encoding_spec.rb

# encoding: utf-8
require 'active_record'

ActiveRecord::Base.establish_connection(
  :adapter  => 'mysql2',
  :host     => '127.0.0.1',
  :port => 13306,
  :username => 'root',
  :database => 'encoding_test',
  :encoding => 'latin1')

conn = ActiveRecord::Base.connection
conn.execute('set sql_mode="STRICT_TRANS_TABLES"')

def create_table(name, charset)
  ActiveRecord::Base.connection.create_table :test, :options => "ENGINE=InnoDB DEFAULT CHARSET=#{charset}" do |t|
    yield t
  end
end

def recreate_test_table(encoding)
  conn.drop_table('test') if conn.table_exists?('test')
  create_table('test', encoding) do |t|
    t.string :name
  end
end

def set_client_encoding(encoding)
  conn.execute("set names #{encoding}")
end

def insert_name
  conn.execute("insert into test (`name`) values ('#{utf8_name}')")
end

describe 'mysql encoding' do
  let(:utf8_name) { 'testş' }
  let(:conn)      { ActiveRecord::Base.connection }
  subject(:name)  { conn.select_value('select name from test limit 1') }
  before { conn.execute('truncate table test') }

  context 'latin1 table' do
    before do
      recreate_test_table('latin1')
    end

    context 'latin1 client' do
      before { set_client_encoding('latin1') }

      it 'stores utf8 data unchanged' do
        insert_name
        name.bytes.should eq utf8_name.bytes
        name.should_not eq utf8_name
      end

      it 'comes back marked as latin1' do
        insert_name
        name.encoding.should eq Encoding::ISO_8859_1
      end

      context 'when marked as utf8' do
        before do
          insert_name
          name.force_encoding('utf-8')
        end

        it { should eq utf8_name }
      end

      context 'when client connection changes to utf8 client' do
        before do
          insert_name
          set_client_encoding('utf8')
        end

        it 'comes back double encoded' do
          name.should eq "testÅŸ"
          name.should eq conn.select_value("SELECT CAST(_latin1'testş' AS CHAR CHARACTER SET utf8)")
        end

        it 'can be recovered' do
          recovered = conn.select_value("SELECT CONVERT(CAST(CONVERT('#{name}' USING latin1) AS BINARY) USING utf8)")
          recovered.should eq utf8_name
        end
      end

      context 'after a mysqldump and restore' do
        # its the same
        before do
          insert_name
          `mysqldump -u root --default-character-set=latin1 encoding_test > /tmp/encoding_test.sql`
          `mysql -u root encoding_test < /tmp/encoding_test.sql`
        end

        it 'stores utf8 data unchanged' do
          name.bytes.should eq utf8_name.bytes
          name.should_not eq utf8_name
        end

        context 'and utf8 client' do
          before { set_client_encoding('utf8') }

          it 'is double encoded' do
            name.should eq "testÅŸ"
          end
        end

        context 'when marked as utf8' do
          before do
            name.force_encoding('utf-8')
          end

          it { should eq utf8_name }
        end
      end
    end

    context 'utf8 client' do
      before { set_client_encoding('utf8') }

      context 'with strict_trans_tables' do
        before do
          conn.execute('set sql_mode="STRICT_TRANS_TABLES"')
        end

        it 'comes back double encoded' do
          expect {
            insert_name
          }.to raise_error(ActiveRecord::StatementInvalid)
        end
      end

      context 'without strict_trans_tables' do
        before do
          conn.execute('set sql_mode=""')
          insert_name
        end

        it 'replaces unknown chars with ?' do
          name.should eq "test?"
        end
      end
    end
  end
end