class PhpTransliterationTest in Zircon Profile 8

Tests Transliteration component functionality.

@group Transliteration

@coversDefaultClass \Drupal\Component\Transliteration\PhpTransliteration

Hierarchy

• class \Drupal\Tests\UnitTestCase extends \Drupal\Tests\PHPUnit_Framework_TestCase
  • class \Drupal\Tests\Component\Transliteration\PhpTransliterationTest

File

core/tests/Drupal/Tests/Component/Transliteration/PhpTransliterationTest.php, line 22

Contains \Drupal\Tests\Component\Transliteration\PhpTransliterationTest.

Namespace

Drupal\Tests\Component\Transliteration

class PhpTransliterationTest extends UnitTestCase {

  /**
   * Tests the PhpTransliteration::removeDiacritics() function.
   *
   * @param string $original
   *   The language code to test.
   * @param string $expected
   *   The expected return from PhpTransliteration::removeDiacritics().
   *
   * @dataProvider providerTestPhpTransliterationRemoveDiacritics
   */
  public function testRemoveDiacritics($original, $expected) {
    $transliterator_class = new PhpTransliteration();
    $result = $transliterator_class
      ->removeDiacritics($original);
    $this
      ->assertEquals($expected, $result);
  }

  /**
   * Provides data for self::testRemoveDiacritics().
   *
   * @return array
   *   An array of arrays, each containing the parameters for
   *   self::testRemoveDiacritics().
   */
  public function providerTestPhpTransliterationRemoveDiacritics() {
    return array(
      // Test all characters in the Unicode range 0x00bf to 0x017f.
      array(
        'ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏ',
        'AAAAAAÆCEEEEIIII',
      ),
      array(
        'ÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞß',
        'ÐNOOOOO×OUUUUYÞß',
      ),
      array(
        'àáâãäåæçèéêëìíîï',
        'aaaaaaæceeeeiiii',
      ),
      array(
        'ðñòóôõö÷øùúûüýþÿ',
        'ðnooooo÷ouuuuyþy',
      ),
      array(
        'ĀāĂ㥹ĆćĈĉĊċČčĎď',
        'AaAaAaCcCcCcCcDd',
      ),
      array(
        'ĐđĒēĔĕĖėĘęĚěĜĝĞğ',
        'DdEeEeEeEeEeGgGg',
      ),
      array(
        'ĠġĢģĤĥĦħĨĩĪīĬĭĮį',
        'GgGgHhHhIiIiIiIi',
      ),
      array(
        'İıIJijĴĵĶķĸĹĺĻļĽľĿ',
        'IiIJijJjKkĸLlLlLlL',
      ),
      array(
        'ŀŁłŃńŅņŇňʼnŊŋŌōŎŏ',
        'lLlNnNnNnʼnŊŋOoOo',
      ),
      array(
        'ŐőŒœŔŕŖŗŘřŚśŜŝŞş',
        'OoŒœRrRrRrSsSsSs',
      ),
      array(
        'ŠšŢţŤťŦŧŨũŪūŬŭŮů',
        'SsTtTtTtUuUuUuUu',
      ),
      array(
        'ŰűŲųŴŵŶŷŸŹźŻżŽž',
        'UuUuWwYyYZzZzZz',
      ),
      // Test all characters in the Unicode range 0x01CD to 0x024F.
      array(
        'ǍǎǏ',
        'AaI',
      ),
      array(
        'ǐǑǒǓǔǕǖǗǘǙǚǛǜǝǞǟ',
        'iOoUuUuUuUuUuǝAa',
      ),
      array(
        'ǠǡǢǣǤǥǦǧǨǩǪǫǬǭǮǯ',
        'AaǢǣGgGgKkOoOoǮǯ',
      ),
      array(
        'ǰDZDzdzǴǵǶǷǸǹǺǻǼǽǾǿ',
        'jDZDzdzGgǶǷNnAaǼǽOo',
      ),
      array(
        'ȀȁȂȃȄȅȆȇȈȉȊȋȌȍȎȏ',
        'AaAaEeEeIiIiOoOo',
      ),
      array(
        'ȐȑȒȓȔȕȖȗȘșȚțȜȝȞȟ',
        'RrRrUuUuSsTtȜȝHh',
      ),
      array(
        'ȠȡȢȣȤȥȦȧȨȩȪȫȬȭȮȯ',
        'ȠȡȢȣZzAaEeOoOoOo',
      ),
      array(
        'ȰȱȲȳȴȵȶȷȸȹȺȻȼȽȾȿ',
        'OoYylntjȸȹACcLTs',
      ),
      array(
        'ɀɁɂɃɄɅɆɇɈɉɊɋɌɍɎɏ',
        'zɁɂBUɅEeJjQqRrYy',
      ),
    );
  }

