|
1
|
|
|
<?php |
|
2
|
|
|
namespace PhpQuery\Dom; |
|
3
|
|
|
|
|
4
|
|
|
use PhpQuery\PhpQuery; |
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* DOMDocumentWrapper class simplifies work with DOMDocument. |
|
8
|
|
|
* |
|
9
|
|
|
* Know bug: |
|
10
|
|
|
* - in XHTML fragments, <br /> changes to <br clear="none" /> |
|
11
|
|
|
* |
|
12
|
|
|
* @todo check XML catalogs compatibility |
|
13
|
|
|
* @author Tobiasz Cudnik <tobiasz.cudnik/gmail.com> |
|
14
|
|
|
* @package PhpQuery |
|
15
|
|
|
*/ |
|
16
|
|
|
class DOMDocumentWrapper |
|
17
|
|
|
{ |
|
18
|
|
|
/** |
|
19
|
|
|
* @var \DOMDocument |
|
20
|
|
|
*/ |
|
21
|
|
|
public $document; |
|
22
|
|
|
public $id; |
|
23
|
|
|
/** |
|
24
|
|
|
* @todo Rewrite as method and quess if null. |
|
25
|
|
|
* @var unknown_type |
|
26
|
|
|
*/ |
|
27
|
|
|
public $contentType = ''; |
|
28
|
|
|
public $xpath; |
|
29
|
|
|
public $uuid = 0; |
|
30
|
|
|
public $data = array(); |
|
31
|
|
|
public $dataNodes = array(); |
|
32
|
|
|
public $events = array(); |
|
33
|
|
|
public $eventsNodes = array(); |
|
34
|
|
|
public $eventsGlobal = array(); |
|
35
|
|
|
/** |
|
36
|
|
|
* @TODO iframes support http://code.google.com/p/phpquery/issues/detail?id=28 |
|
37
|
|
|
* @var unknown_type |
|
38
|
|
|
*/ |
|
39
|
|
|
public $frames = array(); |
|
40
|
|
|
/** |
|
41
|
|
|
* Document root, by default equals to document itself. |
|
42
|
|
|
* Used by documentFragments. |
|
43
|
|
|
* |
|
44
|
|
|
* @var \DOMNode |
|
45
|
|
|
*/ |
|
46
|
|
|
public $root; |
|
47
|
|
|
public $isDocumentFragment; |
|
48
|
|
|
public $isXML = false; |
|
49
|
|
|
public $isXHTML = false; |
|
50
|
|
|
public $isHTML = false; |
|
51
|
|
|
public $charset; |
|
52
|
|
|
|
|
53
|
|
|
public function __construct($markup = null, $contentType = null, $newDocumentID = null) |
|
54
|
|
|
{ |
|
55
|
|
|
if (isset($markup)) { |
|
56
|
|
|
$this->load($markup, $contentType, $newDocumentID); |
|
57
|
|
|
} |
|
58
|
|
|
$this->id = $newDocumentID ? $newDocumentID : md5(microtime()); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
public function load($markup, $contentType = null, $newDocumentID = null) |
|
|
|
|
|
|
62
|
|
|
{ |
|
63
|
|
|
// PhpQuery::$documents[$id] = $this; |
|
|
|
|
|
|
64
|
|
|
$this->contentType = strtolower($contentType); |
|
|
|
|
|
|
65
|
|
|
if ($markup instanceof \DOMDocument) { |
|
66
|
|
|
$this->document = $markup; |
|
67
|
|
|
$this->root = $this->document; |
|
68
|
|
|
$this->charset = $this->document->encoding; |
|
69
|
|
|
// TODO isDocumentFragment |
|
70
|
|
|
$loaded = true; |
|
71
|
|
|
} else { |
|
72
|
|
|
$loaded = $this->loadMarkup($markup); |
|
73
|
|
|
} |
|
74
|
|
|
if ($loaded) { |
|
75
|
|
|
// $this->document->formatOutput = true; |
|
|
|
|
|
|
76
|
|
|
$this->document->preserveWhiteSpace = true; |
|
77
|
|
|
$this->xpath = new \DOMXPath($this->document); |
|
78
|
|
|
$this->afterMarkupLoad(); |
|
79
|
|
|
return true; |
|
80
|
|
|
// remember last loaded document |
|
81
|
|
|
// return PhpQuery::selectDocument($id); |
|
|
|
|
|
|
82
|
|
|
} |
|
83
|
|
|
return false; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
protected function afterMarkupLoad() |
|
87
|
|
|
{ |
|
88
|
|
|
if ($this->isXHTML) { |
|
89
|
|
|
$this->xpath->registerNamespace("html", "http://www.w3.org/1999/xhtml"); |
|
90
|
|
|
} |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
protected function loadMarkup($markup) |
|
94
|
|
|
{ |
|
95
|
|
|
$loaded = false; |
|
96
|
|
|
if ($this->contentType) { |
|
97
|
|
|
self::debug("Load markup for content type {$this->contentType}"); |
|
98
|
|
|
// content determined by contentType |
|
99
|
|
|
list($contentType, $charset) = $this->contentTypeToArray($this->contentType); |
|
100
|
|
|
switch ($contentType) { |
|
101
|
|
|
case 'text/html': |
|
102
|
|
|
PhpQuery::debug("Loading HTML, content type '{$this->contentType}'"); |
|
103
|
|
|
$loaded = $this->loadMarkupHTML($markup, $charset); |
|
104
|
|
|
break; |
|
105
|
|
|
case 'text/xml': |
|
106
|
|
|
case 'application/xhtml+xml': |
|
107
|
|
|
PhpQuery::debug("Loading XML, content type '{$this->contentType}'"); |
|
108
|
|
|
