dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Remove comments literally documenting unit tests being unit tests"
[lhc/web/wiklou.git] / includes / parser / Sanitizer.php
1 <?php
2 /**
3 * HTML sanitizer for %MediaWiki.
4 *
5 * Copyright © 2002-2005 Brion Vibber <brion@pobox.com> et al
6 * https://www.mediawiki.org/
7 *
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2 of the License, or
11 * (at your option) any later version.
12 *
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License along
19 * with this program; if not, write to the Free Software Foundation, Inc.,
20 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
21 * http://www.gnu.org/copyleft/gpl.html
22 *
23 * @file
24 * @ingroup Parser
25 */
26
27 use MediaWiki\MediaWikiServices;
28
29 /**
30 * HTML sanitizer for MediaWiki
31 * @ingroup Parser
32 */
33 class Sanitizer {
34 /**
35 * Regular expression to match various types of character references in
36 * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
37 */
38 const CHAR_REFS_REGEX =
39 '/&([A-Za-z0-9\x80-\xff]+);
40 |&\#([0-9]+);
41 |&\#[xX]([0-9A-Fa-f]+);
42 |(&)/x';
43
44 /**
45 * Acceptable tag name charset from HTML5 parsing spec
46 * https://www.w3.org/TR/html5/syntax.html#tag-open-state
47 */
48 const ELEMENT_BITS_REGEX = '!^(/?)([A-Za-z][^\t\n\v />\0]*+)([^>]*?)(/?>)([^<]*)$!';
49
50 /**
51 * Blacklist for evil uris like javascript:
52 * WARNING: DO NOT use this in any place that actually requires blacklisting
53 * for security reasons. There are NUMEROUS[1] ways to bypass blacklisting, the
54 * only way to be secure from javascript: uri based xss vectors is to whitelist
55 * things that you know are safe and deny everything else.
56 * [1]: http://ha.ckers.org/xss.html
57 */
58 const EVIL_URI_PATTERN = '!(^|\s|\*/\s*)(javascript|vbscript)([^\w]|$)!i';
59 const XMLNS_ATTRIBUTE_PATTERN = "/^xmlns:[:A-Z_a-z-.0-9]+$/";
60
61 /**
62 * Tells escapeUrlForHtml() to encode the ID using the wiki's primary encoding.
63 *
64 * @since 1.30
65 */
66 const ID_PRIMARY = 0;
67
68 /**
69 * Tells escapeUrlForHtml() to encode the ID using the fallback encoding, or return false
70 * if no fallback is configured.
71 *
72 * @since 1.30
73 */
74 const ID_FALLBACK = 1;
75
76 /**
77 * List of all named character entities defined in HTML 4.01
78 * https://www.w3.org/TR/html4/sgml/entities.html
79 * As well as &apos; which is only defined starting in XHTML1.
80 */
81 private static $htmlEntities = [
82 'Aacute' => 193,
83 'aacute' => 225,
84 'Acirc' => 194,
85 'acirc' => 226,
86 'acute' => 180,
87 'AElig' => 198,
88 'aelig' => 230,
89 'Agrave' => 192,
90 'agrave' => 224,
91 'alefsym' => 8501,
92 'Alpha' => 913,
93 'alpha' => 945,
94 'amp' => 38,
95 'and' => 8743,
96 'ang' => 8736,
97 'apos' => 39, // New in XHTML & HTML 5; avoid in output for compatibility with IE.
