|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2017 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
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* This class helps the RulesGenerator by analyzing a given template in order to answer questions |
|
16
|
|
|
* such as "can this tag be a child/descendant of that other tag?" and others related to the HTML5 |
|
17
|
|
|
* content model. |
|
18
|
|
|
* |
|
19
|
|
|
* We use the HTML5 specs to determine which children or descendants should be allowed or denied |
|
20
|
|
|
* based on HTML5 content models. While it does not exactly match HTML5 content models, it gets |
|
21
|
|
|
* pretty close. We also use HTML5 "optional end tag" rules to create closeParent rules. |
|
22
|
|
|
* |
|
23
|
|
|
* Currently, this method does not evaluate elements created with <xsl:element> correctly, or |
|
24
|
|
|
* attributes created with <xsl:attribute> and may never will due to the increased complexity it |
|
25
|
|
|
* would entail. Additionally, it does not evaluate the scope of <xsl:apply-templates/>. For |
|
26
|
|
|
* instance, it will treat <xsl:apply-templates select="LI"/> as if it was <xsl:apply-templates/> |
|
27
|
|
|
* |
|
28
|
|
|
* @link http://dev.w3.org/html5/spec/content-models.html#content-models |
|
29
|
|
|
* @link http://dev.w3.org/html5/spec/syntax.html#optional-tags |
|
30
|
|
|
* @see /scripts/patchTemplateInspector.php |
|
31
|
|
|
*/ |
|
32
|
|
|
class TemplateInspector |
|
33
|
|
|
{ |
|
34
|
|
|
/** |
|
35
|
|
|
* XSL namespace |
|
36
|
|
|
*/ |
|
37
|
|
|
const XMLNS_XSL = 'http://www.w3.org/1999/XSL/Transform'; |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @var string[] allowChild bitfield for each branch |
|
41
|
|
|
*/ |
|
42
|
|
|
protected $allowChildBitfields = []; |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* @var bool Whether elements are allowed as children |
|
46
|
|
|
*/ |
|
47
|
|
|
protected $allowsChildElements; |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* @var bool Whether text nodes are allowed as children |
|
51
|
|
|
*/ |
|
52
|
|
|
protected $allowsText; |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* @var array[] Array of array of DOMElement instances |
|
56
|
|
|
*/ |
|
57
|
|
|
protected $branches; |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* @var string OR-ed bitfield representing all of the categories used by this template |
|
61
|
|
|
*/ |
|
62
|
|
|
protected $contentBitfield = "\0"; |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* @var string denyDescendant bitfield |
|
66
|
|
|
*/ |
|
67
|
|
|
protected $denyDescendantBitfield = "\0"; |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* @var DOMDocument Document containing the template |
|
71
|
|
|
*/ |
|
72
|
|
|
protected $dom; |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* @var bool Whether this template contains any HTML elements |
|
76
|
|
|
*/ |
|
77
|
|
|
protected $hasElements = false; |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @var bool Whether this template renders non-whitespace text nodes at its root |
|
81
|
|
|
*/ |
|
82
|
|
|
protected $hasRootText; |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* @var bool Whether this template should be considered a block-level element |
|
86
|
|
|
*/ |
|
87
|
|
|
protected $isBlock = false; |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* @var bool Whether the template uses the "empty" content model |
|
91
|
|
|
*/ |
|
92
|
|
|
protected $isEmpty; |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* @var bool Whether this template adds to the list of active formatting elements |
|
96
|
|
|
*/ |
|
97
|
|
|
protected $isFormattingElement; |
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* @var bool Whether this template lets content through via an xsl:apply-templates element |
|
101
|
|
|
*/ |
|
102
|
|
|
protected $isPassthrough = false; |
|
103
|
|
|
|
|
104
|
|
|
/** |
|
105
|
|
|
* @var bool Whether all branches use the transparent content model |
|
106
|
|
|
*/ |
|
107
|
|
|
protected $isTransparent = false; |
|
108
|
|
|
|
|
109
|
|
|
/** |
|
110
|
|
|
* @var bool Whether all branches have an ancestor that is a void element |
|
111
|
|
|
*/ |
|
112
|
|
|
protected $isVoid; |
|
113
|
|
|
|
|
114
|
|
|
/** |
|
115
|
|
|
* @var array Names of every last HTML element that precedes an <xsl:apply-templates/> node |
|
116
|
|
|
*/ |
|
117
|
|
|
protected $leafNodes = []; |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* @var bool Whether any branch has an element that preserves new lines by default (e.g. <pre>) |
|
121
|
|
|
*/ |
|
122
|
|
|
protected $preservesNewLines = false; |
|
123
|
|
|
|
|
124
|
|
|
/** |
|
125
|
|
|
* @var array Bitfield of the first HTML element of every branch |
|
126
|
|
|
*/ |
|
127
|
|
|
protected $rootBitfields = []; |
|
128
|
|
|
|
|
129
|
|
|
/** |
|
130
|
|
|
* @var array Names of every HTML element that have no HTML parent |
