Merge branch '342648-add-language-filter-support-for-blob-search' into 'master'

Support language filter for blob searches

See merge request !88674

app/helpers/search_helper.rb

 # frozen_string_literal: true
 
 module SearchHelper
+  # params which should persist when a new tab is selected
   SEARCH_GENERIC_PARAMS = [
     :search,
     :scope,
@@ -129,7 +130,7 @@ def search_blob_title(project, path)
   end
 
   def search_service
-    @search_service ||= ::SearchService.new(current_user, params.merge(confidential: Gitlab::Utils.to_boolean(params[:confidential])))
+    @search_service ||= ::SearchService.new(current_user, sanitized_search_params)
   end
 
   def search_sort_options
@@ -481,6 +482,13 @@ def issuable_state_text(issuable)
   def feature_flag_tab_enabled?(flag)
     @group || Feature.enabled?(flag, current_user, type: :ops)
   end
+
+  def sanitized_search_params
+    sanitized_params = params.dup
+    sanitized_params[:confidential] = Gitlab::Utils.to_boolean(sanitized_params[:confidential]) if sanitized_params.key?(:confidential)
+
+    sanitized_params
+  end
 end
 
 SearchHelper.prepend_mod_with('SearchHelper')

ee/app/services/ee/search/global_service.rb

@@ -18,7 +18,7 @@ def execute
           public_and_internal_projects: elastic_global,
           order_by: params[:order_by],
           sort: params[:sort],
-          filters: { confidential: params[:confidential], state: params[:state] }
+          filters: { confidential: params[:confidential], state: params[:state], language: params[:language] }
         )
       end
 

ee/app/services/ee/search/group_service.rb

@@ -32,7 +32,7 @@ def execute
           public_and_internal_projects: elastic_global,
           order_by: params[:order_by],
           sort: params[:sort],
-          filters: { confidential: params[:confidential], state: params[:state] }
+          filters: { confidential: params[:confidential], state: params[:state], language: params[:language] }
         )
       end
 

ee/app/services/ee/search/project_service.rb

@@ -13,26 +13,31 @@ def execute
         return super if project.respond_to?(:archived?) && project.archived?
         return super unless use_elasticsearch? && use_default_branch?
 
+        search = params[:search]
+        order_by = params[:order_by]
+        sort = params[:sort]
+        filters = { confidential: params[:confidential], state: params[:state], language: params[:language] }
+
         if project.is_a?(Array)
           project_ids = Array(project).map(&:id)
           ::Gitlab::Elastic::SearchResults.new(
             current_user,
-            params[:search],
+            search,
             project_ids,
             public_and_internal_projects: false,
-            order_by: params[:order_by],
-            sort: params[:sort],
-            filters: { confidential: params[:confidential], state: params[:state] }
+            order_by: order_by,
+            sort: sort,
+            filters: filters
           )
         else
           ::Gitlab::Elastic::ProjectSearchResults.new(
             current_user,
-            params[:search],
+            search,
             project: project,
             repository_ref: repository_ref,
-            order_by: params[:order_by],
-            sort: params[:sort],
-            filters: { confidential: params[:confidential], state: params[:state] }
+            order_by: order_by,
+            sort: sort,
+            filters: filters
           )
         end
       end

ee/lib/elastic/latest/git_class_proxy.rb

@@ -57,7 +57,7 @@ def options_filter_context(type, options)
           }
         end
 
-        if languages.any?
+        if type == :blob && languages.any? && allow_aggregations?(type.to_s, options[:count_only], options[:current_user])
           filters << {
             terms: {
               _name: context.name(type, :match, :languages),
@@ -215,7 +215,7 @@ def search_blob(query, type: 'blob', page: 1, per: 20, options: {})
           }
         end
 
-        if include_aggregations?(type, options[:count_only], options[:current_user])
+        if allow_aggregations?(type, options[:count_only], options[:current_user])
           query_hash[:aggs] = {
             language: {
               composite: {
@@ -320,7 +320,7 @@ def apply_simple_query_string(name:, fields:, query:, bool_expr:, count_only:)
         end
       end
 