98 'Aring' => 197,
99 'aring' => 229,
100 'asymp' => 8776,
101 'Atilde' => 195,
102 'atilde' => 227,
103 'Auml' => 196,
104 'auml' => 228,
105 'bdquo' => 8222,
106 'Beta' => 914,
107 'beta' => 946,
108 'brvbar' => 166,
109 'bull' => 8226,
110 'cap' => 8745,
111 'Ccedil' => 199,
112 'ccedil' => 231,
113 'cedil' => 184,
114 'cent' => 162,
115 'Chi' => 935,
116 'chi' => 967,
117 'circ' => 710,
118 'clubs' => 9827,
119 'cong' => 8773,
120 'copy' => 169,
121 'crarr' => 8629,
122 'cup' => 8746,
123 'curren' => 164,
124 'dagger' => 8224,
125 'Dagger' => 8225,
126 'darr' => 8595,
127 'dArr' => 8659,
128 'deg' => 176,
129 'Delta' => 916,
130 'delta' => 948,
131 'diams' => 9830,
132 'divide' => 247,
133 'Eacute' => 201,
134 'eacute' => 233,
135 'Ecirc' => 202,
136 'ecirc' => 234,
137 'Egrave' => 200,
138 'egrave' => 232,
139 'empty' => 8709,
140 'emsp' => 8195,
141 'ensp' => 8194,
142 'Epsilon' => 917,
143 'epsilon' => 949,
144 'equiv' => 8801,
145 'Eta' => 919,
146 'eta' => 951,
147 'ETH' => 208,
148 'eth' => 240,
149 'Euml' => 203,
150 'euml' => 235,
151 'euro' => 8364,
152 'exist' => 8707,
153 'fnof' => 402,
154 'forall' => 8704,
155 'frac12' => 189,
156 'frac14' => 188,
157 'frac34' => 190,
158 'frasl' => 8260,
159 'Gamma' => 915,
160 'gamma' => 947,
161 'ge' => 8805,
162 'gt' => 62,
163 'harr' => 8596,
164 'hArr' => 8660,
165 'hearts' => 9829,
166 'hellip' => 8230,
167 'Iacute' => 205,
168 'iacute' => 237,
169 'Icirc' => 206,
170 'icirc' => 238,
171 'iexcl' => 161,
172 'Igrave' => 204,
173 'igrave' => 236,
174 'image' => 8465,
175 'infin' => 8734,
176 'int' => 8747,
177 'Iota' => 921,
178 'iota' => 953,
179 'iquest' => 191,
180 'isin' => 8712,
181 'Iuml' => 207,
182 'iuml' => 239,
183 'Kappa' => 922,
184 'kappa' => 954,
185 'Lambda' => 923,
186 'lambda' => 955,
187 'lang' => 9001,
188 'laquo' => 171,
189 'larr' => 8592,
190 'lArr' => 8656,
191 'lceil' => 8968,
192 'ldquo' => 8220,
193 'le' => 8804,
194 'lfloor' => 8970,
195 'lowast' => 8727,
196 'loz' => 9674,
197 'lrm' => 8206,
198 'lsaquo' => 8249,
199 'lsquo' => 8216,
200 'lt' => 60,
201 'macr' => 175,
202 'mdash' => 8212,
203 'micro' => 181,
204 'middot' => 183,
205 'minus' => 8722,
206 'Mu' => 924,
207 'mu' => 956,
208 'nabla' => 8711,
209 'nbsp' => 160,
210 'ndash' => 8211,
211 'ne' => 8800,
212 'ni' => 8715,
213 'not' => 172,
214 'notin' => 8713,
215 'nsub' => 8836,
216 'Ntilde' => 209,
217 'ntilde' => 241,
218 'Nu' => 925,
219 'nu' => 957,
220 'Oacute' => 211,
221 'oacute' => 243,
222 'Ocirc' => 212,
223 'ocirc' => 244,
224 'OElig' => 338,
225 'oelig' => 339,
226 'Ograve' => 210,
227 'ograve' => 242,
228 'oline' => 8254,
229 'Omega' => 937,
230 'omega' => 969,
231 'Omicron' => 927,
232 'omicron' => 959,
233 'oplus' => 8853,
234 'or' => 8744,
235 'ordf' => 170,
236 'ordm' => 186,
237 'Oslash' => 216,
238 'oslash' => 248,
239 'Otilde' => 213,
240 'otilde' => 245,
241 'otimes' => 8855,
242 'Ouml' => 214,
243 'ouml' => 246,
244 'para' => 182,
245 'part' => 8706,
246 'permil' => 8240,
