Require user names to be unique after unicode normalisation

As with the previous checks on case sensitivity this only affects
new users, and changes to names of existing users.

diff --git a/app/models/user.rb b/app/models/user.rb
index 1942a25cc237a784394b91f9ba64c844f36e3309..5b48e3a68e53f4656a14e6639bfdc35b48a24f03 100644 (file)
--- a/app/models/user.rb
+++ b/app/models/user.rb
@@ -34,12 +34,13 @@
 #
 # Indexes
 #
-#  users_auth_idx                (auth_provider,auth_uid) UNIQUE
-#  users_display_name_idx        (display_name) UNIQUE
-#  users_display_name_lower_idx  (lower((display_name)::text))
-#  users_email_idx               (email) UNIQUE
-#  users_email_lower_idx         (lower((email)::text))
-#  users_home_idx                (home_tile)
+#  users_auth_idx                    (auth_provider,auth_uid) UNIQUE
+#  users_display_name_canonical_idx  (lower(NORMALIZE(display_name, NFKC)))
+#  users_display_name_idx            (display_name) UNIQUE
+#  users_display_name_lower_idx      (lower((display_name)::text))
+#  users_email_idx                   (email) UNIQUE
+#  users_email_lower_idx             (lower((email)::text))
+#  users_home_idx                    (home_tile)
 #
 
 class User < ApplicationRecord
@@ -91,7 +92,7 @@ class User < ApplicationRecord
   validates :display_name, :presence => true, :length => 3..255,
                            :exclusion => %w[new terms save confirm confirm-email go_public reset-password forgot-password suspended]
   validates :display_name, :if => proc { |u| u.display_name_changed? },
-                           :uniqueness => { :case_sensitive => false }
+                           :normalized_uniqueness => { :case_sensitive => false }
   validates :display_name, :if => proc { |u| u.display_name_changed? },
                            :characters => { :url_safe => true },
                            :whitespace => { :leading => false, :trailing => false }
@@ -129,7 +130,7 @@ class User < ApplicationRecord
       user = find_by("email = ? OR display_name = ?", options[:username].strip, options[:username])
 
       if user.nil?
-        users = where("LOWER(email) = LOWER(?) OR LOWER(display_name) = LOWER(?)", options[:username].strip, options[:username])
+        users = where("LOWER(email) = LOWER(?) OR LOWER(NORMALIZE(display_name, NFKC)) = LOWER(NORMALIZE(?, NFKC))", options[:username].strip, options[:username])
 
         user = users.first if users.count == 1
       end

diff --git a/app/validators/normalized_uniqueness_validator.rb b/app/validators/normalized_uniqueness_validator.rb
new file mode 100644 (file)
index 0000000..eb3600c
--- /dev/null
+++ b/app/validators/normalized_uniqueness_validator.rb
@@ -0,0 +1,18 @@
+class NormalizedUniquenessValidator < ActiveModel::EachValidator
+  def validate_each(record, attribute, value)
+    relation = if options.fetch(:case_sensitive, true)
+                 record.class.where("NORMALIZE(#{attribute}, NFKC) = NORMALIZE(?, NFKC)", value)
+               else
+                 record.class.where("LOWER(NORMALIZE(#{attribute}, NFKC)) = LOWER(NORMALIZE(?, NFKC))", value)
+               end
+
+    relation = relation.where.not(record.class.primary_key => [record.id_in_database]) if record.persisted?
+
+    if relation.exists?
+      error_options = options.except(:case_sensitive)
+      error_options[:value] = value
+
+      record.errors.add(attribute, :taken, **error_options)
+    end
+  end
+end

diff --git a/db/migrate/20231213182102_add_canonical_user_index.rb b/db/migrate/20231213182102_add_canonical_user_index.rb
new file mode 100644 (file)
index 0000000..905fb32
--- /dev/null
+++ b/db/migrate/20231213182102_add_canonical_user_index.rb
@@ -0,0 +1,7 @@
+class AddCanonicalUserIndex < ActiveRecord::Migration[7.1]
+  disable_ddl_transaction!
+
+  def change
+    add_index :users, "LOWER(NORMALIZE(display_name, NFKC))", :name => "users_display_name_canonical_idx", :algorithm => :concurrently
+  end
+end

diff --git a/db/structure.sql b/db/structure.sql
index ba60918f0ee1a93c8333aef7544fe52ed647cf0f..7c5233d67781fec8cf9d4a46d76f2d4e01762b0b 100644 (file)
--- a/db/structure.sql
+++ b/db/structure.sql
@@ -2867,6 +2867,13 @@ CREATE INDEX user_tokens_user_id_idx ON public.user_tokens USING btree (user_id)
 CREATE UNIQUE INDEX users_auth_idx ON public.users USING btree (auth_provider, auth_uid);
 
 
+--
+-- Name: users_display_name_canonical_idx; Type: INDEX; Schema: public; Owner: -
+--
+
+CREATE INDEX users_display_name_canonical_idx ON public.users USING btree (lower(NORMALIZE(display_name, NFKC)));
+
+
 --
 -- Name: users_display_name_idx; Type: INDEX; Schema: public; Owner: -
 --
@@ -3510,6 +3517,7 @@ INSERT INTO "schema_migrations" (version) VALUES
 ('23'),
 ('22'),
 ('21'),
+('20231213182102'),
 ('20231206141457'),
 ('20231117170422'),
 ('20231101222146'),

diff --git a/test/models/user_test.rb b/test/models/user_test.rb
index 5c48bb9698a6a73d56b7ab120b871d24902d18c5..1443e70b057356c770ef021ee6c742a926869f37 100644 (file)
--- a/test/models/user_test.rb
+++ b/test/models/user_test.rb
@@ -27,10 +27,13 @@ class UserTest < ActiveSupport::TestCase
   end
 
   def test_unique_display_name
-    existing_user = create(:user)
-    new_user = build(:user, :display_name => existing_user.display_name)
-    assert_not new_user.save
-    assert_includes new_user.errors[:display_name], "has already been taken"
+    create(:user, :display_name => "H\u{e9}nryIV")
+
+    %W[H\u{e9}nryIV he\u{301}nryiv H\u{c9}nry\u2163 he\u{301}nry\u2173].each do |name|
+      new_user = build(:user, :display_name => name)
+      assert_not new_user.save
+      assert_includes new_user.errors[:display_name], "has already been taken"
+    end
   end
 
   def test_email_valid