Create method for import html

loadHtml parse html string and create either SimpleXmlElement object or a DOMDocument. use of empty($options['loadEntities']) instead of !$options['loadEntities']
cakephp · Jun 2, 2018 · 42aab01 · 42aab01
1 parent d9a61c3
commit 42aab01
Showing 1 changed file with 41 additions and 2 deletions.
diff --git a/src/Utility/Xml.php b/src/Utility/Xml.php
@@ -142,7 +142,7 @@ protected static function _loadXml($input, $options)
     {
         $hasDisable = function_exists('libxml_disable_entity_loader');
         $internalErrors = libxml_use_internal_errors(true);
-        if ($hasDisable && !$options['loadEntities']) {
+        if ($hasDisable && empty($options['loadEntities'])) {
             libxml_disable_entity_loader(true);
         }
         $flags = 0;
@@ -162,7 +162,46 @@ protected static function _loadXml($input, $options)
         } catch (Exception $e) {
             throw new XmlException('Xml cannot be read. ' . $e->getMessage(), null, $e);
         } finally {
-            if ($hasDisable && !$options['loadEntities']) {
+            if ($hasDisable && empty($options['loadEntities'])) {
+                libxml_disable_entity_loader(false);
+            }
+            libxml_use_internal_errors($internalErrors);
+        }
+    }
+
+    /**
+     * Parse the input html string and create either a SimpleXmlElement object or a DOMDocument.
+     *
+     * @param string $input The input html string to load.
+     * @param array $options The options to use. See Xml::build()
+     * @return \SimpleXMLElement|\DOMDocument
+     * @throws \Cake\Utility\Exception\XmlException
+     */
+    public static function loadHtml($input, $options)
+    {
+        $hasDisable = function_exists('libxml_disable_entity_loader');
+        $internalErrors = libxml_use_internal_errors(true);
+        if ($hasDisable && empty($options['loadEntities'])) {
+            libxml_disable_entity_loader(true);
+        }
+        $flags = 0;
+        if (!empty($options['parseHuge'])) {
+            $flags |= LIBXML_PARSEHUGE;
+        }
+        try {
+            $xml = new DOMDocument();
+            $xml->loadHTML($input, $flags);
+
+            if ($options['return'] === 'simplexml' || $options['return'] === 'simplexmlelement') {
+                $flags |= LIBXML_NOCDATA;
+								$xml = simplexml_import_dom($xml);
+            }
+
+            return $xml;
+        } catch (Exception $e) {
+            throw new XmlException('Xml cannot be read. ' . $e->getMessage(), null, $e);
+        } finally {
+            if ($hasDisable && empty($options['loadEntities'])) {
                 libxml_disable_entity_loader(false);
             }
             libxml_use_internal_errors($internalErrors);