  /**
   * Tests the PhpTransliteration class.
   *
   * @param string $langcode
   *   The language code to test.
   * @param string $original
   *   The original string.
   * @param string $expected
   *   The expected return from PhpTransliteration::transliterate().
   * @param string $unknown_character
   *   (optional) The character to substitute for characters in $string without
   *   transliterated equivalents. Defaults to '?'.
   * @param int $max_length
   *   (optional) If provided, return at most this many characters, ensuring
   *   that the transliteration does not split in the middle of an input
   *   character's transliteration.
   *
   * @dataProvider providerTestPhpTransliteration
   */
  public function testPhpTransliteration($langcode, $original, $expected, $unknown_character = '?', $max_length = NULL) {
    $transliterator_class = new PhpTransliteration();
    $actual = $transliterator_class
      ->transliterate($original, $langcode, $unknown_character, $max_length);
    $this
      ->assertSame($expected, $actual);
  }

  /**
   * Provides data for self::testPhpTransliteration().
   *
   * @return array
   *   An array of arrays, each containing the parameters for
   *   self::testPhpTransliteration().
   */
  public function providerTestPhpTransliteration() {
    $random_generator = new Random();
    $random = $random_generator
      ->string(10);

    // Make some strings with two, three, and four-byte characters for testing.
    // Note that the 3-byte character is overridden by the 'kg' language.
    $two_byte = 'Ä Ö Ü Å Ø äöüåøhello';

    // This is a Cyrrillic character that looks something like a u. See
    // http://www.unicode.org/charts/PDF/U0400.pdf
    $three_byte = html_entity_decode('ц', ENT_NOQUOTES, 'UTF-8');

    // This is a Canadian Aboriginal character like a triangle. See
    // http://www.unicode.org/charts/PDF/U1400.pdf
    $four_byte = html_entity_decode('ᐑ', ENT_NOQUOTES, 'UTF-8');

    // These are two Gothic alphabet letters. See
    // http://en.wikipedia.org/wiki/Gothic_alphabet
    // They are not in our tables, but should at least give us '?' (unknown).
    $five_byte = html_entity_decode('𐌰𐌸', ENT_NOQUOTES, 'UTF-8');
    return array(
      // Each test case is (language code, input, output).
      // Test ASCII in English.
      array(
        'en',
        $random,
        $random,
      ),
      // Test ASCII in some other language with no overrides.
      array(
        'fr',
        $random,
        $random,
      ),
      // Test 3 and 4-byte characters in a language without overrides.
      // Note: if the data tables change, these will need to change too! They
      // are set up to test that data table loading works, so values come
      // directly from the data files.
      array(
        'fr',
        $three_byte,
        'c',
      ),
      array(
        'fr',
        $four_byte,
        'wii',
      ),
      // Test 5-byte characters.
      array(
        'en',
        $five_byte,
        '??',
      ),
      // Test a language with no overrides.
      array(
        'en',
        $two_byte,
        'A O U A O aouaohello',
      ),
      // Test language overrides provided by core.
      array(
        'de',
        $two_byte,
        'Ae Oe Ue A O aeoeueaohello',
      ),
      array(
        'de',
        $random,
        $random,
      ),
      array(
        'dk',
        $two_byte,
        'A O U Aa Oe aouaaoehello',
      ),
      array(
        'dk',
        $random,
        $random,
      ),
      array(
        'kg',
        $three_byte,
        'ts',
      ),
      // Test strings in some other languages.
      // Turkish, provided by drupal.org user Kartagis.
      array(
        'tr',
        'Abayı serdiler bize. Söyleyeceğim yüzlerine. Sanırım hepimiz aynı şeyi düşünüyoruz.',
        'Abayi serdiler bize. Soyleyecegim yuzlerine. Sanirim hepimiz ayni seyi dusunuyoruz.',
      ),
      // Illegal/unknown unicode.
      array(
        'en',
        chr(0xf8) . chr(0x80) . chr(0x80) . chr(0x80) . chr(0x80),
        '?',
      ),
      // Max length.
      array(
        'de',
        $two_byte,
        'Ae Oe',
        '?',
        5,
      ),
    );
  }

