Merge branch '439089-switch-issue-title-to-code_analyzer' into 'master'

Switch issue.title to a new analyzer

See merge request https://gitlab.com/gitlab-org/gitlab/-/merge_requests/142580



Merged-by: Terri Chu <tchu@gitlab.com>
Approved-by: Madelein van Niekerk <mvanniekerk@gitlab.com>
Approved-by: Terri Chu <tchu@gitlab.com>
Reviewed-by: Terri Chu <tchu@gitlab.com>
Co-authored-by: rkumar555 <rkumar@gitlab.com>

ee/elastic/docs/20240123181031_reindex_issue_to_update_analyzer_for_title.yml

+---
+name: ReindexIssueToUpdateAnalyzerForTitle
+version: '20240123181031'
+description: This migration reindexes the issues index to start using new analyzer for title
+group: group::global search
+milestone: '16.9'
+introduced_by_url: https://gitlab.com/gitlab-org/gitlab/-/merge_requests/142580
+obsolete: false
+marked_obsolete_by_url:
+marked_obsolete_in_milestone:

ee/elastic/migrate/20240123181031_reindex_issue_to_update_analyzer_for_title.rb

+# frozen_string_literal: true
+
+class ReindexIssueToUpdateAnalyzerForTitle < Elastic::Migration
+  def migrate
+    Elastic::ReindexingTask.create!(targets: %w[Issue], options: { skip_pending_migrations_check: true })
+  end
+
+  def completed?
+    true
+  end
+end

ee/lib/elastic/latest/config.rb

@@ -25,6 +25,11 @@ module Config
           codec: 'best_compression',
           analysis: {
             analyzer: {
+              title_analyzer: {
+                tokenizer: 'standard',
+                filter: %w[word_delimiter_graph_filter asciifolding lowercase stemmer flatten_graph remove_duplicates]
+              },
+
               default: {
                 tokenizer: 'standard',
                 filter: %w[lowercase stemmer]

ee/lib/elastic/latest/issue_config.rb

@@ -22,7 +22,7 @@ module IssueConfig
         indexes :id, type: :integer
         indexes :iid, type: :integer
 
-        indexes :title, type: :text, index_options: 'positions'
+        indexes :title, type: :text, index_options: 'positions', analyzer: :title_analyzer
         indexes :description, type: :text, index_options: 'positions', analyzer: :code_analyzer
         indexes :created_at, type: :date
         indexes :updated_at, type: :date

ee/spec/elastic/migrate/20240123181031_reindex_issue_to_update_analyzer_for_title_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+require_relative 'migration_shared_examples'
+require File.expand_path('ee/elastic/migrate/20240123181031_reindex_issue_to_update_analyzer_for_title.rb')
+
+RSpec.describe ReindexIssueToUpdateAnalyzerForTitle, feature_category: :global_search do
+  let(:version) { 20240123181031 }
+  let(:migration) { described_class.new(version) }
+
+  it 'does not have migration options set', :aggregate_failures do
+    expect(migration).not_to be_batched
+    expect(migration).not_to be_retry_on_failure
+  end
+
+  describe '#migrate' do
+    it 'creates reindexing task with correct target and options' do
+      expect { migration.migrate }.to change { Elastic::ReindexingTask.count }.by(1)
+      task = Elastic::ReindexingTask.last
+      expect(task.targets).to eq(%w[Issue])
+      expect(task.options).to eq({ 'skip_pending_migrations_check' => true })
+    end
+  end
+
+  describe '#completed?' do
+    it 'always returns true' do
+      expect(migration.completed?).to eq(true)
+    end
+  end
+end

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

@@ -364,17 +364,7 @@
       expect(results.issues_count).to eq 0
     end
 
-    it 'handles plural words through algorithmic stemming', :aggregate_failures do
-      issue1 = create(:issue, project: project_1, title: 'remove :title attribute from submit buttons to prevent un-styled tooltips')
-      issue2 = create(:issue, project: project_1, title: 'smarter submit behavior for buttons groups')
-
-      ensure_elasticsearch_index!
-
-      results = described_class.new(user, 'button', limit_project_ids)
-
-      expect(results.objects('issues')).to contain_exactly(issue1, issue2)
-      expect(results.issues_count).to eq 2
-    end
+    it_behaves_like 'can search by title for miscellaneous cases', 'issues'
 
     it 'executes count only queries' do
       results = described_class.new(user, query, limit_project_ids)

ee/spec/support/shared_examples/services/search_service_shared_examples.rb

@@ -118,3 +118,66 @@
     end
   end
 end
+
+RSpec.shared_examples 'can search by title for miscellaneous cases' do |type|
+  let_it_be(:searched_project) { create(:project, :public, :repository, :wiki_repo) }
+  let(:records_count) { 2 }
+
+  def create_records!(type)
+    case type
+    when 'issues'
+      create_list(:issue, records_count, project: searched_project)
+    end
+  end
+
+  # rubocop:disable RSpec/InstanceVariable -- Want to reuse the @records
+  before do
+    @records = create_records!(type)
+  end
+
+  it 'handles plural words through algorithmic stemming', :aggregate_failures do
+    @records[0].update!(title: 'remove :title attribute from submit buttons to prevent un-styled tooltips')
+    @records[1].update!(title: 'smarter submit behavior for buttons groups')
+    ensure_elasticsearch_index!
+    results = described_class.new(user, 'button', [searched_project.id])
+    expect(results.objects(type)).to match_array(@records)
+    expect(results.issues_count).to eq records_count
+  end
+
+  it 'handles if title has umlauts', :aggregate_failures do
+    @records[0].update!(title: 'köln')
+    @records[1].update!(title: 'kǒln')
+    ensure_elasticsearch_index!
+    results = described_class.new(user, 'koln', [searched_project.id])
+    expect(results.objects(type)).to match_array(@records)
+    expect(results.issues_count).to eq records_count
+  end
+
+  it 'handles if title has dots', :aggregate_failures do
+    @records[0].update!(title: 'with.dot.title')
+    @records[1].update!(title: 'there is.dot')
+    ensure_elasticsearch_index!
+    results = described_class.new(user, 'dot', [searched_project.id])
+    expect(results.objects(type)).to match_array(@records)
+    expect(results.issues_count).to eq records_count
+  end
+
+  it 'handles if title has underscore', :aggregate_failures do
+    @records[0].update!(title: 'with_underscore_text')
+    @records[1].update!(title: 'some_underscore')
+    ensure_elasticsearch_index!
+    results = described_class.new(user, 'underscore', [searched_project.id])
+    expect(results.objects(type)).to match_array(@records)
+    expect(results.issues_count).to eq records_count
+  end
+
+  it 'handles if title has camelcase', :aggregate_failures do
+    @records[0].update!(title: 'withCamelcaseTitle')
+    @records[1].update!(title: 'CamelcaseText')
+    ensure_elasticsearch_index!
+    results = described_class.new(user, 'Camelcase', [searched_project.id])
+    expect(results.objects(type)).to match_array(@records)
+    expect(results.issues_count).to eq records_count
+  end
+  # rubocop:enable RSpec/InstanceVariable
+end