Merge branch 'fix-tsvector-timeout' into 'master'

Update full-text regex to strip out strings with @

See merge request gitlab-org/gitlab!83009

app/models/concerns/pg_full_text_searchable.rb

@@ -21,7 +21,7 @@
 module PgFullTextSearchable
   extend ActiveSupport::Concern
 
-  LONG_WORDS_REGEX = %r([A-Za-z0-9+/]{50,}).freeze
+  LONG_WORDS_REGEX = %r([A-Za-z0-9+/@]{50,}).freeze
   TSVECTOR_MAX_LENGTH = 1.megabyte.freeze
   TEXT_SEARCH_DICTIONARY = 'english'
 

lib/gitlab/background_migration/backfill_issue_search_data.rb

@@ -29,7 +29,7 @@ module Gitlab
           SELECT
             project_id,
             id,
-            setweight(to_tsvector('english', LEFT(title, 255)), 'A') || setweight(to_tsvector('english', LEFT(REGEXP_REPLACE(description, '[A-Za-z0-9+/]{50,}', ' ', 'g'), 1048576)), 'B'),
+            setweight(to_tsvector('english', LEFT(title, 255)), 'A') || setweight(to_tsvector('english', LEFT(REGEXP_REPLACE(description, '[A-Za-z0-9+/@]{50,}', ' ', 'g'), 1048576)), 'B'),
             NOW(),
             NOW()
           FROM issues

spec/models/concerns/pg_full_text_searchable_spec.rb

@@ -115,6 +115,21 @@ RSpec.describe PgFullTextSearchable do
       end
     end
 
+    context 'with long words' do
+      let(:model) { model_class.create!(project: project, title: 'title ' + 'long/sequence+1' * 4, description: 'description ' + '@user1' * 20) }
+
+      it 'strips words that are 50 characters or longer' do
+        model.update_search_data!
+
+        expect(model.search_data.search_vector).to match(/'titl':1A/)
+        expect(model.search_data.search_vector).not_to match(/long/)
+        expect(model.search_data.search_vector).not_to match(/sequence/)
+
+        expect(model.search_data.search_vector).to match(/'descript':2B/)
+        expect(model.search_data.search_vector).not_to match(/@user1/)
+      end
+    end
+
     context 'when upsert times out' do
       it 're-raises the exception' do
         expect(Issues::SearchData).to receive(:upsert).once.and_raise(ActiveRecord::StatementTimeout)