247 'perp' => 8869,
248 'Phi' => 934,
249 'phi' => 966,
250 'Pi' => 928,
251 'pi' => 960,
252 'piv' => 982,
253 'plusmn' => 177,
254 'pound' => 163,
255 'prime' => 8242,
256 'Prime' => 8243,
257 'prod' => 8719,
258 'prop' => 8733,
259 'Psi' => 936,
260 'psi' => 968,
261 'quot' => 34,
262 'radic' => 8730,
263 'rang' => 9002,
264 'raquo' => 187,
265 'rarr' => 8594,
266 'rArr' => 8658,
267 'rceil' => 8969,
268 'rdquo' => 8221,
269 'real' => 8476,
270 'reg' => 174,
271 'rfloor' => 8971,
272 'Rho' => 929,
273 'rho' => 961,
274 'rlm' => 8207,
275 'rsaquo' => 8250,
276 'rsquo' => 8217,
277 'sbquo' => 8218,
278 'Scaron' => 352,
279 'scaron' => 353,
280 'sdot' => 8901,
281 'sect' => 167,
282 'shy' => 173,
283 'Sigma' => 931,
284 'sigma' => 963,
285 'sigmaf' => 962,
286 'sim' => 8764,
287 'spades' => 9824,
288 'sub' => 8834,
289 'sube' => 8838,
290 'sum' => 8721,
291 'sup' => 8835,
292 'sup1' => 185,
293 'sup2' => 178,
294 'sup3' => 179,
295 'supe' => 8839,
296 'szlig' => 223,
297 'Tau' => 932,
298 'tau' => 964,
299 'there4' => 8756,
300 'Theta' => 920,
301 'theta' => 952,
302 'thetasym' => 977,
303 'thinsp' => 8201,
304 'THORN' => 222,
305 'thorn' => 254,
306 'tilde' => 732,
307 'times' => 215,
308 'trade' => 8482,
309 'Uacute' => 218,
310 'uacute' => 250,
311 'uarr' => 8593,
312 'uArr' => 8657,
313 'Ucirc' => 219,
314 'ucirc' => 251,
315 'Ugrave' => 217,
316 'ugrave' => 249,
317 'uml' => 168,
318 'upsih' => 978,
319 'Upsilon' => 933,
320 'upsilon' => 965,
321 'Uuml' => 220,
322 'uuml' => 252,
323 'weierp' => 8472,
324 'Xi' => 926,
325 'xi' => 958,
326 'Yacute' => 221,
327 'yacute' => 253,
328 'yen' => 165,
329 'Yuml' => 376,
330 'yuml' => 255,
331 'Zeta' => 918,
332 'zeta' => 950,
333 'zwj' => 8205,
334 'zwnj' => 8204
335 ];
336
337 /**
338 * Character entity aliases accepted by MediaWiki
339 */
340 private static $htmlEntityAliases = [
341 'רלמ' => 'rlm',
342 'رلم' => 'rlm',
343 ];
344
345 /**
346 * Lazy-initialised attributes regex, see getAttribsRegex()
347 */
348 private static $attribsRegex;
349
350 /**
351 * Regular expression to match HTML/XML attribute pairs within a tag.
352 * Based on https://www.w3.org/TR/html5/syntax.html#before-attribute-name-state
353 * Used in Sanitizer::decodeTagAttributes
354 * @return string
355 */
356 static function getAttribsRegex() {
357 if ( self::$attribsRegex === null ) {
358 $spaceChars = '\x09\x0a\x0c\x0d\x20';
359 $space = "[{$spaceChars}]";
360 $attrib = "[^{$spaceChars}\/>=]";
361 $attribFirst = "(?:{$attrib}|=)";
362 self::$attribsRegex =
363 "/({$attribFirst}{$attrib}*)
364 ($space*=$space*
365 (?:
366 # The attribute value: quoted or alone
367 \"([^\"]*)(?:\"|\$)
368 | '([^']*)(?:'|\$)
369 | (((?!$space|>).)*)
370 )
371 )?/sxu";
372 }
373 return self::$attribsRegex;
374 }
375
376 /**
377 * Lazy-initialised attribute name regex, see getAttribNameRegex()
378 */
379 private static $attribNameRegex;
380
381 /**
382 * Used in Sanitizer::decodeTagAttributes to filter attributes.
