1
|
|
|
<?php namespace BetterDOMDocument; |
2
|
|
|
|
3
|
|
|
use Symfony\Component\CssSelector\CssSelectorConverter; |
4
|
|
|
|
5
|
|
|
/** |
6
|
|
|
* Highwire Better DOM Document |
7
|
|
|
* |
8
|
|
|
* Copyright (c) 2010-2011 Board of Trustees, Leland Stanford Jr. University |
9
|
|
|
* This software is open-source licensed under the GNU Public License Version 2 or later |
10
|
|
|
*/ |
11
|
|
|
class DOMDoc extends \DOMDocument { |
12
|
|
|
|
13
|
|
|
private $auto_ns = FALSE; |
14
|
|
|
public $ns = array(); |
15
|
|
|
public $default_ns = FALSE; |
16
|
|
|
public $error_checking = 'strict'; // Can be 'strict', 'warning', 'none' / FALSE |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* Create a new DOMDoc |
20
|
|
|
* |
21
|
|
|
* @param mixed $xml |
22
|
|
|
* $xml can either be an XML string, a DOMDocument, or a DOMElement. |
23
|
|
|
* You can also pass FALSE or NULL (or omit it) and load XML later using loadXML or loadHTML |
24
|
|
|
* |
25
|
|
|
* @param mixed $auto_register_namespaces |
26
|
|
|
* Auto-register namespaces. All namespaces in the root element will be registered for use in xpath queries. |
27
|
|
|
* Namespaces that are not declared in the root element will not be auto-registered |
28
|
|
|
* Defaults to TRUE (Meaning it will auto register all auxiliary namespaces but not the default namespace). |
29
|
|
|
* Pass a prefix string to automatically register the default namespace. |
30
|
|
|
* Pass FALSE to disable auto-namespace registeration |
31
|
|
|
* |
32
|
|
|
* @param bool $error_checking |
33
|
|
|
* Can be 'strict', 'warning', or 'none. Defaults to 'strict'. |
34
|
|
|
* 'none' supresses all errors |
35
|
|
|
* 'warning' is the default behavior in DOMDocument |
36
|
|
|
* 'strict' corresponds to DOMDocument strictErrorChecking TRUE |
37
|
|
|
*/ |
38
|
|
|
public function __construct($xml = FALSE, $auto_register_namespaces = TRUE, $error_checking = 'strict') { |
39
|
|
|
parent::__construct(); |
40
|
|
|
|
41
|
|
|
$this->setErrorChecking($error_checking); |
42
|
|
|
|
43
|
|
|
if(is_object($xml)){ |
44
|
|
|
$class = get_class($xml); |
45
|
|
|
if ($class == 'DOMElement') { |
46
|
|
|
$this->appendChild($this->importNode($xml, true)); |
47
|
|
|
} |
48
|
|
|
if ($class == 'DOMDocument') { |
49
|
|
|
if ($xml->documentElement) { |
50
|
|
|
$this->appendChild($this->importNode($xml->documentElement, true)); |
51
|
|
|
} |
52
|
|
|
} |
53
|
|
|
if ($class == 'BetterDOMDocument\DOMDoc') { |
54
|
|
|
if ($xml->documentElement) { |
55
|
|
|
$this->appendChild($this->importNode($xml->documentElement, true)); |
56
|
|
|
} |
57
|
|
|
$this->ns = $xml->ns; |
58
|
|
|
} |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
if ($xml && is_string($xml)) { |
62
|
|
|
if ($this->error_checking == 'none') { |
63
|
|
|
@$this->loadXML($xml, LIBXML_COMPACT); |
64
|
|
|
} |
65
|
|
|
else { |
66
|
|
|
if (!$this->loadXML($xml, LIBXML_COMPACT)) { |
67
|
|
|
trigger_error('BetterDOMDocument\DOMDoc: Could not load: ' . htmlspecialchars($xml), E_USER_WARNING); |
68
|
|
|
} |
69
|
|
|
} |
70
|
|
|
} |
71
|
|
|
if ($auto_register_namespaces) { |
72
|
|
|
$this->AutoRegisterNamespace($auto_register_namespaces); |
73
|
|
|
} |
74
|
|
|
} |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* Register a namespace to be used in xpath queries |
78
|
|
|
* |
79
|
|
|
* @param string $prefix |
80
|
|
|
* Namespace prefix to register |
81
|
|
|
* |
82
|
|
|
* @param string $url |
83
|
|
|
* Connonical URL for this namespace prefix |
84
|
|
|
*/ |
85
|
|
|
public function registerNamespace($prefix, $url) { |
86
|
|
|
$this->ns[$prefix] = $url; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* Get the list of registered namespaces as an array |
91
|
|
|
*/ |
92
|
|
|
public function getNamespaces() { |
93
|
|
|
return $this->ns; |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* Given a namespace URL, get the prefix |
98
|
|
|
* |
99
|
|
|
* @param string $url |
100
|
|
|
* Connonical URL for this namespace prefix |
101
|
|
|
* |
102
|
|
|
* @return string|false |
103
|
|
|
* The namespace prefix or FALSE if there is no namespace with that URL |
104
|
|
|
*/ |
105
|
|
|