  /**
   * Tests the transliteration with max length.
   */
  public function testTransliterationWithMaxLength() {
    $transliteration = new PhpTransliteration();

    // Test with max length, using German. It should never split up the
    // transliteration of a single character.
    $input = 'Ä Ö Ü Å Ø äöüåøhello';
    $trunc_output = 'Ae Oe Ue A O aeoe';
    $this
      ->assertSame($trunc_output, $transliteration
      ->transliterate($input, 'de', '?', 17), 'Truncating to 17 characters works');
    $this
      ->assertSame($trunc_output, $transliteration
      ->transliterate($input, 'de', '?', 18), 'Truncating to 18 characters works');
  }

  /**
   * Tests inclusion is safe.
   *
   * @covers ::readLanguageOverrides
   */
  public function testSafeInclude() {

    // The overrides in the transliteration data directory transliterates 0x82
    // into "safe" but the overrides one directory higher transliterates the
    // same character into "security hole". So by using "../index" as the
    // language code we can test the ../ is stripped from the langcode.
    vfsStream::setup('transliteration', NULL, [
      'index.php' => '<?php $overrides = ["../index" => [0x82 => "security hole"]];',
      'dir' => [
        'index.php' => '<?php $overrides = ["../index" => [0x82 => "safe"]];',
      ],
    ]);
    $transliteration = new PhpTransliteration(vfsStream::url('transliteration/dir'));
    $transliterated = $transliteration
      ->transliterate(chr(0xc2) . chr(0x82), '../index');
    $this
      ->assertSame($transliterated, 'safe');
  }

}

Members

Name	Modifiers	Type	Description	Overrides
PhpTransliterationTest::providerTestPhpTransliteration	public	function	Provides data for self::testPhpTransliteration().
PhpTransliterationTest::providerTestPhpTransliterationRemoveDiacritics	public	function	Provides data for self::testRemoveDiacritics().
PhpTransliterationTest::testPhpTransliteration	public	function	Tests the PhpTransliteration class.
PhpTransliterationTest::testRemoveDiacritics	public	function	Tests the PhpTransliteration::removeDiacritics() function.
PhpTransliterationTest::testSafeInclude	public	function	Tests inclusion is safe.
PhpTransliterationTest::testTransliterationWithMaxLength	public	function	Tests the transliteration with max length.
UnitTestCase::$randomGenerator	protected	property	The random generator.
UnitTestCase::$root	protected	property	The app root.
UnitTestCase::assertArrayEquals	protected	function	Asserts if two arrays are equal by sorting them first.
UnitTestCase::getBlockMockWithMachineName	protected	function	Mocks a block with a block plugin.
UnitTestCase::getClassResolverStub	protected	function	Returns a stub class resolver.
UnitTestCase::getConfigFactoryStub	public	function	Returns a stub config factory that behaves according to the passed in array.
UnitTestCase::getConfigStorageStub	public	function	Returns a stub config storage that returns the supplied configuration.
UnitTestCase::getContainerWithCacheTagsInvalidator	protected	function	Sets up a container with a cache tags invalidator.
UnitTestCase::getRandomGenerator	protected	function	Gets the random generator for the utility methods.
UnitTestCase::getStringTranslationStub	public	function	Returns a stub translation manager that just returns the passed string.
UnitTestCase::randomMachineName	public	function	Generates a unique random string containing letters and numbers.
UnitTestCase::setUp	protected	function		259