[project/porterstemmer.git] / porterstemmer.test

<?php

/**
 * @file
 * Tests for the Porter Stemmer module.
 * By Jennifer Hodgdon of Poplar ProductivityWare, www.poplarware.com
 * Unit tests are based on sample words from
 *   http://snowball.tartarus.org/algorithms/english/stemmer.html
 * which are stored in a separate file (testwords.txt).
 */

/**
 * Functional test for Porter Stemmer.
 */
class PorterStemmerFunctionalTest extends DrupalWebTestCase {
  public static function getInfo() {
    return array(
      'name' => t('Functional test'),
      'description' => t('Test that stemmed searching works.'),
      'group' => t('Porter Stemmer'),
    );
  }

  public function setUp() {
    parent::setUp('porterstemmer', 'search');

    // Set up a super-user
    $this->superuser = $this->drupalCreateUser( array( 'administer nodes',
        'access content', 'administer content types',
        'administer search', 'search content',
        'access administration pages',
        'administer site configuration' ));
    $this->drupalLogin( $this->superuser );

    // Create some content to search, with the words walk, walking, and walked.
    $info = array(
      'body' => array(LANGUAGE_NONE => array(array('value' => 'I walk through the streets, looking around for trouble'))),
      'type' => 'page',
      'language' => LANGUAGE_NONE,
      'title' => 'first page',
    );
    $node = $this->drupalCreateNode($info);

    $this->drupalGet('node/' . $node->nid);
    $this->assertText('walk');

    $info = array(
      'body' => array(LANGUAGE_NONE => array(array('value' => 'I walked home from work today.'))),
      'type' => 'page',
      'language' => LANGUAGE_NONE,
      'title' => 'second page',
    );
    $node = $this->drupalCreateNode($info);
    $this->drupalGet('node/' . $node->nid);
    $this->assertText('walked');

    $info = array(
      'body' => array(LANGUAGE_NONE => array(array('value' => 'I am always walking everywhere.'))),
      'type' => 'page',
      'language' => LANGUAGE_NONE,
      'title' => 'third page',
    );
    $node = $this->drupalCreateNode($info);
    $this->drupalGet('node/' . $node->nid);
    $this->assertText('walking');

    $this->cronRun();
  }

  /**
   * Tests that all three pages can be found in searches.
   */
  function testSearch() {
    $this->drupalLogin( $this->superuser );

    // Search for 'walk' and verify all three pages were found.
    $this->drupalPost('search',
      array('keys' => 'walk' ),
      'Search'
    );
    $this->assertText('first', 'First page was found with search for walk');
    $this->assertText('second', 'Second page was found with search for walk');
    $this->assertText('third', 'Third page was found with search for walk');

    // Search for 'walking' and verify all three pages were found.
    $this->drupalPost('search',
      array('keys' => 'walking' ),
      'Search'
    );
    $this->assertText('first', 'First page was found with search for walking');
    $this->assertText('second', 'Second page was found with search for walking');
    $this->assertText('third', 'Third page was found with search for walking');

    // Search for 'walked' and verify all three pages were found.
    $this->drupalPost('search',
      array('keys' => 'walked' ),
      'Search'
    );
    $this->assertText('first', 'First page was found with search for walked');
    $this->assertText('second', 'Second page was found with search for walked');
    $this->assertText('third', 'Third page was found with search for walked');

    // Search for 'blahblahblah' and verify none of the pages were found.
    $this->drupalPost('search',
      array('keys' => 'blahblahblah' ),
      'Search'
    );
    $this->assertNoText('first', 'First page was not found with bogus search');
    $this->assertNoText('second', 'Second page was not found with bogus search');
    $this->assertNoText('third', 'Third page was not found with bogus search');
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 1.
 */
class PorterStemmerOutput1UnitTest extends DrupalWebTestCase {

  public $has_pecl_stem = FALSE;

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 1'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  public function setUp() {
    parent::setUp('porterstemmer');

    // See if the PECL stemming library is installed

    $this->has_pecl_stem = _porterstemmer_pecl_loaded();
  }

  /**
   * Tests the first 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 0, 2000 );
  }

  /**
   * Runs a unit test for a portion of words in the test file.
   *
   * The output of the porterstemmer_stem() function is tested against
   * known input/output pairs from
   *   http://snowball.tartarus.org/algorithms/english/stemmer.html
   *
   * Or if $use_pecl is set to TRUE, the PECL stem_english() function is
   * used instead.
   *
   * @param $skipto
   *    Line of file to start on (count starts at 0), not counting short ones.
   * @param $runto
   *    Number of lines to test, not counting short ones.
   * @param $use_pecl
   *    If TRUE, use the PECL library if it is installed. If it isn't installed,
   *    no tests are performed.
   */
  function _run_porterstemmer_stem_test( $skipto = 0, $runto = 2000, $use_pecl = FALSE ) {

    if ( $use_pecl && !$this->has_pecl_stem ) {
      $this->assertTrue( 1, "No PECL stem library found, aborting test" );
      return;
    }

