/**
*
* @param array $terms
* @return string
*/
public function get_highlight_regexp(array $terms)
{
$res = array();
foreach ($this->get_base_forms($terms) as $base) {
$res[] = $base . "[\\pL\\d]{0,{$this->max_remainder_length}}";
}
return nc_search_util::word_regexp("(" . join("|", $res) . ")", "Si");
}
/**
*
*/
protected function get_highlight_regexp($language)
{
if (!$this->highlight_regexp) {
$query_string = $this->get_query_string();
$context = new nc_search_context(array('language' => $language, 'action' => 'searching'));
// Получить слова из запроса.
// (Удалять из запроса термины с префиксом "-" и "NOT" не имеет особого смысла,
// поскольку в результат они как правило не попадают.)
$query_string = preg_replace("/[\\^~][\\d\\.]+/", '', $query_string);
// операторы ^1, ~1
preg_match_all("/[\\pL\\d\\?\\*]+/u", $query_string, $matches);
$terms = $matches[0];
if (strpos($query_string, "*") !== false || strpos($query_string, "?") !== false) {
$wildcards_replacement = nc_search::should('AllowWildcardSearch') ? array("?" => ".", "*" => "[\\S]+") : array("?" => "", "*" => "");
foreach ($terms as $i => $term) {
$terms[$i] = strtr($term, $wildcards_replacement);
}
}
//if ( nc_Core::get_object()->NC_UNICODE ) {
$terms = nc_search_extension_manager::get('nc_search_language_filter', $context)->except('nc_search_language_filter_stopwords')->apply('filter', $terms);
//}
$analyzer = nc_search_extension_manager::get('nc_search_language_analyzer', $context)->first();
if ($analyzer) {
$regexp = $analyzer->get_highlight_regexp($terms);
} else {
$regexp = nc_search_util::word_regexp("(" . join("|", $terms) . ")", "Si");
}
$this->highlight_regexp = $regexp;
}
// of "there was no 'highlight_regexp'"
return $this->highlight_regexp;
}
/**
*
* @param array $terms
* @return string
*/
public function get_highlight_regexp(array $terms)
{
$morphy = $this->get_morphy();
$result = array();
foreach ($terms as $term) {
$paradigms = $morphy->findWord($term);
if ($paradigms) {
foreach ($paradigms as $paradigm) {
$result = array_merge($result, $paradigm->getAllForms());
}
} else {
// phpMorphy doesn't know what to do with that term
$result[] = $term;
}
}
$regexp = "(" . join("|", array_unique($result)) . ")";
return nc_search_util::word_regexp($regexp, "Si");
}