOWED_CONTENT_TYPES = %w(application/gzip application/octet-stream).freeze
+
+    def initialize(
+      configuration:,
+      relative_url:,
+      dir:,
+      file_size_limit: DEFAULT_FILE_SIZE_LIMIT,
+      allowed_content_types: DEFAULT_ALLOWED_CONTENT_TYPES,
+      filename: nil)
       @configuration = configuration
       @relative_url = relative_url
       @filename = filename

lib/bulk_imports/common/extractors/ndjson_extractor.rb

@@ -7,10 +7,6 @@ module BulkImports
         include Gitlab::ImportExport::CommandLineUtil
         include Gitlab::Utils::StrongMemoize
 
-        FILE_SIZE_LIMIT = 5.gigabytes
-        ALLOWED_CONTENT_TYPES = %w(application/gzip application/octet-stream).freeze
-        EXPORT_DOWNLOAD_URL_PATH = "/%{resource}/%{full_path}/export_relations/download?relation=%{relation}"
-
         def initialize(relation:)
           @relation = relation
           @tmp_dir = Dir.mktmpdir
@@ -39,33 +35,19 @@ module BulkImports
         def download_service(tmp_dir, context)
           @download_service ||= BulkImports::FileDownloadService.new(
             configuration: context.configuration,
-            relative_url: relative_resource_url(context),
+            relative_url: context.entity.relation_download_url_path(relation),
             dir: tmp_dir,
-            filename: filename,
-            file_size_limit: FILE_SIZE_LIMIT,
-            allowed_content_types: ALLOWED_CONTENT_TYPES
+            filename: filename
           )
         end
 
         def decompression_service(tmp_dir)
-          @decompression_service ||= BulkImports::FileDecompressionService.new(
-            dir: tmp_dir,
-            filename: filename
-          )
+          @decompression_service ||= BulkImports::FileDecompressionService.new(dir: tmp_dir, filename: filename)
         end
 
         def ndjson_reader(tmp_dir)
           @ndjson_reader ||= Gitlab::ImportExport::Json::NdjsonReader.new(tmp_dir)
         end
-
-        def relative_resource_url(context)
-          strong_memoize(:relative_resource_url) do
-            resource = context.entity.pluralized_name
-            encoded_full_path = context.entity.encoded_source_full_path
-
-            EXPORT_DOWNLOAD_URL_PATH % { resource: resource, full_path: encoded_full_path, relation: relation }
-          end
-        end
       end
     end
   end

lib/bulk_imports/common/pipelines/uploads_pipeline.rb

+# frozen_string_literal: true
+
+module BulkImports
+  module Common
+    module Pipelines
+      class UploadsPipeline
+        include Pipeline
+        include Gitlab::ImportExport::CommandLineUtil
+
+        FILENAME = 'uploads.tar.gz'
+
+        def extract(context)
+          download_service(tmp_dir, context).execute
+          untar_zxf(archive: File.join(tmp_dir, FILENAME), dir: tmp_dir)
+          upload_file_paths = Dir.glob(File.join(tmp_dir, '**', '*'))
+
+          BulkImports::Pipeline::ExtractedData.new(data: upload_file_paths)
+        end
+
+        def load(context, file_path)
+          dynamic_path = FileUploader.extract_dynamic_path(file_path)
+
+          return unless dynamic_path
+          return if File.directory?(file_path)
+
+          named_captures = dynamic_path.named_captures.symbolize_keys
+
+          UploadService.new(context.portable, File.open(file_path, 'r'), FileUploader, **named_captures).execute
+        end
+
+        def after_run(_)
+          FileUtils.remove_entry(tmp_dir)
+        end
+
+        private
+
+        def download_service(tmp_dir, context)
+          BulkImports::FileDownloadService.new(
+            configuration: context.configuration,
+            relative_url: context.entity.relation_download_url_path('uploads'),
+            dir: tmp_dir,
+            filename: FILENAME
+          )
+        end
+
+        def tmp_dir
+          @tmp_dir ||= Dir.mktmpdir('bulk_imports')
+        end
+      end
+    end
+  end
+end

lib/bulk_imports/projects/stage.rb

@@ -27,9 +27,13 @@ module BulkImports
             pipeline: BulkImports::Common::Pipelines::BoardsPipeline,
             stage: 4
           },
+          uploads: {
+            pipeline: BulkImports::Common::Pipelines::UploadsPipeline,
+            stage: 5
+          },
           finisher: {
             pipeline: BulkImports::Common::Pipelines::EntityFinisher,
-            stage: 5
+            stage: 6
           }
         }
       end

