1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace SilverStripe\Forms\HTMLEditor; |
4
|
|
|
|
5
|
|
|
use DOMAttr; |
6
|
|
|
use DOMElement; |
7
|
|
|
use SilverStripe\Core\Config\Configurable; |
8
|
|
|
use SilverStripe\Core\Injector\Injectable; |
9
|
|
|
use SilverStripe\View\Parsers\HTMLValue; |
10
|
|
|
use stdClass; |
11
|
|
|
|
12
|
|
|
/** |
13
|
|
|
* Sanitises an HTMLValue so it's contents are the elements and attributes that are whitelisted |
14
|
|
|
* using the same configuration as TinyMCE |
15
|
|
|
* |
16
|
|
|
* See www.tinymce.com/wiki.php/configuration:valid_elements for details on the spec of TinyMCE's |
17
|
|
|
* whitelist configuration |
18
|
|
|
*/ |
19
|
|
|
class HTMLEditorSanitiser |
20
|
|
|
{ |
21
|
|
|
use Configurable; |
22
|
|
|
use Injectable; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* rel attribute to add to link elements which have a target attribute (usually "_blank") |
26
|
|
|
* this is to done to prevent reverse tabnabbing - see https://www.owasp.org/index.php/Reverse_Tabnabbing |
27
|
|
|
* noopener includes the behaviour we want, though some browsers don't yet support it and rely |
28
|
|
|
* upon using noreferrer instead - see https://caniuse.com/rel-noopener for current browser compatibility |
29
|
|
|
* set this to null if you would like to disable this behaviour |
30
|
|
|
* set this to an empty string if you would like to remove rel attributes that were previously set |
31
|
|
|
* |
32
|
|
|
* @var string |
33
|
|
|
*/ |
34
|
|
|
private static $link_rel_value = 'noopener noreferrer'; |
35
|
|
|
|
36
|
|
|
/** @var [stdClass] - $element => $rule hash for whitelist element rules where the element name isn't a pattern */ |
|
|
|
|
37
|
|
|
protected $elements = array(); |
38
|
|
|
/** @var [stdClass] - Sequential list of whitelist element rules where the element name is a pattern */ |
|
|
|
|
39
|
|
|
protected $elementPatterns = array(); |
40
|
|
|
|
41
|
|
|
/** @var [stdClass] - The list of attributes that apply to all further whitelisted elements added */ |
|
|
|
|
42
|
|
|
protected $globalAttributes = array(); |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* Construct a sanitiser from a given HTMLEditorConfig |
46
|
|
|
* |
47
|
|
|
* Note that we build data structures from the current state of HTMLEditorConfig - later changes to |
48
|
|
|
* the passed instance won't cause this instance to update it's whitelist |
49
|
|
|
* |
50
|
|
|
* @param HTMLEditorConfig $config |
51
|
|
|
*/ |
52
|
|
|
public function __construct(HTMLEditorConfig $config) |
53
|
|
|
{ |
54
|
|
|
$valid = $config->getOption('valid_elements'); |
55
|
|
|
if ($valid) { |
56
|
|
|
$this->addValidElements($valid); |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
$valid = $config->getOption('extended_valid_elements'); |
60
|
|
|
if ($valid) { |
61
|
|
|
$this->addValidElements($valid); |
62
|
|
|
} |
63
|
|
|
} |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* Given a TinyMCE pattern (close to unix glob style), create a regex that does the match |
67
|
|
|
* |
68
|
|
|
* @param $str - The TinyMCE pattern |
|
|
|
|
69
|
|
|
* @return string - The equivalent regex |
70
|
|
|
*/ |
71
|
|
|
protected function patternToRegex($str) |
72
|
|
|
{ |
73
|
|
|
return '/^' . preg_replace('/([?+*])/', '.$1', $str) . '$/'; |
74
|
|
|
} |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* Given a valid_elements string, parse out the actual element and attribute rules and add to the |
78
|
|
|
* internal whitelist |
79
|
|
|
* |
80
|
|
|
* Logic based heavily on javascript version from tiny_mce_src.js |
81
|
|
|
* |
82
|
|
|
* @param string $validElements - The valid_elements or extended_valid_elements string to add to the whitelist |
83
|
|
|
*/ |
84
|
|
|
protected function addValidElements($validElements) |
85
|
|
|
{ |
86
|
|
|
$elementRuleRegExp = '/^([#+\-])?([^\[\/]+)(?:\/([^\[]+))?(?:\[([^\]]+)\])?$/'; |