-      def include_aggregations?(type, count_only, current_user)
+      def allow_aggregations?(type, count_only, current_user)
         type == 'blob' && !count_only && ::Feature.enabled?(:search_blobs_language_aggregation, current_user)
       end
     end

ee/lib/gitlab/elastic/project_search_results.rb

@@ -26,7 +26,7 @@ def blobs(page: 1, per_page: DEFAULT_PER_PAGE, count_only: false, preload_method
             query,
             page: (page || 1).to_i,
             per: per_page,
-            options: base_options.merge(count_only: count_only),
+            options: base_options.merge(count_only: count_only).merge(filters.slice(:language)),
             preload_method: preload_method
           )
         end

ee/lib/gitlab/elastic/search_results.rb

@@ -308,7 +308,7 @@ def blobs(page: 1, per_page: DEFAULT_PER_PAGE, count_only: false, preload_method
             query,
             page: (page || 1).to_i,
             per: per_page,
-            options: base_options.merge(count_only: count_only),
+            options: base_options.merge(count_only: count_only).merge(filters.slice(:language)),
             preload_method: preload_method
           )
         end

ee/spec/lib/gitlab/elastic/group_search_results_spec.rb

@@ -10,7 +10,7 @@
   let(:filters) { {} }
   let(:query) { '*' }
 
-  subject(:results) { described_class.new(user, query, Project.all.pluck_primary_key, group: group, filters: filters) }
+  subject(:results) { described_class.new(user, query, group.projects.pluck_primary_key, group: group, filters: filters) }
 
   before do
     stub_ee_application_setting(elasticsearch_search: true, elasticsearch_indexing: true)
@@ -50,6 +50,12 @@
     end
   end
 
+  context 'blobs' do
+    let!(:project) { create(:project, :public, :repository, group: group) }
+
+    it_behaves_like 'search results filtered by language'
+  end
+
   context 'query performance' do
     include_examples 'does not hit Elasticsearch twice for objects and counts', %w[projects notes blobs wiki_blobs commits issues merge_requests milestones]
     include_examples 'does not load results for count only queries', %w[projects notes blobs wiki_blobs commits issues merge_requests milestones]

ee/spec/lib/gitlab/elastic/project_search_results_spec.rb

@@ -70,7 +70,7 @@
     end
 
     context 'filtering' do
-      let!(:project) { create(:project, :public) }
+      let!(:project) { create(:project, :public, :repository) }
       let(:query) { 'foo' }
 
       context 'issues' do
@@ -100,6 +100,10 @@
 
         include_examples 'search results filtered by state'
       end
+
+      context 'blobs' do
+        it_behaves_like 'search results filtered by language'
+      end
     end
   end
 

ee/spec/lib/gitlab/elastic/search_results_spec.rb

@@ -770,7 +770,7 @@
     end
   end
 
-  describe 'project scoping' do
+  describe 'projects' do
     it "returns items for project" do
       project = create :project, :repository, name: "term"
       project.add_developer(user)
@@ -1014,6 +1014,13 @@ def self.ruby_method_123(ruby_another_method_arg)
         expect(search_for('ruby_call_method_123')).to include(file_name)
       end
     end
+
+    context 'filtering' do
+      let(:project) { project_1 }
+      let(:results) { described_class.new(user, query, [project.id], filters: filters) }
+
+      it_behaves_like 'search results filtered by language'
+    end
   end
 
   describe 'wikis', :sidekiq_inline do

spec/support/shared_examples/lib/gitlab/search_language_filter_shared_examples.rb

+# frozen_string_literal: true
+
+RSpec.shared_examples 'search results filtered by language' do
+  let(:scope) { 'blobs' }
+  let(:filters) { { language: %w[Ruby Markdown] } }
+  let(:query) { 'def | popen | test' }
+
+  before do
+    project.repository.index_commits_and_blobs
+
+    ensure_elasticsearch_index!
+  end
+
+  subject(:blob_results) { results.objects('blobs') }
+
+  it 'filters by language', :sidekiq_inline, :aggregate_failures do
+    expected_paths = %w[
+      files/ruby/popen.rb
+      files/markdown/ruby-style-guide.md
+      files/ruby/regex.rb
+      files/ruby/version_info.rb
+      CONTRIBUTING.md
+    ]
+
+    paths = blob_results.map { |blob| blob.binary_path }
+    expect(blob_results.size).to eq(5)
+    expect(paths).to match_array(expected_paths)
+  end
+
+  context 'when the search_blobs_language_aggregation feature flag is disabled' do
+    before do
+      stub_feature_flags(search_blobs_language_aggregation: false)
+    end
+
+    it 'does not filter by language', :sidekiq_inline, :aggregate_failures do
+      expected_paths = %w[
+        CHANGELOG
+        CONTRIBUTING.md
+        bar/branch-test.txt
+        custom-highlighting/test.gitlab-custom
+        files/ruby/popen.rb
+        files/ruby/regex.rb
+        files/ruby/version_info.rb
+        files/whitespace
+        encoding/test.txt
+        files/markdown/ruby-style-guide.md
+      ]
+
+      paths = blob_results.map { |blob| blob.binary_path }
+      expect(blob_results.size).to eq(10)
+      expect(paths).to match_array(expected_paths)
+    end
+  end
+end