public static function extrairPalavrasChaves($string = NULL, $max_count = 10)
{
#echo PHP_EOL . 'ORIGINAL ========================================' . PHP_EOL;
#echo $string;
$string = Uteis::cleanTextArea($string, true);
#echo PHP_EOL . PHP_EOL . 'CLEANTEXTAREA ========================================' . PHP_EOL . PHP_EOL . PHP_EOL;
#echo $string;
if (Uteis::vazio($string)) {
return '';
}
$stop_words = file(Kernel::get_conf('sys_path') . DIRECTORY_SEPARATOR . '_data' . DIRECTORY_SEPARATOR . 'stopWords_pt-BR3.txt', FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
#echo PHP_EOL . PHP_EOL . 'STOPWORDS ========================================' . PHP_EOL . PHP_EOL . PHP_EOL;
#var_dump($stop_words);
//$string = utf8_decode($string);
$match_words = array();
preg_match_all('/\\b.*?\\b/ium', $string, $match_words);
$match_words = $match_words[0];
#echo PHP_EOL . PHP_EOL . 'WORDS ========================================' . PHP_EOL . PHP_EOL . PHP_EOL;
#var_dump($match_words);
foreach ($match_words as $key => $item) {
if ($item == '' || in_array(mb_strtolower($item), $stop_words) || mb_strlen($item) < 2) {
unset($match_words[$key]);
} else {
$match_words[$key] = mb_strtolower($item);
}
}
#echo PHP_EOL . PHP_EOL . 'WORDS SEM STOPWORDS ========================================' . PHP_EOL . PHP_EOL . PHP_EOL;
#var_dump($match_words);
//$word_count = str_word_count( utf8_decode(implode(" ", $match_words)) , 1, '0123456789');
$word_count = preg_split('/\\W+/iu', implode(" ", $match_words), -1, PREG_SPLIT_NO_EMPTY);
#echo PHP_EOL . PHP_EOL . 'WORD COUNT ========================================' . PHP_EOL . PHP_EOL . PHP_EOL;
#var_dump($word_count);
$frequency = array_count_values($word_count);
arsort($frequency);
$keywords = array_slice($frequency, 0, $max_count, 1);
$keywords = join(',', array_keys($keywords));
#echo PHP_EOL . PHP_EOL . 'KEY WORDS ========================================' . PHP_EOL . PHP_EOL . PHP_EOL;
#var_dump($keywords);
return $keywords;
}
