|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
class AMP_DOM_Utils { |
|
4
|
|
|
public static function get_dom_from_content( $content ) { |
|
5
|
|
|
$libxml_previous_state = libxml_use_internal_errors( true ); |
|
6
|
|
|
|
|
7
|
|
|
$dom = new DOMDocument; |
|
8
|
|
|
// Wrap in dummy tags, since XML needs one parent node. |
|
9
|
|
|
// It also makes it easier to loop through nodes. |
|
10
|
|
|
// We can later use this to extract our nodes. |
|
11
|
|
|
// Add utf-8 charset so loadHTML does not have problems parsing it. See: http://php.net/manual/en/domdocument.loadhtml.php#78243 |
|
12
|
|
|
$result = $dom->loadHTML( '<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body>' . $content . '</body></html>' ); |
|
13
|
|
|
|
|
14
|
|
|
libxml_clear_errors(); |
|
15
|
|
|
libxml_use_internal_errors( $libxml_previous_state ); |
|
16
|
|
|
|
|
17
|
|
|
if ( ! $result ) { |
|
18
|
|
|
return false; |
|
19
|
|
|
} |
|
20
|
|
|
|
|
21
|
|
|
return $dom; |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
|
|
public static function get_content_from_dom( $dom ) { |
|
25
|
|
|
// Only want children of the body tag, since we have a subset of HTML. |
|
26
|
|
|
$out = ''; |
|
27
|
|
|
$body = $dom->getElementsByTagName( 'body' )->item( 0 ); |
|
28
|
|
|
|
|
29
|
|
|
// AMP elements always need closing tags. |
|
30
|
|
|
// To force them, we can't use saveHTML (node support is 5.3+) and LIBXML_NOEMPTYTAG results in issues with self-closing tags like `br` and `hr`. |
|
31
|
|
|
// So, we're manually forcing closing tags. |
|
32
|
|
|
self::recursive_force_closing_tags( $dom, $body ); |
|
33
|
|
|
|
|
34
|
|
|
foreach ( $body->childNodes as $node ) { |
|
35
|
|
|
$out .= $dom->saveXML( $node ); |
|
36
|
|
|
} |
|
37
|
|
|
return $out; |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
public static function create_node( $dom, $tag, $attributes ) { |
|
41
|
|
|
$node = $dom->createElement( $tag ); |
|
42
|
|
|
self::add_attributes_to_node( $node, $attributes ); |
|
43
|
|
|
return $node; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
public static function get_node_attributes_as_assoc_array( $node ) { |
|
47
|
|
|
$attributes = array(); |
|
48
|
|
|
if ( ! $node->hasAttributes() ) { |
|
49
|
|
|
return $attributes; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
foreach ( $node->attributes as $attribute ) { |
|
53
|
|
|
$attributes[ $attribute->nodeName ] = $attribute->nodeValue; |
|
54
|
|
|
} |
|
55
|
|
|
return $attributes; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
public static function add_attributes_to_node( $node, $attributes ) { |
|
59
|
|
|
foreach ( $attributes as $name => $value ) { |
|
60
|
|
|
$node->setAttribute( $name, $value ); |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
public static function is_node_empty( $node ) { |
|
65
|
|
|
return false === $node->hasChildNodes() |
|
66
|
|
|
&& empty( $node->textContent ); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
public static function recursive_force_closing_tags( $dom, $node ) { |
|
70
|
|
|
if ( XML_ELEMENT_NODE !== $node->nodeType ) { |
|
71
|
|
|
return; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
if ( self::is_self_closing_tag( $node->nodeName ) ) { |
|
75
|
|
|
return; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
if ( self::is_node_empty( $node ) ) { |
|
79
|
|
|
$text_node = $dom->createTextNode( '' ); |
|
80
|
|
|
$node->appendChild( $text_node ); |
|
81
|
|
|
return; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
$num_children = $node->childNodes->length; |
|
85
|
|
|
for ( $i = $num_children - 1; $i >= 0; $i-- ) { |
|
86
|
|
|
$child = $node->childNodes->item( $i ); |
|
87
|
|
|
self::recursive_force_closing_tags( $dom, $child ); |
|
88
|
|
|
} |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
private static function is_self_closing_tag( $tag ) { |
|
92
|
|
|
// This function is called a lot; the static var prevents having to re-create the array every time. |
|
93
|
|
|
static $self_closing_tags; |
|
94
|
|
|
if ( ! isset( $self_closing_tags ) ) { |
|
95
|
|
|
// https://www.w3.org/TR/html5/syntax.html#serializing-html-fragments |
|
96
|
|
|
// Not all are valid AMP, but we include them for completeness. |
|
97
|
|
|
$self_closing_tags = array( |
|
98
|
|
|
'area', 'base', 'basefont', 'bgsound', 'br', 'col', 'embed', 'frame', 'hr', 'img', 'input', 'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr', |
|
99
|
|
|
); |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
return in_array( $tag, $self_closing_tags ); |
|
103
|
|
|
} |
|
104
|
|
|
} |
|
105
|
|
|
|
Automattic /
jetpack
