|
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\Node; |
|
12
|
|
|
|
|
13
|
|
|
use SN\HtmlSanitizer\Sanitizer\StringSanitizerTrait; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* A Tag Node. |
|
17
|
|
|
*/ |
|
18
|
|
|
class TagNode extends AbstractNode implements TagNodeInterface |
|
19
|
|
|
{ |
|
20
|
|
|
use HasChildrenTrait; |
|
21
|
|
|
use StringSanitizerTrait; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @var string |
|
25
|
|
|
*/ |
|
26
|
|
|
private $qName; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @var array<string, string|null> |
|
30
|
|
|
*/ |
|
31
|
|
|
private $attributes = []; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @var bool |
|
35
|
|
|
*/ |
|
36
|
|
|
private $isChildless = false; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Default values. |
|
40
|
|
|
* |
|
41
|
|
|
* @param NodeInterface $parent |
|
42
|
|
|
* @param string $qName |
|
43
|
|
|
* @param array $attributes |
|
44
|
|
|
* @param bool $isChildless |
|
45
|
|
|
*/ |
|
46
|
15 |
|
public function __construct(NodeInterface $parent, string $qName, array $attributes = [], bool $isChildless = false) |
|
47
|
|
|
{ |
|
48
|
15 |
|
parent::__construct($parent); |
|
49
|
|
|
|
|
50
|
15 |
|
$this->qName = $qName; |
|
51
|
15 |
|
$this->attributes = $attributes; |
|
52
|
15 |
|
$this->isChildless = $isChildless; |
|
53
|
15 |
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* @return string |
|
57
|
|
|
*/ |
|
58
|
12 |
|
public function getTagName(): string |
|
59
|
|
|
{ |
|
60
|
12 |
|
return $this->qName; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* @param string $name |
|
65
|
|
|
* |
|
66
|
|
|
* @return string|null |
|
67
|
|
|
*/ |
|
68
|
2 |
|
public function getAttribute(string $name): ?string |
|
69
|
|
|
{ |
|
70
|
2 |
|
return $this->attributes[$name] ?? null; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* @param string $name |
|
75
|
|
|
* @param string|null $value |
|
76
|
|
|
*/ |
|
77
|
5 |
|
public function setAttribute(string $name, ?string $value) |
|
78
|
|
|
{ |
|
79
|
|
|
// Always use only the first declaration (ease sanitization) |
|
80
|
5 |
|
if (!\array_key_exists($name, $this->attributes)) { |
|
81
|
5 |
|
$this->attributes[$name] = $value; |
|
82
|
|
|
} |
|
83
|
5 |
|
} |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* @return string |
|
87
|
|
|
*/ |
|
88
|
11 |
|
public function render(): string |
|
89
|
|
|
{ |
|
90
|
11 |
|
$tag = $this->getTagName(); |
|
91
|
|
|
|
|
92
|
11 |
|
if ($this->isChildless) { |
|
93
|
1 |
|
return '<'.$tag.$this->renderAttributes().' />'; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
11 |
|
return '<'.$tag.$this->renderAttributes().'>'.$this->renderChildren().'</'.$tag.'>'; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* @return string |
|
101
|
|
|
*/ |
|
102
|
11 |
|
private function renderAttributes(): string |
|
103
|
|
|
{ |
|
104
|
11 |
|
$rendered = []; |
|
105
|
|
|
|
|
106
|
11 |
|
foreach ($this->attributes as $name => $value) { |
|
107
|
6 |
|
if (null === $value) { |
|
108
|
|
|
// Tag should be removed as a sanitizer found suspect data inside |
|
109
|
1 |
|
continue; |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
6 |
|
$attr = $this->encodeHtmlEntities($name); |
|
113
|
6 |
|
if ('' !== $value) { |
|
114
|
|
|
// In quirks mode, IE8 does a poor job producing innerHTML values. |
|
115
|
|
|
// If JavaScript does: |
|
116
|
|
|
// nodeA.innerHTML = nodeB.innerHTML; |
|
117
|
|
|
// and nodeB contains (or even if ` was encoded properly): |
|
118
|
|
|
// <div attr="``foo=bar"> |
|
119
|
|
|
// then IE8 will produce: |
|
120
|
|
|
// <div attr=``foo=bar> |
|
121
|
|
|
// as the value of nodeB.innerHTML and assign it to nodeA. |
|
122
|
|
|
// IE8's HTML parser treats `` as a blank attribute value and foo=bar becomes a separate attribute. |
|
123
|
|
|
// Adding a space at the end of the attribute prevents this by forcing IE8 to put double |
|
124
|
|
|
// quotes around the attribute when computing nodeB.innerHTML. |
|
125
|
6 |
|
if (false !== \mb_strpos($value, '`')) { |
|
126
|
1 |
|
$value .= ' '; |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
6 |
|
$attr .= '="'.$this->encodeHtmlEntities($value).'"'; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
6 |
|
$rendered[] = $attr; |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
11 |
|
return $rendered ? ' '.\implode(' ', $rendered) : ''; |
|
136
|
|
|
} |
|
137
|
|
|
} |
|
138
|
|
|
|
snebes /
html-sanitizer