383 * @return string
384 */
385 static function getAttribNameRegex() {
386 if ( self::$attribNameRegex === null ) {
387 $attribFirst = "[:_\p{L}\p{N}]";
388 $attrib = "[:_\.\-\p{L}\p{N}]";
389 self::$attribNameRegex = "/^({$attribFirst}{$attrib}*)$/sxu";
390 }
391 return self::$attribNameRegex;
392 }
393
394 /**
395 * Return the various lists of recognized tags
396 * @param array $extratags For any extra tags to include
397 * @param array $removetags For any tags (default or extra) to exclude
398 * @return array
399 */
400 public static function getRecognizedTagData( $extratags = [], $removetags = [] ) {
401 global $wgAllowImageTag;
402
403 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
404 $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
405
406 // Base our staticInitialised variable off of the global config state so that if the globals
407 // are changed (like in the screwed up test system) we will re-initialise the settings.
408 $globalContext = $wgAllowImageTag;
409 if ( !$staticInitialised || $staticInitialised != $globalContext ) {
410 $htmlpairsStatic = [ # Tags that must be closed
411 'b', 'bdi', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
412 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
413 'strike', 'strong', 'tt', 'var', 'div', 'center',
414 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
415 'ruby', 'rb', 'rp', 'rt', 'rtc', 'p', 'span', 'abbr', 'dfn',
416 'kbd', 'samp', 'data', 'time', 'mark'
417 ];
418 $htmlsingle = [
419 'br', 'wbr', 'hr', 'li', 'dt', 'dd', 'meta', 'link'
420 ];
421
422 # Elements that cannot have close tags. This is (not coincidentally)
423 # also the list of tags for which the HTML 5 parsing algorithm
424 # requires you to "acknowledge the token's self-closing flag", i.e.
425 # a self-closing tag like <br/> is not an HTML 5 parse error only
426 # for this list.
427 $htmlsingleonly = [
428 'br', 'wbr', 'hr', 'meta', 'link'
429 ];
430
431 $htmlnest = [ # Tags that can be nested--??
432 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
433 'li', 'dl', 'dt', 'dd', 'font', 'big', 'small', 'sub', 'sup', 'span',
434 'var', 'kbd', 'samp', 'em', 'strong', 'q', 'ruby', 'bdo'
435 ];
436 $tabletags = [ # Can only appear inside table, we will close them
437 'td', 'th', 'tr',
438 ];
439 $htmllist = [ # Tags used by list
440 'ul', 'ol',
441 ];
442 $listtags = [ # Tags that can appear in a list
443 'li',
444 ];
445
446 if ( $wgAllowImageTag ) {
447 $htmlsingle[] = 'img';
448 $htmlsingleonly[] = 'img';
449 }
450
451 $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
452 $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
453
454 # Convert them all to hashtables for faster lookup
455 $vars = [ 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
456 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' ];
457 foreach ( $vars as $var ) {
458 $$var = array_flip( $$var );
459 }
460 $staticInitialised = $globalContext;
461 }
462
463 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
464 $extratags = array_flip( $extratags );
465 $removetags = array_flip( $removetags );
466 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
467 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ), $removetags );
468
469 return [
470 'htmlpairs' => $htmlpairs,
471 'htmlsingle' => $htmlsingle,
472 'htmlsingleonly' => $htmlsingleonly,
473 'htmlnest' => $htmlnest,
474 'tabletags' => $tabletags,
475 'htmllist' => $htmllist,
476 'listtags' => $listtags,
477 'htmlsingleallowed' => $htmlsingleallowed,
478 'htmlelements' => $htmlelements,
479 ];
480 }
481
482 /**
483 * Cleans up HTML, removes dangerous tags and attributes, and
484 * removes HTML comments
485 * @param string $text
486 * @param callable|null $processCallback Callback to do any variable or parameter
487 * replacements in HTML attribute values
488 * @param array|bool $args Arguments for the processing callback
489 * @param array $extratags For any extra tags to include
490 * @param array $removetags For any tags (default or extra) to exclude
491 * @param callable|null $warnCallback (Deprecated) Callback allowing the
492 * addition of a tracking category when bad input is encountered.
493 * DO NOT ADD NEW PARAMETERS AFTER $warnCallback, since it will be
494 * removed shortly.