    // Open word file
    $file = drupal_get_path('module', 'porterstemmer') . '/testwords.txt';

    $handle = @fopen($file, "r");
    $this->assertTrue($handle, "Open file containing words to test", 'Startup');
    if ( !$handle ) {
      // no point doing rest of test...
      return;
    }

    // Set min characters for search to 2 so we test whole word list
    variable_set( 'minimum_word_size', 2 );
    porterstemmer_too_short( '', TRUE );

    $ran = 0;
    $skipped = 0;

    $start = time();
    $mstart = 0;
    if ( function_exists( 'microtime' )) {
      $mstart = microtime(TRUE);
    }

    while ( !feof( $handle ) && $ran < $runto ) {
      // Read a line of the file, and split into words
      $line = fgets($handle, 4096);
      $words = preg_split( "/\s+/", $line, -1, PREG_SPLIT_NO_EMPTY );
      if ( count( $words ) < 2 ) {
        continue;
      }

      $in = $words[0];
      $right = $words[1];
      // Skip words less than 2 characters, which is minimum for Porter
      if ( drupal_strlen( $in ) < 2 ||
        drupal_strlen( $right ) < 2 ) {
        continue;
      }

      $skipped++;
      if ( $skipped < $skipto ) {
        continue;
      }

      // Stem the word
      if ( $use_pecl && $this->has_pecl_stem ) {
        $stem = stem_english( $in );
      }
      else {
        $stem = porterstemmer_stem( $in );
      }

      // Test correctness
      $this->assertEqual( $right, $stem, "Stemming $in results in $right (was $stem)",
        t( 'Stemming test' ));
      $ran++;
    }

    $start = time() - $start;
    if ( function_exists( 'microtime' )) {
      $mstart = microtime(TRUE) - $mstart;
    }
    $this->assertTrue( 1, "Elapsed time: $start seconds ($mstart)" );

    fclose($handle);
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 2.
 */
class PorterStemmerOutput2UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 2'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 2000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 3.
 */
class PorterStemmerOutput3UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 3'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 4000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 4.
 */
class PorterStemmerOutput4UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 4'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 6000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 5.
 */
class PorterStemmerOutput5UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 5'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 8000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 6.
 */
class PorterStemmerOutput6UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 6'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 10000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 7.
 */
class PorterStemmerOutput7UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 7'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 12000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 8.
 */
class PorterStemmerOutput8UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 8'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 14000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 9.
 */
class PorterStemmerOutput9UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 9'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 16000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 10.
 */
class PorterStemmerOutput10UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 10'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 18000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 11.
 */
class PorterStemmerOutput11UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 11'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 20000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 12.
 */
class PorterStemmerOutput12UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 12'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 22000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 13.
 */
class PorterStemmerOutput13UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 13'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 24000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 14.
 */
class PorterStemmerOutput14UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 14'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 26000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 15.
 */
class PorterStemmerOutput15UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 15'),
      'description' => t('Test that the stemming function returns the correct stemmed words'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 28000, 2000 );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming internals.
 */
class PorterStemmerInternalsUnitTest extends DrupalWebTestCase {

  public static function getInfo() {
    return array(
      'name' => t('Stemming internals'),
      'description' => t('Test that various algorithm steps and internal functions are working correctly'),
      'group' => t('Porter Stemmer'),
    );
  }

  public function setUp() {
    parent::setUp('porterstemmer');
  }

  /**
   * Verify that short words are not stemmed, and longer ones are.
   */
  function testStemLength3UnitTest() {
    // Words 3 letters or less should not be stemmed if min word length is 3
    variable_set( 'minimum_word_size', 3 );
    porterstemmer_too_short( '', TRUE );

    $words = array(
      'a' => 'a',
      'at' => 'at',
      'say' => 'say',
      'fished' => 'fish',
      'saying' => 'say',
    );

    foreach ( $words as $in => $out ) {
      $stem = porterstemmer_stem( $in );
      $this->assertEqual( $out, $stem, "Stemming length 3 test for $in gives $out (was $stem)", t( 'Stemming length' ));
    }
  }

  /**
   * Verify that short words are not stemmed, and longer ones are.
   */
  function testStemLength4UnitTest() {
    // Words 4 letters or less should not be stemmed if min word length is 4
    variable_set( 'minimum_word_size', 4 );
    porterstemmer_too_short( '', TRUE );

    $words = array(
      'a' => 'a',
      'at' => 'at',
      'say' => 'say',
      'fished' => 'fish',
      'saying' => 'saying',
    );

    foreach ( $words as $in => $out ) {
      $stem = porterstemmer_stem( $in );
      $this->assertEqual( $out, $stem, "Stemming length 4 test for $in gives $out (was $stem)", t( 'Stemming length' ));
    }
  }