|
131
|
|
|
*/ |
|
132
|
|
|
protected $rootNodes = []; |
|
133
|
|
|
|
|
134
|
|
|
/** |
|
135
|
|
|
* @var DOMXPath XPath engine associated with $this->dom |
|
136
|
|
|
*/ |
|
137
|
|
|
protected $xpath; |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* Constructor |
|
141
|
|
|
* |
|
142
|
|
|
* @param string $template Template content |
|
143
|
|
|
*/ |
|
144
|
93 |
|
public function __construct($template) |
|
145
|
|
|
{ |
|
146
|
93 |
|
$this->dom = TemplateHelper::loadTemplate($template); |
|
147
|
93 |
|
$this->xpath = new DOMXPath($this->dom); |
|
148
|
|
|
|
|
149
|
93 |
|
$this->analyseRootNodes(); |
|
150
|
93 |
|
$this->analyseBranches(); |
|
151
|
93 |
|
$this->analyseContent(); |
|
152
|
93 |
|
} |
|
153
|
|
|
|
|
154
|
|
|
/** |
|
155
|
|
|
* Return the value of a boolean property |
|
156
|
|
|
* |
|
157
|
|
|
* @param string $k |
|
158
|
|
|
* @return bool |
|
159
|
|
|
*/ |
|
160
|
|
|
public function __get($k) |
|
161
|
|
|
{ |
|
162
|
|
|
return (isset($this->$propName) && is_bool($this->$propName)) ? $this->$propName : false; |
|
|
|
|
|
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
/** |
|
166
|
|
|
* Return whether this template allows a given child |
|
167
|
|
|
* |
|
168
|
|
|
* @param TemplateInspector $child |
|
169
|
|
|
* @return bool |
|
170
|
|
|
*/ |
|
171
|
26 |
|
public function allowsChild(TemplateInspector $child) |
|
172
|
|
|
{ |
|
173
|
|
|
// Sometimes, a template can technically be allowed as a child but denied as a descendant |
|
174
|
26 |
|
if (!$this->allowsDescendant($child)) |
|
175
|
|
|
{ |
|
176
|
3 |
|
return false; |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
23 |
|
foreach ($child->rootBitfields as $rootBitfield) |
|
180
|
|
|
{ |
|
181
|
18 |
|
foreach ($this->allowChildBitfields as $allowChildBitfield) |
|
182
|
|
|
{ |
|
183
|
18 |
|
if (!self::match($rootBitfield, $allowChildBitfield)) |
|
184
|
|
|
{ |
|
185
|
18 |
|
return false; |
|
186
|
|
|
} |
|
187
|
|
|
} |
|
188
|
|
|
} |
|
189
|
|
|
|
|
190
|
15 |
|
return ($this->allowsText || !$child->hasRootText); |
|
191
|
|
|
} |
|
192
|
|
|
|
|
193
|
|
|
/** |
|
194
|
|
|
* Return whether this template allows a given descendant |
|
195
|
|
|
* |
|
196
|
|
|
* @param TemplateInspector $descendant |
|
197
|
|
|
* @return bool |
|
198
|
|
|
*/ |
|
199
|
29 |
|
public function allowsDescendant(TemplateInspector $descendant) |
|
200
|
|
|
{ |
|
201
|
|
|
// Test whether the descendant is explicitly disallowed |
|
202
|
29 |
|
if (self::match($descendant->contentBitfield, $this->denyDescendantBitfield)) |
|
203
|
|
|
{ |
|
204
|
3 |
|
return false; |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
// Test whether the descendant contains any elements and we disallow elements |
|
208
|
26 |
|
return ($this->allowsChildElements || !$descendant->hasElements); |
|
209
|
|
|
} |
|
210
|
|
|
|
|
211
|
|
|
/** |
|
212
|
|
|
* Return whether this template automatically closes given parent template |
|
213
|
|
|
* |
|
214
|
|
|
* @param TemplateInspector $parent |
|
215
|
|
|
* @return bool |
|
216
|
|
|
*/ |
|
217
|
10 |
|
public function closesParent(TemplateInspector $parent) |
|
218
|
|
|
{ |
|
219
|
10 |
|
foreach ($this->rootNodes as $rootName) |
|
220
|
|
|
{ |
|
221
|
10 |
|
if (empty(self::$htmlElements[$rootName]['cp'])) |
|
222
|
|
|
{ |
|
223
|
1 |
|
continue; |
|
224
|
|
|
} |
|
225
|
|
|
|
|
226
|
9 |
|
foreach ($parent->leafNodes as $leafName) |
|
227
|
|
|
{ |
|
228
|
9 |
|
if (in_array($leafName, self::$htmlElements[$rootName]['cp'], true)) |
|
229
|
|
|
{ |
|
230
|
|
|
// If any of this template's root node closes one of the parent's leaf node, we |
|
231
|
|
|
// consider that this template closes the other one |
|
232
|
9 |
|
return true; |
|
233
|
|
|
} |
|
234
|
|
|
} |
|
235
|
|
|
} |
|
236
|
|
|
|
|
237
|
3 |
|
return false; |
|
238
|
|
|
} |
|
239
|
|
|
|
|
240
|
|
|
/** |
|
241
|
|
|
* Evaluate an XPath expression |
|
242
|
|
|
* |
|
243
|
|
|
* @param string $expr XPath expression |
|
244
|
|
|
* @param DOMElement $node Context node |
|
245
|
|
|
* @return mixed |
|
246
|
|
|
*/ |
|
247
|
93 |
|
public function evaluate($expr, DOMElement $node = null) |
|
248
|
|
|
{ |
|
249
|
93 |
|
return $this->xpath->evaluate($expr, $node); |
|
250
|
|
|
} |
|
251
|
|
|
|
|
252
|
|
|
/** |
|
253
|
|
|
* Analyses the content of the whole template and set $this->contentBitfield accordingly |
|
254
|
|
|
*/ |
|
255
|
93 |
|
protected function analyseContent() |
|
256
|
|
|
{ |
|
257
|
|
|
// Get all non-XSL elements |
|
258
|
93 |
|
$query = '//*[namespace-uri() != "' . self::XMLNS_XSL . '"]'; |
|
259
|
93 |
|
foreach ($this->xpath->query($query) as $node) |
|
260
|
|
|
{ |
|
261
|
88 |
|
$this->contentBitfield |= $this->getBitfield($node, 'c'); |
|
262
|
88 |
|
$this->hasElements = true; |
|
263
|
|
|
} |
|
264
|
|
|
|
|
265
|
|
|
// Test whether this template is passthrough |
|
266
|
93 |
|
