/**
* {@inheritdoc}
*/
public function normalize(array $tokens)
{
foreach ($tokens as $k => $token) {
$tokens[$k] = stemword($token, $this->lang, $this->charset);
}
return $tokens;
}
public function filter_keywords(array $keywords)
{
$keywords = parent::filter_keywords($keywords);
$tagger = __NAMESPACE__ . '\\POS\\' . ucfirst($this->lang) . '\\Tagger';
if (class_exists($tagger)) {
$keywords = $tagger::get($keywords);
}
$keywords = array_filter($keywords, function ($word) {
$word = mb_strtolower($word);
$normalized_word = stemword($word, 'french', 'UTF_8');
return empty($this->stopword[$word]) && empty($this->common_words[$normalized_word]) && stristr($word, 'www') === false && stristr($word, '@') === false && stristr($word, '.com') === false && stristr($word, '.fr') === false;
});
return $keywords;
}
private function getPhpStemmerPseudoRoot($word)
{
$word = mb_strtolower($word, 'utf-8');
return stemword($word, $this->languageDetect($word), 'UTF_8');
}