public function lookupPrefix($url) { |
106
|
|
|
return array_search($url, $this->ns); |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
/** |
110
|
|
|
* Given a namespace prefix, get the URL |
111
|
|
|
* |
112
|
|
|
* @param string $prefix |
113
|
|
|
* namespace prefix |
114
|
|
|
* |
115
|
|
|
* return string|false |
116
|
|
|
* The namespace URL or FALSE if there is no namespace with that prefix |
117
|
|
|
*/ |
118
|
|
|
public function lookupURL($prefix) { |
119
|
|
|
if (isset($this->ns[$prefix])) { |
120
|
|
|
return $this->ns[$prefix]; |
121
|
|
|
} |
122
|
|
|
else { |
123
|
|
|
return FALSE; |
124
|
|
|
} |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* Given an xpath, get a list of nodes. |
129
|
|
|
* |
130
|
|
|
* @param string $xpath |
131
|
|
|
* xpath to be used for query |
132
|
|
|
* |
133
|
|
|
* @param mixed $context |
134
|
|
|
* $context can either be an xpath string, or a DOMElement |
135
|
|
|
* Provides context for the xpath query |
136
|
|
|
* |
137
|
|
|
* @return DOMList|false |
138
|
|
|
* A DOMList object, which is very similar to a DOMNodeList, but with better iterabilility. |
139
|
|
|
*/ |
140
|
|
|
public function xpath($xpath, $context = NULL) { |
141
|
|
|
$this->createContext($context, 'xpath', FALSE); |
142
|
|
|
|
143
|
|
|
if ($context === FALSE) { |
144
|
|
|
return FALSE; |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
$xob = new \DOMXPath($this); |
148
|
|
|
|
149
|
|
|
// Register the namespaces |
150
|
|
|
foreach ($this->ns as $namespace => $url) { |
151
|
|
|
$xob->registerNamespace($namespace, $url); |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
if ($context) { |
155
|
|
|
$result = $xob->query($xpath, $context); |
156
|
|
|
} |
157
|
|
|
else { |
158
|
|
|
$result = $xob->query($xpath); |
159
|
|
|
} |
160
|
|
|
|
161
|
|
|
if ($result) { |
162
|
|
|
return new DOMList($result, $this); |
163
|
|
|
} |
164
|
|
|
else { |
165
|
|
|
return FALSE; |
166
|
|
|
} |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
|
170
|
|
|
/** |
171
|
|
|
* Given an xpath, get a single node (first one found) |
172
|
|
|
* |
173
|
|
|
* @param string $xpath |
174
|
|
|
* xpath to be used for query |
175
|
|
|
* |
176
|
|
|
* @param mixed $context |
177
|
|
|
* $context can either be an xpath string, or a DOMElement |
178
|
|
|
* Provides context for the xpath query |
179
|
|
|
* |
180
|
|
|
* @return mixed |
181
|
|
|
* The first node found by the xpath query |
182
|
|
|
*/ |
183
|
|
|
public function xpathSingle($xpath, $context = NULL) { |
184
|
|
|
$result = $this->xpath($xpath, $context); |
185
|
|
|
|
186
|
|
|
if (empty($result) || !count($result)) { |
187
|
|
|
return FALSE; |
188
|
|
|
} |
189
|
|
|
else { |
190
|
|
|
return $result->item(0); |
191
|
|
|
} |
192
|
|
|
} |
193
|
|
|
|
194
|
|
|
|
195
|
|
|
/** |
196
|
|
|
* Given an CSS selector, get a list of nodes. |
197
|
|
|
* |
198
|
|
|
* @param string $css_selector |
199
|
|
|
* CSS Selector to be used for query |
200
|
|
|
* |
201
|
|
|
* @param mixed $context |
202
|
|
|
* $context can either be an xpath string, or a DOMElement |
203
|
|
|
* Provides context for the CSS selector |
204
|
|
|
* |
205
|
|
|
* @return DOMList |
206
|
|
|
* A DOMList object, which is very similar to a DOMNodeList, but with better iterabilility. |
207
|
|
|
*/ |
208
|
|
|
public function select($css_selector, $context = NULL) { |
209
|
|
|
$converter = new CssSelectorConverter(); |
210
|
|
|
$xpath = $converter->toXPath($css_selector); |
211
|
|
|
|
212
|
|
|
return $this->xpath($xpath, $context); |
213
|
|
|
} |
214
|
|
|
|
215
|
|
|
/** |
216
|
|
|
* Given an CSS selector, get a single node. |
217
|
|
|
* |
218
|
|
|
* @param string $css_selector |
219
|
|
|
* CSS Selector to be used for query |
220
|
|
|
* |
221
|
|
|
* @param mixed $context |
222
|
|
|
* $context can either be an xpath string, or a DOMElement |
223
|
|
|
* Provides context for the CSS selector |
224
|
|
|
* |
225
|
|
|
* @return DOMList |
226
|
|
|
* A DOMList object, which is very similar to a DOMNodeList, but with better iterabilility. |
227
|
|
|
*/ |
228
|
|
|
public function selectSingle($css_selector, $context = NULL) { |
229
|
|