$this->isPassthrough = (bool) $this->evaluate('count(//xsl:apply-templates)'); |
|
267
|
93 |
|
} |
|
268
|
|
|
|
|
269
|
|
|
/** |
|
270
|
|
|
* Records the HTML elements (and their bitfield) rendered at the root of the template |
|
271
|
|
|
*/ |
|
272
|
93 |
|
protected function analyseRootNodes() |
|
273
|
|
|
{ |
|
274
|
|
|
// Get every non-XSL element with no non-XSL ancestor. This should return us the first |
|
275
|
|
|
// HTML element of every branch |
|
276
|
93 |
|
$query = '//*[namespace-uri() != "' . self::XMLNS_XSL . '"]' |
|
277
|
93 |
|
. '[not(ancestor::*[namespace-uri() != "' . self::XMLNS_XSL . '"])]'; |
|
278
|
93 |
|
foreach ($this->xpath->query($query) as $node) |
|
279
|
|
|
{ |
|
280
|
|
|
// Save the actual name of the root node |
|
281
|
88 |
|
$this->rootNodes[] = $node->localName; |
|
282
|
|
|
|
|
283
|
|
|
// If any root node is a block-level element, we'll mark the template as such |
|
284
|
88 |
|
if ($this->elementIsBlock($node)) |
|
285
|
|
|
{ |
|
286
|
45 |
|
$this->isBlock = true; |
|
287
|
|
|
} |
|
288
|
|
|
|
|
289
|
88 |
|
$this->rootBitfields[] = $this->getBitfield($node, 'c'); |
|
290
|
|
|
} |
|
291
|
|
|
|
|
292
|
|
|
// Test for non-whitespace text nodes at the root. For that we need a predicate that filters |
|
293
|
|
|
// out: nodes with a non-XSL ancestor, |
|
294
|
93 |
|
$predicate = '[not(ancestor::*[namespace-uri() != "' . self::XMLNS_XSL . '"])]'; |
|
295
|
|
|
|
|
296
|
|
|
// ..and nodes with an <xsl:attribute/>, <xsl:comment/> or <xsl:variable/> ancestor |
|
297
|
93 |
|
$predicate .= '[not(ancestor::xsl:attribute | ancestor::xsl:comment | ancestor::xsl:variable)]'; |
|
298
|
|
|
|
|
299
|
93 |
|
$query = '//text()[normalize-space() != ""]' . $predicate |
|
300
|
93 |
|
. '|' |
|
301
|
93 |
|
. '//xsl:text[normalize-space() != ""]' . $predicate |
|
302
|
93 |
|
. '|' |
|
303
|
93 |
|
. '//xsl:value-of' . $predicate; |
|
304
|
|
|
|
|
305
|
93 |
|
$this->hasRootText = (bool) $this->evaluate('count(' . $query . ')'); |
|
306
|
93 |
|
} |
|
307
|
|
|
|
|
308
|
|
|
/** |
|
309
|
|
|
* Analyses each branch that leads to an <xsl:apply-templates/> tag |
|
310
|
|
|
*/ |
|
311
|
93 |
|
protected function analyseBranches() |
|
312
|
|
|
{ |
|
313
|
93 |
|
$this->branches = []; |
|
314
|
93 |
|
foreach ($this->xpath->query('//xsl:apply-templates') as $applyTemplates) |
|
315
|
|
|
{ |
|
316
|
85 |
|
$query = 'ancestor::*[namespace-uri() != "' . self::XMLNS_XSL . '"]'; |
|
317
|
85 |
|
$this->branches[] = iterator_to_array($this->xpath->query($query, $applyTemplates)); |
|
318
|
|
|
} |
|
319
|
|
|
|
|
320
|
93 |
|
$this->computeAllowsChildElements(); |
|
321
|
93 |
|
$this->computeAllowsText(); |
|
322
|
93 |
|
$this->computeBitfields(); |
|
323
|
93 |
|
$this->computeFormattingElement(); |
|
324
|
93 |
|
$this->computeIsEmpty(); |
|
325
|
93 |
|
$this->computeIsTransparent(); |
|
326
|
93 |
|
$this->computeIsVoid(); |
|
327
|
93 |
|
$this->computePreservesNewLines(); |
|
328
|
93 |
|
$this->storeLeafNodes(); |
|
329
|
93 |
|
} |
|
330
|
|
|
|
|
331
|
|
|
/** |
|
332
|
|
|
* Test whether any branch of this template has an element that has given property |
|
333
|
|
|
* |
|
334
|
|
|
* @param string $propName |
|
335
|
|
|
* @return bool |
|
336
|
|
|
*/ |
|
337
|
93 |
|
protected function anyBranchHasProperty($propName) |
|
338
|
|
|
{ |
|
339
|
93 |
|
foreach ($this->branches as $branch) |
|
340
|
|
|
{ |
|
341
|
85 |
|
foreach ($branch as $element) |
|
342
|
|
|
{ |
|
343
|
81 |
|
if ($this->hasProperty($element->nodeName, $propName, $element)) |
|
344
|
|
|
{ |
|
345
|
85 |
|
return true; |
|
346
|
|
|
} |
|
347
|
|
|
} |
|
348
|
|
|
} |
|
349
|
|
|
|
|
350
|
93 |
|
return false; |
|
351
|
|
|
} |
|
352
|
|
|
|
|
353
|
|
|
/** |
|
354
|
|
|
* Compute the allowChildBitfields and denyDescendantBitfield properties |
|
355
|
|
|
* |
|
356
|
|
|
* @return void |
|
357
|
|
|
*/ |
|
358
|
93 |
|
protected function computeBitfields() |
|
359
|
|
|
{ |
|
360
|
93 |
|
if (empty($this->branches)) |
|
361
|
|
|
{ |
|
362
|
86 |
|
$this->allowChildBitfields = ["\0"]; |
|
363
|
|
|
|
|
364
|
86 |
|
return; |
|
365
|
|
|
} |
|
366
|
85 |
|
foreach ($this->branches as $branch) |
|
367
|
|
|
{ |
|
368
|
|
|
/** |
|
369
|
|
|
* @var string allowChild bitfield for current branch. Starts with the value associated |
|
370
|
|
|
* with <div> in order to approximate a value if the whole branch uses the |
|
371
|
|
|
* transparent content model |
|
372
|
|
|
*/ |
|
373
|
85 |
|
$branchBitfield = self::$htmlElements['div']['ac']; |
|
374
|
|
|
|
|
375
|
85 |
|
foreach ($branch as $element) |
|
376
|
|
|
{ |
|
377
|
81 |
|
$elName = $element->localName; |
|
378
|
81 |
|
if (!$this->hasProperty($elName, 't', $element)) |
|
379
|
|
|
{ |
|
380
|
|
|
// If the element isn't transparent, we reset its bitfield |
|
381
|
74 |
|
$branchBitfield = "\0"; |
|
382
|
|
|
} |
|
383
|
|
|
|
|
384
|
|
|
// allowChild rules are cumulative if transparent, and reset above otherwise |
