|
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 DOMXPath; |
|
12
|
|
|
|
|
13
|
|
|
abstract class NodeLocator |
|
14
|
|
|
{ |
|
15
|
|
|
/** |
|
16
|
|
|
* Return all attributes (literal or generated) that match given regexp |
|
17
|
|
|
* |
|
18
|
|
|
* @param DOMDocument $dom Document |
|
19
|
|
|
* @param string $regexp Regexp |
|
20
|
|
|
* @return DOMNode[] List of DOMNode instances |
|
21
|
|
|
*/ |
|
22
|
|
|
public static function getAttributesByRegexp(DOMDocument $dom, $regexp) |
|
23
|
|
|
{ |
|
24
|
|
|
return self::getNodesByRegexp($dom, $regexp, 'attribute', '@'); |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Return all DOMNodes whose content is CSS |
|
29
|
|
|
* |
|
30
|
|
|
* @param DOMDocument $dom Document |
|
31
|
|
|
* @return DOMNode[] List of DOMNode instances |
|
32
|
|
|
*/ |
|
33
|
|
|
public static function getCSSNodes(DOMDocument $dom) |
|
34
|
|
|
{ |
|
35
|
|
|
$regexp = '/^style$/i'; |
|
36
|
|
|
$nodes = array_merge( |
|
37
|
|
|
self::getAttributesByRegexp($dom, $regexp), |
|
38
|
|
|
self::getElementsByRegexp($dom, '/^style$/i') |
|
39
|
|
|
); |
|
40
|
|
|
|
|
41
|
|
|
return $nodes; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* Return all elements (literal or generated) that match given regexp |
|
46
|
|
|
* |
|
47
|
|
|
* @param DOMDocument $dom Document |
|
48
|
|
|
* @param string $regexp Regexp |
|
49
|
|
|
* @return DOMNode[] List of DOMNode instances |
|
50
|
|
|
*/ |
|
51
|
|
|
public static function getElementsByRegexp(DOMDocument $dom, $regexp) |
|
52
|
|
|
{ |
|
53
|
|
|
return self::getNodesByRegexp($dom, $regexp, 'element', ''); |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* Return all DOMNodes whose content is JavaScript |
|
58
|
|
|
* |
|
59
|
|
|
* @param DOMDocument $dom Document |
|
60
|
|
|
* @return DOMNode[] List of DOMNode instances |
|
61
|
|
|
*/ |
|
62
|
|
|
public static function getJSNodes(DOMDocument $dom) |
|
63
|
|
|
{ |
|
64
|
|
|
$regexp = '/^(?:data-s9e-livepreview-postprocess$|on)/i'; |
|
65
|
|
|
$nodes = array_merge( |
|
66
|
|
|
self::getAttributesByRegexp($dom, $regexp), |
|
67
|
|
|
self::getElementsByRegexp($dom, '/^script$/i') |
|
68
|
|
|
); |
|
69
|
|
|
|
|
70
|
|
|
return $nodes; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* Return all elements (literal or generated) that match given regexp |
|
75
|
|
|
* |
|
76
|
|
|
* Will return all <param/> descendants of <object/> and all attributes of <embed/> whose name |
|
77
|
|
|
* matches given regexp. This method will NOT catch <param/> elements whose 'name' attribute is |
|
78
|
|
|
* set via an <xsl:attribute/> |
|
79
|
|
|
* |
|
80
|
|
|
* @param DOMDocument $dom Document |
|
81
|
|
|
* @param string $regexp |
|
82
|
|
|
* @return DOMNode[] List of DOMNode instances |
|
83
|
|
|
*/ |
|
84
|
|
|
public static function getObjectParamsByRegexp(DOMDocument $dom, $regexp) |
|
85
|
|
|
{ |
|
86
|
|
|
$xpath = new DOMXPath($dom); |
|
87
|
|
|
$nodes = []; |
|
88
|
|
|
|
|
89
|
|
|
// Collect attributes from <embed/> elements |
|
90
|
|
|
foreach (self::getAttributesByRegexp($dom, $regexp) as $attribute) |
|
91
|
|
|
{ |
|
92
|
|
|
if ($attribute->nodeType === XML_ATTRIBUTE_NODE) |
|
93
|
|
|
{ |
|
94
|
|
|
if (strtolower($attribute->parentNode->localName) === 'embed') |
|
95
|
|
|
{ |
|
96
|
|
|
$nodes[] = $attribute; |
|
97
|
|
|
} |
|
98
|
|
|
} |
|
99
|
|
|
elseif ($xpath->evaluate('count(ancestor::embed)', $attribute)) |
|
100
|
|
|
{ |
|
101
|
|
|
