1: <?php
 2: /**
 3:  * This file is part of the FastFeed package.
 4:  *
 5:  * (c) Daniel González <daniel@desarrolla2.com>
 6:  *
 7:  * For the full copyright and license information, please view the LICENSE
 8:  * file that was distributed with this source code.
 9:  */
10: namespace FastFeed\Processor;
11: 
12: use FastFeed\Exception\InvalidArgumentException;
13: use FastFeed\Item;
14: 
15: /**
16:  * SanitizerProcessor
17:  * Remove malicious HTML
18:  */
19: class SanitizerProcessor implements ProcessorInterface
20: {
21:     /**
22:      * @var HTMLPurifier
23:      */
24:     protected $purifier;
25: 
26:     /**
27:      * @param null $cacheDirectory
28:      *
29:      * @throws \FastFeed\Exception\InvalidArgumentException
30:      */
31:     public function __construct($cacheDirectory = null)
32:     {
33:         if (!$cacheDirectory) {
34:             $cacheDirectory = realpath(sys_get_temp_dir());
35:         }
36: 
37:         if (!is_writable($cacheDirectory)) {
38:             throw new InvalidArgumentException($cacheDirectory . ' is not writable');
39:         }
40:         // require to configure some CONSTANST
41:         new \HTMLPurifier_Bootstrap();
42:         $config = \HTMLPurifier_Config::createDefault();
43:         $config->set('Cache.SerializerPath', $cacheDirectory);
44:         $this->purifier = new \HTMLPurifier($config);
45:     }
46: 
47:     /**
48:      * Execute processor
49:      *
50:      * @param array $items
51:      *
52:      * @return array
53:      */
54:     public function process(array $items)
55:     {
56:         foreach ($items as $key => $item) {
57:             $items[$key] = $this->doClean($item);
58:         }
59: 
60:         return $items;
61:     }
62: 
63:     /**
64:      * @param Item $item
65:      *
66:      * @return string
67:      */
68:     protected function doClean(Item $item)
69:     {
70:         $item->setIntro(
71:             $this->purifier->purify($item->getIntro())
72:         );
73:         $item->setContent(
74:             $this->purifier->purify($item->getContent())
75:         );
76: 
77:         return $item;
78:     }
79: }
80: