Merge branch '210513-introduce-ndjson-reader-for-project-import' into 'master'

Introduce ndjson reader for project import

Closes #210513

See merge request gitlab-org/gitlab!27206

lib/gitlab/import_export/group/tree_restorer.rb

@@ -20,6 +20,7 @@ module Gitlab
         def restore
           @group_attributes = relation_reader.consume_attributes(nil)
           @group_members = relation_reader.consume_relation(nil, 'members')
+            .map(&:first)
 
           # We need to remove `name` and `path` as we did consume it in previous pass
           @group_attributes.delete('name')

lib/gitlab/import_export/json/legacy_reader.rb

@@ -53,6 +53,7 @@ module Gitlab
 
         def initialize(relation_names:, allowed_path:)
           @relation_names = relation_names.map(&:to_s)
+          @consumed_relations = Set.new
 
           # This is legacy reader, to be used in transition
           # period before `.ndjson`,
@@ -81,17 +82,19 @@ module Gitlab
             raise ArgumentError, "Invalid #{importable_name} passed to `consume_relation`. Use #{@allowed_path} instead."
           end
 
-          value = relations.delete(key)
+          Enumerator.new do |documents|
+            next unless @consumed_relations.add?("#{importable_path}/#{key}")
 
-          return value unless block_given?
-          return if value.nil?
+            value = relations.delete(key)
+            next if value.nil?
 
-          if value.is_a?(Array)
-            value.each.with_index do |item, idx|
-              yield(item, idx)
+            if value.is_a?(Array)
+              value.each.with_index do |item, idx|
+                documents << [item, idx]
+              end
+            else
+              documents << [value, 0]
             end
-          else
-            yield(value, 0)
           end
         end
 

lib/gitlab/import_export/json/ndjson_reader.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module ImportExport
+    module JSON
+      class NdjsonReader
+        MAX_JSON_DOCUMENT_SIZE = 50.megabytes
+
+        attr_reader :dir_path
+
+        def initialize(dir_path)
+          @dir_path = dir_path
+          @consumed_relations = Set.new
+        end
+
+        def exist?
+          Dir.exist?(@dir_path)
+        end
+
+        # This can be removed once legacy_reader is deprecated.
+        def legacy?
+          false
+        end
+
+        def consume_attributes(importable_path)
+          # This reads from `tree/project.json`
+          path = file_path("#{importable_path}.json")
+          data = File.read(path, MAX_JSON_DOCUMENT_SIZE)
+          json_decode(data)
+        end
+
+        def consume_relation(importable_path, key)
+          Enumerator.new do |documents|
+            next unless @consumed_relations.add?("#{importable_path}/#{key}")
+
+            # This reads from `tree/project/merge_requests.ndjson`
+            path = file_path(importable_path, "#{key}.ndjson")
+            next unless File.exist?(path)
+
+            File.foreach(path, MAX_JSON_DOCUMENT_SIZE).with_index do |line, line_num|
+              documents << [json_decode(line), line_num]
+            end
+          end
+        end
+
+        private
+
+        def json_decode(string)
+          ActiveSupport::JSON.decode(string)
+        rescue ActiveSupport::JSON.parse_error => e
+          Gitlab::ErrorTracking.log_exception(e)
+          raise Gitlab::ImportExport::Error, 'Incorrect JSON format'
+        end
+
+        def file_path(*path)
+          File.join(dir_path, *path)
+        end
+      end
+    end
+  end
+end

lib/gitlab/import_export/project/tree_restorer.rb

@@ -17,8 +17,13 @@ module Gitlab
         end
 
         def restore
+          unless relation_reader
+            raise Gitlab::ImportExport::Error, 'invalid import format'
+          end
+
           @project_attributes = relation_reader.consume_attributes(importable_path)
           @project_members = relation_reader.consume_relation(importable_path, 'project_members')
+            .map(&:first)
 
           if relation_tree_restorer.restore
             import_failure_service.with_retry(action: 'set_latest_merge_request_diff_ids!') do
@@ -38,14 +43,27 @@ module Gitlab
 
         def relation_reader
           strong_memoize(:relation_reader) do
-            ImportExport::JSON::LegacyReader::File.new(
-              File.join(shared.export_path, 'project.json'),
-              relation_names: reader.project_relation_names,
-              allowed_path: importable_path
-            )
+            [ndjson_relation_reader, legacy_relation_reader]
+              .compact.find(&:exist?)
           end
         end
 
