diff --git a/src/Syllable.php b/src/Syllable.php
index 7985661..2d3f7bd 100644
--- a/src/Syllable.php
+++ b/src/Syllable.php
@@ -62,6 +62,11 @@ class Syllable
private $excludes = array();
private $includes = array();
+ /**
+ * @var int
+ */
+ private $libxmlOptions = 0;
+
/**
* Create a new Syllable class, with defaults
*
@@ -195,6 +200,16 @@ public function getMinWordLength()
return $this->min_word_length;
}
+ /**
+ * Options to use for HTML parsing by libxml
+ * @param integer $libxmlOptions
+ * @see https://www.php.net/manual/de/libxml.constants.php
+ */
+ public function setLibxmlOptions($libxmlOptions)
+ {
+ $this->libxmlOptions = $libxmlOptions;
+ }
+
private static function initEncoding()
{
if (self::$encoding) {
@@ -436,7 +451,7 @@ public function hyphenateHtml($html)
{
$dom = new \DOMDocument();
$dom->resolveExternals = true;
- $dom->loadHTML($html);
+ $dom->loadHTML($html, $this->libxmlOptions);
// filter excludes
$xpath = new \DOMXPath($dom);
diff --git a/tests/SyllableTest.php b/tests/SyllableTest.php
index 86c72d2..1c03d42 100644
--- a/tests/SyllableTest.php
+++ b/tests/SyllableTest.php
@@ -222,6 +222,10 @@ public function testHyphenateHtml()
$this->assertEquals(''
. "\n" . 'Ridicu-lous-ly com-pli-cat-ed meta-text
'
. "\n", $this->object->hyphenateHtml('Ridiculously complicated metatext'));
+
+ $this->object->setLibxmlOptions(LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
+ $this->assertEquals('Ridicu-lous-ly com-pli-cat-ed meta-text
'
+ . "\n", $this->object->hyphenateHtml('Ridiculously complicated metatext'));
}
public function testCaseInsensitivity()