transliteration.test

<?php
/**
 * Tests the transliteration class.
 */
class TransliterationTest extends BackdropUnitTestCase {
  /**
   * Tests transliteration API.
   */
  public function testTransliteration() {
    $random = $this->randomName(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('&#x446;', 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('&#x1411;', ENT_NOQUOTES, 'UTF-8');
    // A character that doesn't have a replacement value.
    $unknown_character = html_entity_decode('&#xfff1;', ENT_NOQUOTES, 'UTF-8');

    $cases = 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 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('da', $two_byte, 'A O U Aa Oe aouaaoehello'),
      array('da', $random, $random),
      array('kg', $three_byte, 'ts'),

      // Unknown characters.
      array('en', 'foo' . $unknown_character . 'foo', 'foo?foo'),
      array('en', 'foo' . $unknown_character . $unknown_character . $unknown_character . 'foo', 'foo???foo'),

      // Chinese characters.
      array('en', '一個簡單的句子', 'yigejiandandejuzi'),

      // Bulgarian characters.
      array('bg', 'първа статия', 'parva statiya'),
    );

    include_once BACKDROP_ROOT . '/core/includes/transliteration.inc';
    foreach ($cases as $case) {
      list($langcode, $before, $after) = $case;
      $actual = transliteration_get($before, '?', $langcode);
      $this->assertEqual($after, $actual, format_string('@before is correctly transliterated to @after (actual: @actual) in @langcode langcode.', array('@before' => $before, '@after' => $after, '@actual' => $actual, '@langcode' => $langcode)));
    }
  }

  /**
   * Tests machine name transliteration provided by System module.
   */
  public function testMachineNameTransliteration() {
    $unknown_character = html_entity_decode('&#xfff1;', ENT_NOQUOTES, 'UTF-8');
    $string = "  ä $unknown_character test $unknown_character$unknown_character string - gap -_ ";
    $cases = array(
      // Keys are langcode, replace, maxlength, replace_pattern, input, output.
      array('en', '_', NULL, NULL, $string, 'a_test_string_gap'),
      array('de', '_', NULL, NULL, $string, 'ae_test_string_gap'),
      array('en', '_', NULL, '[^a-zA-Z0-9_-]+', $string, 'a_test_string_-_gap_-'),
      array('en', '',  NULL, NULL, $string, 'ateststringgap'),
      array('en', '',  10,   NULL, $string, 'ateststrin'),
    );
    include_once BACKDROP_ROOT . '/core/modules/system/system.admin.inc';
    foreach ($cases as $case) {
      list($langcode, $replace, $maxlength, $replace_pattern, $before, $after) = $case;
      $options = array(
        'langcode' => $langcode,
        'replace' => $replace,
        'maxlength' => $maxlength,
        'replace_pattern' => $replace_pattern,
      );
      // Remove NULL options.
      $options = array_filter($options, function($variable) { return !is_null($variable); });
      $actual = system_transliterate_machine_name($before, $options);
      $this->assertEqual($after, $actual, format_string('@before is correctly transliterated to @after (actual: @actual) in @langcode langcode with options @options.', array('@before' => $before, '@after' => $after, '@actual' => $actual, '@langcode' => $langcode, '@options' => print_r($options, 1))));
    }
  }
}