/**
* Validate that the provided value is a valid alphabetic string (strictly a-zA-Z).
*
* @since 1.0
*/
public function testIsAlpha()
{
$this->assertTrue(Validator::isAlpha('test'));
$this->assertTrue(Validator::isAlpha('Test'));
$this->assertTrue(Validator::isAlpha('TEST'));
$this->assertFalse(Validator::isAlpha('number5'));
$this->assertFalse(Validator::isAlpha('!-++#'));
$this->assertFalse(Validator::isAlpha('100'));
}
/**
* Splits the passed content by spaces, filters (removes) stop words from stopwords.ini,
* and returns an array of Tag instances.
*
* @param $content
* @param $taggedClass Optionally provide a BO class name (with namespace)
* @param $taggedOID Optionally provide a BO instance OID
* @param $applyStopwords Defaults true, set to false if you want to ignore the stopwords.
*
* @return array
*
* @throws Alpha\Exception\AlphaException
*
* @since 1.0
*/
public static function tokenize($content, $taggedClass = '', $taggedOID = '', $applyStopwords = true)
{
if (self::$logger == null) {
self::$logger = new Logger('Tag');
}
$config = ConfigProvider::getInstance();
// apply stop words
$lowerWords = preg_split("/[\\s,.:-]+/", $content);
array_walk($lowerWords, 'Alpha\\Model\\Tag::lowercaseArrayElement');
if ($applyStopwords) {
if (file_exists($config->get('app.root') . 'config/stopwords-' . $config->get('search.stop.words.size') . '.ini')) {
$stopwords = file($config->get('app.root') . 'config/stopwords-' . $config->get('search.stop.words.size') . '.ini', FILE_IGNORE_NEW_LINES);
} elseif (file_exists($config->get('app.root') . 'Alpha/stopwords-' . $config->get('search.stop.words.size') . '.ini')) {
$stopwords = file($config->get('app.root') . 'Alpha/stopwords-' . $config->get('search.stop.words.size') . '.ini', FILE_IGNORE_NEW_LINES);
} else {
throw new AlphaException('Unable to find a stopwords-' . $config->get('search.stop.words.size') . '.ini file in the application!');
}
array_walk($stopwords, 'Alpha\\Model\\Tag::lowercaseArrayElement');
$filtered = array_diff($lowerWords, $stopwords);
} else {
$filtered = $lowerWords;
}
$tagObjects = array();
$tagContents = array();
foreach ($filtered as $tagContent) {
// we only want to create word tags
if (Validator::isAlpha($tagContent)) {
// just making sure that we haven't added this one in already
if (!in_array($tagContent, $tagContents) && !empty($tagContent)) {
$tag = new self();
$tag->set('content', trim(mb_strtolower($tagContent)));
if (!empty($taggedClass)) {
$tag->set('taggedClass', $taggedClass);
}
if (!empty($taggedOID)) {
$tag->set('taggedOID', $taggedOID);
}
array_push($tagObjects, $tag);
array_push($tagContents, $tagContent);
}
}
}
self::$logger->debug('Tags generated: [' . var_export($tagContents, true) . ']');
return $tagObjects;
}