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

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe BulkImports::Common::Pipelines::UploadsPipeline do
+  let_it_be(:tmpdir) { Dir.mktmpdir }
+  let_it_be(:project) { create(:project) }
+  let_it_be(:entity) { create(:bulk_import_entity, :project_entity, project: project, source_full_path: 'test') }
+  let_it_be(:tracker) { create(:bulk_import_tracker, entity: entity) }
+  let_it_be(:context) { BulkImports::Pipeline::Context.new(tracker) }
+  let_it_be(:uploads_dir_path) { File.join(tmpdir, '72a497a02fe3ee09edae2ed06d390038') }
+  let_it_be(:upload_file_path) { File.join(uploads_dir_path, 'upload.txt')}
+
+  subject(:pipeline) { described_class.new(context) }
+
+  before do
+    stub_uploads_object_storage(FileUploader)
+
+    FileUtils.mkdir_p(uploads_dir_path)
+    FileUtils.touch(upload_file_path)
+  end
+
+  after do
+    FileUtils.remove_entry(tmpdir) if Dir.exist?(tmpdir)
+  end
+
+  describe '#run' do
+    it 'imports uploads into destination portable and removes tmpdir' do
+      allow(Dir).to receive(:mktmpdir).with('bulk_imports').and_return(tmpdir)
+      allow(pipeline).to receive(:extract).and_return(BulkImports::Pipeline::ExtractedData.new(data: [upload_file_path]))
+
+      pipeline.run
+
+      expect(project.uploads.map { |u| u.retrieve_uploader.filename }).to include('upload.txt')
+
+      expect(Dir.exist?(tmpdir)).to eq(false)
+    end
+  end
+
+  describe '#extract' do
+    it 'downloads & extracts upload paths' do
+      allow(Dir).to receive(:mktmpdir).and_return(tmpdir)
+      expect(pipeline).to receive(:untar_zxf)
+      file_download_service = instance_double("BulkImports::FileDownloadService")
+
+      expect(BulkImports::FileDownloadService)
+        .to receive(:new)
+        .with(
+          configuration: context.configuration,
+          relative_url: "/projects/test/export_relations/download?relation=uploads",
+          dir: tmpdir,
+          filename: 'uploads.tar.gz')
+        .and_return(file_download_service)
+
+      expect(file_download_service).to receive(:execute)
+
+      extracted_data = pipeline.extract(context)
+
+      expect(extracted_data.data).to contain_exactly(uploads_dir_path, upload_file_path)
+    end
+  end
+
+  describe '#load' do
+    it 'creates a file upload' do
+      expect { pipeline.load(context, upload_file_path) }.to change { project.uploads.count }.by(1)
+    end
+
+    context 'when dynamic path is nil' do
+      it 'returns' do
+        expect { pipeline.load(context, File.join(tmpdir, 'test')) }.not_to change { project.uploads.count }
+      end
+    end
+
+    context 'when path is a directory' do
+      it 'returns' do
+        expect { pipeline.load(context, uploads_dir_path) }.not_to change { project.uploads.count }
+      end
+    end
+  end
+end

spec/lib/bulk_imports/projects/stage_spec.rb

@@ -10,7 +10,8 @@ RSpec.describe BulkImports::Projects::Stage do
       [2, BulkImports::Common::Pipelines::LabelsPipeline],
       [3, BulkImports::Projects::Pipelines::IssuesPipeline],
       [4, BulkImports::Common::Pipelines::BoardsPipeline],
-      [5, BulkImports::Common::Pipelines::EntityFinisher]
+      [5, BulkImports::Common::Pipelines::UploadsPipeline],
+      [6, BulkImports::Common::Pipelines::EntityFinisher]
     ]
   end
 

spec/models/bulk_imports/entity_spec.rb

@@ -243,4 +243,13 @@ RSpec.describe BulkImports::Entity, type: :model do
       end
     end
   end
+
+  describe '#relation_download_url_path' do
+    it 'returns export relations url with download query string' do
+      entity = build(:bulk_import_entity)
+
+      expect(entity.relation_download_url_path('test'))
+        .to eq("/groups/#{entity.encoded_source_full_path}/export_relations/download?relation=test")
+    end
+  end
 end