+        def ndjson_relation_reader
+          return unless Feature.enabled?(:project_import_ndjson, project.namespace)
+
+          ImportExport::JSON::NdjsonReader.new(
+            File.join(shared.export_path, 'tree')
+          )
+        end
+
+        def legacy_relation_reader
+          ImportExport::JSON::LegacyReader::File.new(
+            File.join(shared.export_path, 'project.json'),
+            relation_names: reader.project_relation_names,
+            allowed_path: importable_path
+          )
+        end
+
         def relation_tree_restorer
           @relation_tree_restorer ||= RelationTreeRestorer.new(
             user: @user,

lib/gitlab/import_export/relation_tree_restorer.rb

@@ -67,7 +67,7 @@ module Gitlab
       end
 
       def process_relation!(relation_key, relation_definition)
-        @relation_reader.consume_relation(@importable_path, relation_key) do |data_hash, relation_index|
+        @relation_reader.consume_relation(@importable_path, relation_key).each do |data_hash, relation_index|
           process_relation_item!(relation_key, relation_definition, relation_index, data_hash)
         end
       end

spec/fixtures/lib/gitlab/import_export/light/project.json

@@ -186,5 +186,23 @@
     }
   ],
   "snippets": [],
-  "hooks": []
+  "hooks": [],
+  "custom_attributes": [
+    {
+      "id": 201,
+      "project_id": 5,
+      "created_at": "2016-06-14T15:01:51.315Z",
+      "updated_at": "2016-06-14T15:01:51.315Z",
+      "key": "color",
+      "value": "red"
+    },
+    {
+      "id": 202,
+      "project_id": 5,
+      "created_at": "2016-06-14T15:01:51.315Z",
+      "updated_at": "2016-06-14T15:01:51.315Z",
+      "key": "size",
+      "value": "small"
+    }
+  ]
 }

spec/lib/gitlab/import_export/json/legacy_reader/shared_example.rb

@@ -15,7 +15,6 @@ RSpec.shared_examples 'import/export json legacy reader' do
       subject { legacy_reader.consume_attributes("project") }
 
       context 'no excluded attributes' do
-        let(:excluded_attributes) { [] }
         let(:relation_names) { [] }
 
         it 'returns the whole tree from parsed JSON' do
@@ -42,60 +41,53 @@ RSpec.shared_examples 'import/export json legacy reader' do
 
   describe '#consume_relation' do
     context 'when valid path is passed' do
-      let(:key) { 'description' }
+      let(:key) { 'labels' }
 
-      context 'block not given' do
-        it 'returns value of the key' do
-          expect(legacy_reader).to receive(:relations).and_return({ key => 'test value' })
-          expect(legacy_reader.consume_relation("project", key)).to eq('test value')
-        end
-      end
+      subject { legacy_reader.consume_relation("project", key) }
 
-      context 'key has been consumed' do
-        before do
-          legacy_reader.consume_relation("project", key)
+      context 'key has not been consumed' do
+        it 'returns an Enumerator' do
+          expect(subject).to be_an_instance_of(Enumerator)
         end
 
-        it 'does not yield' do
-          expect do |blk|
-            legacy_reader.consume_relation("project", key, &blk)
-          end.not_to yield_control
-        end
-      end
+        context 'value is nil' do
+          before do
+            expect(legacy_reader).to receive(:relations).and_return({ key => nil })
+          end
 
-      context 'value is nil' do
-        before do
-          expect(legacy_reader).to receive(:relations).and_return({ key => nil })
+          it 'yields nothing to the Enumerator' do
+            expect(subject.to_a).to eq([])
+          end
         end
 
-        it 'does not yield' do
-          expect do |blk|
-            legacy_reader.consume_relation("project", key, &blk)
-          end.not_to yield_control
-        end
-      end
+        context 'value is an array' do
+          before do
+            expect(legacy_reader).to receive(:relations).and_return({ key => %w[label1 label2] })
+          end
 
-      context 'value is not array' do
-        before do
-          expect(legacy_reader).to receive(:relations).and_return({ key => 'value' })
+          it 'yields every relation value to the Enumerator' do
+            expect(subject.to_a).to eq([['label1', 0], ['label2', 1]])
+          end
         end
 
-        it 'yield the value with index 0' do
-          expect do |blk|
-            legacy_reader.consume_relation("project", key, &blk)
-          end.to yield_with_args('value', 0)
+        context 'value is not array' do
+          before do
+            expect(legacy_reader).to receive(:relations).and_return({ key => 'non-array value' })
+          end
+
+          it 'yields the value with index 0 to the Enumerator' do
+            expect(subject.to_a).to eq([['non-array value', 0]])
+          end
         end
       end
 
-      context 'value is an array' do
+      context 'key has been consumed' do
         before do
-          expect(legacy_reader).to receive(:relations).and_return({ key => %w[item1 item2 item3] })
+          legacy_reader.consume_relation("project", key).first
         end
 
-        it 'yield each array element with index' do
-          expect do |blk|
-            legacy_reader.consume_relation("project", key, &blk)
-          end.to yield_successive_args(['item1', 0], ['item2', 1], ['item3', 2])
+        it 'yields nothing to the Enumerator' do
+          expect(subject.to_a).to eq([])
         end
       end
     end