87
|
|
|
$attrRuleRegExp = '/^([!\-])?(\w+::\w+|[^=:<]+)?(?:([=:<])(.*))?$/'; |
88
|
|
|
$hasPatternsRegExp = '/[*?+]/'; |
89
|
|
|
|
90
|
|
|
foreach (explode(',', $validElements) as $validElement) { |
91
|
|
|
if (preg_match($elementRuleRegExp, $validElement, $matches)) { |
92
|
|
|
$prefix = isset($matches[1]) ? $matches[1] : null; |
93
|
|
|
$elementName = isset($matches[2]) ? $matches[2] : null; |
94
|
|
|
$outputName = isset($matches[3]) ? $matches[3] : null; |
95
|
|
|
$attrData = isset($matches[4]) ? $matches[4] : null; |
96
|
|
|
|
97
|
|
|
// Create the new element |
98
|
|
|
$element = new stdClass(); |
99
|
|
|
$element->attributes = array(); |
100
|
|
|
$element->attributePatterns = array(); |
101
|
|
|
|
102
|
|
|
$element->attributesRequired = array(); |
103
|
|
|
$element->attributesDefault = array(); |
104
|
|
|
$element->attributesForced = array(); |
105
|
|
|
|
106
|
|
|
foreach (array('#' => 'paddEmpty', '-' => 'removeEmpty') as $match => $means) { |
107
|
|
|
$element->$means = ($prefix === $match); |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
// Copy attributes from global rule into current rule |
111
|
|
|
if ($this->globalAttributes) { |
|
|
|
|
112
|
|
|
$element->attributes = array_merge($element->attributes, $this->globalAttributes); |
113
|
|
|
} |
114
|
|
|
|
115
|
|
|
// Attributes defined |
116
|
|
|
if ($attrData) { |
117
|
|
|
foreach (explode('|', $attrData) as $attr) { |
118
|
|
|
if (preg_match($attrRuleRegExp, $attr, $matches)) { |
119
|
|
|
$attr = new stdClass(); |
120
|
|
|
|
121
|
|
|
$attrType = isset($matches[1]) ? $matches[1] : null; |
122
|
|
|
$attrName = isset($matches[2]) ? str_replace('::', ':', $matches[2]) : null; |
123
|
|
|
$prefix = isset($matches[3]) ? $matches[3] : null; |
124
|
|
|
$value = isset($matches[4]) ? $matches[4] : null; |
125
|
|
|
|
126
|
|
|
// Required |
127
|
|
|
if ($attrType === '!') { |
128
|
|
|
$element->attributesRequired[] = $attrName; |
129
|
|
|
$attr->required = true; |
130
|
|
|
} elseif ($attrType === '-') { |
131
|
|
|
// Denied from global |
132
|
|
|
unset($element->attributes[$attrName]); |
133
|
|
|
continue; |
134
|
|
|
} |
135
|
|
|
|
136
|
|
|
// Default value |
137
|
|
|
if ($prefix) { |
138
|
|
|
if ($prefix === '=') { // Default value |
139
|
|
|
$element->attributesDefault[$attrName] = $value; |
140
|
|
|
$attr->defaultValue = $value; |
141
|
|
|
} elseif ($prefix === ':') { |
142
|
|
|
// Forced value |
143
|
|
|
$element->attributesForced[$attrName] = $value; |
144
|
|
|
$attr->forcedValue = $value; |
145
|
|
|
} elseif ($prefix === '<') { |
146
|
|
|
// Required values |
147
|
|
|
$attr->validValues = explode('?', $value); |
148
|
|
|
} |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
// Check for attribute patterns |
152
|
|
|
if (preg_match($hasPatternsRegExp, $attrName)) { |
153
|
|
|
$attr->pattern = $this->patternToRegex($attrName); |
154
|
|
|
$element->attributePatterns[] = $attr; |
155
|
|
|
} else { |
156
|
|
|
$element->attributes[$attrName] = $attr; |
157
|
|
|
} |
158
|
|
|
} |
159
|
|
|
} |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
// Global rule, store away these for later usage |
163
|
|
|
if (!$this->globalAttributes && $elementName == '@') { |
|
|
|
|
164
|
|
|
$this->globalAttributes = $element->attributes; |
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
// Handle substitute elements such as b/strong |
168
|
|
|
if ($outputName) { |
169
|
|
|
$element->outputName = $elementName; |
170
|
|
|
$this->elements[$outputName] = $element; |
171
|
|
|
} |
172
|
|
|
|
173
|
|
|
// Add pattern or exact element |
174
|
|
|
if (preg_match($hasPatternsRegExp, $elementName)) { |
175
|
|
|
$element->pattern = $this->patternToRegex($elementName); |
176
|
|
|
$this->elementPatterns[] = $element; |
177
|
|
|
} else { |
178
|
|
|
$this->elements[$elementName] = $element; |
179
|
|
|
} |
180
|
|
|
} |
181
|
|
|
