Implement Balanced Full-Text Search for Diary Entries

Gemfile

@@ -10,6 +10,9 @@ gem "json"
 # Use postgres as the database
 gem "pg"
 
+# Use postgres search
+gem "pg_search"
+
 # Use SCSS for stylesheets
 gem "dartsass-sprockets"
 # Pin the dependentent sass-embedded to avoid deprecation warnings in bootstrap

Gemfile.lock

@@ -414,6 +414,9 @@ GEM
       ast (~> 2.4.1)
       racc
     pg (1.5.7)
+    pg_search (2.3.7)
+      activerecord (>= 6.1)
+      activesupport (>= 6.1)
     popper_js (2.11.8)
     progress (3.6.0)
     psych (5.1.2)
@@ -680,6 +683,7 @@ DEPENDENCIES
   openstreetmap-deadlock_retry (>= 1.3.1)
   overcommit
   pg
+  pg_search
   puma (~> 5.6)
   quad_tile (~> 1.0.1)
   rack-cors

app/controllers/diary_entries_controller.rb

@@ -48,6 +48,9 @@ def index
       if params[:language]
         @title = t ".in_language_title", :language => Language.find(params[:language]).english_name
         entries = entries.where(:language_code => params[:language])
+      elsif params[:query].present?
+        @title = t ".search_results_title"
+        entries = DiaryEntry.search_by_title_and_body(params[:query])
       else
         candidate_codes = preferred_languages.flat_map(&:candidates).uniq.map(&:to_s)
         @languages = Language.where(:code => candidate_codes).in_order_of(:code, candidate_codes)
@@ -241,3 +244,47 @@ def set_map_location
     end
   end
 end
+
+def test_index_search
+  user = create(:user)
+  create(:diary_entry, :user => user, :title => "First Entry", :body => "This is the first diary entry.")
+  create(:diary_entry, :user => user, :title => "Second Entry", :body => "This is the second diary entry.")
+  create(:diary_entry, :user => user, :title => "Third Entry", :body => "Nothing related to the search term.")
+
+  # Test search by title
+  get diary_entries_path(:query => "First")
+  assert_response :success
+  assert_select "article.diary_post", 1
+  assert_select "h2", :text => /First Entry/
+
+  # Test search by body
+  get diary_entries_path(:query => "second")
+  assert_response :success
+  assert_select "article.diary_post", 1
+  assert_select "h2", :text => /Second Entry/
+
+  # Test no results
+  get diary_entries_path(:query => "Nonexistent")
+  assert_response :success
+  assert_select "h4", :text => /No entries/, :count => 1
+end
+
+def test_index_search_language
+  user = create(:user)
+  create(:language, :code => "en")
+  create(:language, :code => "de")
+
+  create(:diary_entry, :user => user, :title => "English Entry", :language_code => "en")
+  create(:diary_entry, :user => user, :title => "German Entry", :language_code => "de")
+
+  # Search within a specific language
+  get diary_entries_path(:query => "Entry", :language => "en")
+  assert_response :success
+  assert_select "article.diary_post", 1
+  assert_select "h2", :text => /English Entry/
+
+  get diary_entries_path(:query => "Entry", :language => "de")
+  assert_response :success
+  assert_select "article.diary_post", 1
+  assert_select "h2", :text => /German Entry/
+end

app/models/diary_entry.rb

@@ -35,6 +35,16 @@ class DiaryEntry < ApplicationRecord
   has_many :subscriptions, :class_name => "DiaryEntrySubscription"
   has_many :subscribers, :through => :subscriptions, :source => :user
 
+  include PgSearch::Model
+  pg_search_scope :search_by_title_and_body,
+                  :against => [:title, :body],
+                  :using => {
+                    :tsearch => {
+                      :dictionary => "simple",
+                      :tsvector_column => "searchable"
+                    }
+                  }
+
   scope :visible, -> { where(:visible => true) }
 
   validates :title, :presence => true, :length => 1..255, :characters => true

db/migrate/20240903014508_add_searchable_column_to_diary_entries.rb

@@ -0,0 +1,22 @@
+class AddSearchableColumnToDiaryEntries < ActiveRecord::Migration[7.1]
+  def up
+    safety_assured do
+      execute <<-SQL.squish
+        ALTER TABLE diary_entries
+        ADD COLUMN searchable tsvector GENERATED ALWAYS AS (
+          setweight(to_tsvector('simple', coalesce(title, '')), 'A') ||
+          setweight(to_tsvector('simple', coalesce(body, '')), 'B')
+        ) STORED;
+      SQL
+    end
+  end
+
+  def down
+    safety_assured do
+      execute <<-SQL.squish
+        ALTER TABLE diary_entries
+        DROP COLUMN IF EXISTS searchable;
+      SQL
+    end
+  end
+end

db/migrate/20240903180252_add_index_to_searchable_diary_entries.rb

@@ -0,0 +1,15 @@
+class AddIndexToSearchableDiaryEntries < ActiveRecord::Migration[7.1]
+  disable_ddl_transaction!
+
+  def up
+    safety_assured do
+      add_index :diary_entries, :searchable, :using => :gin, :algorithm => :concurrently
+    end
+  end
+
+  def down
+    safety_assured do
+      remove_index :diary_entries, :searchable, :algorithm => :concurrently
+    end
+  end
+end

db/structure.sql

@@ -717,7 +717,8 @@ CREATE TABLE public.diary_entries (
     longitude double precision,
     language_code character varying DEFAULT 'en'::character varying NOT NULL,
     visible boolean DEFAULT true NOT NULL,
-    body_format public.format_enum DEFAULT 'markdown'::public.format_enum NOT NULL
+    body_format public.format_enum DEFAULT 'markdown'::public.format_enum NOT NULL,
+    searchable tsvector GENERATED ALWAYS AS ((setweight(to_tsvector('simple'::regconfig, (COALESCE(title, ''::character varying))::text), 'A'::"char") || setweight(to_tsvector('simple'::regconfig, COALESCE(body, ''::text)), 'B'::"char"))) STORED
 );
 
 
@@ -2414,6 +2415,13 @@ CREATE INDEX index_changesets_subscribers_on_changeset_id ON public.changesets_s
 CREATE UNIQUE INDEX index_changesets_subscribers_on_subscriber_id_and_changeset_id ON public.changesets_subscribers USING btree (subscriber_id, changeset_id);
 
 
+--
+-- Name: index_diary_entries_on_searchable; Type: INDEX; Schema: public; Owner: -
+--
+
+CREATE INDEX index_diary_entries_on_searchable ON public.diary_entries USING gin (searchable);
+
+
 --
 -- Name: index_diary_entry_subscriptions_on_diary_entry_id; Type: INDEX; Schema: public; Owner: -
 --
@@ -3349,6 +3357,8 @@ INSERT INTO "schema_migrations" (version) VALUES
 ('23'),
 ('22'),
 ('21'),
+('20240903180252'),
+('20240903014508'),
 ('20240822121603'),
 ('20240813070506'),
 ('20240724194738'),