|
385
|
81 |
|
$branchBitfield |= $this->getBitfield($element, 'ac'); |
|
386
|
|
|
|
|
387
|
|
|
// denyDescendant rules are cumulative |
|
388
|
81 |
|
$this->denyDescendantBitfield |= $this->getBitfield($element, 'dd'); |
|
389
|
|
|
} |
|
390
|
|
|
|
|
391
|
|
|
// Add this branch's bitfield to the list |
|
392
|
85 |
|
$this->allowChildBitfields[] = $branchBitfield; |
|
393
|
|
|
} |
|
394
|
85 |
|
} |
|
395
|
|
|
|
|
396
|
|
|
/** |
|
397
|
|
|
* Compute the allowsChildElements property |
|
398
|
|
|
* |
|
399
|
|
|
* A template allows child Elements if it has at least one xsl:apply-templates and none of its |
|
400
|
|
|
* ancestors have the text-only ("to") property |
|
401
|
|
|
* |
|
402
|
|
|
* @return void |
|
403
|
|
|
*/ |
|
404
|
93 |
|
protected function computeAllowsChildElements() |
|
405
|
|
|
{ |
|
406
|
93 |
|
$this->allowsChildElements = ($this->anyBranchHasProperty('to')) ? false : !empty($this->branches); |
|
407
|
93 |
|
} |
|
408
|
|
|
|
|
409
|
|
|
/** |
|
410
|
|
|
* Compute the allowsText property |
|
411
|
|
|
* |
|
412
|
|
|
* A template is said to allow text if none of the leaf elements disallow text |
|
413
|
|
|
* |
|
414
|
|
|
* @return void |
|
415
|
|
|
*/ |
|
416
|
93 |
|
protected function computeAllowsText() |
|
417
|
|
|
{ |
|
418
|
93 |
|
foreach (array_filter($this->branches) as $branch) |
|
419
|
|
|
{ |
|
420
|
81 |
|
$element = end($branch); |
|
421
|
81 |
|
if ($this->hasProperty($element->nodeName, 'nt', $element)) |
|
422
|
|
|
{ |
|
423
|
16 |
|
$this->allowsText = false; |
|
424
|
|
|
|
|
425
|
81 |
|
return; |
|
426
|
|
|
} |
|
427
|
|
|
} |
|
428
|
93 |
|
$this->allowsText = true; |
|
429
|
93 |
|
} |
|
430
|
|
|
|
|
431
|
|
|
/** |
|
432
|
|
|
* Compute the isFormattingElement property |
|
433
|
|
|
* |
|
434
|
|
|
* A template is said to be a formatting element if all (non-zero) of its branches are entirely |
|
435
|
|
|
* composed of formatting elements |
|
436
|
|
|
* |
|
437
|
|
|
* @return void |
|
438
|
|
|
*/ |
|
439
|
93 |
|
protected function computeFormattingElement() |
|
440
|
|
|
{ |
|
441
|
93 |
|
foreach ($this->branches as $branch) |
|
442
|
|
|
{ |
|
443
|
85 |
|
foreach ($branch as $element) |
|
444
|
|
|
{ |
|
445
|
81 |
|
if (!$this->hasProperty($element->nodeName, 'fe', $element) && !$this->isFormattingSpan($element)) |
|
446
|
|
|
{ |
|
447
|
66 |
|
$this->isFormattingElement = false; |
|
448
|
|
|
|
|
449
|
85 |
|
return; |
|
450
|
|
|
} |
|
451
|
|
|
} |
|
452
|
|
|
} |
|
453
|
86 |
|
$this->isFormattingElement = (bool) count(array_filter($this->branches)); |
|
454
|
86 |
|
} |
|
455
|
|
|
|
|
456
|
|
|
/** |
|
457
|
|
|
* Compute the isEmpty property |
|
458
|
|
|
* |
|
459
|
|
|
* A template is said to be empty if it has no xsl:apply-templates elements or any there is a empty |
|
460
|
|
|
* element ancestor to an xsl:apply-templates element |
|
461
|
|
|
* |
|
462
|
|
|
* @return void |
|
463
|
|
|
*/ |
|
464
|
93 |
|
protected function computeIsEmpty() |
|
465
|
|
|
{ |
|
466
|
93 |
|
$this->isEmpty = ($this->anyBranchHasProperty('e')) || empty($this->branches); |
|
467
|
93 |
|
} |
|
468
|
|
|
|
|
469
|
|
|
/** |
|
470
|
|
|
* Compute the isTransparent property |
|
471
|
|
|
* |
|
472
|
|
|
* A template is said to be transparent if it has at least one branch and no non-transparent |
|
473
|
|
|
* elements in its path |
|
474
|
|
|
* |
|
475
|
|
|
* @return void |
|
476
|
|
|
*/ |
|
477
|
93 |
|
protected function computeIsTransparent() |
|
478
|
|
|
{ |
|
479
|
93 |
|
foreach ($this->branches as $branch) |
|
480
|
|
|
{ |
|
481
|
85 |
|
foreach ($branch as $element) |
|
482
|
|
|
{ |
|
483
|
81 |
|
if (!$this->hasProperty($element->nodeName, 't', $element)) |
|
484
|
|
|
{ |
|
485
|
74 |
|
$this->isTransparent = false; |
|
486
|
|
|
|
|
487
|
85 |
|
return; |
|
488
|
|
|
} |
|
489
|
|
|
} |
|
490
|
|
|
} |
|
491
|
86 |
|
$this->isTransparent = !empty($this->branches); |
|
492
|
86 |
|
} |
|
493
|
|
|
|
|
494
|
|
|
/** |
|
495
|
|
|
* Compute the isVoid property |
|
496
|
|
|
* |
|
497
|
|
|
* A template is said to be void if it has no xsl:apply-templates elements or any there is a void |
|
498
|
|
|
* element ancestor to an xsl:apply-templates element |
|
499
|
|
|
* |
|
500
|
|
|
* @return void |
|
501
|
|
|
*/ |
|
502
|
93 |
|
protected function computeIsVoid() |
|
503
|
|
|
{ |
|
504
|
93 |
|
$this->isVoid = ($this->anyBranchHasProperty('v')) || empty($this->branches); |
|
505
|
93 |
|
} |
|
506
|
|
|
|
|
507
|
|
|
/** |
|
508
|
|
|
* Compute the preservesNewLines property |
|
509
|
|
|
* |
|
510
|
|
|
* @return void |
|
511
|
|
|
*/ |
|
512
|
93 |
|
protected function computePreservesNewLines() |
|
513
|
|
|
{ |
|
514
|
93 |
|
foreach ($this->branches as $branch) |
|
515
|
|
|
{ |
|
516
|
85 |
|
$style = ''; |
|
517
|
85 |
|
foreach ($branch as $element) |
|
518
|
|
|
{ |
|
519
|
81 |
|
$style .= $this->getStyle($element, true); |
|
520
|
|
|
} |
|
521
|
|
|
|
|
522
|