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

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe Gitlab::ImportExport::JSON::NdjsonReader do
+  include ImportExport::CommonUtil
+
+  let(:fixture) { 'spec/fixtures/lib/gitlab/import_export/light/tree' }
+  let(:root_tree) { JSON.parse(File.read(File.join(fixture, 'project.json'))) }
+  let(:ndjson_reader) { described_class.new(dir_path) }
+  let(:importable_path) { 'project' }
+
+  before :all do
+    extract_archive('spec/fixtures/lib/gitlab/import_export/light', 'tree.tar.gz')
+  end
+
+  after :all do
+    cleanup_artifacts_from_extract_archive('light')
+  end
+
+  describe '#exist?' do
+    subject { ndjson_reader.exist? }
+
+    context 'given valid dir_path' do
+      let(:dir_path) { fixture }
+
+      it { is_expected.to be true }
+    end
+
+    context 'given invalid dir_path' do
+      let(:dir_path) { 'invalid-dir-path' }
+
+      it { is_expected.to be false }
+    end
+  end
+
+  describe '#legacy?' do
+    let(:dir_path) { fixture }
+
+    subject { ndjson_reader.legacy? }
+
+    it { is_expected.to be false }
+  end
+
+  describe '#consume_attributes' do
+    let(:dir_path) { fixture }
+
+    subject { ndjson_reader.consume_attributes(importable_path) }
+
+    it 'returns the whole root tree from parsed JSON' do
+      expect(subject).to eq(root_tree)
+    end
+  end
+
+  describe '#consume_relation' do
+    let(:dir_path) { fixture }
+
+    subject { ndjson_reader.consume_relation(importable_path, key) }
+
+    context 'given any key' do
+      let(:key) { 'any-key' }
+
+      it 'returns an Enumerator' do
+        expect(subject).to be_an_instance_of(Enumerator)
+      end
+    end
+
+    context 'key has been consumed' do
+      let(:key) { 'issues' }
+
+      before do
+        ndjson_reader.consume_relation(importable_path, key).first
+      end
+
+      it 'yields nothing to the Enumerator' do
+        expect(subject.to_a).to eq([])
+      end
+    end
+
+    context 'key has not been consumed' do
+      context 'relation file does not exist' do
+        let(:key) { 'non-exist-relation-file-name' }
+
+        before do
+          relation_file_path = File.join(dir_path, importable_path, "#{key}.ndjson")
+          expect(File).to receive(:exist?).with(relation_file_path).and_return(false)
+        end
+
+        it 'yields nothing to the Enumerator' do
+          expect(subject.to_a).to eq([])
+        end
+      end
+
+      context 'relation file is empty' do
+        let(:key) { 'empty' }
+
+        it 'yields nothing to the Enumerator' do
+          expect(subject.to_a).to eq([])
+        end
+      end
+
+      context 'relation file contains multiple lines' do
+        let(:key) { 'custom_attributes' }
+        let(:attr_1) { JSON.parse('{"id":201,"project_id":5,"created_at":"2016-06-14T15:01:51.315Z","updated_at":"2016-06-14T15:01:51.315Z","key":"color","value":"red"}') }
+        let(:attr_2) { JSON.parse('{"id":202,"project_id":5,"created_at":"2016-06-14T15:01:51.315Z","updated_at":"2016-06-14T15:01:51.315Z","key":"size","value":"small"}') }
+
+        it 'yields every relation value to the Enumerator' do
+          expect(subject.to_a).to eq([[attr_1, 0], [attr_2, 1]])
+        end
+      end
+    end
+  end
+end

spec/lib/gitlab/import_export/relation_tree_restorer_spec.rb

@@ -14,7 +14,7 @@ describe Gitlab::ImportExport::RelationTreeRestorer do
 
   let(:user) { create(:user) }
   let(:shared) { Gitlab::ImportExport::Shared.new(importable) }
-  let(:attributes) { {} }
+  let(:attributes) { relation_reader.consume_attributes(importable_name) }
 
   let(:members_mapper) do
     Gitlab::ImportExport::MembersMapper.new(exported_members: {}, user: user, importable: importable)
@@ -30,7 +30,7 @@ describe Gitlab::ImportExport::RelationTreeRestorer do
       relation_factory:      relation_factory,
       reader:                reader,
       importable:            importable,
-      importable_path:       nil,
+      importable_path:       importable_path,
       importable_attributes: attributes
     )
   end
@@ -94,21 +94,24 @@ describe Gitlab::ImportExport::RelationTreeRestorer do
   end
 
   context 'when restoring a project' do
-    let(:path) { 'spec/fixtures/lib/gitlab/import_export/complex/project.json' }
     let(:importable) { create(:project, :builds_enabled, :issues_disabled, name: 'project', path: 'project') }
+    let(:importable_name) { 'project' }
+    let(:importable_path) { 'project' }
     let(:object_builder) { Gitlab::ImportExport::Project::ObjectBuilder }
     let(:relation_factory) { Gitlab::ImportExport::Project::RelationFactory }
     let(:reader) { Gitlab::ImportExport::Reader.new(shared: shared) }
 
     context 'using legacy reader' do
+      let(:path) { 'spec/fixtures/lib/gitlab/import_export/complex/project.json' }
       let(:relation_reader) do
         Gitlab::ImportExport::JSON::LegacyReader::File.new(
           path,
-          relation_names: reader.project_relation_names
+          relation_names: reader.project_relation_names,
+          allowed_path: 'project'
         )
       end
 
-      let(:attributes) { relation_reader.consume_attributes(nil) }
+      let(:attributes) { relation_reader.consume_attributes('project') }
 
       it_behaves_like 'import project successfully'
 
@@ -118,6 +121,21 @@ describe Gitlab::ImportExport::RelationTreeRestorer do
 
         include_examples 'logging of relations creation'
       end
+
+      context 'using ndjson reader' do
+        let(:path) { 'spec/fixtures/lib/gitlab/import_export/complex/tree' }
+        let(:relation_reader) { Gitlab::ImportExport::JSON::NdjsonReader.new(path) }
+
+        before :all do
+          extract_archive('spec/fixtures/lib/gitlab/import_export/complex', 'tree.tar.gz')
+        end
+
+        after :all do
+          cleanup_artifacts_from_extract_archive('complex')
+        end
+
+        it_behaves_like 'import project successfully'
+      end
     end
   end
 
@@ -125,9 +143,16 @@ describe Gitlab::ImportExport::RelationTreeRestorer do
     let(:path) { 'spec/fixtures/lib/gitlab/import_export/group_exports/no_children/group.json' }
     let(:group) { create(:group) }
     let(:importable) { create(:group, parent: group) }
+    let(:importable_name) { nil }
+    let(:importable_path) { nil }
     let(:object_builder) { Gitlab::ImportExport::Group::ObjectBuilder }
     let(:relation_factory) { Gitlab::ImportExport::Group::RelationFactory }
-    let(:relation_reader) { Gitlab::ImportExport::JSON::LegacyReader::File.new(path, relation_names: reader.group_relation_names) }
+    let(:relation_reader) do
+      Gitlab::ImportExport::JSON::LegacyReader::File.new(
+        path,
+        relation_names: reader.group_relation_names)
+    end
+
     let(:reader) do
       Gitlab::ImportExport::Reader.new(
         shared: shared,
@@ -135,6 +160,10 @@ describe Gitlab::ImportExport::RelationTreeRestorer do
       )
     end
 
+    it 'restores group tree' do
+      expect(subject).to eq(true)
+    end
+
     include_examples 'logging of relations creation'
   end
 end

spec/support/import_export/common_util.rb

@@ -15,9 +15,39 @@ module ImportExport
       export_path = [prefix, 'spec', 'fixtures', 'lib', 'gitlab', 'import_export', name].compact
       export_path = File.join(*export_path)
 
+      extract_archive(export_path, 'tree.tar.gz')
+
       allow_any_instance_of(Gitlab::ImportExport).to receive(:export_path) { export_path }
     end
 
+    def extract_archive(path, archive)
+      if File.exist?(File.join(path, archive))
+        system("cd #{path}; tar xzvf #{archive} &> /dev/null")
+      end
+    end
+
+    def cleanup_artifacts_from_extract_archive(name, prefix = nil)
+      export_path = [prefix, 'spec', 'fixtures', 'lib', 'gitlab', 'import_export', name].compact
+      export_path = File.join(*export_path)
+
+      if File.exist?(File.join(export_path, 'tree.tar.gz'))
+        system("cd #{export_path}; rm -fr tree &> /dev/null")
+      end
+    end
+
+    def setup_reader(reader)
+      case reader
+      when :legacy_reader
+        allow_any_instance_of(Gitlab::ImportExport::JSON::LegacyReader::File).to receive(:exist?).and_return(true)
+        allow_any_instance_of(Gitlab::ImportExport::JSON::NdjsonReader).to receive(:exist?).and_return(false)
+      when :ndjson_reader
+        allow_any_instance_of(Gitlab::ImportExport::JSON::LegacyReader::File).to receive(:exist?).and_return(false)
+        allow_any_instance_of(Gitlab::ImportExport::JSON::NdjsonReader).to receive(:exist?).and_return(true)
+      else
+        raise "invalid reader #{reader}. Supported readers: :legacy_reader, :ndjson_reader"
+      end
+    end
+
     def fixtures_path
       "spec/fixtures/lib/gitlab/import_export"
     end