/**
* Get text before first heading.
* @param string $text
* @return string|null
*/
private function extractHeadingBeforeFirstHeading($text)
{
$matches = [];
if (!preg_match('/<h[123456]>/', $text, $matches, PREG_OFFSET_CAPTURE)) {
// There isn't a first heading so we interpret this as the article
// being entirely without heading.
return null;
}
$text = substr($text, 0, $matches[0][1]);
if (!$text) {
// There isn't any text before the first heading so we declare there isn't
// a first heading.
return null;
}
$formatter = new HtmlFormatter($text);
$formatter->remove($this->excludedElementSelectors);
$formatter->remove($this->auxiliaryElementSelectors);
$formatter->filterContent();
$text = trim(Sanitizer::stripAllTags($formatter->getText()));
if (!$text) {
// There isn't any text after filtering before the first heading so we declare
// that there isn't a first heading.
return null;
}
return $text;
}