85 |
|
if (preg_match('(.*white-space\\s*:\\s*(no|pre))is', $style, $m) && strtolower($m[1]) === 'pre') |
|
523
|
|
|
{ |
|
524
|
6 |
|
$this->preservesNewLines = true; |
|
525
|
|
|
|
|
526
|
85 |
|
return; |
|
527
|
|
|
} |
|
528
|
|
|
} |
|
529
|
93 |
|
$this->preservesNewLines = false; |
|
530
|
93 |
|
} |
|
531
|
|
|
|
|
532
|
|
|
/** |
|
533
|
|
|
* Test whether given element is a block-level element |
|
534
|
|
|
* |
|
535
|
|
|
* @param DOMElement $element |
|
536
|
|
|
* @return bool |
|
537
|
|
|
*/ |
|
538
|
88 |
|
protected function elementIsBlock(DOMElement $element) |
|
539
|
|
|
{ |
|
540
|
88 |
|
$style = $this->getStyle($element); |
|
541
|
88 |
|
if (preg_match('(\\bdisplay\\s*:\\s*block)i', $style)) |
|
542
|
|
|
{ |
|
543
|
1 |
|
return true; |
|
544
|
|
|
} |
|
545
|
87 |
|
if (preg_match('(\\bdisplay\\s*:\\s*(?:inli|no)ne)i', $style)) |
|
546
|
|
|
{ |
|
547
|
3 |
|
return false; |
|
548
|
|
|
} |
|
549
|
|
|
|
|
550
|
84 |
|
return $this->hasProperty($element->nodeName, 'b', $element); |
|
551
|
|
|
} |
|
552
|
|
|
|
|
553
|
|
|
/** |
|
554
|
|
|
* Retrieve and return the inline style assigned to given element |
|
555
|
|
|
* |
|
556
|
|
|
* @param DOMElement $node Context node |
|
557
|
|
|
* @param bool $deep Whether to retrieve the content of all xsl:attribute descendants |
|
558
|
|
|
* @return string |
|
559
|
|
|
*/ |
|
560
|
88 |
|
protected function getStyle(DOMElement $node, $deep = false) |
|
561
|
|
|
{ |
|
562
|
88 |
|
$style = ''; |
|
563
|
88 |
|
if ($this->hasProperty($node->nodeName, 'pre', $node)) |
|
564
|
|
|
{ |
|
565
|
4 |
|
$style .= 'white-space:pre;'; |
|
566
|
|
|
} |
|
567
|
88 |
|
$style .= $node->getAttribute('style'); |
|
568
|
|
|
|
|
569
|
|
|
// Add the content of any descendant/child xsl:attribute named "style" |
|
570
|
88 |
|
$query = (($deep) ? './/' : './') . 'xsl:attribute[@name="style"]'; |
|
571
|
88 |
|
foreach ($this->xpath->query($query, $node) as $attribute) |
|
572
|
|
|
{ |
|
573
|
2 |
|
$style .= ';' . $attribute->textContent; |
|
574
|
|
|
} |
|
575
|
|
|
|
|
576
|
88 |
|
return $style; |
|
577
|
|
|
} |
|
578
|
|
|
|
|
579
|
|
|
/** |
|
580
|
|
|
* Test whether given node is a span element used for formatting |
|
581
|
|
|
* |
|
582
|
|
|
* Will return TRUE if the node is a span element with a class attribute and/or a style attribute |
|
583
|
|
|
* and no other attributes |
|
584
|
|
|
* |
|
585
|
|
|
* @param DOMElement $node |
|
586
|
|
|
* @return boolean |
|
587
|
|
|
*/ |
|
588
|
72 |
|
protected function isFormattingSpan(DOMElement $node) |
|
589
|
|
|
{ |
|
590
|
72 |
|
if ($node->nodeName !== 'span') |
|
591
|
|
|
{ |
|
592
|
50 |
|
return false; |
|
593
|
|
|
} |
|
594
|
|
|
|
|
595
|
24 |
|
if ($node->getAttribute('class') === '' && $node->getAttribute('style') === '') |
|
596
|
|
|
{ |
|
597
|
17 |
|
return false; |
|
598
|
|
|
} |
|
599
|
|
|
|
|
600
|
7 |
|
foreach ($node->attributes as $attrName => $attribute) |
|
601
|
|
|
{ |
|
602
|
7 |
|
if ($attrName !== 'class' && $attrName !== 'style') |
|
603
|
|
|
{ |
|
604
|
7 |
|
return false; |
|
605
|
|
|
} |
|
606
|
|
|
} |
|
607
|
|
|
|
|
608
|
6 |
|
return true; |
|
609
|
|
|
} |
|
610
|
|
|
|
|
611
|
|
|
/** |
|
612
|
|
|
* Store the names of every leaf node |
|
613
|
|
|
* |
|
614
|
|
|
* A leaf node is defined as the closest non-XSL ancestor to an xsl:apply-templates element |
|
615
|
|
|
* |
|
616
|
|
|
* @return void |
|
617
|
|
|
*/ |
|
618
|
93 |
|
protected function storeLeafNodes() |
|
619
|
|
|
{ |
|
620
|
93 |
|
foreach (array_filter($this->branches) as $branch) |
|
621
|
|
|
{ |
|
622
|
81 |
|
$this->leafNodes[] = end($branch)->nodeName; |
|
623
|
|
|
} |
|
624
|
93 |
|
} |
|
625
|
|
|
|
|
626
|
|
|
/** |
|
627
|
|
|
* "What is this?" you might ask. This is basically a compressed version of the HTML5 content |
|
628
|
|
|
* models and rules, with some liberties taken. |
|
629
|
|
|
* |
|
630
|
|
|
* For each element, up to three bitfields are defined: "c", "ac" and "dd". Bitfields are stored |
|
631
|
|
|
* as raw bytes, formatted using the octal notation to keep the sources ASCII. |
|
632
|
|
|
* |
|
633
|
|
|
* "c" represents the categories the element belongs to. The categories are comprised of HTML5 |
|
634
|
|
|
* content models (such as "phrasing content" or "interactive content") plus a few special |
|
635
|
|
|
* categories created to cover the parts of the specs that refer to "a group of X and Y |
|
636
|
|
|
* elements" rather than a specific content model. |
|
637
|
|
|
* |
|
638
|
|
|
* "ac" represents the categories that are allowed as children of given element. |
|
639
|
|
|
* |
|
640
|
|
|
* "dd" represents the categories that must not appear as a descendant of given element. |
|
641
|
|
|
* |
|
642
|
|
|
* Sometimes, HTML5 specifies some restrictions on when an element can accept certain children, |
|
643
|
|
|
* or what categories the element belongs to. For example, an <img> element is only part of the |
