/**
* Returns a string where HTML entities have been converted back to their original characters
*
* @param string String to be parsed
* @param boolean Convert unicode characters back from HTML entities?
*
* @return string
*/
public static function unHtmlSpecialChars($text, $doUniCode = false)
{
if ($doUniCode) {
$text = preg_replace_callback('/&#([0-9]+);/siU', function ($matches) {
return vB5_String::convertIntToUtf8($matches[1]);
}, $text);
}
return str_replace(array('<', '>', '"', '&'), array('<', '>', '"', '&'), $text);
}
/**
* Break the keyword search into words
* @param string keywords -- keyword string as entered by the user
* @return array -- array of word records
* array('word' => $word, 'joiner' => {'', 'NOT', 'AND', 'OR'})
* The search implementation is expected to use these to build the search
* query.
*/
private function get_words($keywords)
{
$is_mb = preg_match('/&#([0-9]+);|[^\\x00-\\x7F]/siU', $keywords);
// @todo handleing for thousand and decimal separators for numbers
// removing punctuation
$origKeywords = $keywords;
$keywords = preg_replace('#(?!-)[\\p{Pd}\\p{Pe}\\p{Pf}\\p{Pi}\\p{Po}\\p{Ps}]#' . ($is_mb ? 'u' : ''), ' ', $keywords);
// a tokenizing based approach to building a search query
preg_match_all('#("[^"]*"|[^\\s]+)#', $keywords, $matches, PREG_SET_ORDER);
$token_joiner = null;
$words = array();
foreach ($matches as $match) {
if ($is_mb) {
$match = preg_replace_callback('/&#([0-9]+);/siU', function ($matches) {
return vB5_String::convertIntToUtf8($matches[1]);
}, $match);
}
if ($is_mb) {
$token = vB_String::vBStrToLower($match[1]);
} else {
$token = strtolower($match[1]);
}
//this means that we implicitly have a not joiner.
if ($token[0] == '-') {
//this effectively means two joiners, which is bad.
if ($token_joiner) {
$this->add_error('invalid_search_syntax');
} else {
$token = substr($token, 1);
$token_joiner = 'not';
}
}
switch ($token) {
case 'or':
case 'and':
case 'not':
// this isn't a searchable word, but a joiner
$token_joiner = strtoupper($token);
break;
default:
//$lowWord = strtolower($token);
if (vB_Api_Search::is_index_word($token, true)) {
$words[] = array('word' => $token, 'joiner' => strtoupper($token_joiner));
} else {
$this->ignored_keywords[] = $match[1];
}
$token_joiner = null;
break;
}
}
if (empty($matches) and !empty($origKeywords)) {
$this->ignored_keywords[] = $origKeywords;
}
return $words;
}