|
$converter = new CssSelectorConverter(); |
230
|
|
|
$xpath = $converter->toXPath($css_selector); |
231
|
|
|
|
232
|
|
|
return $this->xpathSingle($xpath, $context); |
233
|
|
|
} |
234
|
|
|
|
235
|
|
|
/** |
236
|
|
|
* Get the document (or an element) as an array |
237
|
|
|
* |
238
|
|
|
* @param string $raw |
239
|
|
|
* Can be either FALSE, 'full', or 'inner'. Defaults to FALSE. |
240
|
|
|
* When set to 'full' every node's full XML is also attached to the array |
241
|
|
|
* When set to 'inner' every node's inner XML is attached to the array. |
242
|
|
|
* |
243
|
|
|
* @param mixed $context |
244
|
|
|
* Optional context node. Can pass an DOMElement object or an xpath string. |
245
|
|
|
* If passed, only the given node will be used when generating the array |
246
|
|
|
*/ |
247
|
|
|
public function getArray($raw = FALSE, $context = NULL) { |
248
|
|
|
$array = false; |
249
|
|
|
|
250
|
|
|
$this->createContext($context, 'xpath', FALSE); |
251
|
|
|
|
252
|
|
|
if ($context) { |
253
|
|
|
if ($raw == 'full') { |
254
|
|
|
$array['#raw'] = $this->saveXML($context); |
255
|
|
|
} |
256
|
|
|
if ($raw == 'inner') { |
257
|
|
|
$array['#raw'] = $this->innerText($context); |
258
|
|
|
} |
259
|
|
|
if ($context->hasAttributes()) { |
260
|
|
|
foreach ($context->attributes as $attr) { |
261
|
|
|
$array['@'.$attr->nodeName] = $attr->nodeValue; |
262
|
|
|
} |
263
|
|
|
} |
264
|
|
|
|
265
|
|
|
if ($context->hasChildNodes()) { |
266
|
|
|
if ($context->childNodes->length == 1 && $context->firstChild->nodeType == XML_TEXT_NODE) { |
267
|
|
|
$array['#text'] = $context->firstChild->nodeValue; |
268
|
|
|
} |
269
|
|
|
else { |
270
|
|
|
foreach ($context->childNodes as $childNode) { |
271
|
|
|
if ($childNode->nodeType == XML_ELEMENT_NODE) { |
272
|
|
|
$array[$childNode->nodeName][] = $this->getArray($raw, $childNode); |
273
|
|
|
} |
274
|
|
|
elseif ($childNode->nodeType == XML_CDATA_SECTION_NODE) { |
275
|
|
|
$array['#text'] = $childNode->textContent; |
276
|
|
|
} |
277
|
|
|
} |
278
|
|
|
} |
279
|
|
|
} |
280
|
|
|
} |
281
|
|
|
// Else no node was passed, which means we are processing the entire domDocument |
282
|
|
|
else { |
283
|
|
|
foreach ($this->childNodes as $childNode) { |
284
|
|
|
if ($childNode->nodeType == XML_ELEMENT_NODE) { |
285
|
|
|
$array[$childNode->nodeName][] = $this->getArray($raw, $childNode); |
286
|
|
|
} |
287
|
|
|
} |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
return $array; |
291
|
|
|
} |
292
|
|
|
|
293
|
|
|
/** |
294
|
|
|
* Get the inner text of an element |
295
|
|
|
* |
296
|
|
|
* @param mixed $context |
297
|
|
|
* Optional context node. Can pass an DOMElement object or an xpath string. |
298
|
|
|
*/ |
299
|
|
|
public function innerText($context = NULL) { |
300
|
|
|
$this->createContext($context, 'xpath'); |
301
|
|
|
|
302
|
|
|
$pattern = "/<".preg_quote($context->nodeName)."\b[^>]*>(.*)<\/".preg_quote($context->nodeName).">/s"; |
303
|
|
|
$matches = array(); |
304
|
|
|
if (preg_match($pattern, $this->saveXML($context), $matches)) { |
305
|
|
|
return $matches[1]; |
306
|
|
|
} |
307
|
|
|
else { |
308
|
|
|
return ''; |
309
|
|
|
} |
310
|
|
|
} |
311
|
|
|
|
312
|
|
|
/** |
313
|
|
|
* Create an DOMElement from XML and attach it to the DOMDocument |
314
|
|
|
* |
315
|
|
|
* Note that this does not place it anywhere in the dom tree, it merely imports it. |
316
|
|
|
* |
317
|
|
|
* @param string $xml |
318
|
|
|
* XML string to import |
319
|
|
|
*/ |
320
|
|
|
public function createElementFromXML($xml) { |
321
|
|
|
|
322
|
|
|
// To make thing easy and make sure namespaces work properly, we add the root namespace delcarations if it is not declared |
323
|
|
|
$namespaces = $this->ns; |
324
|
|
|
$xml = preg_replace_callback('/<[^\?^!].+?>/s', function($root_match) use ($namespaces) { |
325
|
|
|
preg_match('/<([^ <>]+)[\d\s]?.*?>/s', $root_match[0], $root_tag); |
326
|
|
|
$new_root = $root_tag[1]; |
327
|
|
|
if (strpos($new_root, ':')) { |
328
|
|
|
$parts = explode(':', $new_root); |
329
|
|
|
$prefix = $parts[0]; |
330
|
|
|
if (isset($namespaces[$prefix])) { |
331
|
|
|
if (!strpos($root_match[0], "xmlns:$prefix")) { |
