Refactor on namespace and repository checks and added specs

lib/system_check/orphans/namespace_check.rb

@@ -4,20 +4,13 @@ module SystemCheck
       set_name 'Orphaned namespaces:'
 
       def multi_check
-        Gitlab.config.repositories.storages.each do |name, repository_storage|
+        Gitlab.config.repositories.storages.each do |storage_name, repository_storage|
           $stdout.puts
-          $stdout.puts "* Storage: #{name} (#{repository_storage['path']})".color(:yellow)
-          toplevel_namespace_dirs = Dir.glob(File.join(repository_storage['path'], '*')).map{|p| File.basename(p)}
+          $stdout.puts "* Storage: #{storage_name} (#{repository_storage['path']})".color(:yellow)
+          toplevel_namespace_dirs = disk_namespaces(repository_storage['path'])
 
           orphans = (toplevel_namespace_dirs - existing_namespaces)
-          if orphans.empty?
-            $stdout.puts "* No orphaned namespaces for #{name} storage".color(:green)
-            next
-          end
-
-          orphans.each do |orphan|
-            $stdout.puts " - #{orphan}".color(:red)
-          end
+          print_orphans(orphans, storage_name)
         end
 
         clear_namespaces! # releases memory when check finishes
@@ -25,8 +18,32 @@ module SystemCheck
 
       private
 
+      def print_orphans(orphans, storage_name)
+        if orphans.empty?
+          $stdout.puts "* No orphaned namespaces for #{storage_name} storage".color(:green)
+          return
+        end
+
+        orphans.each do |orphan|
+          $stdout.puts " - #{orphan}".color(:red)
+        end
+      end
+
+      def disk_namespaces(storage_path)
+        fetch_disk_namespaces(storage_path).each_with_object([]) do |namespace_path, result|
+          namespace = File.basename(namespace_path)
+          next if namespace.eql?('@hashed')
+
+          result << namespace
+        end
+      end
+
+      def fetch_disk_namespaces(storage_path)
+        Dir.glob(File.join(storage_path, '*'))
+      end
+
       def existing_namespaces
-        @namespaces ||= Namespace.all.pluck(:path)
+        @namespaces ||= Namespace.where(parent: nil).all.pluck(:path)
       end
 
       def clear_namespaces!

lib/system_check/orphans/repository_check.rb

@@ -2,49 +2,67 @@ module SystemCheck
   module Orphans
     class RepositoryCheck < SystemCheck::BaseCheck
       set_name 'Orphaned repositories:'
+      attr_accessor :orphans
 
       def multi_check
-        Gitlab.config.repositories.storages.each do |name, repository_storage|
+        Gitlab.config.repositories.storages.each do |storage_name, repository_storage|
           $stdout.puts
-          $stdout.puts "* Storage: #{name} (#{repository_storage['path']})".color(:yellow)
+          $stdout.puts "* Storage: #{storage_name} (#{repository_storage['path']})".color(:yellow)
 
-          repositories = toplevel_namespace_dirs(repository_storage['path']).map do |path|
-            namespace = File.basename(path)
-            Dir.glob(File.join(path, '*')).map {|repo| "#{namespace}/#{File.basename(repo)}"}
-          end.try(:flatten!)
+          repositories = disk_repositories(repository_storage['path'])
+          orphans = (repositories - fetch_repositories(storage_name))
 
-          orphans = (repositories - list_repositories(name))
-          if orphans.empty?
-            $stdout.puts "* No orphaned repositories for #{name} storage".color(:green)
-            next
-          end
-
-          orphans.each do |orphan|
-            $stdout.puts " - #{orphan}".color(:red)
-          end
+          print_orphans(orphans, storage_name)
         end
       end
 
       private
 
-      def list_repositories(storage_name)
+      def print_orphans(orphans, storage_name)
+        if orphans.empty?
+          $stdout.puts "* No orphaned repositories for #{storage_name} storage".color(:green)
+          return
+        end
+
+        orphans.each do |orphan|
+          $stdout.puts " - #{orphan}".color(:red)
+        end
+      end
+
+      def disk_repositories(storage_path)
+        fetch_disk_namespaces(storage_path).each_with_object([]) do |namespace_path, result|
+          namespace = File.basename(namespace_path)
+          next if namespace.eql?('@hashed')
+
+          fetch_disk_repositories(namespace_path).each do |repo|
+            result << "#{namespace}/#{File.basename(repo)}"
+          end
+        end
+      end
+
+      def fetch_repositories(storage_name)
         sql = "
           SELECT
             CONCAT(n.path, '/', p.path, '.git') repo,
             CONCAT(n.path, '/', p.path, '.wiki.git') wiki
           FROM projects p
             JOIN namespaces n
-              ON (p.namespace_id = n.id)
+              ON (p.namespace_id = n.id AND
+                  n.parent_id IS NULL)
           WHERE (p.repository_storage LIKE ?)
         "
 
-        query = ActiveRecord::Base.send(:sanitize_sql_array, [sql, storage_name])
-        ActiveRecord::Base.connection.select_all(query).rows.try(:flatten!)
+        query = ActiveRecord::Base.send(:sanitize_sql_array, [sql, storage_name]) # rubocop:disable GitlabSecurity/PublicSend
+        ActiveRecord::Base.connection.select_all(query).rows.try(:flatten!) || []
       end
 
-      def toplevel_namespace_dirs(storage_path)
+      def fetch_disk_namespaces(storage_path)
         Dir.glob(File.join(storage_path, '*'))
       end
+
+      def fetch_disk_repositories(namespace_path)
+        Dir.glob(File.join(namespace_path, '*'))
+      end
     end
   end
 end

lib/tasks/gitlab/check.rake

@@ -427,7 +427,6 @@ namespace :gitlab do
     end
   end
 
-
   namespace :user do
     desc "GitLab | Check the integrity of a specific user's repositories"
     task :check_repos, [:username] => :environment do |t, args|

spec/lib/system_check/orphans/namespace_check_spec.rb

+require 'spec_helper'
+require 'rake_helper'
+
+describe SystemCheck::Orphans::NamespaceCheck do
+  let(:storages) { Gitlab.config.repositories.storages.reject { |key, _| key.eql? 'broken' } }
+
+  before do
+    allow(Gitlab.config.repositories).to receive(:storages).and_return(storages)
+    allow(subject).to receive(:fetch_disk_namespaces).and_return(disk_namespaces)
+    silence_output
+  end
+
+  describe '#multi_check' do
+    context 'all orphans' do
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 repos/@hashed) }
+
+      it 'prints list of all orphaned namespaces except @hashed' do
+        expect_list_of_orphans(%w(orphan1 orphan2))
+
+        subject.multi_check
+      end
+    end
+
+    context 'few orphans with existing namespace' do
+      let!(:first_level) { create(:group, path: 'my-namespace') }
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 /repos/my-namespace /repos/@hashed) }
+
+      it 'prints list of orphaned namespaces' do
+        expect_list_of_orphans(%w(orphan1 orphan2))
+
+        subject.multi_check
+      end
+    end
+
+    context 'few orphans with existing namespace and parents with same name as orphans' do
+      let!(:first_level) { create(:group, path: 'my-namespace') }
+      let!(:second_level) { create(:group, path: 'second-level', parent: first_level) }
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 /repos/my-namespace /repos/second-level /repos/@hashed) }
+
+      it 'prints list of orphaned namespaces ignoring parents with same namespace as orphans' do
+        expect_list_of_orphans(%w(orphan1 orphan2 second-level))
+
+        subject.multi_check
+      end
+    end
+
+    context 'no orphans' do
+      let(:disk_namespaces) { %w(@hashed) }
+
+      it 'prints an empty list ignoring @hashed' do
+        expect_list_of_orphans([])
+
+        subject.multi_check
+      end
+    end
+  end
+
+  def expect_list_of_orphans(orphans)
+    expect(subject).to receive(:print_orphans).with(orphans, 'default')
+  end
+end

spec/lib/system_check/orphans/repository_check_spec.rb

+require 'spec_helper'
+require 'rake_helper'
+
+describe SystemCheck::Orphans::RepositoryCheck do
+  let(:storages) { Gitlab.config.repositories.storages.reject { |key, _| key.eql? 'broken' } }
+
+  before do
+    allow(Gitlab.config.repositories).to receive(:storages).and_return(storages)
+    allow(subject).to receive(:fetch_disk_namespaces).and_return(disk_namespaces)
+    allow(subject).to receive(:fetch_disk_repositories).and_return(disk_repositories)
+    # silence_output
+  end
+
+  describe '#multi_check' do
+    context 'all orphans' do
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 repos/@hashed) }
+      let(:disk_repositories) { %w(repo1.git repo2.git) }
+
+      it 'prints list of all orphaned namespaces except @hashed' do
+        expect_list_of_orphans(%w(orphan1/repo1.git orphan1/repo2.git orphan2/repo1.git orphan2/repo2.git))
+
+        subject.multi_check
+      end
+    end
+
+    context 'few orphans with existing namespace' do
+      let!(:first_level) { create(:group, path: 'my-namespace') }
+      let!(:project) { create(:project, path: 'repo', namespace: first_level) }
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 /repos/my-namespace /repos/@hashed) }
+      let(:disk_repositories) { %w(repo.git) }
+
+      it 'prints list of orphaned namespaces' do
+        expect_list_of_orphans(%w(orphan1/repo.git orphan2/repo.git))
+
+        subject.multi_check
+      end
+    end
+
+    context 'few orphans with existing namespace and parents with same name as orphans' do
+      let!(:first_level) { create(:group, path: 'my-namespace') }
+      let!(:second_level) { create(:group, path: 'second-level', parent: first_level) }
+      let!(:project) { create(:project, path: 'repo', namespace: first_level) }
+      let(:disk_namespaces) { %w(/repos/orphan1 /repos/orphan2 /repos/my-namespace /repos/second-level /repos/@hashed) }
+      let(:disk_repositories) { %w(repo.git) }
+
+      it 'prints list of orphaned namespaces ignoring parents with same namespace as orphans' do
+        expect_list_of_orphans(%w(orphan1/repo.git orphan2/repo.git second-level/repo.git))
+
+        subject.multi_check
+      end
+    end
+
+    context 'no orphans' do
+      let(:disk_namespaces) { %w(@hashed) }
+      let(:disk_repositories) { %w(repo.git) }
+
+      it 'prints an empty list ignoring @hashed' do
+        expect_list_of_orphans([])
+
+        subject.multi_check
+      end
+    end
+  end
+
+  def expect_list_of_orphans(orphans)
+    expect(subject).to receive(:print_orphans).with(orphans, 'default')
+  end
+end