495 * @return string
496 */
497 public static function removeHTMLtags( $text, $processCallback = null,
498 $args = [], $extratags = [], $removetags = [], $warnCallback = null
499 ) {
500 $tagData = self::getRecognizedTagData( $extratags, $removetags );
501 $htmlpairs = $tagData['htmlpairs'];
502 $htmlsingle = $tagData['htmlsingle'];
503 $htmlsingleonly = $tagData['htmlsingleonly'];
504 $htmlnest = $tagData['htmlnest'];
505 $tabletags = $tagData['tabletags'];
506 $htmllist = $tagData['htmllist'];
507 $listtags = $tagData['listtags'];
508 $htmlsingleallowed = $tagData['htmlsingleallowed'];
509 $htmlelements = $tagData['htmlelements'];
510
511 # Remove HTML comments
512 $text = self::removeHTMLcomments( $text );
513 $bits = explode( '<', $text );
514 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
515 if ( !MWTidy::isEnabled() ) {
516 wfDeprecated( 'disabling tidy', '1.33' );
517 $tagstack = $tablestack = [];
518 foreach ( $bits as $x ) {
519 $regs = [];
520 # $slash: Does the current element start with a '/'?
521 # $t: Current element name
522 # $params: String between element name and >
523 # $brace: Ending '>' or '/>'
524 # $rest: Everything until the next element of $bits
525 if ( preg_match( self::ELEMENT_BITS_REGEX, $x, $regs ) ) {
526 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
527 } else {
528 $slash = $t = $params = $brace = $rest = null;
529 }
530
531 $badtag = false;
532 $t = strtolower( $t );
533 if ( isset( $htmlelements[$t] ) ) {
534 # Check our stack
535 if ( $slash && isset( $htmlsingleonly[$t] ) ) {
536 $badtag = true;
537 } elseif ( $slash ) {
538 # Closing a tag... is it the one we just opened?
539 Wikimedia\suppressWarnings();
540 $ot = array_pop( $tagstack );
541 Wikimedia\restoreWarnings();
542
543 if ( $ot != $t ) {
544 if ( isset( $htmlsingleallowed[$ot] ) ) {
545 # Pop all elements with an optional close tag
546 # and see if we find a match below them
547 $optstack = [];
548 array_push( $optstack, $ot );
549 Wikimedia\suppressWarnings();
550 $ot = array_pop( $tagstack );
551 Wikimedia\restoreWarnings();
552 while ( $ot != $t && isset( $htmlsingleallowed[$ot] ) ) {
553 array_push( $optstack, $ot );
554 Wikimedia\suppressWarnings();
555 $ot = array_pop( $tagstack );
556 Wikimedia\restoreWarnings();
557 }
558 if ( $t != $ot ) {
559 # No match. Push the optional elements back again
560 $badtag = true;
561 Wikimedia\suppressWarnings();
562 $ot = array_pop( $optstack );
563 Wikimedia\restoreWarnings();
564 while ( $ot ) {
565 array_push( $tagstack, $ot );
566 Wikimedia\suppressWarnings();
567 $ot = array_pop( $optstack );
568 Wikimedia\restoreWarnings();
569 }
570 }
571 } else {
572 Wikimedia\suppressWarnings();
573 array_push( $tagstack, $ot );
574 Wikimedia\restoreWarnings();
575
576 # <li> can be nested in <ul> or <ol>, skip those cases:
577 if ( !isset( $htmllist[$ot] ) || !isset( $listtags[$t] ) ) {
578 $badtag = true;
579 }
580 }
581 } else {
582 if ( $t == 'table' ) {
583 $tagstack = array_pop( $tablestack );
584 }
585 }
586 $newparams = '';
587 } else {
588 # Keep track for later
589 if ( isset( $tabletags[$t] ) && !in_array( 'table', $tagstack ) ) {
590 $badtag = true;
591 } elseif ( in_array( $t, $tagstack ) && !isset( $htmlnest[$t] ) ) {
592 $badtag = true;
593 #  Is it a self closed htmlpair ? (T7487)
594 } elseif ( $brace == '/>' && isset( $htmlpairs[$t] ) ) {
595 // Eventually we'll just remove the self-closing
596 // slash, in order to be consistent with HTML5
597 // semantics.
598 // $brace = '>';
599 // For now, let's just warn authors to clean up.
