|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* (c) Steve Nebes <[email protected]>. |
|
4
|
|
|
* |
|
5
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
6
|
|
|
* file that was distributed with this source code. |
|
7
|
|
|
*/ |
|
8
|
|
|
|
|
9
|
|
|
declare(strict_types=1); |
|
10
|
|
|
|
|
11
|
|
|
namespace SN\HtmlSanitizer; |
|
12
|
|
|
|
|
13
|
|
|
use SN\HtmlSanitizer\Model\Cursor; |
|
14
|
|
|
use SN\HtmlSanitizer\Node\DocumentNode; |
|
15
|
|
|
use SN\HtmlSanitizer\Node\TextNode; |
|
16
|
|
|
use SN\HtmlSanitizer\NodeVisitor\NodeVisitorInterface; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* The DomVisitor iterate over the parsed DOM tree and visit nodes using NodeVisitorInterface objects. |
|
20
|
|
|
* For performance reasons, these objects are split in 2 groups: generic ones and node-specific ones. |
|
21
|
|
|
*/ |
|
22
|
|
|
class DomVisitor |
|
23
|
|
|
{ |
|
24
|
|
|
/** |
|
25
|
|
|
* @var NodeVisitorInterface[] |
|
26
|
|
|
*/ |
|
27
|
|
|
private $nodeVisitors = []; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @param NodeVisitorInterface[] $visitors |
|
31
|
|
|
*/ |
|
32
|
5 |
|
public function __construct(array $visitors = []) |
|
33
|
|
|
{ |
|
34
|
5 |
|
foreach ($visitors as $visitor) { |
|
35
|
5 |
|
if ($visitor instanceof NodeVisitorInterface) { |
|
36
|
5 |
|
foreach ($visitor->getSupportedNodeNames() as $nodeName) { |
|
37
|
5 |
|
$this->nodeVisitors[$nodeName][] = $visitor; |
|
38
|
|
|
} |
|
39
|
|
|
} |
|
40
|
|
|
} |
|
41
|
5 |
|
} |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* @param \DOMNode $node |
|
45
|
|
|
* |
|
46
|
|
|
* @return DocumentNode |
|
47
|
|
|
*/ |
|
48
|
3 |
|
public function visit(\DOMNode $node): DocumentNode |
|
49
|
|
|
{ |
|
50
|
3 |
|
$cursor = new Cursor(); |
|
51
|
3 |
|
$cursor->node = new DocumentNode(); |
|
52
|
|
|
|
|
53
|
3 |
|
$this->visitNode($node, $cursor); |
|
54
|
|
|
|
|
55
|
3 |
|
return $cursor->node; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @param \DOMNode $node |
|
60
|
|
|
* @param Cursor $cursor |
|
61
|
|
|
*/ |
|
62
|
3 |
|
private function visitNode(\DOMNode $node, Cursor $cursor) |
|
63
|
|
|
{ |
|
64
|
|
|
/** @var NodeVisitorInterface[] $supportedVisitors */ |
|
65
|
3 |
|
$supportedVisitors = $this->nodeVisitors[$node->nodeName] ?? []; |
|
66
|
|
|
|
|
67
|
3 |
|
foreach ($supportedVisitors as $visitor) { |
|
68
|
3 |
|
if ($visitor->supports($node, $cursor)) { |
|
69
|
3 |
|
$visitor->enterNode($node, $cursor); |
|
70
|
|
|
} |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** @var \DOMNode $child */ |
|
74
|
3 |
|
foreach ($node->childNodes ?? [] as $child) { |
|
75
|
3 |
|
if ('#text' === $child->nodeName) { |
|
76
|
|
|
// Add text in the safe tree without a visitor for performance |
|
77
|
1 |
|
$cursor->node->addChild(new TextNode($cursor->node, $child->nodeValue)); |
|
78
|
3 |
|
} elseif (!$child instanceof \DOMText) { |
|
79
|
|
|
// Ignore comments for security reasons (interpreted differently by browsers) |
|
80
|
3 |
|
$this->visitNode($child, $cursor); |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
3 |
|
foreach (\array_reverse($supportedVisitors) as $visitor) { |
|
85
|
3 |
|
if ($visitor->supports($node, $cursor)) { |
|
86
|
3 |
|
$visitor->leaveNode($node, $cursor); |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
3 |
|
} |
|
90
|
|
|
} |
|
91
|
|
|
|
snebes /
html-sanitizer
