Fix Wiki Pipeline network error if wiki does not exist

  - Add a preliminary check on source to see if there are
    any wiki pages available before doing wiki repository import
  - Update `BulkImports::Clients::HTTP` to have a more detailed
    error message when a non-successful http response is returned

Changelog: fixed

app/models/bulk_imports/entity.rb

@@ -20,8 +20,6 @@
 class BulkImports::Entity < ApplicationRecord
   self.table_name = 'bulk_import_entities'
 
-  EXPORT_RELATIONS_URL = '/%{resource}/%{full_path}/export_relations'
-
   belongs_to :bulk_import, optional: false
   belongs_to :parent, class_name: 'BulkImports::Entity', optional: true
 
@@ -112,14 +110,22 @@ class BulkImports::Entity < ApplicationRecord
     entity_type.pluralize
   end
 
+  def base_resource_url_path
+    "/#{pluralized_name}/#{encoded_source_full_path}"
+  end
+
   def export_relations_url_path
-    @export_relations_url_path ||= EXPORT_RELATIONS_URL % { resource: pluralized_name, full_path: encoded_source_full_path }
+    "#{base_resource_url_path}/export_relations"
   end
 
   def relation_download_url_path(relation)
     "#{export_relations_url_path}/download?relation=#{relation}"
   end
 
+  def wikis_url_path
+    "#{base_resource_url_path}/wikis"
+  end
+
   def project?
     source_type == 'project_entity'
   end

lib/bulk_imports/clients/http.rb

@@ -123,7 +123,7 @@ module BulkImports
       def with_error_handling
         response = yield
 
-        raise ::BulkImports::NetworkError.new(response: response) unless response.success?
+        raise ::BulkImports::NetworkError.new("Unsuccessful response #{response.code} from #{response.request.path.path}", response: response) unless response.success?
 
         response
       rescue *Gitlab::HTTP::HTTP_ERRORS => e

lib/bulk_imports/common/pipelines/wiki_pipeline.rb

@@ -7,7 +7,9 @@ module BulkImports
         include Pipeline
 
         def extract(*)
-          BulkImports::Pipeline::ExtractedData.new(data: { url: url_from_parent_path(context.entity.source_full_path) })
+          url = url_from_parent_path(context.entity.source_full_path) if source_wiki_exists?
+
+          BulkImports::Pipeline::ExtractedData.new(data: { url: url })
         end
 
         def transform(_, data)
@@ -15,14 +17,15 @@ module BulkImports
         end
 
         def load(context, data)
-          return unless context.portable.wiki
+          return unless data&.dig(:url)
 
+          wiki = context.portable.wiki
           url = data[:url].sub("://", "://oauth2:#{context.configuration.access_token}@")
 
           Gitlab::UrlBlocker.validate!(url, allow_local_network: allow_local_requests?, allow_localhost: allow_local_requests?)
 
-          context.portable.wiki.ensure_repository
-          context.portable.wiki.repository.fetch_as_mirror(url)
+          wiki.ensure_repository
+          wiki.repository.fetch_as_mirror(url)
         end
 
         private
@@ -36,6 +39,16 @@ module BulkImports
         def allow_local_requests?
           Gitlab::CurrentSettings.allow_local_requests_from_web_hooks_and_services?
         end
+
+        def source_wiki_exists?
+          wikis = client.get(context.entity.wikis_url_path).parsed_response
+
+          wikis.any?
+        end
+
+        def client
+          BulkImports::Clients::HTTP.new(url: context.configuration.url, token: context.configuration.access_token)
+        end
       end
     end
   end

spec/lib/bulk_imports/clients/http_spec.rb

@@ -38,11 +38,11 @@ RSpec.describe BulkImports::Clients::HTTP do
 
       context 'when response is not success' do
         it 'raises BulkImports::Error' do
-          response_double = double(code: 503, success?: false)
+          response_double = double(code: 503, success?: false, request: double(path: double(path: '/test')))
 
           allow(Gitlab::HTTP).to receive(method).and_return(response_double)
 
-          expect { subject.public_send(method, resource) }.to raise_exception(BulkImports::NetworkError)
+          expect { subject.public_send(method, resource) }.to raise_exception(BulkImports::NetworkError, 'Unsuccessful response 503 from /test')
         end
       end
     end

spec/models/bulk_imports/entity_spec.rb

@@ -282,4 +282,20 @@ RSpec.describe BulkImports::Entity, type: :model do
       expect(project_entity.group?).to eq(false)
     end
   end
+
+  describe '#base_resource_url_path' do
+    it 'returns base entity url path' do
+      entity = build(:bulk_import_entity)
+
+      expect(entity.base_resource_url_path).to eq("/groups/#{entity.encoded_source_full_path}")
+    end
+  end
+
+  describe '#wiki_url_path' do
+    it 'returns entity wiki url path' do
+      entity = build(:bulk_import_entity)
+
+      expect(entity.wikis_url_path).to eq("/groups/#{entity.encoded_source_full_path}/wikis")
+    end
+  end
 end

spec/support/shared_examples/bulk_imports/common/pipelines/wiki_pipeline_examples.rb

@@ -9,16 +9,18 @@ RSpec.shared_examples 'wiki pipeline imports a wiki for an entity' do
 
     let(:extracted_data) { BulkImports::Pipeline::ExtractedData.new(data: {}) }
 
-    context 'successfully imports wiki for an entity' do
-      subject { described_class.new(context) }
+    subject { described_class.new(context) }
 
-      before do
-        allow_next_instance_of(BulkImports::Common::Extractors::GraphqlExtractor) do |extractor|
-          allow(extractor).to receive(:extract).and_return(extracted_data)
-        end
+    before do
+      allow_next_instance_of(BulkImports::Common::Extractors::GraphqlExtractor) do |extractor|
+        allow(extractor).to receive(:extract).and_return(extracted_data)
       end
+    end
 
+    context 'when wiki exists' do
       it 'imports new wiki into destination project' do
+        expect(subject).to receive(:source_wiki_exists?).and_return(true)
+
         expect_next_instance_of(Gitlab::GitalyClient::RepositoryService) do |repository_service|
           url = "https://oauth2:token@gitlab.example/#{entity.source_full_path}.wiki.git"
           expect(repository_service).to receive(:fetch_remote).with(url, any_args).and_return 0
@@ -27,5 +29,16 @@ RSpec.shared_examples 'wiki pipeline imports a wiki for an entity' do
         subject.run
       end
     end
+
+    context 'when wiki does not exist' do
+      it 'does not import wiki' do
+        expect(subject).to receive(:source_wiki_exists?).and_return(false)
+
+        expect(parent.wiki).not_to receive(:ensure_repository)
+        expect(parent.wiki.repository).not_to receive(:ensure_repository)
+
+        expect { subject.run }.not_to raise_error
+      end
+    end
   end
 end