Improve tag search query (#16104)

diff --git a/app/models/tag.rb b/app/models/tag.rb
index bb93a52e2e56e6717a6b50c651a069e8367f50a6..efffc7eee4c2411d5d06eda27e573ed0b018a0f7 100644 (file)
--- a/app/models/tag.rb
+++ b/app/models/tag.rb
@@ -40,7 +40,8 @@ class Tag < ApplicationRecord
   scope :trendable, -> { Setting.trendable_by_default ? where(trendable: [true, nil]) : where(trendable: true) }
   scope :discoverable, -> { listable.joins(:account_tag_stat).where(AccountTagStat.arel_table[:accounts_count].gt(0)).order(Arel.sql('account_tag_stats.accounts_count desc')) }
   scope :recently_used, ->(account) { joins(:statuses).where(statuses: { id: account.statuses.select(:id).limit(1000) }).group(:id).order(Arel.sql('count(*) desc')) }
-  scope :matches_name, ->(value) { where(arel_table[:name].matches("#{value}%")) }
+  # Search with case-sensitive to use B-tree index.
+  scope :matches_name, ->(term) { where(arel_table[:name].lower.matches(arel_table.lower("#{sanitize_sql_like(Tag.normalize(term))}%"), nil, true)) }
 
   delegate :accounts_count,
            :accounts_count=,
@@ -126,10 +127,9 @@ class Tag < ApplicationRecord
     end
 
     def search_for(term, limit = 5, offset = 0, options = {})
-      normalized_term = normalize(term.strip)
-      pattern         = sanitize_sql_like(normalized_term) + '%'
-      query           = Tag.listable.where(arel_table[:name].lower.matches(pattern))
-      query           = query.where(arel_table[:name].lower.eq(normalized_term).or(arel_table[:reviewed_at].not_eq(nil))) if options[:exclude_unreviewed]
+      striped_term = term.strip
+      query = Tag.listable.matches_name(striped_term)
+      query = query.merge(matching_name(striped_term).or(where.not(reviewed_at: nil))) if options[:exclude_unreviewed]
 
       query.order(Arel.sql('length(name) ASC, name ASC'))
            .limit(limit)
@@ -145,7 +145,7 @@ class Tag < ApplicationRecord
     end
 
     def matching_name(name_or_names)
-      names = Array(name_or_names).map { |name| normalize(name).mb_chars.downcase.to_s }
+      names = Array(name_or_names).map { |name| arel_table.lower(normalize(name)) }
 
       if names.size == 1
         where(arel_table[:name].lower.eq(names.first))
@@ -154,8 +154,6 @@ class Tag < ApplicationRecord
       end
     end
 
-    private
-
     def normalize(str)
       str.gsub(/\A#/, '')
     end

diff --git a/db/migrate/20210421121431_add_case_insensitive_btree_index_to_tags.rb b/db/migrate/20210421121431_add_case_insensitive_btree_index_to_tags.rb
new file mode 100644 (file)
index 0000000..ed359e8
--- /dev/null
+++ b/db/migrate/20210421121431_add_case_insensitive_btree_index_to_tags.rb
@@ -0,0 +1,13 @@
+class AddCaseInsensitiveBtreeIndexToTags < ActiveRecord::Migration[5.2]
+  disable_ddl_transaction!
+
+  def up
+    safety_assured { execute 'CREATE UNIQUE INDEX CONCURRENTLY index_tags_on_name_lower_btree ON tags (lower(name) text_pattern_ops)' }
+    remove_index :tags, name: 'index_tags_on_name_lower'
+  end
+
+  def down
+    safety_assured { execute 'CREATE UNIQUE INDEX CONCURRENTLY index_tags_on_name_lower ON tags (lower(name))' }
+    remove_index :tags, name: 'index_tags_on_name_lower_btree'
+  end
+end

diff --git a/db/schema.rb b/db/schema.rb
index dcbbf4aea99283d85b1e878776941e00b007fe72..8dc0661fc92376236713a3e9be5df922662b39c8 100644 (file)
--- a/db/schema.rb
+++ b/db/schema.rb
@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema.define(version: 2021_04_16_200740) do
+ActiveRecord::Schema.define(version: 2021_04_21_121431) do
 
   # These are extensions that must be enabled in order to support this database
   enable_extension "plpgsql"
@@ -862,7 +862,7 @@ ActiveRecord::Schema.define(version: 2021_04_16_200740) do
     t.datetime "last_status_at"
     t.float "max_score"
     t.datetime "max_score_at"
-    t.index "lower((name)::text)", name: "index_tags_on_name_lower", unique: true
+    t.index "lower((name)::text) text_pattern_ops", name: "index_tags_on_name_lower_btree", unique: true
   end
 
   create_table "tombstones", force: :cascade do |t|

diff --git a/spec/models/tag_spec.rb b/spec/models/tag_spec.rb
index df876593c9e026d9b1f0c103041a34719e112234..3949dbce548ee1d9507dc8d74fd8ae0d0546719b 100644 (file)
--- a/spec/models/tag_spec.rb
+++ b/spec/models/tag_spec.rb
@@ -96,6 +96,20 @@ RSpec.describe Tag, type: :model do
     end
   end
 
+  describe '.matches_name' do
+    it 'returns tags for multibyte case-insensitive names' do
+      upcase_string   = 'abcABCａｂｃＡＢＣやゆよ'
+      downcase_string = 'abcabcａｂｃａｂｃやゆよ';
+
+      tag = Fabricate(:tag, name: downcase_string)
+      expect(Tag.matches_name(upcase_string)).to eq [tag]
+    end
+
+    it 'uses the LIKE operator' do
+      expect(Tag.matches_name('100%abc').to_sql).to eq %q[SELECT "tags".* FROM "tags" WHERE LOWER("tags"."name") LIKE LOWER('100\\%abc%')]
+    end
+  end
+
   describe '.matching_name' do
     it 'returns tags for multibyte case-insensitive names' do
       upcase_string   = 'abcABCａｂｃＡＢＣやゆよ'