332
|
|
|
$new_root .= " xmlns:$prefix='" . $namespaces[$prefix] . "'"; |
333
|
|
|
} |
334
|
|
|
} |
335
|
|
|
} |
336
|
|
|
return str_replace($root_tag[1], $new_root, $root_match[0]); |
337
|
|
|
}, $xml, 1); |
338
|
|
|
|
339
|
|
|
$dom = new DOMDoc($xml, $this->auto_ns); |
340
|
|
|
if (!$dom->documentElement) { |
341
|
|
|
trigger_error('BetterDomDocument\DOMDoc Error: Invalid XML: ' . $xml); |
342
|
|
|
} |
343
|
|
|
$element = $dom->documentElement; |
344
|
|
|
|
345
|
|
|
// Merge the namespaces |
346
|
|
|
foreach ($dom->getNamespaces() as $prefix => $url) { |
347
|
|
|
$this->registerNamespace($prefix, $url); |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
return $this->importNode($element, true); |
351
|
|
|
} |
352
|
|
|
|
353
|
|
|
/** |
354
|
|
|
* Append a child to the context node, make it the last child |
355
|
|
|
* |
356
|
|
|
* @param mixed $newnode |
357
|
|
|
* $newnode can either be an XML string, a DOMDocument, or a DOMElement. |
358
|
|
|
* |
359
|
|
|
* @param mixed $context |
360
|
|
|
* $context can either be an xpath string, or a DOMElement |
361
|
|
|
* Omiting $context results in using the root document element as the context |
362
|
|
|
* |
363
|
|
|
* @return DOMElement|false |
364
|
|
|
* The $newnode, properly attached to DOMDocument. If you passed $newnode as a DOMElement |
365
|
|
|
* then you should replace your DOMElement with the returned one. |
366
|
|
|
*/ |
367
|
|
|
public function append($newnode, $context = NULL) { |
368
|
|
|
$this->createContext($newnode, 'xml'); |
369
|
|
|
$this->createContext($context, 'xpath'); |
370
|
|
|
|
371
|
|
|
if (!$context || !$newnode) { |
372
|
|
|
return FALSE; |
373
|
|
|
} |
374
|
|
|
|
375
|
|
|
if ($newnode->ownerDocument === $this) { |
376
|
|
|
$appendnode = $newnode; |
377
|
|
|
} |
378
|
|
|
else { |
379
|
|
|
$appendnode = $this->importNode($newnode, true); |
380
|
|
|
} |
381
|
|
|
|
382
|
|
|
return $context->appendChild($appendnode); |
383
|
|
|
} |
384
|
|
|
|
385
|
|
|
/** |
386
|
|
|
* Append a child to the context node, make it the first child |
387
|
|
|
* |
388
|
|
|
* @param mixed $newnode |
389
|
|
|
* $newnode can either be an XML string, a DOMDocument, or a DOMElement. |
390
|
|
|
* |
391
|
|
|
* @param mixed $context |
392
|
|
|
* $context can either be an xpath string, or a DOMElement |
393
|
|
|
* Omiting $context results in using the root document element as the context |
394
|
|
|
* |
395
|
|
|
* @return DOMElement|false |
396
|
|
|
* The $newnode, properly attached to DOMDocument. If you passed $newnode as a DOMElement |
397
|
|
|
* then you should replace your DOMElement with the returned one. |
398
|
|
|
*/ |
399
|
|
|
public function prepend($newnode, $context = NULL) { |
400
|
|
|
$this->createContext($newnode, 'xml'); |
401
|
|
|
$this->createContext($context, 'xpath'); |
402
|
|
|
|
403
|
|
|
if (!$context || !$newnode) { |
404
|
|
|
return FALSE; |
405
|
|
|
} |
406
|
|
|
|
407
|
|
|
return $context->insertBefore($newnode, $context->firstChild); |
408
|
|
|
} |
409
|
|
|
|
410
|
|
|
/** |
411
|
|
|
* Prepend a sibling to the context node, put it just before the context node |
412
|
|
|
* |
413
|
|
|
* @param mixed $newnode |
414
|
|
|
* $newnode can either be an XML string, a DOMDocument, or a DOMElement. |
415
|
|
|
* |
416
|
|
|
* @param mixed $context |
417
|
|
|
* $context can either be an xpath string, or a DOMElement |
418
|
|
|
* Omiting $context results in using the root document element as the context |
419
|
|
|
* |
420
|
|
|
* @return DOMElement|false |
421
|
|
|
* The $newnode, properly attached to DOMDocument. If you passed $newnode as a DOMElement |
422
|
|
|
* then you should replace your DOMElement with the returned one. |
423
|
|
|
*/ |
424
|
|
|
public function prependSibling($newnode, $context = NULL) { |
425
|
|
|
$this->createContext($newnode, 'xml'); |
426
|
|
|
$this->createContext($context, 'xpath'); |
427
|
|
|
|
428
|
|
|
if (!$context || !$newnode) { |
429
|
|
|
return FALSE; |
430
|
|
|
} |
431
|
|
|
|
432
|
|
|
return $context->parentNode->insertBefore($newnode, $context); |
433
|
|
|
} |
434
|
|
|
|
435
|
|
|
/** |
436
|
|
|