|
644
|
|
|
* "interactive content" category if it has a "usemap" attribute. Those restrictions are |
|
645
|
|
|
* expressed as an XPath expression and stored using the concatenation of the key of the bitfield |
|
646
|
|
|
* plus the bit number of the category. For instance, if "interactive content" got assigned to |
|
647
|
|
|
* bit 2, the definition of the <img> element will contain a key "c2" with value "@usemap". |
|
648
|
|
|
* |
|
649
|
|
|
* Additionally, other flags are set: |
|
650
|
|
|
* |
|
651
|
|
|
* "t" indicates that the element uses the "transparent" content model. |
|
652
|
|
|
* "e" indicates that the element uses the "empty" content model. |
|
653
|
|
|
* "v" indicates that the element is a void element. |
|
654
|
|
|
* "nt" indicates that the element does not accept text nodes. (no text) |
|
655
|
|
|
* "to" indicates that the element should only contain text. (text-only) |
|
656
|
|
|
* "fe" indicates that the element is a formatting element. It will automatically be reopened |
|
657
|
|
|
* when closed by an end tag of a different name. |
|
658
|
|
|
* "b" indicates that the element is not phrasing content, which makes it likely to act like |
|
659
|
|
|
* a block element. |
|
660
|
|
|
* |
|
661
|
|
|
* Finally, HTML5 defines "optional end tag" rules, where one element automatically closes its |
|
662
|
|
|
* predecessor. Those are used to generate closeParent rules and are stored in the "cp" key. |
|
663
|
|
|
* |
|
664
|
|
|
* @var array |
|
665
|
|
|
* @see /scripts/patchTemplateInspector.php |
|
666
|
|
|
*/ |
|
667
|
|
|
protected static $htmlElements = [ |
|
668
|
|
|
'a'=>['c'=>"\17\0\0\0\0\1",'c3'=>'@href','ac'=>"\0",'dd'=>"\10\0\0\0\0\1",'t'=>1,'fe'=>1], |
|
669
|
|
|
'abbr'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
670
|
|
|
'address'=>['c'=>"\3\40",'ac'=>"\1",'dd'=>"\0\45",'b'=>1,'cp'=>['p']], |
|
671
|
|
|
'article'=>['c'=>"\3\4",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
672
|
|
|
'aside'=>['c'=>"\3\4",'ac'=>"\1",'dd'=>"\0\0\0\0\10",'b'=>1,'cp'=>['p']], |
|
673
|
|
|
'audio'=>['c'=>"\57",'c3'=>'@controls','c1'=>'@controls','ac'=>"\0\0\0\104",'ac26'=>'not(@src)','dd'=>"\0\0\0\0\0\2",'dd41'=>'@src','t'=>1], |
|
674
|
|
|
'b'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0",'fe'=>1], |
|
675
|
|
|
'base'=>['c'=>"\20",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1], |
|
676
|
|
|
'bdi'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
677
|
|
|
'bdo'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
678
|
|
|
'blockquote'=>['c'=>"\203",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
679
|
|
|
'body'=>['c'=>"\200\0\4",'ac'=>"\1",'dd'=>"\0",'b'=>1], |
|
680
|
|
|
'br'=>['c'=>"\5",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1], |
|
681
|
|
|
'button'=>['c'=>"\117",'ac'=>"\4",'dd'=>"\10"], |
|
682
|
|
|
'canvas'=>['c'=>"\47",'ac'=>"\0",'dd'=>"\0",'t'=>1], |
|
683
|
|
|
'caption'=>['c'=>"\0\2",'ac'=>"\1",'dd'=>"\0\0\0\200",'b'=>1], |
|
684
|
|
|
'cite'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
685
|
|
|
'code'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0",'fe'=>1], |
|
686
|
|
|
'col'=>['c'=>"\0\0\20",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1], |
|
687
|
|
|
'colgroup'=>['c'=>"\0\2",'ac'=>"\0\0\20",'ac20'=>'not(@span)','dd'=>"\0",'nt'=>1,'e'=>1,'e0'=>'@span','b'=>1], |
|
688
|
|
|
'data'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
689
|
|
|
'datalist'=>['c'=>"\5",'ac'=>"\4\200\0\10",'dd'=>"\0"], |
|
690
|
|
|
'dd'=>['c'=>"\0\0\200",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['dd','dt']], |
|
691
|
|
|
'del'=>['c'=>"\5",'ac'=>"\0",'dd'=>"\0",'t'=>1], |
|
692
|
|
|
'details'=>['c'=>"\213",'ac'=>"\1\0\0\2",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
693
|
|
|
'dfn'=>['c'=>"\7\0\0\0\40",'ac'=>"\4",'dd'=>"\0\0\0\0\40"], |
|
694
|
|
|
'div'=>['c'=>"\3",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
695
|
|
|
'dl'=>['c'=>"\3",'c1'=>'dt and dd','ac'=>"\0\200\200",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['p']], |
|
696
|
|
|
'dt'=>['c'=>"\0\0\200",'ac'=>"\1",'dd'=>"\0\5\0\40",'b'=>1,'cp'=>['dd','dt']], |
|
697
|
|
|
'em'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0",'fe'=>1], |
|
698
|
|
|
'embed'=>['c'=>"\57",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1], |
|
699
|
|
|
'fieldset'=>['c'=>"\303",'ac'=>"\1\0\0\20",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
700
|
|
|
'figcaption'=>['c'=>"\0\0\0\0\0\4",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
701
|
|
|
'figure'=>['c'=>"\203",'ac'=>"\1\0\0\0\0\4",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
702
|
|
|
'footer'=>['c'=>"\3\40",'ac'=>"\1",'dd'=>"\0\0\0\0\10",'b'=>1,'cp'=>['p']], |
|
703
|
|
|
'form'=>['c'=>"\3\0\0\0\20",'ac'=>"\1",'dd'=>"\0\0\0\0\20",'b'=>1,'cp'=>['p']], |
|
704
|
|
|
'h1'=>['c'=>"\3\1",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
705
|
|
|