600 if ( is_callable( $warnCallback ) ) {
601 call_user_func_array( $warnCallback, [ 'deprecated-self-close-category' ] );
602 }
603 $badtag = true;
604 } elseif ( isset( $htmlsingleonly[$t] ) ) {
605 # Hack to force empty tag for unclosable elements
606 $brace = '/>';
607 } elseif ( isset( $htmlsingle[$t] ) ) {
608 # Hack to not close $htmlsingle tags
609 $brace = null;
610 # Still need to push this optionally-closed tag to
611 # the tag stack so that we can match end tags
612 # instead of marking them as bad.
613 array_push( $tagstack, $t );
614 } elseif ( isset( $tabletags[$t] ) && in_array( $t, $tagstack ) ) {
615 // New table tag but forgot to close the previous one
616 $text .= "</$t>";
617 } else {
618 if ( $t == 'table' ) {
619 array_push( $tablestack, $tagstack );
620 $tagstack = [];
621 }
622 array_push( $tagstack, $t );
623 }
624
625 # Replace any variables or template parameters with
626 # plaintext results.
627 if ( is_callable( $processCallback ) ) {
628 call_user_func_array( $processCallback, [ &$params, $args ] );
629 }
630
631 if ( !self::validateTag( $params, $t ) ) {
632 $badtag = true;
633 }
634
635 # Strip non-approved attributes from the tag
636 $newparams = self::fixTagAttributes( $params, $t );
637 }
638 if ( !$badtag ) {
639 $rest = str_replace( '>', '&gt;', $rest );
640 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
641 $text .= "<$slash$t$newparams$close>$rest";
642 continue;
643 }
644 }
645 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
646 }
647 # Close off any remaining tags
648 while ( is_array( $tagstack ) && ( $t = array_pop( $tagstack ) ) ) {
649 $text .= "</$t>\n";
650 if ( $t == 'table' ) {
651 $tagstack = array_pop( $tablestack );
652 }
653 }
654 } else {
655 # this might be possible using tidy itself
656 foreach ( $bits as $x ) {
657 if ( preg_match( self::ELEMENT_BITS_REGEX, $x, $regs ) ) {
658 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
659
660 $badtag = false;
661 $t = strtolower( $t );
662 if ( isset( $htmlelements[$t] ) ) {
663 if ( is_callable( $processCallback ) ) {
664 call_user_func_array( $processCallback, [ &$params, $args ] );
665 }
666
667 if ( $brace == '/>' && !( isset( $htmlsingle[$t] ) || isset( $htmlsingleonly[$t] ) ) ) {
668 // Eventually we'll just remove the self-closing
669 // slash, in order to be consistent with HTML5
670 // semantics.
671 // $brace = '>';
672 // For now, let's just warn authors to clean up.
673 if ( is_callable( $warnCallback ) ) {
674 call_user_func_array( $warnCallback, [ 'deprecated-self-close-category' ] );
675 }
676 }
677 if ( !self::validateTag( $params, $t ) ) {
678 $badtag = true;
679 }
680
681 $newparams = self::fixTagAttributes( $params, $t );
682 if ( !$badtag ) {
683 if ( $brace === '/>' && !isset( $htmlsingleonly[$t] ) ) {
684 # Interpret self-closing tags as empty tags even when
685 # HTML 5 would interpret them as start tags. Such input
686 # is commonly seen on Wikimedia wikis with this intention.
687 $brace = "></$t>";
688 }
689
690 $rest = str_replace( '>', '&gt;', $rest );
691 $text .= "<$slash$t$newparams$brace$rest";
692 continue;
693 }
694 }
695 }
696 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
697 }
698 }
699 return $text;
700 }
701
702 /**
703 * Remove '<!--', '-->', and everything between.
704 * To avoid leaving blank lines, when a comment is both preceded
705 * and followed by a newline (ignoring spaces), trim leading and
706 * trailing spaces and one of the newlines.