} |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* Given an element tag, return the rule structure for that element |
186
|
|
|
* @param string $tag The element tag |
187
|
|
|
* @return stdClass The element rule |
188
|
|
|
*/ |
189
|
|
|
protected function getRuleForElement($tag) |
190
|
|
|
{ |
191
|
|
|
if (isset($this->elements[$tag])) { |
192
|
|
|
return $this->elements[$tag]; |
193
|
|
|
} |
194
|
|
|
foreach ($this->elementPatterns as $element) { |
195
|
|
|
if (preg_match($element->pattern, $tag)) { |
196
|
|
|
return $element; |
197
|
|
|
} |
198
|
|
|
} |
199
|
|
|
return null; |
200
|
|
|
} |
201
|
|
|
|
202
|
|
|
/** |
203
|
|
|
* Given an attribute name, return the rule structure for that attribute |
204
|
|
|
* |
205
|
|
|
* @param object $elementRule |
206
|
|
|
* @param string $name The attribute name |
207
|
|
|
* @return stdClass The attribute rule |
208
|
|
|
*/ |
209
|
|
|
protected function getRuleForAttribute($elementRule, $name) |
210
|
|
|
{ |
211
|
|
|
if (isset($elementRule->attributes[$name])) { |
212
|
|
|
return $elementRule->attributes[$name]; |
213
|
|
|
} |
214
|
|
|
foreach ($elementRule->attributePatterns as $attribute) { |
215
|
|
|
if (preg_match($attribute->pattern, $name)) { |
216
|
|
|
return $attribute; |
217
|
|
|
} |
218
|
|
|
} |
219
|
|
|
return null; |
220
|
|
|
} |
221
|
|
|
|
222
|
|
|
/** |
223
|
|
|
* Given a DOMElement and an element rule, check if that element passes the rule |
224
|
|
|
* @param DOMElement $element The element to check |
225
|
|
|
* @param stdClass $rule The rule to check against |
226
|
|
|
* @return bool True if the element passes (and so can be kept), false if it fails (and so needs stripping) |
227
|
|
|
*/ |
228
|
|
|
protected function elementMatchesRule($element, $rule = null) |
229
|
|
|
{ |
230
|
|
|
// If the rule doesn't exist at all, the element isn't allowed |
231
|
|
|
if (!$rule) { |
232
|
|
|
return false; |
233
|
|
|
} |
234
|
|
|
|
235
|
|
|
// If the rule has attributes required, check them to see if this element has at least one |
236
|
|
|
if ($rule->attributesRequired) { |
237
|
|
|
$hasMatch = false; |
238
|
|
|
|
239
|
|
|
foreach ($rule->attributesRequired as $attr) { |
240
|
|
|
if ($element->getAttribute($attr)) { |
241
|
|
|
$hasMatch = true; |
242
|
|
|
break; |
243
|
|
|
} |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
if (!$hasMatch) { |
247
|
|
|
return false; |
248
|
|
|
} |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
// If the rule says to remove empty elements, and this element is empty, remove it |
252
|
|
|
if ($rule->removeEmpty && !$element->firstChild) { |
253
|
|
|
return false; |
254
|
|
|
} |
255
|
|
|
|
256
|
|
|
// No further tests required, element passes |
257
|
|
|
return true; |
258
|
|
|
} |
259
|
|
|
|
260
|
|
|
/** |
261
|
|
|
* Given a DOMAttr and an attribute rule, check if that attribute passes the rule |
262
|
|
|
* @param DOMAttr $attr - the attribute to check |
263
|
|
|
* @param stdClass $rule - the rule to check against |
264
|
|
|
* @return bool - true if the attribute passes (and so can be kept), false if it fails (and so needs stripping) |
265
|
|
|
*/ |
266
|
|
|
protected function attributeMatchesRule($attr, $rule = null) |
267
|
|
|
{ |
268
|
|
|
// If the rule doesn't exist at all, the attribute isn't allowed |
269
|
|
|
if (!$rule) { |
270
|
|
|
return false; |
271
|
|
|
} |
272
|
|
|
|
273
|
|
|
// If the rule has a set of valid values, check them to see if this attribute is one |
274
|
|
|
if (isset($rule->validValues) && !in_array($attr->value, $rule->validValues)) { |
275
|
|
|
return false; |
276
|
|
|
} |
277
|
|
|
|
278
|
|
|
// No further tests required, attribute passes |
279
|
|
|
return true; |
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
/** |
283
|
|
|
* Given an SS_HTMLValue instance, will remove and elements and attributes that are |
284
|
|
|
