Merge branch 'es_project_search' into 'master'

Fix: Elasticsearch does not find partial matches in project names

Closes https://gitlab.com/gitlab-org/gitlab-ee/issues/960

See merge request !726

CHANGELOG-EE

@@ -3,6 +3,7 @@ v 8.12.0 (Unreleased)
   - Reduce UPDATE queries when moving between import states on projects
   - [ES] Instrument Elasticsearch::Git::Repository
   - [ES] Instrument other Gitlab::Elastic classes
+  - [ES] Fix: Elasticsearch does not find partial matches in project names
 
 v 8.11.6
   - Exclude blocked users from potential MR approvers

app/models/concerns/elastic/application_search.rb

@@ -13,14 +13,26 @@ module Elastic
           analysis: {
             analyzer: {
               default: {
-                tokenizer: "standard",
-                filter: ["standard", "lowercase", "my_stemmer"]
+                tokenizer: 'standard',
+                filter: ['standard', 'lowercase', 'my_stemmer']
+              },
+              my_ngram_analyzer: {
+                tokenizer: 'my_ngram_tokenizer',
+                filter: ['lowercase']
               }
             },
             filter: {
               my_stemmer: {
-                type: "stemmer",
-                name: "light_english"
+                type: 'stemmer',
+                name: 'light_english'
+              }
+            },
+            tokenizer: {
+              my_ngram_tokenizer: {
+                type: 'nGram',
+                min_gram: 2,
+                max_gram: 3,
+                token_chars: [ 'letter', 'digit' ]
               }
             }
           }

app/models/concerns/elastic/projects_search.rb

@@ -12,7 +12,8 @@ module Elastic
         indexes :path,                type: :string,
                                       index_options: 'offsets'
         indexes :name_with_namespace, type: :string,
-                                      index_options: 'offsets'
+                                      index_options: 'offsets',
+                                      analyzer: :my_ngram_analyzer
         indexes :path_with_namespace, type: :string,
                                       index_options: 'offsets'
         indexes :description,         type: :string,

lib/tasks/gitlab/elastic.rake

@@ -58,7 +58,7 @@ namespace :gitlab do
 
     desc "GitLab | Elasticsearch | Index wiki repositories"
     task index_wikis: :environment  do
-      projects = apply_project_filters(Project.where(wiki_enabled: true))
+      projects = apply_project_filters(Project.with_wiki_enabled)
 
       projects.find_each do |project|
         unless project.wiki.empty?

spec/models/concerns/elastic/project_spec.rb

@@ -11,7 +11,7 @@ describe Project, elastic: true do
     stub_application_setting(elasticsearch_search: false, elasticsearch_indexing: false)
   end
 
-  it "searches projects" do
+  it "finds projects" do
     project_ids = []
 
     Sidekiq::Testing.inline! do
@@ -29,6 +29,20 @@ describe Project, elastic: true do
     expect(described_class.elastic_search('someone_elses_project', options: { pids: project_ids }).total_count).to eq(0)
   end
 
+  it "finds partial matches in project names" do
+    project_ids = []
+
+    Sidekiq::Testing.inline! do
+      project = create :empty_project, name: 'tesla-model-s'
+      project1 = create :empty_project, name: 'tesla_model_s'
+      project_ids += [project.id, project1.id]
+
+      Gitlab::Elastic::Helper.refresh_index
+    end
+
+    expect(described_class.elastic_search('tesla', options: { pids: project_ids }).total_count).to eq(2)
+  end
+
   it "returns json with all needed elements" do
     project = create :empty_project