|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2018 The s9e Authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Configurator\Helpers; |
|
9
|
|
|
|
|
10
|
|
|
use DOMDocument; |
|
11
|
|
|
use DOMElement; |
|
12
|
|
|
use DOMXPath; |
|
13
|
|
|
use RuntimeException; |
|
14
|
|
|
|
|
15
|
|
|
abstract class TemplateLoader |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* XSL namespace |
|
19
|
|
|
*/ |
|
20
|
|
|
const XMLNS_XSL = 'http://www.w3.org/1999/XSL/Transform'; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* Get the XML content of an element |
|
24
|
|
|
* |
|
25
|
|
|
* @private |
|
26
|
|
|
* |
|
27
|
|
|
* @param DOMElement $element |
|
28
|
|
|
* @return string |
|
29
|
|
|
*/ |
|
30
|
|
|
public static function innerXML(DOMElement $element) |
|
31
|
|
|
{ |
|
32
|
|
|
// Serialize the XML then remove the outer element |
|
33
|
|
|
$xml = $element->ownerDocument->saveXML($element); |
|
34
|
|
|
$pos = 1 + strpos($xml, '>'); |
|
35
|
|
|
$len = strrpos($xml, '<') - $pos; |
|
36
|
|
|
|
|
37
|
|
|
// If the template is empty, return an empty string |
|
38
|
|
|
return ($len < 1) ? '' : substr($xml, $pos, $len); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Load a template as an xsl:template node |
|
43
|
|
|
* |
|
44
|
|
|
* Will attempt to load it as XML first, then as HTML as a fallback. Either way, an xsl:template |
|
45
|
|
|
* node is returned |
|
46
|
|
|
* |
|
47
|
|
|
* @param string $template |
|
48
|
|
|
* @return DOMDocument |
|
49
|
|
|
*/ |
|
50
|
|
|
public static function load($template) |
|
51
|
|
|
{ |
|
52
|
|
|
$dom = self::loadAsXML($template) ?: self::loadAsXML(self::fixEntities($template)); |
|
53
|
|
|
if ($dom) |
|
54
|
|
|
{ |
|
55
|
|
|
return $dom; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
// If the template contains an XSL element, abort now. Otherwise, try reparsing it as HTML |
|
59
|
|
|
if (strpos($template, '<xsl:') !== false) |
|
60
|
|
|
{ |
|
61
|
|
|
$error = libxml_get_last_error(); |
|
62
|
|
|
|
|
63
|
|
|
throw new RuntimeException('Invalid XSL: ' . $error->message); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
return self::loadAsHTML($template); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* Serialize a loaded template back into a string |
|
71
|
|
|
* |
|
72
|
|
|
* NOTE: removes the root node created by load() |
|
73
|
|
|
* |
|
74
|
|
|
* @param DOMDocument $dom |
|
75
|
|
|
* @return string |
|
76
|
|
|
*/ |
|
77
|
|
|
public static function save(DOMDocument $dom) |
|
78
|
|
|
{ |
|
79
|
|
|
return self::innerXML($dom->documentElement); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* Replace HTML entities and unescaped ampersands in given template |
|
84
|
|
|
* |
|
85
|
|
|
* @param string $template |
|
86
|
|
|
* @return string |
|
87
|
|
|
*/ |
|
88
|
|
|
protected static function fixEntities($template) |
|
89
|
|
|
{ |
|
90
|
|
|
return preg_replace_callback( |
|
91
|
|
|
'(&(?!quot;|amp;|apos;|lt;|gt;)\\w+;)', |
|
92
|
|
|
function ($m) |
|
93
|
|
|
{ |
|
94
|
|
|
return html_entity_decode($m[0], ENT_NOQUOTES, 'UTF-8'); |
|
95
|
|
|
}, |
|
96
|
|
|
preg_replace('(&(?![A-Za-z0-9]+;|#\\d+;|#x[A-Fa-f0-9]+;))', '&', $template) |
|
97
|
|
|
); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* Load given HTML template in a DOM document |
|
102
|
|
|
* |
|
103
|
|
|
* @param string $template Original template |
|
104
|
|
|
* @return DOMDocument |
|
105
|
|
|
*/ |
|
106
|
|
|
protected static function loadAsHTML($template) |
|
107
|
|
|
{ |
|
108
|
|
|
$dom = new DOMDocument; |
|
109
|
|
|
$html = '<?xml version="1.0" encoding="utf-8" ?><html><body><div>' . $template . '</div></body></html>'; |
|
110
|
|
|
|
|
111
|
|
|
$useErrors = libxml_use_internal_errors(true); |
|
112
|
|
|
$dom->loadHTML($html); |
|
113
|
|
|
self::removeInvalidAttributes($dom); |
|
114
|
|
|
libxml_use_internal_errors($useErrors); |
|
115
|
|
|
|
|
116
|
|
|
// Now dump the thing as XML then reload it with the proper root element |
|
117
|
|
|
$xml = '<?xml version="1.0" encoding="utf-8" ?><xsl:template xmlns:xsl="' . self::XMLNS_XSL . '">' . self::innerXML($dom->documentElement->firstChild->firstChild) . '</xsl:template>'; |
|
118
|
|
|
|
|
119
|
|
|
$useErrors = libxml_use_internal_errors(true); |
|
120
|
|
|
$dom->loadXML($xml); |
|
121
|
|
|
libxml_use_internal_errors($useErrors); |
|
122
|
|
|
|
|
123
|
|
|
return $dom; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
/** |
|
127
|
|
|
* Load given XSL template in a DOM document |
|
128
|
|
|
* |
|
129
|
|
|
* @param string $template Original template |
|
130
|
|
|
* @return bool|DOMDocument DOMDocument on success, FALSE otherwise |
|
131
|
|
|
*/ |
|
132
|
|
|
protected static function loadAsXML($template) |
|
133
|
|
|
{ |
|
134
|
|
|
$xml = '<?xml version="1.0" encoding="utf-8" ?><xsl:template xmlns:xsl="' . self::XMLNS_XSL . '">' . $template . '</xsl:template>'; |
|
135
|
|
|
|
|
136
|
|
|
$useErrors = libxml_use_internal_errors(true); |
|
137
|
|
|
$dom = new DOMDocument; |
|
138
|
|
|
$success = $dom->loadXML($xml); |
|
139
|
|
|
self::removeInvalidAttributes($dom); |
|
140
|
|
|
libxml_use_internal_errors($useErrors); |
|
141
|
|
|
|
|
142
|
|
|
return ($success) ? $dom : false; |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
/** |
|
146
|
|
|
* Remove attributes with an invalid name from given DOM document |
|
147
|
|
|
* |
|
148
|
|
|
* @param DOMDocument $dom |
|
149
|
|
|
* @return void |
|
150
|
|
|
*/ |
|
151
|
|
|
protected static function removeInvalidAttributes(DOMDocument $dom) |
|
152
|
|
|
{ |
|
153
|
|
|
$xpath = new DOMXPath($dom); |
|
154
|
|
|
foreach ($xpath->query('//@*') as $attribute) |
|
155
|
|
|
{ |
|
156
|
|
|
if (!preg_match('(^(?:[-\\w]+:)?(?!\\d)[-\\w]+$)D', $attribute->nodeName)) |
|
157
|
|
|
{ |
|
158
|
|
|
$attribute->parentNode->removeAttributeNode($attribute); |
|
159
|
|
|
} |
|
160
|
|
|
} |
|
161
|
|
|
} |
|
162
|
|
|
} |
s9e /
TextFormatter