$loaded = $this->loadMarkupXML($markup, $charset); |
|
109
|
|
|
break; |
|
110
|
|
|
default: |
|
111
|
|
|
// for feeds or anything that sometimes doesn't use text/xml |
|
112
|
|
|
if (strpos('xml', $this->contentType) !== false) { |
|
113
|
|
|
PhpQuery::debug("Loading XML, content type '{$this->contentType}'"); |
|
114
|
|
|
$loaded = $this->loadMarkupXML($markup, $charset); |
|
115
|
|
|
} else { |
|
116
|
|
|
PhpQuery::debug("Could not determine document type from content type '{$this->contentType}'"); |
|
117
|
|
|
} |
|
118
|
|
|
} |
|
119
|
|
|
} else { |
|
120
|
|
|
// content type autodetection |
|
121
|
|
|
if ($this->isXML($markup)) { |
|
122
|
|
|
PhpQuery::debug("Loading XML, isXML() == true"); |
|
123
|
|
|
$loaded = $this->loadMarkupXML($markup); |
|
124
|
|
|
if (!$loaded && $this->isXHTML) { |
|
125
|
|
|
PhpQuery::debug('Loading as XML failed, trying to load as HTML, isXHTML == true'); |
|
126
|
|
|
$loaded = $this->loadMarkupHTML($markup); |
|
127
|
|
|
} |
|
128
|
|
|
} else { |
|
129
|
|
|
PhpQuery::debug("Loading HTML, isXML() == false"); |
|
130
|
|
|
$loaded = $this->loadMarkupHTML($markup); |
|
131
|
|
|
} |
|
132
|
|
|
} |
|
133
|
|
|
return $loaded; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
protected function loadMarkupReset() |
|
137
|
|
|
{ |
|
138
|
|
|
$this->isXML = $this->isXHTML = $this->isHTML = false; |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
protected function documentCreate($charset, $version = '1.0') |
|
142
|
|
|
{ |
|
143
|
|
|
if (!$version) { |
|
144
|
|
|
$version = '1.0'; |
|
145
|
|
|
} |
|
146
|
|
|
$this->document = new \DOMDocument($version, $charset); |
|
147
|
|
|
$this->charset = $this->document->encoding; |
|
148
|
|
|
// $this->document->encoding = $charset; |
|
|
|
|
|
|
149
|
|
|
$this->document->formatOutput = true; |
|
150
|
|
|
$this->document->preserveWhiteSpace = true; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
protected function loadMarkupHTML($markup, $requestedCharset = null) |
|
154
|
|
|
{ |
|
155
|
|
|
if (PhpQuery::$debug) { |
|
156
|
|
|
PhpQuery::debug('Full markup load (HTML): ' . substr($markup, 0, 250)); |
|
157
|
|
|
} |
|
158
|
|
|
$this->loadMarkupReset(); |
|
159
|
|
|
$this->isHTML = true; |
|
160
|
|
|
if (!isset($this->isDocumentFragment)) { |
|
161
|
|
|
$this->isDocumentFragment = self::isDocumentFragmentHTML($markup); |
|
162
|
|
|
} |
|
163
|
|
|
$charset = null; |
|
164
|
|
|
$documentCharset = $this->charsetFromHTML($markup); |
|
165
|
|
|
$addDocumentCharset = false; |
|
166
|
|
|
if ($documentCharset) { |
|
167
|
|
|
$charset = $documentCharset; |
|
168
|
|
|
$markup = $this->charsetFixHTML($markup); |
|
169
|
|
|
} else { |
|
170
|
|
|
if ($requestedCharset) { |
|
171
|
|
|
$charset = $requestedCharset; |
|
172
|
|
|
} |
|
173
|
|
|
} |
|
174
|
|
|
if (!$charset) { |
|
175
|
|
|
$charset = PhpQuery::$defaultCharset; |
|
176
|
|
|
} |
|
177
|
|
|
// HTTP 1.1 says that the default charset is ISO-8859-1 |
|
178
|
|
|
// @see http://www.w3.org/International/O-HTTP-charset |
|
179
|
|
|
if (!$documentCharset) { |
|
180
|
|
|
$documentCharset = 'ISO-8859-1'; |
|
181
|
|
|
$addDocumentCharset = true; |
|
182
|
|
|
} |
|
183
|
|
|
// Should be careful here, still need 'magic encoding detection' since lots of pages have other 'default encoding' |
|
184
|
|
|
// Worse, some pages can have mixed encodings... we'll try not to worry about that |
|
185
|
|
|
$requestedCharset = strtoupper($requestedCharset); |
|
186
|
|
|
$documentCharset = strtoupper($documentCharset); |
|
187
|
|
|
PhpQuery::debug("DOC: $documentCharset REQ: $requestedCharset"); |
|
188
|
|
|
if ($requestedCharset && $documentCharset |
|
189
|
|
|
&& $requestedCharset !== $documentCharset |
|
190
|
|
|
) { |
|
191
|
|
|
PhpQuery::debug("CHARSET CONVERT"); |
|
192
|
|
|
// Document Encoding Conversion |
|
193
|
|
|
// http://code.google.com/p/phpquery/issues/detail?id=86 |
|
194
|
|
|
if (function_exists('mb_detect_encoding')) { |
|
195
|
|
|
$possibleCharsets = array( |
|
196
|
|
|
$documentCharset, |
|
197
|
|
|
$requestedCharset, |
|
198
|
|
|
'AUTO' |
|
199
|
|
|
); |
|
200
|
|
|
$docEncoding = mb_detect_encoding($markup, implode(', ', $possibleCharsets)); |
|
201
|
|
|
if (!$docEncoding) { |
|
202
|
|
|
$docEncoding = $documentCharset; |
|
203
|
|
|
} |
|
204
|
|
|
// ok trust the document |
|
205
|
|
|
PhpQuery::debug("DETECTED '$docEncoding'"); |
|
206
|
|
|
// Detected does not match what document says... |
|
207
|
|
|
if ($docEncoding !== $documentCharset) { |
|
|
|
|
|
|
208
|
|
|
// Tricky.. |
|
209
|
|
|
} |
|
210
|
|
|
if ($docEncoding !== $requestedCharset) { |
|
211
|
|
|
PhpQuery::debug("CONVERT $docEncoding => $requestedCharset"); |
|
212
|
|
|
$markup = mb_convert_encoding($markup, $requestedCharset, $docEncoding); |
|
213
|
|
|
$markup = $this->charsetAppendToHTML($markup, $requestedCharset); |
|
214
|
|
|
$charset = $requestedCharset; |
|
215
|
|
|
} |
|
216
|
|
|
} else { |
|
217
|
|
|
PhpQuery::debug("TODO: charset conversion without mbstring..."); |
|
218
|
|
|
} |
|
219
|
|
|
} |
|
220
|
|
|
$return = false; |
|
|
|
|
|
|
221
|
|
|
if ($this->isDocumentFragment) { |
|
222
|
|
|
PhpQuery::debug("Full markup load (HTML), DocumentFragment detected, using charset '$charset'"); |
|
223
|
|
|
$return = $this->documentFragmentLoadMarkup($this, $charset, $markup); |
|
224
|
|
|
} else { |
|
225
|
|
|
if ($addDocumentCharset) { |
|
226
|
|
|
PhpQuery::debug("Full markup load (HTML), appending charset: '$charset'"); |
|
227
|
|
|
$markup = $this->charsetAppendToHTML($markup, $charset); |
|
228
|
|
|
} |
|
229
|
|
|
PhpQuery::debug("Full markup load (HTML), documentCreate('$charset')"); |
|
230
|
|
|
$this->documentCreate($charset); |
|
231
|
|
|
$return = PhpQuery::$debug === 2 ? $this->document->loadHTML($markup) |
|
232
|
|
|
: @$this->document->loadHTML($markup); |
|
233
|
|
|
if ($return) { |
|
234
|
|
|
$this->root = $this->document; |
|
235
|
|
|
} |
|
236
|
|
|
} |
|
237
|
|
|
if ($return && !$this->contentType) { |
|
238
|
|
|
$this->contentType = 'text/html'; |
|
|
|
|
|
|
239
|
|
|
} |
|
240
|
|
|
return $return; |
|
241
|
|
|
} |
|
242
|
|
|
|
|
243
|
|
|
protected function loadMarkupXML($markup, $requestedCharset = null) |
|
244
|
|
|
{ |
|
245
|
|
|
if (PhpQuery::$debug) { |
|
246
|
|
|
PhpQuery::debug('Full markup load (XML): ' . substr($markup, 0, 250)); |
|
247
|
|
|
} |
|
248
|
|
|
$this->loadMarkupReset(); |
|
249
|
|
|
$this->isXML = true; |
|
250
|
|
|
// check agains XHTML in contentType or markup |
|
251
|
|
|
$isContentTypeXHTML = $this->isXHTML(); |
|
252
|
|
|
$isMarkupXHTML = $this->isXHTML($markup); |
|
253
|
|
|
if ($isContentTypeXHTML || $isMarkupXHTML) { |
|
254
|
|
|
self::debug('Full markup load (XML), XHTML detected'); |
|
255
|
|
|
$this->isXHTML = true; |
|
256
|
|
|
} |
|
257
|
|
|
// determine document fragment |
|
258
|
|
|
if (!isset($this->isDocumentFragment)) { |
|
259
|
|
|
$this->isDocumentFragment = $this->isXHTML ? self::isDocumentFragmentXHTML($markup) |
|
260
|
|
|
: self::isDocumentFragmentXML($markup); |
|
261
|
|
|
} |
|
262
|
|
|
// this charset will be used |
|
263
|
|
|
$charset = null; |
|
264
|
|
|
// charset from XML declaration @var string |
|
265
|
|
|
$documentCharset = $this->charsetFromXML($markup); |
|
266
|
|
|
if (!$documentCharset) { |
|
|
|
|
|
|
267
|
|
|
if ($this->isXHTML) { |
|
268
|
|
|
// this is XHTML, try to get charset from content-type meta header |
|
269
|
|
|
$documentCharset = $this->charsetFromHTML($markup); |
|
270
|
|
|
if ($documentCharset) { |
|
271
|
|
|
PhpQuery::debug("Full markup load (XML), appending XHTML charset '$documentCharset'"); |
|
272
|
|
|
$this->charsetAppendToXML($markup, $documentCharset); |
|
|
|
|
|
|
273
|
|
|
$charset = $documentCharset; |
|
274
|
|
|
} |
|
275
|
|
|
} |
|
276
|
|
|
if (!$documentCharset) { |
|
277
|
|
|
// if still no document charset... |
|
278
|
|
|
$charset = $requestedCharset; |
|
279
|
|
|
} |
|
280
|
|
|
} else { |
|
281
|
|
|
if ($requestedCharset) { |
|
282
|
|
|
$charset = $requestedCharset; |
|
283
|
|
|
} |
|
284
|
|
|
} |
|
285
|
|
|
if (!$charset) { |
|
286
|
|
|
$charset = PhpQuery::$defaultCharset; |
|
287
|
|
|
} |
|
288
|
|
|
if ($requestedCharset && $documentCharset |
|
|
|
|
|
|
289
|
|
|
&& $requestedCharset != $documentCharset |
|
290
|
|
|
) { |
|
291
|
|
|
