/**
* Truncate given HTML string to specified length.
* If length_in_chars is false it's trimmed by number
* of words, otherwise by number of characters.
*
* @param string $html
* @param integer $length
* @param string|array $opts
* @return string
*/
public static function truncate($html, $length, $opts = array())
{
if (is_string($opts)) {
$opts = array('ellipsis' => $opts);
}
$opts = array_merge(static::$default_options, $opts);
// wrap the html in case it consists of adjacent nodes like <p>foo</p><p>bar</p>
$html = mb_convert_encoding("<div>" . $html . "</div>", 'HTML-ENTITIES', 'UTF-8');
$root_node = null;
// Parse using HTML5Lib if it's available.
if (class_exists('HTML5Lib\\Parser')) {
try {
$doc = \HTML5Lib\Parser::parse($html);
$root_node = $doc->documentElement->lastChild->lastChild;
} catch (\Exception $e) {
}
}
if ($root_node === null) {
// HTML5Lib not available so we'll have to use DOMDocument
// We'll only be able to parse HTML5 if it's valid XML
$doc = new DOMDocument('4.01', 'utf-8');
$doc->formatOutput = false;
$doc->preserveWhiteSpace = true;
// loadHTML will fail with HTML5 tags (article, nav, etc)
// so we need to suppress errors and if it fails to parse we
// retry with the XML parser instead
$prev_use_errors = libxml_use_internal_errors(true);
if ($doc->loadHTML($html)) {
$root_node = $doc->documentElement->lastChild->lastChild;
} else {
if ($doc->loadXML($html)) {
$root_node = $doc->documentElement;
} else {
libxml_use_internal_errors($prev_use_errors);
throw new \RuntimeException();
}
}
libxml_use_internal_errors($prev_use_errors);
}
list($text, $_, $opts) = static::truncateNode($doc, $root_node, $length, $opts);
$text = mb_substr(mb_substr($text, 0, -6), 5);
return $text;
}
public function testParse()
{
$result = Parser::parse('<html><body></body></html>');
$this->assertIsA($result, 'DOMDocument');
}