Merge branch 'LukasRos-master'

Author: paquettg

.gitignore

@@ -0,0 +1,2 @@
+vendor
+composer.lock

src/PHPHtmlParser/Dom.php

@@ -359,6 +359,12 @@ protected function isLoaded()
 	 */
 	protected function clean($str)
 	{
+		if ($this->options->get('cleanupInput') != true)
+		{
+			// skip entire cleanup step
+			return $str;
+		}
+
 		// clean out the \n\r
 		$str = str_replace(["\r\n", "\r", "\n"], ' ', $str);
 
@@ -372,14 +378,20 @@ protected function clean($str)
 		$str = mb_eregi_replace("<!\[CDATA\[(.*?)\]\]>", '', $str);
 
 		// strip out <script> tags
-		$str = mb_eregi_replace("<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>", '', $str);
-		$str = mb_eregi_replace("<\s*script\s*>(.*?)<\s*/\s*script\s*>", '', $str);
+		if ($this->options->get('removeScripts') == true)
+		{
+			$str = mb_eregi_replace("<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>", '', $str);
+			$str = mb_eregi_replace("<\s*script\s*>(.*?)<\s*/\s*script\s*>", '', $str);
+		}
 
 		// strip out <style> tags
-		$str = mb_eregi_replace("<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>", '', $str);
-		$str = mb_eregi_replace("<\s*style\s*>(.*?)<\s*/\s*style\s*>", '', $str);
+		if ($this->options->get('removeStyles') == true)
+		{
+			$str = mb_eregi_replace("<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>", '', $str);
+			$str = mb_eregi_replace("<\s*style\s*>(.*?)<\s*/\s*style\s*>", '', $str);
+		}
 
-		// strip out pre-formatted tags
+		// strip out preformatted tags
 		$str = mb_eregi_replace("<\s*(?:code)[^>]*>(.*?)<\s*/\s*(?:code)\s*>", '', $str);
 
 		// strip out server side scripts

src/PHPHtmlParser/Options.php

@@ -20,6 +20,9 @@ class Options {
 		'whitespaceTextNode' => true,
 		'strict'             => false,
 		'enforceEncoding'    => null,
+		'cleanupInput'       => true,
+		'removeScripts'      => true,
+		'removeStyles'       => true
 	];
 
 	/**

tests/Options/CleanupTest.php

@@ -0,0 +1,73 @@
+<?php
+
+use PHPHtmlParser\Dom;
+
+class CleanupTest extends PHPUnit_Framework_TestCase {
+
+	public function testCleanupInputTrue()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'cleanupInput' => true,
+		]);
+		$dom->loadFromFile('tests/files/horrible.html');
+		$this->assertEquals(0, count($dom->find('style')));
+		$this->assertEquals(0, count($dom->find('script')));
+	}
+
+	public function testCleanupInputFalse()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'cleanupInput' => false,
+		]);
+		$dom->loadFromFile('tests/files/horrible.html');
+		$this->assertEquals(1, count($dom->find('style')));
+		$this->assertEquals(1, count($dom->find('script')));
+	}
+
+	public function testRemoveStylesTrue()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'removeStyles' => true,
+		]);
+		$dom->loadFromFile('tests/files/horrible.html');
+		$this->assertEquals(0, count($dom->find('style')));
+	}
+
+	public function testRemoveStylesFalse()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'removeStyles' => false,
+		]);
+		$dom->loadFromFile('tests/files/horrible.html');
+		$this->assertEquals(1, count($dom->find('style')));
+		$this->assertEquals('text/css',
+			$dom->find('style')->getAttribute('type'));
+	}
+
+	public function testRemoveScriptsTrue()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'removeScripts' => true,
+		]);
+		$dom->loadFromFile('tests/files/horrible.html');
+		$this->assertEquals(0, count($dom->find('script')));
+	}
+
+	public function testRemoveScriptsFalse()
+	{
+		$dom = new Dom;
+		$dom->setOptions([
+			'removeScripts' => false,
+		]);
+		$dom->loadFromFile('tests/files/horrible.html');
+		$this->assertEquals(1, count($dom->find('script')));
+		$this->assertEquals('text/JavaScript',
+			$dom->find('script')->getAttribute('type'));
+	}
+
+}