'h2'=>['c'=>"\3\1",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
706
|
|
|
'h3'=>['c'=>"\3\1",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
707
|
|
|
'h4'=>['c'=>"\3\1",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
708
|
|
|
'h5'=>['c'=>"\3\1",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
709
|
|
|
'h6'=>['c'=>"\3\1",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
710
|
|
|
'head'=>['c'=>"\0\0\4",'ac'=>"\20",'dd'=>"\0",'nt'=>1,'b'=>1], |
|
711
|
|
|
'header'=>['c'=>"\3\40\0\40",'ac'=>"\1",'dd'=>"\0\0\0\0\10",'b'=>1,'cp'=>['p']], |
|
712
|
|
|
'hr'=>['c'=>"\1\100",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1,'cp'=>['p']], |
|
713
|
|
|
'html'=>['c'=>"\0",'ac'=>"\0\0\4",'dd'=>"\0",'nt'=>1,'b'=>1], |
|
714
|
|
|
'i'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0",'fe'=>1], |
|
715
|
|
|
'iframe'=>['c'=>"\57",'ac'=>"\4",'dd'=>"\0"], |
|
716
|
|
|
'img'=>['c'=>"\57\20\10",'c3'=>'@usemap','ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1], |
|
717
|
|
|
'input'=>['c'=>"\17\20",'c3'=>'@type!="hidden"','c12'=>'@type!="hidden" or @type="hidden"','c1'=>'@type!="hidden"','ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1], |
|
718
|
|
|
'ins'=>['c'=>"\7",'ac'=>"\0",'dd'=>"\0",'t'=>1], |
|
719
|
|
|
'kbd'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
720
|
|
|
'keygen'=>['c'=>"\117",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1], |
|
721
|
|
|
'label'=>['c'=>"\17\20\0\0\4",'ac'=>"\4",'dd'=>"\0\0\1\0\4"], |
|
722
|
|
|
'legend'=>['c'=>"\0\0\0\20",'ac'=>"\4",'dd'=>"\0",'b'=>1], |
|
723
|
|
|
'li'=>['c'=>"\0\0\0\0\200",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['li']], |
|
724
|
|
|
'link'=>['c'=>"\20",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1], |
|
725
|
|
|
'main'=>['c'=>"\3\0\0\0\10",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
726
|
|
|
'mark'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
727
|
|
|
'media element'=>['c'=>"\0\0\0\0\0\2",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'b'=>1], |
|
728
|
|
|
'menu'=>['c'=>"\1\100",'ac'=>"\0\300",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['p']], |
|
729
|
|
|
'menuitem'=>['c'=>"\0\100",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1], |
|
730
|
|
|
'meta'=>['c'=>"\20",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1], |
|
731
|
|
|
'meter'=>['c'=>"\7\0\1\0\2",'ac'=>"\4",'dd'=>"\0\0\0\0\2"], |
|
732
|
|
|
'nav'=>['c'=>"\3\4",'ac'=>"\1",'dd'=>"\0\0\0\0\10",'b'=>1,'cp'=>['p']], |
|
733
|
|
|
'noscript'=>['c'=>"\25",'ac'=>"\0",'dd'=>"\0",'nt'=>1], |
|
734
|
|
|
'object'=>['c'=>"\147",'ac'=>"\0\0\0\0\1",'dd'=>"\0",'t'=>1], |
|
735
|
|
|
'ol'=>['c'=>"\3",'c1'=>'li','ac'=>"\0\200\0\0\200",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['p']], |
|
736
|
|
|
'optgroup'=>['c'=>"\0\0\2",'ac'=>"\0\200\0\10",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['optgroup','option']], |
|
737
|
|
|
'option'=>['c'=>"\0\0\2\10",'ac'=>"\0",'dd'=>"\0",'b'=>1,'cp'=>['option']], |
|
738
|
|
|
'output'=>['c'=>"\107",'ac'=>"\4",'dd'=>"\0"], |
|
739
|
|
|
'p'=>['c'=>"\3",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
740
|
|
|
'param'=>['c'=>"\0\0\0\0\1",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1], |
|
741
|
|
|
'picture'=>['c'=>"\45",'ac'=>"\0\200\10",'dd'=>"\0",'nt'=>1], |
|
742
|
|
|
'pre'=>['c'=>"\3",'ac'=>"\4",'dd'=>"\0",'pre'=>1,'b'=>1,'cp'=>['p']], |
|
743
|
|
|
'progress'=>['c'=>"\7\0\1\1",'ac'=>"\4",'dd'=>"\0\0\0\1"], |
|
744
|
|
|
'q'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
745
|
|
|
'rb'=>['c'=>"\0\10",'ac'=>"\4",'dd'=>"\0",'b'=>1], |
|
746
|
|
|
'rp'=>['c'=>"\0\10\100",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['rp','rt']], |
|
747
|
|
|
'rt'=>['c'=>"\0\10\100",'ac'=>"\4",'dd'=>"\0",'b'=>1,'cp'=>['rp','rt']], |
|
748
|
|
|
'rtc'=>['c'=>"\0\10",'ac'=>"\4\0\100",'dd'=>"\0",'b'=>1], |
|
749
|
|
|
'ruby'=>['c'=>"\7",'ac'=>"\4\10",'dd'=>"\0"], |
|
750
|
|
|
's'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0",'fe'=>1], |
|
751
|
|
|
'samp'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
752
|
|
|
'script'=>['c'=>"\25\200",'ac'=>"\0",'dd'=>"\0",'to'=>1], |
|
753
|
|
|
'section'=>['c'=>"\3\4",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['p']], |
|
754
|
|
|
'select'=>['c'=>"\117",'ac'=>"\0\200\2",'dd'=>"\0",'nt'=>1], |
|
755
|
|
|
'small'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0",'fe'=>1], |
|
756
|
|
|
'source'=>['c'=>"\0\0\10\4",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1], |
|
757
|
|
|
'span'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
758
|
|
|
'strong'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0",'fe'=>1], |
|
759
|
|
|
'style'=>['c'=>"\20",'ac'=>"\0",'dd'=>"\0",'to'=>1,'b'=>1], |
|
760
|
|
|
'sub'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
761
|
|
|
'summary'=>['c'=>"\0\0\0\2",'ac'=>"\4\1",'dd'=>"\0",'b'=>1], |
|
762
|
|
|
'sup'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
763
|
|
|
'table'=>['c'=>"\3\0\0\200",'ac'=>"\0\202",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['p']], |
|
764
|
|
|
'tbody'=>['c'=>"\0\2",'ac'=>"\0\200\0\0\100",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['tbody','td','tfoot','th','thead','tr']], |
|
765
|
|
|
'td'=>['c'=>"\200\0\40",'ac'=>"\1",'dd'=>"\0",'b'=>1,'cp'=>['td','th']], |
|
766
|
|
|
'template'=>['c'=>"\25\200\20",'ac'=>"\0",'dd'=>"\0",'nt'=>1], |
|
767
|
|
|
'textarea'=>['c'=>"\117",'ac'=>"\0",'dd'=>"\0",'pre'=>1,'to'=>1], |
|
768
|
|
|
'tfoot'=>['c'=>"\0\2",'ac'=>"\0\200\0\0\100",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['tbody','td','th','thead','tr']], |
|
769
|
|
|
'th'=>['c'=>"\0\0\40",'ac'=>"\1",'dd'=>"\0\5\0\40",'b'=>1,'cp'=>['td','th']], |
|
770
|
|
|
'thead'=>['c'=>"\0\2",'ac'=>"\0\200\0\0\100",'dd'=>"\0",'nt'=>1,'b'=>1], |
|
771
|
|
|
'time'=>['c'=>"\7",'ac'=>"\4",'ac2'=>'@datetime','dd'=>"\0"], |
|
772
|
|
|
'title'=>['c'=>"\20",'ac'=>"\0",'dd'=>"\0",'to'=>1,'b'=>1], |
|
773
|
|
|
'tr'=>['c'=>"\0\2\0\0\100",'ac'=>"\0\200\40",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['td','th','tr']], |
|
774
|
|
|
'track'=>['c'=>"\0\0\0\100",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1,'b'=>1], |
|
775
|
|
|
'u'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0",'fe'=>1], |
|
776
|
|
|
'ul'=>['c'=>"\3",'c1'=>'li','ac'=>"\0\200\0\0\200",'dd'=>"\0",'nt'=>1,'b'=>1,'cp'=>['p']], |
|
777
|
|
|
'var'=>['c'=>"\7",'ac'=>"\4",'dd'=>"\0"], |
|
778
|
|
|
'video'=>['c'=>"\57",'c3'=>'@controls','ac'=>"\0\0\0\104",'ac26'=>'not(@src)','dd'=>"\0\0\0\0\0\2",'dd41'=>'@src','t'=>1], |
|
779
|
|
|
'wbr'=>['c'=>"\5",'ac'=>"\0",'dd'=>"\0",'nt'=>1,'e'=>1,'v'=>1] |
|
780
|
|
|
]; |
|
781
|
|
|
|
|
782
|
|
|
/** |
|
783
|
|
|
* Get the bitfield value for a given element in a given context |
|
784
|
|
|
* |
|
785
|
|
|
* @param DOMElement $element Context node |
|
786
|
|
|
* @param string $k Bitfield name: either 'c', 'ac' or 'dd' |
|
787
|
|
|
* @return string |
|
788
|
|
|
*/ |
|
789
|
88 |
|
protected function getBitfield(DOMElement $element, $k) |
|
790
|
|
|
{ |
|
791
|
88 |
|
$elName = $element->nodeName; |
|
792
|
88 |
|
if (!isset(self::$htmlElements[$elName])) |
|
793
|
|
|
{ |
|
794
|
2 |
|
$elName = 'span'; |
|
795
|
|
|
} |
|
796
|
|
|
|
|
797
|
88 |
|
$bitfield = self::$htmlElements[$elName][$k]; |
|
798
|
88 |
|
foreach (str_split($bitfield, 1) as $byteNumber => $char) |
|
799
|
|
|
{ |
|
800
|
88 |
|
$byteValue = ord($char); |
|
801
|
88 |
|
for ($bitNumber = 0; $bitNumber < 8; ++$bitNumber) |
|
802
|
|
|
{ |
|
803
|
88 |
|
$bitValue = 1 << $bitNumber; |
|
804
|
88 |
|
if (!($byteValue & $bitValue)) |
|
805
|
|
|
{ |
|
806
|
|
|
// The bit is not set |
|
807
|
88 |
|
continue; |
|
808
|
|
|
} |
|
809
|
|
|
|
|
810
|
88 |
|
$n = $byteNumber * 8 + $bitNumber; |
|
811
|
|
|
|
|
812
|
|
|
// Test for an XPath condition for that category |
|
813
|
88 |
|
if (isset(self::$htmlElements[$elName][$k . $n])) |
|
814
|
|
|
{ |
|
815
|
18 |
|
$xpath = 'boolean(' . self::$htmlElements[$elName][$k . $n] . ')'; |
|
816
|
|
|
|
|
817
|
|
|
// If the XPath condition is not fulfilled... |
|
818
|
18 |
|
if (!$this->evaluate($xpath, $element)) |
|
819
|
|
|
{ |
|
820
|
|
|
// ...turn off the corresponding bit |
|
821
|
16 |
|
$byteValue ^= $bitValue; |
|
822
|
|
|
|
|
823
|
|
|
// Update the original bitfield |
|
824
|
16 |
|
$bitfield[$byteNumber] = chr($byteValue); |
|
825
|
|
|
} |
|
826
|
|
|
} |
|
827
|
|
|
} |
|
828
|
|
|
} |
|
829
|
|
|
|
|
830
|
88 |
|
return $bitfield; |
|
831
|
|
|
} |
|
832
|
|
|
|
|
833
|
|
|
/** |
|
834
|
|
|
* Test whether given element has given property in context |
|
835
|
|
|
* |
|
836
|
|
|
* @param string $elName Element name |
|
837
|
|
|
* @param string $propName Property name, see self::$htmlElements |
|
838
|
|
|
* @param DOMElement $node Context node |
|
839
|
|
|
* @return bool |
|
840
|
|
|
*/ |
|
841
|
88 |
|
protected function hasProperty($elName, $propName, DOMElement $node) |
|
842
|
|
|
{ |
|
843
|
88 |
|
if (!empty(self::$htmlElements[$elName][$propName])) |
|
844
|
|
|
{ |
|
845
|
|
|
// Test the XPath condition |
|
846
|
60 |
|
if (!isset(self::$htmlElements[$elName][$propName . '0']) |
|
847
|
60 |
|
|| $this->evaluate('boolean(' . self::$htmlElements[$elName][$propName . '0'] . ')', $node)) |
|
848
|
|
|
{ |
|
849
|
60 |
|
return true; |
|
850
|
|
|
} |
|
851
|
|
|
} |
|
852
|
|
|
|
|
853
|
88 |
|
return false; |
|
854
|
|
|
} |
|
855
|
|
|
|
|
856
|
|
|
/** |
|
857
|
|
|
* Test whether two bitfields have any bits in common |
|
858
|
|
|
* |
|
859
|
|
|
* @param string $bitfield1 |
|
860
|
|
|
* @param string $bitfield2 |
|
861
|
|
|
* @return bool |
|
862
|
|
|
*/ |
|
863
|
29 |
|
protected static function match($bitfield1, $bitfield2) |
|
864
|
|
|
{ |
|
865
|
29 |
|
return (trim($bitfield1 & $bitfield2, "\0") !== ''); |
|
866
|
|
|
} |
|
867
|
|
|
} |
s9e /
TextFormatter
This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug.