function SearchSimplifyTestCase::testSearchSimplifyUnicode

Tests that all Unicode characters simplify correctly.

File

core/modules/search/tests/search.test, line 1091

Tests for search.module.

Class

SearchSimplifyTestCase

Test search_simplify() on every Unicode character, and some other cases.

Code

function testSearchSimplifyUnicode() {
  // This test uses a file that was constructed so that the even lines are
  // boundary characters, and the odd lines are valid word characters. (It
  // was generated as a sequence of all the Unicode characters, and then the
  // boundary characters (punctuation, spaces, etc.) were split off into
  // their own lines).  So the even-numbered lines should simplify to nothing,
  // and the odd-numbered lines we need to split into shorter chunks and
  // verify that simplification doesn't lose any characters.
  $input = file_get_contents(BACKDROP_ROOT . '/core/modules/search/tests/UnicodeTest.txt');
  $base_strings = explode(chr(10), $input);
  $strings = array();
  foreach ($base_strings as $key => $string) {
    if ($key % 2) {
      // Even line - should simplify down to a space.
      $simplified = search_simplify($string);
      $this->assertIdentical($simplified, ' ', "Line $key is excluded from the index");
    }
    else {
      // Odd line, should be word characters.
      // Split this into 30-character chunks, so we don't run into limits
      // of truncation in search_simplify().
      $start = 0;
      while ($start < backdrop_strlen($string)) {
        $new_string = backdrop_substr($string, $start, 30);
        // Special case: leading zeros are removed from numeric strings,
        // and there's one string in this file that is numbers starting with
        // zero, so prepend a 1 on that string.
        if (preg_match('/^[0-9]+$/', $new_string)) {
          $new_string = '1' . $new_string;
        }
        $strings[] = $new_string;
        $start += 30;
      }
    }
  }
  foreach ($strings as $key => $string) {
    $simplified = search_simplify($string);
    $this->assertTrue(backdrop_strlen($simplified) >= backdrop_strlen($string), "Nothing is removed from string $key.");
  }

  // Test the low-numbered ASCII control characters separately. They are not
  // in the text file because they are problematic for diff, especially \0.
  $string = '';
  for ($i = 0; $i < 32; $i++) {
    $string .= chr($i);
  }
  $this->assertIdentical(' ', search_simplify($string), 'Search simplify works for ASCII control characters.');
}