/**
* @param string $regexp
* @param nc_search_document_parser_html_fragment|null $recipient
* @param boolean $remove
* @return boolean success
*/
public function extract_regexp($regexp, self $recipient = null, $remove = false)
{
$html = $this->get_html();
if (!preg_match_all($regexp . "u", $html, $matches)) {
return false;
}
if ($recipient) {
$recipient->append_fragment(new self(join("\n", $matches[1])));
}
if ($remove) {
// it's easier to dismiss the old DOMDocument than to delete all nodes properly
$this->set_html(str_replace($matches[0], "", $html));
}
return true;
}
/**
*
* @param string $query_string
* @param nc_search_document_parser_html_fragment $source
* @param nc_search_document_parser_html_fragment|null $recipient
* @param boolean $remove
* @param boolean $only_first
* @return boolean success
*/
protected function execute_queries($query_string, nc_search_document_parser_html_fragment $source, $recipient, $remove, $only_first)
{
if (!trim($query_string)) {
return false;
}
if (preg_match("/^[\\w \\-]+\$/", $query_string)) {
// tag names delimited with a space
$queries = explode(" ", $query_string);
} else {
$queries = explode("\n", $query_string);
}
$matched = false;
foreach ($queries as $query) {
$query = trim($query);
if ($query[0] == "#") {
$matched = $source->extract_regexp($query, $recipient, $remove);
} elseif (strpos($query, "/") !== false) {
$matched = $source->extract_xpath($query, $recipient, $remove);
} elseif (preg_match("/^[\\w\\-]+\$/", $query)) {
$matched = $source->extract_tag($query, $recipient, $remove);
}
if ($matched && $only_first) {
break;
}
}
return $matched;
}