* Append a sibling to the context node, put it just after the context node |
437
|
|
|
* |
438
|
|
|
* @param mixed $newnode |
439
|
|
|
* $newnode can either be an XML string, a DOMDocument, or a DOMElement. |
440
|
|
|
* |
441
|
|
|
* @param mixed $context |
442
|
|
|
* $context can either be an xpath string, or a DOMElement |
443
|
|
|
* Omiting $context results in using the root document element as the context |
444
|
|
|
* |
445
|
|
|
* @return DOMElement|false |
446
|
|
|
* The $newnode, properly attached to DOMDocument. If you passed $newnode as a DOMElement |
447
|
|
|
* then you should replace your DOMElement with the returned one. |
448
|
|
|
*/ |
449
|
|
|
public function appendSibling($newnode, $context) { |
450
|
|
|
$this->createContext($newnode, 'xml'); |
451
|
|
|
$this->createContext($context, 'xpath'); |
452
|
|
|
|
453
|
|
|
if (!$context){ |
454
|
|
|
return FALSE; |
455
|
|
|
} |
456
|
|
|
|
457
|
|
|
if ($context->nextSibling) { |
458
|
|
|
// $context has an immediate sibling : insert newnode before this one |
459
|
|
|
return $context->parentNode->insertBefore($newnode, $context->nextSibling); |
460
|
|
|
} |
461
|
|
|
else { |
462
|
|
|
// $context has no sibling next to it : insert newnode as last child of it's parent |
463
|
|
|
if ($newnode->ownerDocument === $this) { |
464
|
|
|
$appendnode = $newnode; |
465
|
|
|
} |
466
|
|
|
else { |
467
|
|
|
$appendnode = $this->importNode($newnode, true); |
468
|
|
|
} |
469
|
|
|
return $context->parentNode->appendChild($appendnode); |
470
|
|
|
} |
471
|
|
|
} |
472
|
|
|
|
473
|
|
|
/** |
474
|
|
|
* Given an xpath or DOMElement, return a new DOMDoc. |
475
|
|
|
* |
476
|
|
|
* @param mixed $node |
477
|
|
|
* $node can either be an xpath string or a DOMElement. |
478
|
|
|
* |
479
|
|
|
* @return DOMDoc |
480
|
|
|
* A new DOMDoc created from the xpath or DOMElement |
481
|
|
|
*/ |
482
|
|
|
public function extract($node, $auto_register_namespaces = TRUE, $error_checking = 'none') { |
483
|
|
|
$this->createContext($node, 'xpath'); |
484
|
|
|
$dom = new DOMDoc($node, $auto_register_namespaces, $error_checking); |
485
|
|
|
$dom->ns = $this->ns; |
486
|
|
|
return $dom; |
487
|
|
|
} |
488
|
|
|
|
489
|
|
|
/** |
490
|
|
|
* Given a pair of nodes, replace the first with the second |
491
|
|
|
* |
492
|
|
|
* @param mixed $node |
493
|
|
|
* Node to be replaced. Can either be an xpath string or a DOMDocument (or even a DOMNode). |
494
|
|
|
* |
495
|
|
|
* @param mixed $replace |
496
|
|
|
* Replace $node with $replace. Replace can be an XML string, or a DOMNode |
497
|
|
|
* |
498
|
|
|
* @return mixed |
499
|
|
|
* The overwritten / replaced node. |
500
|
|
|
*/ |
501
|
|
|
public function replace($node, $replace) { |
502
|
|
|
$this->createContext($node, 'xpath'); |
503
|
|
|
$this->createContext($replace, 'xml'); |
504
|
|
|
|
505
|
|
|
if (!$node || !$replace) { |
506
|
|
|
return FALSE; |
507
|
|
|
} |
508
|
|
|
|
509
|
|
|
if (!$replace->ownerDocument->documentElement->isSameNode($this->documentElement)) { |
510
|
|
|
$replace = $this->importNode($replace, true); |
511
|
|
|
} |
512
|
|
|
$node->parentNode->replaceChild($replace, $node); |
513
|
|
|
$node = $replace; |
514
|
|
|
return $node; |
515
|
|
|
} |
516
|
|
|
|
517
|
|
|
/** |
518
|
|
|
* Given a node(s), remove / delete them |
519
|
|
|
* |
520
|
|
|
* @param mixed $node |
521
|
|
|
* Can pass a DOMNode, a NodeList, DOMNodeList, an xpath string, or an array of any of these. |
522
|
|
|
*/ |
523
|
|
|
public function remove($node) { |
524
|
|
|
// We can't use createContext here because we want to use the entire nodeList (not just a single element) |
525
|
|
|
if (is_string($node)) { |
526
|
|
|
$node = $this->xpath($node); |
527
|
|
|
} |
528
|
|
|
|
529
|
|
|
if ($node) { |
530
|
|
|
if (is_array($node) || get_class($node) == 'BetterDOMDocument\DOMList') { |
531
|
|
|
foreach($node as $item) { |
532
|
|
|
$this->remove($item); |
533
|
|
|
} |
534
|
|
|
} |
535
|
|
|
else if (get_class($node) == 'DOMNodeList') { |
536
|
|
|
$this->remove(new DOMList($node, $this)); |
537
|
|
|
} |
538
|
|
|
else { |
539
|
|
|
$parent = $node->parentNode; |
540
|
|
|
$parent->removeChild($node); |
541
|
|
|
} |
542
|
|
|
} |
543
|
|
|
} |
544
|
|
|
|
545
|
|
|
/** |
546
|
|
|
* Given an XSL string, transform the DOMDoc (or a passed context node) |
547
|
|
|
* |
548
|
|
|
* @param string $xsl |
549
|
|
|
* XSL Transormation |
550
|
|
|
* |
551
|
|
|
* @param mixed $context |
552
|
|
|
* $context can either be an xpath string, or a DOMElement. Ommiting it |
553
|
|
|
* results in transforming the entire document |
554
|
|
|
* |
555
|
|
|
* @return a new DOMDoc |
556
|
|
|
*/ |
557
|
|
|
public function tranform($xsl, $context = NULL) { |
558
|
|
|
if (!$context) { |
559
|
|
|
$doc = $this; |
560
|
|
|
} |
561
|
|
|
else { |
562
|
|
|
if (is_string($context)) { |
563
|
|
|
$context = $this->xpathSingle($context); |
564
|
|
|
} |
565
|
|
|
$doc = new DOMDoc($context); |
566
|
|
|
} |
567
|
|
|
|
568
|
|
|
$xslDoc = new DOMDoc($xsl); |
569
|
|
|
$xslt = new \XSLTProcessor(); |
570
|
|
|
$xslt->importStylesheet($xslDoc); |
571
|
|
|
|
572
|
|
|
return new DOMDoc($xslt->transformToDoc($doc)); |
573
|
|
|
} |
574
|
|
|
|
575
|
|
|
/** |
576
|
|
|
* Given a node, change it's namespace to the specified namespace in situ |
577
|
|
|
* |
578
|
|
|
* @param mixed $node |
579
|
|
|
* Node to be changed. Can either be an xpath string or a DOMElement. |
580
|
|
|
* |
581
|
|
|
* @param mixed $prefix |
582
|
|
|
* prefix for the new namespace |
583
|
|
|
* |
584
|
|
|
* @param mixed $url |
585
|
|
|
* The URL for the new namespace |
586
|
|
|
* |
587
|
|
|
* @return mixed |
588
|
|
|
* The node with the new namespace. The node will also be changed in-situ in the document as well. |
589
|
|
|
*/ |
590
|
|
|
public function changeNamespace($node, $prefix, $url) { |
591
|
|
|
$this->createContext($node, 'xpath'); |
592
|
|
|
|
593
|
|
|
if (!$node) { |
594
|
|
|
return FALSE; |
595
|
|
|
} |
596
|
|
|
|
597
|
|
|
$this->registerNamespace($prefix, $url); |
598
|
|
|
|
599
|
|
|
if (get_class($node) == 'DOMElement') { |
600
|
|
|
$elemname = array_pop(explode(':', $node->tagName)); |
601
|
|
|
|
602
|
|
|
$xsl = ' |
603
|
|
|
<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"> |
604
|
|
|
<xsl:template match="*"> |
605
|
|
|
<xsl:element name="' . $prefix . ':{local-name()}" namespace="' . $url . '"> |
606
|
|
|
<xsl:copy-of select="@*"/> |
607
|
|
|
<xsl:apply-templates/> |
608
|
|
|
</xsl:element> |
609
|
|
|
</xsl:template> |
610
|
|
|
</xsl:stylesheet>'; |
611
|
|
|
|
612
|
|
|
$transformed = $this->tranform($xsl, $node); |
613
|
|
|
return $this->replace($node, $transformed->documentElement); |
614
|
|
|
} |
615
|
|
|
else { |
616
|
|
|
// @@TODO: Report the correct calling file and number |
617
|
|
|
throw new Exception("Changing the namespace of a " . get_class($node) . " is not supported"); |
618
|
|
|
} |
619
|
|
|
} |
620
|
|
|
|
621
|
|
|
/** |
622
|
|
|
* Get a lossless HTML representation of the XML |
623
|
|
|
* |
624
|
|
|
* Transforms the document (or passed context) into a set of HTML spans. |
625
|
|
|
* The element name becomes the class, all other attributes become HTML5 |
626
|
|
|
* "data-" attributes. |
627
|
|
|
* |
628
|
|
|
* @param mixed $context |
629
|
|
|
* $context can either be an xpath string, or a DOMElement. Ommiting it |
630
|
|
|
* results in transforming the entire document |
631
|
|
|
* |
632
|
|
|
* @param array $options |
633
|
|
|
* Options for transforming the HTML into XML. The following options are supported: |
634
|
|
|
* 'xlink' => {TRUE or xpath} |
635
|
|
|
* Transform xlink links into <a href> elements. If you specify 'xlink' => TRUE then |
636
|
|
|
* it will transform all elements with xlink:type = simple into a <a href> element. |
637
|
|
|
* Alternatively you may specify your own xpath for selecting which elements get transformed |
638
|
|
|
* into <a href> tags. |
639
|
|
|
* @return HTML string |
640
|
|
|
*/ |
641
|
|
|
public function asHTML($context = NULL, $options = array()) { |
642
|
|
|
$xslSimple = ' |
643
|
|
|
<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"> |
644
|
|
|
<xsl:template match="*"> |
645
|
|
|
<span class="{translate(name(.),\':\',\'-\')}"> |
646
|
|
|
<xsl:for-each select="./@*"> |
647
|
|
|
<xsl:attribute name="data-{translate(name(.),\':\',\'-\')}"> |
648
|
|
|
<xsl:value-of select="." /> |
649
|
|
|
</xsl:attribute> |
650
|
|
|
</xsl:for-each> |
651
|
|
|
<xsl:apply-templates/> |
652
|
|
|
</span> |
653
|
|
|
</xsl:template> |
654
|
|
|
</xsl:stylesheet>'; |
655
|
|
|
|
656
|
|
|
$xslOptions = ' |
657
|
|
|
<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:xlink="http://www.w3.org/1999/xlink" ||namespaces||> |
658
|
|
|
<xsl:template match="*"> |
659
|
|
|
<xsl:choose> |
660
|
|
|
<xsl:when test="||xlink||"> |
661
|
|
|
<a class="{translate(name(.),\':\',\'-\')}"> |
662
|
|
|
<xsl:for-each select="./@*"> |
663
|
|
|
<xsl:attribute name="data-{translate(name(.),\':\',\'-\')}"> |
664
|
|
|
<xsl:value-of select="."/> |
665
|
|
|
</xsl:attribute> |
666
|
|
|
</xsl:for-each> |
667
|
|
|
<xsl:attribute name="href"> |
668
|
|
|
<xsl:value-of select="@xlink:href"/> |
669
|
|
|
</xsl:attribute> |
670
|
|
|
<xsl:apply-templates/> |
671
|
|
|
</a> |
672
|
|
|
</xsl:when> |
673
|
|
|
<xsl:otherwise> |
674
|
|
|
<span class="{translate(name(.),\':\',\'-\')}"> |
675
|
|
|
<xsl:for-each select="./@*"> |
676
|
|
|
<xsl:attribute name="data-{translate(name(.),\':\',\'-\')}"> |
677
|
|
|
<xsl:value-of select="." /> |
678
|
|
|
</xsl:attribute> |
679
|
|
|
</xsl:for-each> |
680
|
|
|
<xsl:apply-templates/> |
681
|
|
|
</span> |
682
|
|
|
</xsl:otherwise> |
683
|
|
|
</xsl:choose> |
684
|
|
|
</xsl:template> |
685
|
|
|
</xsl:stylesheet>'; |
686
|
|
|
|
687
|
|
|
if (!empty($options)) { |
688
|
|
|
// Add in the namespaces |
689
|
|
|
foreach ($this->getNamespaces() as $prefix => $url) { |
690
|
|
|
$namespaces = ''; |
691
|
|
|
if ($prefix != 'xsl' && $prefix != 'xlink') { |
692
|
|
|
$namespaces .= 'xmlns:' . $prefix . '="' . $url. '" '; |
693
|
|
|
} |
694
|
|
|
$xslOptions = str_replace("||namespaces||", $namespaces, $xslOptions); |
695
|
|
|
} |
696
|
|
|
|
697
|
|
|
// Add in xlink options |
698
|
|
|
if ($options['xlink'] === TRUE) { |
699
|
|
|
$options['xlink'] = "@xlink:type = 'simple'"; |
700
|
|
|
} |
701
|
|
|
else if (empty($options['xlink'])) { |
702
|
|
|
$options['xlink'] = "false()"; |
703
|
|
|
} |
704
|
|
|
$xslOptions = str_replace("||xlink||", $options['xlink'], $xslOptions); |
705
|
|
|
$transformed = $this->tranform($xslOptions, $context); |
706
|
|
|
} |
707
|
|
|
else { |
708
|
|
|
$transformed = $this->tranform($xslSimple, $context); |
709
|
|
|
} |
710
|
|
|
|
711
|
|
|
return $transformed->out(); |
712
|
|
|
} |
713
|
|
|
|
714
|
|
|
/** |
715
|
|
|
* Output the DOMDoc as an XML string |
716
|
|
|
* |
717
|
|
|
* @param mixed $context |
718
|
|
|
* $context can either be an xpath string, or a DOMElement. Ommiting it |
719
|
|
|
* results in outputting the entire document |
720
|
|
|
* |
721
|
|
|
* @return XML string |
722
|
|
|
*/ |
723
|
|
|
public function out($context = NULL) { |
724
|
|
|
$this->createContext($context, 'xpath'); |
725
|
|
|
if (!$context) { |
726
|
|
|
return ''; |
727
|
|
|
} |
728
|
|
|
|
729
|
|
|
// Copy namespace prefixes |
730
|
|
|
foreach ($this->ns as $prefix => $namespace) { |
731
|
|
|
if (!$context->hasAttribute('xmlns:' . $prefix)) { |
732
|
|
|
$context->setAttribute('xmlns:' . $prefix, $namespace); |
733
|
|
|
} |
734
|
|
|
} |
735
|
|
|
|
736
|
|
|
// Check to seee if it's HTML, if it is we need to fix broken html void elements. |
737
|
|
|
if ($this->documentElement->lookupNamespaceURI(NULL) == 'http://www.w3.org/1999/xhtml' || $this->documentElement->tagName == 'html') { |
738
|
|
|
$output = $this->saveXML($context, LIBXML_NOEMPTYTAG); |
739
|
|
|
// The types listed are html "void" elements. |
740
|
|
|
// Find any of these elements that have no child nodes and are therefore candidates for self-closing, replace them with a self-closed version. |
741
|
|
|
$pattern = '<(area|base|br|col|command|embed|hr|img|input|keygen|link|meta|param|source|track|wbr)(\b[^<]*)><\/\1>'; |
742
|
|
|
return preg_replace('/' . $pattern . '/', '<$1$2/>', $output); |
743
|
|
|
} |
744
|
|
|
else { |
745
|
|
|
return $this->saveXML($context, LIBXML_NOEMPTYTAG); |
746
|
|
|
} |
747
|
|
|
} |
748
|
|
|
|
749
|
|
|
/** |
750
|
|
|
* Magic method for casting a DOMDoc as a string |
751
|
|