// Assuming <xsl:attribute/> or <xsl:copy-of/> |
|
102
|
|
|
$nodes[] = $attribute; |
|
103
|
|
|
} |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
// Collect <param/> descendants of <object/> elements |
|
107
|
|
|
foreach ($xpath->query('//object//param') as $param) |
|
108
|
|
|
{ |
|
109
|
|
|
if (preg_match($regexp, $param->getAttribute('name'))) |
|
110
|
|
|
{ |
|
111
|
|
|
$nodes[] = $param; |
|
112
|
|
|
} |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
return $nodes; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* Return all DOMNodes whose content is an URL |
|
120
|
|
|
* |
|
121
|
|
|
* NOTE: it will also return HTML4 nodes whose content is an URI |
|
122
|
|
|
* |
|
123
|
|
|
* @param DOMDocument $dom Document |
|
124
|
|
|
* @return DOMNode[] List of DOMNode instances |
|
125
|
|
|
*/ |
|
126
|
|
|
public static function getURLNodes(DOMDocument $dom) |
|
127
|
|
|
{ |
|
128
|
|
|
$regexp = '/(?:^(?:action|background|c(?:ite|lassid|odebase)|data|formaction|href|icon|longdesc|manifest|p(?:ing|luginspage|oster|rofile)|usemap)|src)$/i'; |
|
129
|
|
|
$nodes = self::getAttributesByRegexp($dom, $regexp); |
|
130
|
|
|
|
|
131
|
|
|
/** |
|
132
|
|
|
* @link http://helpx.adobe.com/flash/kb/object-tag-syntax-flash-professional.html |
|
133
|
|
|
* @link http://www.sitepoint.com/control-internet-explorer/ |
|
134
|
|
|
*/ |
|
135
|
|
|
foreach (self::getObjectParamsByRegexp($dom, '/^(?:dataurl|movie)$/i') as $param) |
|
136
|
|
|
{ |
|
137
|
|
|
$node = $param->getAttributeNode('value'); |
|
138
|
|
|
if ($node) |
|
139
|
|
|
{ |
|
140
|
|
|
$nodes[] = $node; |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
return $nodes; |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* Return all nodes (literal or generated) that match given regexp |
|
149
|
|
|
* |
|
150
|
|
|
* @param DOMDocument $dom Document |
|
151
|
|
|
* @param string $regexp Regexp |
|
152
|
|
|
* @param string $type Node type ('element' or 'attribute') |
|
153
|
|
|
* @param string $prefix Prefix used in XPath ('' or '@') |
|
154
|
|
|
* @return DOMNode[] List of DOMNode instances |
|
155
|
|
|
*/ |
|
156
|
|
|
protected static function getNodesByRegexp(DOMDocument $dom, $regexp, $type, $prefix) |
|
157
|
|
|
{ |
|
158
|
|
|
$candidates = []; |
|
159
|
|
|
$xpath = new DOMXPath($dom); |
|
160
|
|
|
|
|
161
|
|
|
// Get natural nodes |
|
162
|
|
|
foreach ($xpath->query('//' . $prefix . '*') as $node) |
|
163
|
|
|
{ |
|
164
|
|
|
$candidates[] = [$node, $node->nodeName]; |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
// Get XSL-generated nodes |
|
168
|
|
|
foreach ($xpath->query('//xsl:' . $type) as $node) |
|
169
|
|
|
{ |
|
170
|
|
|
$candidates[] = [$node, $node->getAttribute('name')]; |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
// Get xsl:copy-of nodes |
|
174
|
|
|
foreach ($xpath->query('//xsl:copy-of') as $node) |
|
175
|
|
|
{ |
|
176
|
|
|
if (preg_match('/^' . $prefix . '(\\w+)$/', $node->getAttribute('select'), $m)) |
|
177
|
|
|
{ |
|
178
|
|
|
$candidates[] = [$node, $m[1]]; |
|
179
|
|
|
} |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
|
|
// Filter candidate nodes |
|
183
|
|
|
$nodes = []; |
|
184
|
|
|
foreach ($candidates as list($node, $name)) |
|
185
|
|
|
{ |
|
186
|
|
|
if (preg_match($regexp, $name)) |
|
187
|
|
|
{ |
|
188
|
|
|
$nodes[] = $node; |
|
189
|
|
|
} |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
|
|
return $nodes; |
|
193
|
|
|
} |
|
194
|
|
|
} |
s9e /
TextFormatter