* not explicitly included in the whitelist passed to __construct on instance creation |
285
|
|
|
* |
286
|
|
|
* @param HTMLValue $html - The HTMLValue to remove any non-whitelisted elements & attributes from |
287
|
|
|
*/ |
288
|
|
|
public function sanitise(HTMLValue $html) |
289
|
|
|
{ |
290
|
|
|
if (!$this->elements && !$this->elementPatterns) { |
|
|
|
|
291
|
|
|
return; |
292
|
|
|
} |
293
|
|
|
|
294
|
|
|
$linkRelValue = $this->config()->get('link_rel_value'); |
295
|
|
|
$doc = $html->getDocument(); |
296
|
|
|
|
297
|
|
|
/** @var DOMElement $el */ |
298
|
|
|
foreach ($html->query('//body//*') as $el) { |
299
|
|
|
$elementRule = $this->getRuleForElement($el->tagName); |
300
|
|
|
|
301
|
|
|
// If this element isn't allowed, strip it |
302
|
|
|
if (!$this->elementMatchesRule($el, $elementRule)) { |
303
|
|
|
// If it's a script or style, we don't keep contents |
304
|
|
|
if ($el->tagName === 'script' || $el->tagName === 'style') { |
305
|
|
|
$el->parentNode->removeChild($el); |
306
|
|
|
} else { |
307
|
|
|
// Otherwise we replace this node with all it's children |
308
|
|
|
// First, create a new fragment with all of $el's children moved into it |
309
|
|
|
$frag = $doc->createDocumentFragment(); |
310
|
|
|
while ($el->firstChild) { |
311
|
|
|
$frag->appendChild($el->firstChild); |
312
|
|
|
} |
313
|
|
|
|
314
|
|
|
// Then replace $el with the frags contents (which used to be it's children) |
315
|
|
|
$el->parentNode->replaceChild($frag, $el); |
316
|
|
|
} |
317
|
|
|
} else { |
318
|
|
|
// Otherwise tidy the element |
319
|
|
|
// First, if we're supposed to pad & this element is empty, fix that |
320
|
|
|
if ($elementRule->paddEmpty && !$el->firstChild) { |
321
|
|
|
$el->nodeValue = ' '; |
322
|
|
|
} |
323
|
|
|
|
324
|
|
|
// Then filter out any non-whitelisted attributes |
325
|
|
|
$children = $el->attributes; |
326
|
|
|
$i = $children->length; |
327
|
|
|
while ($i--) { |
328
|
|
|
$attr = $children->item($i); |
329
|
|
|
$attributeRule = $this->getRuleForAttribute($elementRule, $attr->name); |
330
|
|
|
|
331
|
|
|
// If this attribute isn't allowed, strip it |
332
|
|
|
if (!$this->attributeMatchesRule($attr, $attributeRule)) { |
333
|
|
|
$el->removeAttributeNode($attr); |
334
|
|
|
} |
335
|
|
|
} |
336
|
|
|
|
337
|
|
|
// Then enforce any default attributes |
338
|
|
|
foreach ($elementRule->attributesDefault as $attr => $default) { |
339
|
|
|
if (!$el->getAttribute($attr)) { |
340
|
|
|
$el->setAttribute($attr, $default); |
341
|
|
|
} |
342
|
|
|
} |
343
|
|
|
|
344
|
|
|
// And any forced attributes |
345
|
|
|
foreach ($elementRule->attributesForced as $attr => $forced) { |
346
|
|
|
$el->setAttribute($attr, $forced); |
347
|
|
|
} |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
if ($el->tagName === 'a' && $linkRelValue !== null) { |
351
|
|
|
$this->addRelValue($el, $linkRelValue); |
352
|
|
|
} |
353
|
|
|
} |
354
|
|
|
} |
355
|
|
|
|
356
|
|
|
/** |
357
|
|
|
* Adds rel="noopener noreferrer" to link elements with a target attribute |
358
|
|
|
* |
359
|
|
|
* @param DOMElement $el |
360
|
|
|
* @param string|null $linkRelValue |
361
|
|
|
*/ |
362
|
|
|
private function addRelValue(DOMElement $el, $linkRelValue) |
363
|
|
|
{ |
364
|
|
|
// user has checked the checkbox 'open link in new window' |
365
|
|
|
if ($el->getAttribute('target') && $el->getAttribute('rel') !== $linkRelValue) { |
366
|
|
|
if ($linkRelValue !== '') { |
367
|
|
|
$el->setAttribute('rel', $linkRelValue); |
368
|
|
|
} else { |
369
|
|
|
$el->removeAttribute('rel'); |
370
|
|
|
} |
371
|
|
|
} elseif ($el->getAttribute('rel') === $linkRelValue && !$el->getAttribute('target')) { |
372
|
|
|
// user previously checked 'open link in new window' and noopener was added, |
373
|
|
|
// now user has unchecked the checkbox so we can remove noopener |
374
|
|
|
$el->removeAttribute('rel'); |
375
|
|
|
} |
376
|
|
|
} |
377
|
|
|
} |
378
|
|
|
|