|
*/ |
752
|
|
|
public function __toString() { |
753
|
|
|
return $this->out(); |
754
|
|
|
} |
755
|
|
|
|
756
|
|
|
public public function setErrorChecking($error_checking) { |
|
|
|
|
757
|
|
|
// Check up error-checking |
758
|
|
|
if ($error_checking == FALSE) { |
759
|
|
|
$this->error_checking = 'none'; |
760
|
|
|
} |
761
|
|
|
else { |
762
|
|
|
$this->error_checking = $error_checking; |
763
|
|
|
} |
764
|
|
|
if ($this->error_checking != 'strict') { |
765
|
|
|
$this->strictErrorChecking = FALSE; |
766
|
|
|
} |
767
|
|
|
} |
768
|
|
|
|
769
|
|
|
public static function loadFile($file_or_url, $auto_register_namespaces = TRUE) { |
770
|
|
|
$dom = @parent::load($file_or_url, LIBXML_COMPACT); |
771
|
|
|
if (empty($dom)) { |
772
|
|
|
return FALSE; |
773
|
|
|
} |
774
|
|
|
|
775
|
|
|
return new DOMDoc($dom, $auto_register_namespaces); |
776
|
|
|
} |
777
|
|
|
|
778
|
|
|
private function AutoRegisterNamespace($auto_register_namespaces) { |
779
|
|
|
$this->auto_ns = TRUE; |
780
|
|
|
|
781
|
|
|
// If it's an "XML" document, then get namespaces via xpath |
782
|
|
|
$xpath = new \DOMXPath($this); |
783
|
|
|
foreach($xpath->query('namespace::*') as $namespace) { |
784
|
|
|
if (!empty($namespace->prefix)) { |
785
|
|
|
if ($namespace->prefix != 'xml' && $namespace->nodeValue != 'http://www.w3.org/XML/1998/namespace') { |
786
|
|
|
$this->registerNamespace($namespace->prefix, $namespace->nodeValue); |
787
|
|
|
} |
788
|
|
|
} |
789
|
|
|
else { |
790
|
|
|
$this->default_ns = $namespace->nodeValue; |
791
|
|
|
if (is_string($auto_register_namespaces)) { |
792
|
|
|
$this->registerNamespace($auto_register_namespaces, $namespace->nodeValue); |
793
|
|
|
} |
794
|
|
|
// Otherwise, automatically set-up the root element tag name as the prefix for the default namespace |
795
|
|
|
else { |
796
|
|
|
$tagname = $this->documentElement->tagName; |
797
|
|
|
if (empty($this->ns[$tagname])) { |
798
|
|
|
$this->registerNamespace($tagname, $this->documentElement->getAttribute('xmlns')); |
799
|
|
|
} |
800
|
|
|
} |
801
|
|
|
} |
802
|
|
|
} |
803
|
|
|
|
804
|
|
|
// If it's an "HTML" document, we get namespaces via attributes |
805
|
|
|
if (empty($this->ns)) { |
806
|
|
|
foreach ($this->documentElement->attributes as $attr) { |
807
|
|
|
if ($attr->name == 'xmlns') { |
808
|
|
|
$this->default_ns = $attr->value; |
809
|
|
|
// If auto_register_namespaces is a prefix string, then we register the default namespace to that string |
810
|
|
|
if (is_string($auto_register_namespaces)) { |
811
|
|
|
$this->registerNamespace($auto_register_namespaces, $attr->value); |
812
|
|
|
} |
813
|
|
|
// Otherwise, automatically set-up the root element tag name as the prefix for the default namespace |
814
|
|
|
else { |
815
|
|
|
$tagname = $this->documentElement->tagName; |
816
|
|
|
if (empty($this->ns[$tagname])) { |
817
|
|
|
$this->registerNamespace($tagname, $attr->value); |
818
|
|
|
} |
819
|
|
|
} |
820
|
|
|
} |
821
|
|
|
else if (substr($attr->name,0,6) == 'xmlns:') { |
822
|
|
|
$prefix = substr($attr->name,6); |
823
|
|
|
$this->registerNamespace($prefix, $attr->value); |
824
|
|
|
} |
825
|
|
|
} |
826
|
|
|
} |
827
|
|
|
} |
828
|
|
|
|
829
|
|
|
private function createContext(&$context, $type = 'xpath', $createDocument = TRUE) { |
830
|
|
|
if (!$context && $createDocument) { |
831
|
|
|
$context = $this->documentElement; |
832
|
|
|
return; |
833
|
|
|
} |
834
|
|
|
|
835
|
|
|
if (!$context) { |
836
|
|
|
return FALSE; |
837
|
|
|
} |
838
|
|
|
|
839
|
|
|
if ($context && is_string($context)) { |
840
|
|
|
if ($type == 'xpath') { |
841
|
|
|
$context = $this->xpathSingle($context); |
842
|
|
|
return; |
843
|
|
|
} |
844
|
|
|
if ($type == 'xml') { |
845
|
|
|
$context = $this->createElementFromXML($context); |
846
|
|
|
return; |
847
|
|
|
} |
848
|
|
|
} |
849
|
|
|
|
850
|
|
|
if (is_object($context)) { |
851
|
|
|
if (is_a($context, 'DOMElement')) { |
852
|
|
|
return $context; |
853
|
|
|
} |
854
|
|
|
if (is_a($context, 'DOMDocument')) { |
855
|
|
|
return $context->documentElement; |
856
|
|
|
} |
857
|
|
|
} |
858
|
|
|
} |
859
|
|
|
} |
860
|
|
|
|
861
|
|
|
|
862
|
|
|
|
863
|
|
|
|