Merge branch '31101-project-export-with-ndjson' into 'master'

Introduce ndjson writer

See merge request gitlab-org/gitlab!26995

ee/spec/lib/ee/gitlab/import_export/project/tree_saver_spec.rb

@@ -3,50 +3,55 @@
 require 'spec_helper'
 
 describe Gitlab::ImportExport::Project::TreeSaver do
-  describe 'saves the project tree into a json object' do
-    let_it_be(:user) { create(:user) }
-    let_it_be(:group) { create(:group) }
-    let_it_be(:project) { create(:project, group: group) }
-    let_it_be(:issue) { create(:issue, project: project) }
-
-    let_it_be(:design) { create(:design, :with_file, versions_count: 2, issue: issue) }
-    let_it_be(:note) { create(:diff_note_on_design, noteable: design, project: project, author: user) }
-    let_it_be(:note2) { create(:note, noteable: issue, project: project, author: user) }
-
-    let_it_be(:epic) { create(:epic, group: group) }
-    let_it_be(:epic_issue) { create(:epic_issue, issue: issue, epic: epic) }
-
-    let(:shared) { project.import_export_shared }
-    let(:export_path) { "#{Dir.tmpdir}/project_tree_saver_spec_ee" }
-    let(:project_tree_saver) { described_class.new(project: project, current_user: user, shared: shared) }
-    let(:saved_project_json) do
-      project_tree_saver.save
-      project_json(project_tree_saver.full_path)
-    end
+  let_it_be(:user) { create(:user) }
+  let_it_be(:group) { create(:group) }
+  let_it_be(:project) { create(:project, group: group) }
+  let_it_be(:issue) { create(:issue, project: project) }
+  let_it_be(:shared) { project.import_export_shared }
 
-    before do
-      project.add_maintainer(user)
-    end
+  let_it_be(:design) { create(:design, :with_file, versions_count: 2, issue: issue) }
+  let_it_be(:note) { create(:diff_note_on_design, noteable: design, project: project, author: user) }
+  let_it_be(:note2) { create(:note, noteable: issue, project: project, author: user) }
+
+  let_it_be(:epic) { create(:epic, group: group) }
+  let_it_be(:epic_issue) { create(:epic_issue, issue: issue, epic: epic) }
+
+  let_it_be(:export_path) { "#{Dir.tmpdir}/project_tree_saver_spec_ee" }
+
+  after :all do
+    FileUtils.rm_rf(export_path)
+  end
+
+  shared_examples 'EE saves project tree successfully' do |ndjson_enabled|
+    include ::ImportExport::CommonUtil
 
-    after do
-      FileUtils.rm_rf(export_path)
+    let_it_be(:project_tree_saver) { described_class.new(project: project, current_user: user, shared: shared) }
+
+    let_it_be(:full_path) do
+      if ndjson_enabled
+        File.join(shared.export_path, 'tree')
+      else
+        File.join(shared.export_path, Gitlab::ImportExport.project_filename)
+      end
     end
 
-    it 'saves successfully' do
+    let_it_be(:exportable_path) { 'project' }
+
+    before_all do
+      Feature.enable(:project_export_as_ndjson) if ndjson_enabled
+      project.add_maintainer(user)
       expect(project_tree_saver.save).to be true
     end
 
-    describe 'the designs json' do
-      let(:issue_json) { saved_project_json['issues'].first }
+    let_it_be(:issue_json) { get_json(full_path, exportable_path, :issues, ndjson_enabled).first }
 
+    describe 'the designs json' do
       it 'saves issue.designs correctly' do
         expect(issue_json['designs'].size).to eq(1)
       end
 
       it 'saves issue.design_versions correctly' do
-        actions = issue_json['design_versions'].map do |v|
-          v['actions']
-        end.flatten
+        actions = issue_json['design_versions'].flat_map { |v| v['actions'] }
 
         expect(issue_json['design_versions'].size).to eq(2)
         issue_json['design_versions'].each do |version|
@@ -61,25 +66,29 @@ describe Gitlab::ImportExport::Project::TreeSaver do
 
     context 'epics' do
       it 'has epic_issue' do
-        expect(saved_project_json['issues'].first['epic_issue']).not_to be_empty
-        expect(saved_project_json['issues'].first['epic_issue']['id']).to eql(epic_issue.id)
+        expect(issue_json['epic_issue']).not_to be_empty
+        expect(issue_json['epic_issue']['id']).to eql(epic_issue.id)
       end
 
       it 'has epic' do
-        expect(saved_project_json['issues'].first['epic_issue']['epic']['title']).to eql(epic.title)
+        expect(issue_json['epic_issue']['epic']['title']).to eql(epic.title)
       end
 
       it 'does not have epic_id' do
-        expect(saved_project_json['issues'].first['epic_issue']['epic_id']).to be_nil
+        expect(issue_json['epic_issue']['epic_id']).to be_nil
       end
 
       it 'does not have issue_id' do
-        expect(saved_project_json['issues'].first['epic_issue']['issue_id']).to be_nil
+        expect(issue_json['epic_issue']['issue_id']).to be_nil
       end
     end
   end
 
-  def project_json(filename)
-    ::JSON.parse(IO.read(filename))
+  context 'with JSON' do
+    it_behaves_like "EE saves project tree successfully", false
+  end
+
+  context 'with NDJSON' do
+    it_behaves_like "EE saves project tree successfully", true
   end
 end

lib/gitlab/import_export/json/ndjson_writer.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module ImportExport
+    module JSON
+      class NdjsonWriter
+        include Gitlab::ImportExport::CommandLineUtil
+
+        def initialize(dir_path)
+          @dir_path = dir_path
+        end
+
+        def close
+        end
+
+        def write_attributes(exportable_path, hash)
+          # It will create:
+          # tree/project.json
+          with_file("#{exportable_path}.json") do |file|
+            file.write(hash.to_json)
+          end
+        end
+
+        def write_relation(exportable_path, relation, value)
+          # It will create:
+          # tree/project/ci_cd_setting.ndjson
+          with_file(exportable_path, "#{relation}.ndjson") do |file|
+            file.write(value.to_json)
+          end
+        end
+
+        def write_relation_array(exportable_path, relation, items)
+          # It will create:
+          # tree/project/merge_requests.ndjson
+          with_file(exportable_path, "#{relation}.ndjson") do |file|
+            items.each do |item|
+              file.write(item.to_json)
+              file.write("\n")
+            end
+          end
+        end
+
+        private
+
+        def with_file(*path)
+          file_path = File.join(@dir_path, *path)
+          raise ArgumentError, "The #{file_path} already exist" if File.exist?(file_path)
+
+          # ensure that path is created
+          mkdir_p(File.dirname(file_path))
+
+          File.open(file_path, "wb") do |file|
+            yield(file)
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/import_export/project/tree_saver.rb

@@ -11,15 +11,9 @@ module Gitlab
           @project      = project
           @current_user = current_user
           @shared       = shared
-          @full_path    = File.join(@shared.export_path, ImportExport.project_filename)
         end
 
         def save
-          json_writer = ImportExport::JSON::LegacyWriter.new(
-            @full_path,
-            allowed_path: "project"
-          )
-
           ImportExport::JSON::StreamingSerializer.new(
             exportable,
             reader.project_tree,
@@ -57,6 +51,18 @@ module Gitlab
         def presenter_class
           Projects::ImportExport::ProjectExportPresenter
         end
+
+        def json_writer
+          @json_writer ||= begin
+            if ::Feature.enabled?(:project_export_as_ndjson, @project.namespace)
+              full_path = File.join(@shared.export_path, 'tree')
+              Gitlab::ImportExport::JSON::NdjsonWriter.new(full_path)
+            else
+              full_path = File.join(@shared.export_path, ImportExport.project_filename)
+              Gitlab::ImportExport::JSON::LegacyWriter.new(full_path, allowed_path: 'project')
+            end
+          end
+        end
       end
     end
   end

spec/features/projects/import_export/export_file_spec.rb

@@ -38,51 +38,109 @@ describe 'Import/Export - project export integration test', :js do
       sign_in(user)
     end
 
-    it 'exports a project successfully', :sidekiq_might_not_need_inline do
-      visit edit_project_path(project)
+    shared_examples 'export file without sensitive words' do
+      it 'exports a project successfully', :sidekiq_inline do
+        export_project_and_download_file(page, project)
 
-      expect(page).to have_content('Export project')
+        in_directory_with_expanded_export(project) do |exit_status, tmpdir|
+          expect(exit_status).to eq(0)
 
-      find(:link, 'Export project').send_keys(:return)
+          project_json_path = File.join(tmpdir, 'project.json')
+          expect(File).to exist(project_json_path)
 
-      visit edit_project_path(project)
+          project_hash = JSON.parse(IO.read(project_json_path))
 
-      expect(page).to have_content('Download export')
+          sensitive_words.each do |sensitive_word|
+            found = find_sensitive_attributes(sensitive_word, project_hash)
 
-      expect(project.export_status).to eq(:finished)
-      expect(project.export_file.path).to include('tar.gz')
+            expect(found).to be_nil, failure_message(found.try(:key_found), found.try(:parent), sensitive_word)
+          end
+        end
+      end
+    end
+
+    context "with legacy export" do
+      before do
+        stub_feature_flags(streaming_serializer: false)
+        stub_feature_flags(project_export_as_ndjson: false)
+      end
+
+      it_behaves_like "export file without sensitive words"
+    end
+
+    context "with streaming serializer" do
+      before do
+        stub_feature_flags(streaming_serializer: true)
+        stub_feature_flags(project_export_as_ndjson: false)
+      end
+
+      it_behaves_like "export file without sensitive words"
+    end
 
-      in_directory_with_expanded_export(project) do |exit_status, tmpdir|
-        expect(exit_status).to eq(0)
+    context "with ndjson" do
+      before do
+        stub_feature_flags(streaming_serializer: true)
+        stub_feature_flags(project_export_as_ndjson: true)
+      end
+
+      it 'exports a project successfully', :sidekiq_inline do
+        export_project_and_download_file(page, project)
+
+        in_directory_with_expanded_export(project) do |exit_status, tmpdir|
+          expect(exit_status).to eq(0)
 
-        project_json_path = File.join(tmpdir, 'project.json')
-        expect(File).to exist(project_json_path)
+          project_json_path = File.join(tmpdir, 'tree', 'project.json')
+          expect(File).to exist(project_json_path)
 
-        project_hash = JSON.parse(IO.read(project_json_path))
+          relations = []
+          relations << JSON.parse(IO.read(project_json_path))
+          Dir.glob(File.join(tmpdir, 'tree/project', '*.ndjson')) do |rb_filename|
+            File.foreach(rb_filename) do |line|
+              json = ActiveSupport::JSON.decode(line)
+              relations << json
+            end
+          end
 
-        sensitive_words.each do |sensitive_word|
-          found = find_sensitive_attributes(sensitive_word, project_hash)
+          relations.each do |relation_hash|
+            sensitive_words.each do |sensitive_word|
+              found = find_sensitive_attributes(sensitive_word, relation_hash)
 
-          expect(found).to be_nil, failure_message(found.try(:key_found), found.try(:parent), sensitive_word)
+              expect(found).to be_nil, failure_message(found.try(:key_found), found.try(:parent), sensitive_word)
+            end
+          end
         end
       end
     end
+  end
 
-    def failure_message(key_found, parent, sensitive_word)
-      <<-MSG
-        Found a new sensitive word <#{key_found}>, which is part of the hash #{parent.inspect}
+  def export_project_and_download_file(page, project)
+    visit edit_project_path(project)
 
-        If you think this information shouldn't get exported, please exclude the model or attribute in IMPORT_EXPORT_CONFIG.
+    expect(page).to have_content('Export project')
 
-        Otherwise, please add the exception to +safe_list+ in CURRENT_SPEC using #{sensitive_word} as the key and the
-        correspondent hash or model as the value.
+    find(:link, 'Export project').send_keys(:return)
 
-        Also, if the attribute is a generated unique token, please add it to RelationFactory::TOKEN_RESET_MODELS if it needs to be
-        reset (to prevent duplicate column problems while importing to the same instance).
+    visit edit_project_path(project)
 
-        IMPORT_EXPORT_CONFIG: #{Gitlab::ImportExport.config_file}
-        CURRENT_SPEC: #{__FILE__}
-      MSG
-    end
+    expect(page).to have_content('Download export')
+    expect(project.export_status).to eq(:finished)
+    expect(project.export_file.path).to include('tar.gz')
+  end
+
+  def failure_message(key_found, parent, sensitive_word)
+    <<-MSG
+      Found a new sensitive word <#{key_found}>, which is part of the hash #{parent.inspect}
+
+      If you think this information shouldn't get exported, please exclude the model or attribute in IMPORT_EXPORT_CONFIG.
+
+      Otherwise, please add the exception to +safe_list+ in CURRENT_SPEC using #{sensitive_word} as the key and the
+      correspondent hash or model as the value.
+
+      Also, if the attribute is a generated unique token, please add it to RelationFactory::TOKEN_RESET_MODELS if it needs to be
+      reset (to prevent duplicate column problems while importing to the same instance).
+
+      IMPORT_EXPORT_CONFIG: #{Gitlab::ImportExport.config_file}
+      CURRENT_SPEC: #{__FILE__}
+    MSG
   end
 end

spec/lib/gitlab/import_export/fork_spec.rb

@@ -32,6 +32,8 @@ describe 'forked project import' do
   end
 
   before do
+    stub_feature_flags(project_export_as_ndjson: false)
+
     allow_next_instance_of(Gitlab::ImportExport) do |instance|
       allow(instance).to receive(:storage_path).and_return(export_path)
     end

spec/lib/gitlab/import_export/import_export_equivalence_spec.rb

@@ -20,6 +20,10 @@ describe Gitlab::ImportExport do
 
   let(:json_fixture) { 'complex' }
 
+  before do
+    stub_feature_flags(project_export_as_ndjson: false)
+  end
+
   it 'yields the initial tree when importing and exporting it again' do
     project = create(:project, creator: create(:user, :admin))
 

spec/lib/gitlab/import_export/json/ndjson_writer_spec.rb

+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Gitlab::ImportExport::JSON::NdjsonWriter do
+  include ImportExport::CommonUtil
+
+  let(:path) { "#{Dir.tmpdir}/ndjson_writer_spec/tree" }
+  let(:exportable_path) { 'projects' }
+
+  subject { described_class.new(path) }
+
+  after do
+    FileUtils.rm_rf(path)
+  end
+
+  describe "#write_attributes" do
+    it "writes correct json to root" do
+      expected_hash = { "key" => "value_1", "key_1" => "value_2" }
+      subject.write_attributes(exportable_path, expected_hash)
+
+      expect(consume_attributes(path, exportable_path)).to eq(expected_hash)
+    end
+  end
+
+  describe "#write_relation" do
+    context "when single relation is serialized" do
+      it "appends json in correct file " do
+        relation = "relation"
+        value =  { "key" => "value_1", "key_1" => "value_1" }
+        subject.write_relation(exportable_path, relation, value)
+
+        expect(consume_relations(path, exportable_path, relation)).to eq([value])
+      end
+    end
+
+    context "when single relation is already serialized" do
+      it "raise exception" do
+        values = [{ "key" => "value_1", "key_1" => "value_1" }, { "key" => "value_2", "key_1" => "value_2" }]
+        relation = "relation"
+        file_path = File.join(path, exportable_path, "#{relation}.ndjson")
+        subject.write_relation(exportable_path, relation, values[0])
+
+        expect {subject.write_relation(exportable_path, relation, values[1])}.to raise_exception("The #{file_path} already exist")
+      end
+    end
+  end
+
+  describe "#write_relation_array" do
+    it "writes json in correct files" do
+      values =  [{ "key" => "value_1", "key_1" => "value_1" }, { "key" => "value_2", "key_1" => "value_2" }]
+      relations = %w(relation1 relation2)
+      relations.each do |relation|
+        subject.write_relation_array(exportable_path, relation, values.to_enum)
+      end
+      subject.close
+
+      relations.each do |relation|
+        expect(consume_relations(path, exportable_path, relation)).to eq(values)
+      end
+    end
+  end
+end

spec/support/import_export/common_util.rb

@@ -26,6 +26,21 @@ module ImportExport
       "tmp/tests/gitlab-test/import_export"
     end
 
+    def get_json(path, exportable_path, key, ndjson_enabled)
+      if ndjson_enabled
+        json = if key == :projects
+                 consume_attributes(path, exportable_path)
+               else
+                 consume_relations(path, exportable_path, key)
+               end
+      else
+        json = project_json(path)
+        json = json[key.to_s] unless key == :projects
+      end
+
+      json
+    end
+
     def restore_then_save_project(project, import_path:, export_path:)
       project_restorer = get_project_restorer(project, import_path)
       project_saver = get_project_saver(project, export_path)
@@ -50,5 +65,30 @@ module ImportExport
         allow(shared).to receive(:export_path).and_return(path)
       end
     end
+
+    def consume_attributes(dir_path, exportable_path)
+      path = File.join(dir_path, "#{exportable_path}.json")
+      return unless File.exist?(path)
+
+      ActiveSupport::JSON.decode(IO.read(path))
+    end
+
+    def consume_relations(dir_path, exportable_path, key)
+      path = File.join(dir_path, exportable_path, "#{key}.ndjson")
+      return unless File.exist?(path)
+
+      relations = []
+
+      File.foreach(path) do |line|
+        json = ActiveSupport::JSON.decode(line)
+        relations << json
+      end
+
+      key == :project_feature ? relations.first : relations.flatten
+    end
+
+    def project_json(filename)
+      ActiveSupport::JSON.decode(IO.read(filename))
+    end
   end
 end