  /**
   * Tests the function that determines if a word is "short".
   */
  function testShortWord() {
    // Test "shortness", using examples from algorithm web page,
    // as well as several variations on the word "administ...".

    variable_set( 'minimum_word_size', 2 );
    porterstemmer_too_short( '', TRUE );

    $this->assertFalse( porterstemmer_short_word( 'administered', 2, 'Stemmer steps' ),
      "administered should not be a short word" );

    $this->assertFalse( porterstemmer_short_word( 'administer', 2, 'Stemmer steps'),
      "administer should not be a short word" );

    $this->assertFalse( porterstemmer_short_word( 'admin', 2, 'Stemmer steps' ),
      "admin should not be a short word" );

    $this->assertTrue( porterstemmer_short_word( 'bed', 3, 'Stemmer steps' ),
      "bed should be a short word" );

    $this->assertTrue( porterstemmer_short_word( 'shed', 4, 'Stemmer steps' ),
      "shed should be a short word" );

    $this->assertTrue( porterstemmer_short_word( 'shred', 5, 'Stemmer steps' ),
      "shred should be a short word" );

    $this->assertFalse( porterstemmer_short_word( 'bead', 4, 'Stemmer steps' ),
      "bead should not be a short word" );

    $this->assertFalse( porterstemmer_short_word( 'beds', 3, 'Stemmer steps' ),
      "beds should not be a short word" );

    $this->assertFalse( porterstemmer_short_word( 'bake', 3, 'Stemmer steps' ),
      "bake should not be a short word" );

    $this->assertTrue( porterstemmer_short_word( 'bak', 3, 'Stemmer steps' ),
      "bak should be a short word" );
  }

  /**
   * Test internal steps on the word "administered".
   */
  function testAdministered() {

    variable_set( 'minimum_word_size', 2 );
    porterstemmer_too_short( '', TRUE );

    $r1 = 0;
    $r2 = 0;
    $word = 'administered';

    porterstemmer_prestemming( $word, $r1, $r2 );

    // Test calculation of R1 and R2
    $this->assertEqual( $r1, 2, "R1 for administered should be 2, was $r1", 'Stemmer steps' );
    $this->assertEqual( $r2, 5, "R2 for administered should be 5, was $r2", 'Stemmer steps' );

    // Test step 1b of the algorithm
    porterstemmer_step1b($word, $r1);
    $this->assertEqual( $word, 'administer', "Step1b should be administer, was $word", 'Stemmer steps' );

    // Test step 4 of the algorithm
    porterstemmer_step4($word, $r2);
    $this->assertEqual( $word, 'administ', "Step4 should be administ, was $word", 'Stemmer steps' );
  }

  /**
   * Test internal steps on the word "baked".
   */
  function testBaked() {

    variable_set( 'minimum_word_size', 2 );
    porterstemmer_too_short( '', TRUE );

    $r1 = 0;
    $r2 = 0;
    $word = 'baked';

    porterstemmer_prestemming( $word, $r1, $r2 );

    // Test calculation of R1 and R2
    $this->assertEqual( $r1, 3, "R1 for baked should be 3, was $r1", 'Stemmer steps' );
    $this->assertEqual( $r2, 5, "R2 for baked should be 5, was $r2", 'Stemmer steps' );

    // Test step 1b of the algorithm
    porterstemmer_step1b($word, $r1);
    $this->assertEqual( $word, 'bake', "Step1b should be bake, was $word", 'Stemmer steps' );

    // Test step 5 of the algorithm
    porterstemmer_step5($word, $r1, $r2);
    $this->assertEqual( $word, 'bake', "Step5 should be bake, was $word", 'Stemmer steps' );
  }

  /**
   * Test internal steps on the word "geology".
   */
  function testGeology() {

    $r1 = 0;
    $r2 = 0;
    $word = 'geology';

    variable_set( 'minimum_word_size', 2 );
    porterstemmer_too_short( '', TRUE );

    porterstemmer_prestemming( $word, $r1, $r2 );

    // Test calculation of R1 and R2
    $this->assertEqual( $r1, 4, "R1 for geology should be 4, was $r1", 'Stemmer steps' );
    $this->assertEqual( $r2, 6, "R2 for geology should be 6, was $r2", 'Stemmer steps' );

    // Test step 1c of the algorithm
    porterstemmer_step1c($word);
    $this->assertEqual( $word, 'geologi', "Step1c should be geologi, was $word", 'Stemmer steps' );

    // Test step 2 of the algorithm
    porterstemmer_step2($word, $r1);
    $this->assertEqual( $word, 'geolog', "Step2 should be geolog, was $word", 'Stemmer steps' );
  }