// TODO place for charset conversion |
|
292
|
|
|
// $charset = $requestedCharset; |
|
|
|
|
|
|
293
|
|
|
} |
|
294
|
|
|
$return = false; |
|
|
|
|
|
|
295
|
|
|
if ($this->isDocumentFragment) { |
|
296
|
|
|
PhpQuery::debug("Full markup load (XML), DocumentFragment detected, using charset '$charset'"); |
|
297
|
|
|
$return = $this->documentFragmentLoadMarkup($this, $charset, $markup); |
|
298
|
|
|
} else { |
|
299
|
|
|
// FIXME ??? |
|
|
|
|
|
|
300
|
|
|
if ($isContentTypeXHTML && !$isMarkupXHTML) { |
|
301
|
|
|
if (!$documentCharset) { |
|
302
|
|
|
PhpQuery::debug("Full markup load (XML), appending charset '$charset'"); |
|
303
|
|
|
$markup = $this->charsetAppendToXML($markup, $charset); |
|
304
|
|
|
} |
|
305
|
|
|
} |
|
306
|
|
|
// see http://pl2.php.net/manual/en/book.dom.php#78929 |
|
307
|
|
|
// LIBXML_DTDLOAD (>= PHP 5.1) |
|
308
|
|
|
// does XML ctalogues works with LIBXML_NONET |
|
309
|
|
|
// $this->document->resolveExternals = true; |
|
|
|
|
|
|
310
|
|
|
// TODO test LIBXML_COMPACT for performance improvement |
|
311
|
|
|
// create document |
|
312
|
|
|
$this->documentCreate($charset); |
|
313
|
|
|
if (phpversion() < 5.1) { |
|
314
|
|
|
$this->document->resolveExternals = true; |
|
315
|
|
|
$return = PhpQuery::$debug === 2 ? $this->document->loadXML($markup) |
|
|
|
|
|
|
316
|
|
|
: @$this->document->loadXML($markup); |
|
|
|
|
|
|
317
|
|
|
} else { |
|
318
|
|
|
/** @link http://pl2.php.net/manual/en/libxml.constants.php */ |
|
319
|
|
|
$libxmlStatic = PhpQuery::$debug === 2 ? LIBXML_DTDLOAD |
|
320
|
|
|
| LIBXML_DTDATTR | LIBXML_NONET |
|
321
|
|
|
: LIBXML_DTDLOAD | LIBXML_DTDATTR | LIBXML_NONET | LIBXML_NOWARNING |
|
322
|
|
|
| LIBXML_NOERROR; |
|
323
|
|
|
$return = $this->document->loadXML($markup, $libxmlStatic); |
|
|
|
|
|
|
324
|
|
|
// if (! $return) |
|
|
|
|
|
|
325
|
|
|
// $return = $this->document->loadHTML($markup); |
|
|
|
|
|
|
326
|
|
|
} |
|
327
|
|
|
if ($return) { |
|
328
|
|
|
$this->root = $this->document; |
|
329
|
|
|
} |
|
330
|
|
|
} |
|
331
|
|
|
if ($return) { |
|
332
|
|
|
if (!$this->contentType) { |
|
333
|
|
|
if ($this->isXHTML) { |
|
334
|
|
|
$this->contentType = 'application/xhtml+xml'; |
|
|
|
|
|
|
335
|
|
|
} else { |
|
336
|
|
|
$this->contentType = 'text/xml'; |
|
|
|
|
|
|
337
|
|
|
} |
|
338
|
|
|
} |
|
339
|
|
|
return $return; |
|
340
|
|
|
} else { |
|
341
|
|
|
throw new \Exception("Error loading XML markup"); |
|
342
|
|
|
} |
|
343
|
|
|
} |
|
344
|
|
|
|
|
345
|
|
|
protected function isXHTML($markup = null) |
|
346
|
|
|
{ |
|
347
|
|
|
if (!isset($markup)) { |
|
348
|
|
|
return strpos($this->contentType, 'xhtml') !== false; |
|
349
|
|
|
} |
|
350
|
|
|
// XXX ok ? |
|
351
|
|
|
return strpos($markup, "<!DOCTYPE html") !== false; |
|
352
|
|
|
// return stripos($doctype, 'xhtml') !== false; |
|
|
|
|
|
|
353
|
|
|
// $doctype = isset($dom->doctype) && is_object($dom->doctype) |
|
|
|
|
|
|
354
|
|
|
// ? $dom->doctype->publicId |
|
|
|
|
|
|
355
|
|
|
// : self::$defaultDoctype; |
|
|
|
|
|
|
356
|
|
|
} |
|
357
|
|
|
|
|
358
|
|
|
protected function isXML($markup) |
|
359
|
|
|
{ |
|
360
|
|
|
// return strpos($markup, '<?xml') !== false && stripos($markup, 'xhtml') === false; |
|
|
|
|
|
|
361
|
|
|
return strpos(substr($markup, 0, 100), '<' . '?xml') !== false; |
|
362
|
|
|
} |
|
363
|
|
|
|
|
364
|
|
|
protected function contentTypeToArray($contentType) |
|
365
|
|
|
{ |
|
366
|
|
|
$test = null; |
|
|
|
|
|
|
367
|
|
|
$test = $matches = explode(';', trim(strtolower($contentType))); |
|
|
|
|
|
|
368
|
|
|
if (isset($matches[1])) { |
|
369
|
|
|
$matches[1] = explode('=', $matches[1]); |
|
370
|
|
|
// strip 'charset=' |
|
371
|
|
|
$matches[1] = isset($matches[1][1]) && trim($matches[1][1]) ? $matches[1][1] |
|
372
|
|
|
: $matches[1][0]; |
|
373
|
|
|
} else { |
|
374
|
|
|
$matches[1] = null; |
|
375
|
|
|
} |
|
376
|
|
|
return $matches; |
|
377
|
|
|
} |
|
378
|
|
|
|
|
379
|
|
|
/** |
|
380
|
|
|
* |
|
381
|
|
|
* @param $markup |
|
382
|
|
|
* @return array contentType, charset |
|
383
|
|
|
*/ |
|
384
|
|
|
protected function contentTypeFromHTML($markup) |
|
385
|
|
|
{ |
|
386
|
|
|