  /**
   * Test internal steps on the words "ies" and "ied".
   */
  function testIesIed() {

    $r1a = 0;
    $r2a = 0;
    $r1b = 0;
    $r2b = 0;
    $worda = 'ied';
    $wordb = 'ies';

    variable_set( 'minimum_word_size', 2 );
    porterstemmer_too_short( '', TRUE );

    porterstemmer_prestemming( $worda, $r1a, $r2a );
    porterstemmer_prestemming( $wordb, $r1b, $r2b );

    // Test step 1a of the algorithm

    porterstemmer_step1a( $worda );
    porterstemmer_step1a( $wordb );

    $this->assertEqual( $worda, 'ie', "Step1a should be ie, was $worda", 'Stemmer steps' );
    $this->assertEqual( $wordb, 'ie', "Step1a should be ie, was $wordb", 'Stemmer steps' );
  }

  /**
   * Tests the excerpt function porterstemmer_sbp_excerpt_match().
   */
  function testExcerpts() {
    // Test with simple stemmed match.
    $key = 'walking';
    $text = 'I walked to the Grand Walks yesterday.';
    $offset = 0;
    $boundary = '[ ]+';
    $result1 = porterstemmer_sbp_excerpt_match($key, $text, $offset, $boundary);
    $this->assertEqual($result1['where'], 2, 'Found match 1 in right place', 'Excerpt');
    $this->assertEqual($result1['keyword'], 'walked', 'Found right keyword for match 1', 'Excerpt');

    // Test with upper-case in the keyword.
    $result1b = porterstemmer_sbp_excerpt_match('waLk', $text, $offset, $boundary);
    $this->assertEqual($result1b['where'], 2, 'Found match 1b in right place', 'Excerpt');
    $this->assertEqual($result1b['keyword'], 'walked', 'Found right keyword for match 1b', 'Excerpt');

    // Test with upper-case in the text.
    $offset = 12;
    $result2 = porterstemmer_sbp_excerpt_match($key, $text, $offset, $boundary);
    $this->assertEqual($result2['where'], 22, 'Found match 2 in right place', 'Excerpt');
    $this->assertEqual($result2['keyword'], 'Walks', 'Found right keyword for match 2', 'Excerpt');

    // Test with a false match followed by a real match.
    $text = 'I walknotawordhere to walk to school.';
    $offset = 0;
    $result3 = porterstemmer_sbp_excerpt_match($key, $text, $offset, $boundary);
    $this->assertEqual($result3['where'], 22, 'Found match 3 in right place', 'Excerpt');
    $this->assertEqual($result3['keyword'], 'walk', 'Found right keyword for match 3', 'Excerpt');

  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 1 PECL.
 */
class PorterStemmerPECLOutput1UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 1 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 0, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 2 PECL.
 */
class PorterStemmerPECLOutput2UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 2 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 2000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 3 PECL.
 */
class PorterStemmerPECLOutput3UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 3 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 4000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 4 PECL.
 */
class PorterStemmerPECLOutput4UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 4 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 6000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 5 PECL.
 */
class PorterStemmerPECLOutput5UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 5 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 8000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 6 PECL.
 */
class PorterStemmerPECLOutput6UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 6 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 10000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 7 PECL.
 */
class PorterStemmerPECLOutput7UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 7 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 12000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 8 PECL.
 */
class PorterStemmerPECLOutput8UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 8 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 14000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 9 PECL.
 */
class PorterStemmerPECLOutput9UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 9 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 16000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 10 PECL.
 */
class PorterStemmerPECLOutput10UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 10 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 18000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 11 PECL.
 */
class PorterStemmerPECLOutput11UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 11 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 20000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 12 PECL.
 */
class PorterStemmerPECLOutput12UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 12 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 22000, 2000, TRUE );
  }
}
/**
 * Unit tests for Porter Stemmer - Stemming output part 13 PECL.
 */
class PorterStemmerPECLOutput13UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 13 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 24000, 2000, TRUE );
  }
}
/**
 * Unit tests for Porter Stemmer - Stemming output part 14 PECL.
 */
class PorterStemmerPECLOutput14UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 14 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 26000, 2000, TRUE );
  }
}

/**
 * Unit tests for Porter Stemmer - Stemming output part 15 PECL.
 */
class PorterStemmerPECLOutput15UnitTest extends PorterStemmerOutput1UnitTest {

  public static function getInfo() {
    return array(
      'name' => t('Stemming output 15 PECL'),
      'description' => t('Test that the stemming function returns the correct stemmed words with PECL library'),
      'group' => t('Porter Stemmer'),
    );
  }

  /**
   * Tests the next 2000 words in the file for stemming accuracy.
   */
  function testStemmingUnitTest() {
    $this->_run_porterstemmer_stem_test( 28000, 2000, TRUE );
  }
}