$matches = array(); |
|
387
|
|
|
// find meta tag |
|
388
|
|
|
preg_match('@<meta[^>]+http-equiv\\s*=\\s*(["|\'])Content-Type\\1([^>]+?)>@i', $markup, $matches); |
|
389
|
|
|
if (!isset($matches[0])) { |
|
390
|
|
|
return array( |
|
391
|
|
|
null, |
|
392
|
|
|
null |
|
393
|
|
|
); |
|
394
|
|
|
} |
|
395
|
|
|
// get attr 'content' |
|
396
|
|
|
preg_match('@content\\s*=\\s*(["|\'])(.+?)\\1@', $matches[0], $matches); |
|
397
|
|
|
if (!isset($matches[0])) { |
|
398
|
|
|
return array( |
|
399
|
|
|
null, |
|
400
|
|
|
null |
|
401
|
|
|
); |
|
402
|
|
|
} |
|
403
|
|
|
return $this->contentTypeToArray($matches[2]); |
|
404
|
|
|
} |
|
405
|
|
|
|
|
406
|
|
|
protected function charsetFromHTML($markup) |
|
407
|
|
|
{ |
|
408
|
|
|
$contentType = $this->contentTypeFromHTML($markup); |
|
409
|
|
|
return $contentType[1]; |
|
410
|
|
|
} |
|
411
|
|
|
|
|
412
|
|
|
protected function charsetFromXML($markup) |
|
413
|
|
|
{ |
|
414
|
|
|
$matches; |
|
|
|
|
|
|
415
|
|
|
// find declaration |
|
416
|
|
|
preg_match('@<' . '?xml[^>]+encoding\\s*=\\s*(["|\'])(.*?)\\1@i', $markup, $matches); |
|
417
|
|
|
return isset($matches[2]) ? strtolower($matches[2]) : null; |
|
418
|
|
|
} |
|
419
|
|
|
|
|
420
|
|
|
/** |
|
421
|
|
|
* Repositions meta[type=charset] at the start of head. Bypasses \DOMDocument bug. |
|
422
|
|
|
* |
|
423
|
|
|
* @link http://code.google.com/p/phpquery/issues/detail?id=80 |
|
424
|
|
|
* @param $html |
|
425
|
|
|
*/ |
|
426
|
|
|
protected function charsetFixHTML($markup) |
|
427
|
|
|
{ |
|
428
|
|
|
$matches = array(); |
|
429
|
|
|
// find meta tag |
|
430
|
|
|
preg_match( |
|
431
|
|
|
'@\s*<meta[^>]+http-equiv\\s*=\\s*(["|\'])Content-Type\\1([^>]+?)>@i', |
|
432
|
|
|
$markup, |
|
433
|
|
|
$matches, |
|
434
|
|
|
PREG_OFFSET_CAPTURE |
|
435
|
|
|
); |
|
436
|
|
|
if (!isset($matches[0])) { |
|
437
|
|
|
return; |
|
438
|
|
|
} |
|
439
|
|
|
$metaContentType = $matches[0][0]; |
|
440
|
|
|
$markup = substr($markup, 0, $matches[0][1]) |
|
441
|
|
|
. substr($markup, $matches[0][1] + strlen($metaContentType)); |
|
442
|
|
|
$headStart = stripos($markup, '<head>'); |
|
443
|
|
|
$markup = substr($markup, 0, $headStart + 6) . $metaContentType |
|
444
|
|
|
. substr($markup, $headStart + 6); |
|
445
|
|
|
return $markup; |
|
446
|
|
|
} |
|
447
|
|
|
|
|
448
|
|
|
protected function charsetAppendToHTML($html, $charset, $xhtml = false) |
|
449
|
|
|
{ |
|
450
|
|
|
// remove existing meta[type=content-type] |
|
451
|
|
|
$html = preg_replace('@\s*<meta[^>]+http-equiv\\s*=\\s*(["|\'])Content-Type\\1([^>]+?)>@i', '', $html); |
|
452
|
|
|
$meta = '<meta http-equiv="Content-Type" content="text/html;charset=' |
|
453
|
|
|
. $charset . '" ' . ($xhtml ? '/' : '') . '>'; |
|
454
|
|
|
if (strpos($html, '<head') === false) { |
|
455
|
|
|
if (strpos($html, '<html') === false) { |
|
456
|
|
|
return $meta . $html; |
|
457
|
|
|
} else { |
|
458
|
|
|
return preg_replace('@<html(.*?)(?(?<!\?)>)@s', "<html\\1><head>{$meta}</head>", $html); |
|
459
|
|
|
} |
|
460
|
|
|
} else { |
|
461
|
|
|
return preg_replace('@<head(.*?)(?(?<!\?)>)@s', '<head\\1>' . $meta, $html); |
|
462
|
|
|
} |
|
463
|
|
|
} |
|
464
|
|
|
|
|
465
|
|
|
protected function charsetAppendToXML($markup, $charset) |
|
466
|
|
|
{ |
|
467
|
|
|
$declaration = '<' . '?xml version="1.0" encoding="' . $charset . '"?' |
|
468
|
|
|
. '>'; |
|
469
|
|
|
return $declaration . $markup; |
|
470
|
|
|
} |
|
471
|
|
|
|
|
472
|
|
|
public static function isDocumentFragmentHTML($markup) |
|
473
|
|
|
{ |
|
474
|
|
|
return stripos($markup, '<html') === false |
|
475
|
|
|
&& stripos($markup, '<!doctype') === false; |
|
476
|
|
|
} |
|
477
|
|
|
|
|
478
|
|
|
public static function isDocumentFragmentXML($markup) |
|
479
|
|
|
{ |
|
480
|
|
|
return stripos($markup, '<' . '?xml') === false; |
|
481
|
|
|
} |
|
482
|
|
|
|
|
483
|
|
|
public static function isDocumentFragmentXHTML($markup) |
|
484
|
|
|
{ |
|
485
|
|
|
return self::isDocumentFragmentHTML($markup); |
|
486
|
|
|
} |
|
487
|
|
|
|
|
488
|
|
|
public function importAttr($value) |
|
|
|
|
|
|
489
|
|
|
{ |
|
490
|
|
|
// TODO |
|
491
|
|
|
} |
|
492
|
|
|
|
|
493
|
|
|
/** |
|
494
|
|
|
* |
|
495
|
|
|
* @param $source |
|
496
|
|
|
* @param $target |
|
497
|
|
|
* @param $sourceCharset |
|
498
|
|
|
* @return array Array of imported nodes. |
|
499
|
|
|
*/ |
|
500
|
|
|
public function import($source, $sourceCharset = null) |
|
501
|
|
|
{ |
|
502
|
|
|
// TODO charset conversions |
|
503
|
|
|
$return = array(); |
|
504
|
|
|
if ($source instanceof \DOMNode && !($source instanceof \DOMNodeList)) { |
|
505
|
|
|
$source = array( |
|
506
|
|
|
$source |
|
507
|
|
|
); |
|
508
|
|
|
} |
|
509
|
|
|
// if (is_array($source)) { |
|
|
|
|
|
|
510
|
|
|
// foreach($source as $node) { |
|
|
|
|
|
|
511
|
|
|
// if (is_string($node)) { |
|
|
|
|
|
|
512
|
|
|
// // string markup |
|
513
|
|
|
// $fake = $this->documentFragmentCreate($node, $sourceCharset); |
|
|
|
|
|
|
514
|
|
|
// if ($fake === false) |
|
|
|
|
|
|
515
|
|
|
// throw new \Exception("Error loading documentFragment markup"); |
|
|
|
|
|
|
516
|
|
|
// else |
|
517
|
|
|
// $return = array_merge($return, |
|
|
|
|
|
|
518
|
|
|
// $this->import($fake->root->childNodes) |
|
|
|
|
|
|
519
|
|
|
// ); |
|
520
|
|
|
// } else { |
|
|
|
|
|
|
521
|
|
|
// $return[] = $this->document->importNode($node, true); |
|
|
|
|
|
|
522
|
|
|
// } |
|
523
|
|
|
// } |
|
524
|
|
|
// return $return; |
|
525
|
|
|
// } else { |
|
|
|
|
|
|
526
|
|
|
// // string markup |
|
527
|
|
|
// $fake = $this->documentFragmentCreate($source, $sourceCharset); |
|
|
|
|
|
|
528
|
|
|
// if ($fake === false) |
|
|
|
|
|
|
529
|
|
|
// throw new \Exception("Error loading documentFragment markup"); |
|
|
|
|
|
|
530
|
|
|
// else |
|
531
|
|
|
// return $this->import($fake->root->childNodes); |
|
|
|
|
|
|
532
|
|
|
// } |
|
533
|
|
|
if (is_array($source) || $source instanceof \DOMNodeList) { |
|
534
|
|
|
// dom nodes |
|
535
|
|
|
self::debug('Importing nodes to document'); |
|
536
|
|
|
foreach ($source as $node) { |
|
537
|
|
|
$return[] = $this->document->importNode($node, true); |
|
538
|
|
|
} |
|
539
|
|
|
} else { |
|
540
|
|
|
// string markup |
|
541
|
|
|
$fake = $this->documentFragmentCreate($source, $sourceCharset); |
|
542
|
|
|
if ($fake === false) { |
|
543
|
|
|
throw new \Exception("Error loading documentFragment markup"); |
|
544
|
|
|
} else { |
|
545
|
|
|
return $this->import($fake->root->childNodes); |
|
546
|
|
|
} |
|
547
|
|
|
} |
|
548
|
|
|
return $return; |
|
549
|
|
|
} |
|
550
|
|
|
|
|
551
|
|
|
/** |
|
552
|
|
|
* Creates new document fragment. |
|
553
|
|
|
* |
|
554
|
|
|
* @param $source |
|
555
|
|
|
* @return DOMDocumentWrapper |
|
556
|
|
|
*/ |
|
557
|
|
|
protected function documentFragmentCreate($source, $charset = null) |
|
558
|
|
|
{ |
|
559
|
|
|
$fake = new DOMDocumentWrapper(); |
|
560
|
|
|
$fake->contentType = $this->contentType; |
|
561
|
|
|
$fake->isXML = $this->isXML; |
|
562
|
|
|
$fake->isHTML = $this->isHTML; |
|
563
|
|
|
$fake->isXHTML = $this->isXHTML; |
|
564
|
|
|
$fake->root = $fake->document; |
|
565
|
|
|
if (!$charset) { |
|
566
|
|
|
$charset = $this->charset; |
|
567
|
|
|
} |
|
568
|
|
|
// $fake->documentCreate($this->charset); |
|
|
|
|
|
|
569
|
|
|
if ($source instanceof \DOMNode && !($source instanceof \DOMNodeList)) { |
|
570
|
|
|
$source = array( |
|
571
|
|
|
$source |
|
572
|
|
|
); |
|
573
|
|
|
} |
|
574
|
|
|
if (is_array($source) || $source instanceof \DOMNodeList) { |
|
575
|
|
|
// dom nodes |
|
576
|
|
|
// load fake document |
|
577
|
|
|
if (!$this->documentFragmentLoadMarkup($fake, $charset)) { |
|
578
|
|
|
return false; |
|
|
|
|
|
|
579
|
|
|
} |
|
580
|
|
|
$nodes = $fake->import($source); |
|
581
|
|
|
foreach ($nodes as $node) { |
|
582
|
|
|
$fake->root->appendChild($node); |
|
583
|
|
|
} |
|
584
|
|
|
} else { |
|
585
|
|
|
// string markup |
|
586
|
|
|
$this->documentFragmentLoadMarkup($fake, $charset, $source); |
|
587
|
|
|
} |
|
588
|
|
|
return $fake; |
|
589
|
|
|
} |
|
590
|
|
|
|
|
591
|
|
|
/** |
|
592
|
|
|
* |
|
593
|
|
|
* @param $document DOMDocumentWrapper |
|
594
|
|
|
* @param $markup |
|
595
|
|
|
* @return $document |
|
|
|
|
|
|
596
|
|
|
*/ |
|
597
|
|
|
private function documentFragmentLoadMarkup($fragment, $charset, $markup = null) |
|
598
|
|
|
{ |
|
599
|
|
|
// TODO error handling |
|
600
|
|
|
// TODO copy doctype |
|
601
|
|
|
// tempolary turn off |
|
602
|
|
|
$fragment->isDocumentFragment = false; |
|
603
|
|
|
if ($fragment->isXML) { |
|
604
|
|
|
if ($fragment->isXHTML) { |
|
605
|
|
|
// add FAKE element to set default namespace |
|
606
|
|
|
$fragment->loadMarkupXML( |
|
607
|
|
|
'<?xml version="1.0" encoding="' . $charset |
|
608
|
|
|
. '"?>' |
|
609
|
|
|
. '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" ' |
|
610
|
|
|
. '"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">' |
|
611
|
|
|
. '<fake xmlns="http://www.w3.org/1999/xhtml">' . $markup . '</fake>' |
|
612
|
|
|
); |
|
613
|
|
|
$fragment->root = $fragment->document->firstChild->nextSibling; |
|
614
|
|
|
} else { |
|
615
|
|
|
$fragment->loadMarkupXML( |
|
616
|
|
|
'<?xml version="1.0" encoding="' . $charset |
|
617
|
|
|
. '"?><fake>' . $markup . '</fake>' |
|
618
|
|
|
); |
|
619
|
|
|
$fragment->root = $fragment->document->firstChild; |
|
620
|
|
|
} |
|
621
|
|
|
} else { |
|
622
|
|
|
$markup2 = PhpQuery::$defaultDoctype |
|
623
|
|
|
. '<html><head><meta http-equiv="Content-Type" content="text/html;charset=' |
|
624
|
|
|
. $charset . '"></head>'; |
|
625
|
|
|
$noBody = strpos($markup, '<body') === false; |
|
626
|
|
|
if ($noBody) { |
|
627
|
|
|
$markup2 .= '<body>'; |
|
628
|
|
|
} |
|
629
|
|
|
$markup2 .= $markup; |
|
630
|
|
|
if ($noBody) { |
|
631
|
|
|
$markup2 .= '</body>'; |
|
632
|
|
|
} |
|
633
|
|
|
$markup2 .= '</html>'; |
|
634
|
|
|
$fragment->loadMarkupHTML($markup2); |
|
635
|
|
|
// TODO resolv body tag merging issue |
|
636
|
|
|
$fragment->root = $noBody ? $fragment->document->firstChild->nextSibling->firstChild->nextSibling |
|
637
|
|
|
: $fragment->document->firstChild->nextSibling->firstChild->nextSibling; |
|
638
|
|
|
} |
|
639
|
|
|
if (!$fragment->root) { |
|
640
|
|
|
return false; |
|
641
|
|
|
} |
|
642
|
|
|
$fragment->isDocumentFragment = true; |
|
643
|
|
|
return true; |
|
644
|
|
|
} |
|
645
|
|
|
|
|
646
|
|
|
protected function documentFragmentToMarkup($fragment) |
|
647
|
|
|
{ |
|
648
|
|
|
PhpQuery::debug('documentFragmentToMarkup'); |
|
649
|
|
|
$tmp = $fragment->isDocumentFragment; |
|
650
|
|
|
$fragment->isDocumentFragment = false; |
|
651
|
|
|
$markup = $fragment->markup(); |
|
652
|
|
|
if ($fragment->isXML) { |
|
653
|
|
|
$markup = substr($markup, 0, strrpos($markup, '</fake>')); |
|
654
|
|
View Code Duplication |
if ($fragment->isXHTML) { |
|
|
|
|
|
|
655
|
|
|
$markup = substr($markup, strpos($markup, '<fake') + 43); |
|
656
|
|
|
} else { |
|
657
|
|
|
$markup = substr($markup, strpos($markup, '<fake>') + 6); |
|
658
|
|
|
} |
|
659
|
|
View Code Duplication |
} else { |
|
|
|
|
|
|
660
|
|
|
$markup = substr($markup, strpos($markup, '<body>') + 6); |
|
661
|
|
|
$markup = substr($markup, 0, strrpos($markup, '</body>')); |
|
662
|
|
|
} |
|
663
|
|
|
$fragment->isDocumentFragment = $tmp; |
|
664
|
|
|
if (PhpQuery::$debug) { |
|
665
|
|
|
PhpQuery::debug('documentFragmentToMarkup: ' . substr($markup, 0, 150)); |
|
666
|
|
|
} |
|
667
|
|
|
return $markup; |
|
668
|
|
|
} |
|
669
|
|
|
|
|
670
|
|
|
/** |
|
671
|
|
|
* Return document markup, starting with optional $nodes as root. |
|
672
|
|
|
* |
|
673
|
|
|
* @param $nodes \DOMNode|\DOMNodeList |
|
674
|
|
|
* @return string |
|
675
|
|
|
*/ |
|
676
|
|
|
public function markup($nodes = null, $innerMarkup = false) |
|
677
|
|
|
{ |
|
678
|
|
|
if (isset($nodes) && count($nodes) == 1 |
|
679
|
|
|
&& $nodes[0] instanceof \DOMDocument |
|
680
|
|
|
) { |
|
681
|
|
|
$nodes = null; |
|
682
|
|
|
} |
|
683
|
|
|
if (isset($nodes)) { |
|
684
|
|
|
$markup = ''; |
|
685
|
|
|
if (!is_array($nodes) && !($nodes instanceof \DOMNodeList)) { |
|
686
|
|
|
$nodes = array( |
|
687
|
|
|
$nodes |
|
688
|
|
|
); |
|
689
|
|
|
} |
|
690
|
|
|
if ($this->isDocumentFragment && !$innerMarkup) { |
|
691
|
|
|
foreach ($nodes as $i => $node) { |
|
692
|
|
|
if ($node->isSameNode($this->root)) { |
|
693
|
|
|
// var_dump($node); |
|
|
|
|
|
|
694
|
|
|
$nodes = array_slice($nodes, 0, $i) |
|
695
|
|
|
+ PhpQuery::DOMNodeListToArray($node->childNodes) |
|
696
|
|
|
+ array_slice($nodes, $i + 1); |
|
697
|
|
|
} |
|
698
|
|
|
} |
|
699
|
|
|
} |
|
700
|
|
|
if ($this->isXML && !$innerMarkup) { |
|
701
|
|
|
self::debug("Getting outerXML with charset '{$this->charset}'"); |
|
702
|
|
|
// we need outerXML, so we can benefit from |
|
703
|
|
|
// $node param support in saveXML() |
|
704
|
|
|
foreach ($nodes as $node) { |
|
705
|
|
|
$markup .= $this->document->saveXML($node); |
|
706
|
|
|
} |
|
707
|
|
|
} else { |
|
708
|
|
|
$loop = array(); |
|
709
|
|
|
if ($innerMarkup) { |
|
710
|
|
|
foreach ($nodes as $node) { |
|
711
|
|
|
if ($node->childNodes) { |
|
712
|
|
|
foreach ($node->childNodes as $child) { |
|
713
|
|
|
$loop[] = $child; |
|
714
|
|
|
} |
|
715
|
|
|
} else { |
|
716
|
|
|
$loop[] = $node; |
|
717
|
|
|
} |
|
718
|
|
|
} |
|
719
|
|
|
} else { |
|
720
|
|
|
$loop = $nodes; |
|
721
|
|
|
} |
|
722
|
|
|
self::debug( |
|
723
|
|
|
"Getting markup, moving selected nodes (" . count($loop) |
|
724
|
|
|
. ") to new DocumentFragment" |
|
725
|
|
|
); |
|
726
|
|
|
$fake = $this->documentFragmentCreate($loop); |
|
727
|
|
|
$markup = $this->documentFragmentToMarkup($fake); |
|
728
|
|
|
} |
|
729
|
|
|
if ($this->isXHTML) { |
|
730
|
|
|
self::debug("Fixing XHTML"); |
|
731
|
|
|
$markup = self::markupFixXHTML($markup); |
|
732
|
|
|
} |
|
733
|
|
|
self::debug("Markup: " . substr($markup, 0, 250)); |
|
734
|
|
|
return $markup; |
|
735
|
|
|
} else { |
|
736
|
|
|
if ($this->isDocumentFragment) { |
|
737
|
|
|
// documentFragment, html only... |
|
738
|
|
|
self::debug("Getting markup, DocumentFragment detected"); |
|
739
|
|
|
// return $this->markup( |
|
|
|
|
|
|
740
|
|
|
//// $this->document->getElementsByTagName('body')->item(0) |
|
741
|
|
|
// $this->document->root, true |
|
|
|
|
|
|
742
|
|
|
// ); |
|
743
|
|
|
$markup = $this->documentFragmentToMarkup($this); |
|
744
|
|
|
// no need for markupFixXHTML, as it's done thought markup($nodes) method |
|
745
|
|
|
return $markup; |
|
746
|
|
|
} else { |
|
747
|
|
|
self::debug( |
|
748
|
|
|
"Getting markup (" . ($this->isXML ? 'XML' : 'HTML') |
|
749
|
|
|
. "), final with charset '{$this->charset}'" |
|
750
|
|
|
); |
|
751
|
|
|
$markup = $this->isXML ? $this->document->saveXML() |
|
752
|
|
|
: $this->document->saveHTML(); |
|
753
|
|
|
if ($this->isXHTML) { |
|
754
|
|
|
self::debug("Fixing XHTML"); |
|
755
|
|
|
$markup = self::markupFixXHTML($markup); |
|
756
|
|
|
} |
|
757
|
|
|
self::debug("Markup: " . substr($markup, 0, 250)); |
|
758
|
|
|
return $markup; |
|
759
|
|
|
} |
|
760
|
|
|
} |
|
761
|
|
|
} |
|
762
|
|
|
|
|
763
|
|
|
protected static function markupFixXHTML($markup) |
|
764
|
|
|
{ |
|
765
|
|
|
$markup = self::expandEmptyTag('script', $markup); |
|
766
|
|
|
$markup = self::expandEmptyTag('select', $markup); |
|
767
|
|
|
$markup = self::expandEmptyTag('textarea', $markup); |
|
768
|
|
|
return $markup; |
|
769
|
|
|
} |
|
770
|
|
|
|
|
771
|
|
|
public static function debug($text) |
|
772
|
|
|
{ |
|
773
|
|
|
PhpQuery::debug($text); |
|
774
|
|
|
} |
|
775
|
|
|
|
|
776
|
|
|
/** |
|
777
|
|
|
* expandEmptyTag |
|
778
|
|
|
* |
|
779
|
|
|
* @param $tag |
|
780
|
|
|
* @param $xml |
|
781
|
|
|
* @return string |
|
782
|
|
|
* @author mjaque at ilkebenson dot com |
|
783
|
|
|
* @link http://php.net/manual/en/domdocument.savehtml.php#81256 |
|
784
|
|
|
*/ |
|
785
|
|
|
public static function expandEmptyTag($tag, $xml) |
|
786
|
|
|
{ |
|
787
|
|
|
$indice = 0; |
|
788
|
|
|
while ($indice < strlen($xml)) { |
|
789
|
|
|
$pos = strpos($xml, "<$tag ", $indice); |
|
790
|
|
|
if ($pos) { |
|
791
|
|
|
$posCierre = strpos($xml, ">", $pos); |
|
792
|
|
|
if ($xml[$posCierre - 1] == "/") { |
|
793
|
|
|
$xml = substr_replace($xml, "></$tag>", $posCierre - 1, 2); |
|
794
|
|
|
} |
|
795
|
|
|
$indice = $posCierre; |
|
796
|
|
|
} else { |
|
797
|
|
|
break; |
|
798
|
|
|
} |
|
799
|
|
|
} |
|
800
|
|
|
return $xml; |
|
801
|
|
|
} |
|
802
|
|
|
} |
|
803
|
|
|
|
This check looks from parameters that have been defined for a function or method, but which are not used in the method body.