707 *
708 * @param string $text
709 * @return string
710 */
711 public static function removeHTMLcomments( $text ) {
712 while ( ( $start = strpos( $text, '<!--' ) ) !== false ) {
713 $end = strpos( $text, '-->', $start + 4 );
714 if ( $end === false ) {
715 # Unterminated comment; bail out
716 break;
717 }
718
719 $end += 3;
720
721 # Trim space and newline if the comment is both
722 # preceded and followed by a newline
723 $spaceStart = max( $start - 1, 0 );
724 $spaceLen = $end - $spaceStart;
725 while ( substr( $text, $spaceStart, 1 ) === ' ' && $spaceStart > 0 ) {
726 $spaceStart--;
727 $spaceLen++;
728 }
729 while ( substr( $text, $spaceStart + $spaceLen, 1 ) === ' ' ) {
730 $spaceLen++;
731 }
732 if ( substr( $text, $spaceStart, 1 ) === "\n"
733 && substr( $text, $spaceStart + $spaceLen, 1 ) === "\n" ) {
734 # Remove the comment, leading and trailing
735 # spaces, and leave only one newline.
736 $text = substr_replace( $text, "\n", $spaceStart, $spaceLen + 1 );
737 } else {
738 # Remove just the comment.
739 $text = substr_replace( $text, '', $start, $end - $start );
740 }
741 }
742 return $text;
743 }
744
745 /**
746 * Takes attribute names and values for a tag and the tag name and
747 * validates that the tag is allowed to be present.
748 * This DOES NOT validate the attributes, nor does it validate the
749 * tags themselves. This method only handles the special circumstances
750 * where we may want to allow a tag within content but ONLY when it has
751 * specific attributes set.
752 *
753 * @param string $params
754 * @param string $element
755 * @return bool
756 */
757 static function validateTag( $params, $element ) {
758 $params = self::decodeTagAttributes( $params );
759
760 if ( $element == 'meta' || $element == 'link' ) {
761 if ( !isset( $params['itemprop'] ) ) {
762 // <meta> and <link> must have an itemprop="" otherwise they are not valid or safe in content
763 return false;
764 }
765 if ( $element == 'meta' && !isset( $params['content'] ) ) {
766 // <meta> must have a content="" for the itemprop
767 return false;
768 }
769 if ( $element == 'link' && !isset( $params['href'] ) ) {
770 // <link> must have an associated href=""
771 return false;
772 }
773 }
774
775 return true;
776 }
777
778 /**
779 * Take an array of attribute names and values and normalize or discard
780 * illegal values for the given element type.
781 *
782 * - Discards attributes not on a whitelist for the given element
783 * - Unsafe style attributes are discarded
784 * - Invalid id attributes are re-encoded
785 *
786 * @param array $attribs
787 * @param string $element
788 * @return array
789 *
790 * @todo Check for legal values where the DTD limits things.
791 * @todo Check for unique id attribute :P
792 */
793 static function validateTagAttributes( $attribs, $element ) {
794 return self::validateAttributes( $attribs,
795 self::attributeWhitelist( $element ) );
796 }
797
798 /**
799 * Take an array of attribute names and values and normalize or discard
800 * illegal values for the given whitelist.
801 *
802 * - Discards attributes not on the given whitelist
803 * - Unsafe style attributes are discarded
804 * - Invalid id attributes are re-encoded
805 *
806 * @param array $attribs
807 * @param array $whitelist List of allowed attribute names
808 * @return array
809 *
810 * @todo Check for legal values where the DTD limits things.
811 * @todo Check for unique id attribute :P
812 */
813 static function validateAttributes( $attribs, $whitelist ) {
814 $whitelist = array_flip( $whitelist );
815 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
816
817 $out = [];
818 foreach ( $attribs as $attribute => $value ) {
819 # Allow XML namespace declaration to allow RDFa
820 if ( preg_match( self::XMLNS_ATTRIBUTE_PATTERN, $attribute ) ) {
821 if ( !preg_match( self::EVIL_URI_PATTERN, $value ) ) {
822 $out[$attribute] = $value;
823 }
824
825 continue;
826 }
827
828 # Allow any attribute beginning with "data-"
829 # However:
830 # * Disallow data attributes used by MediaWiki code
831 # * Ensure that the attribute is not namespaced by banning
832 # colons.
833 if ( !preg_match( '/^data-[^:]*$/i', $attribute )
834 && !isset( $whitelist[$attribute] )
835 || self::isReservedDataAttribute( $attribute )
836 ) {
837 continue;
838 }
839
840 # Strip javascript "expression" from stylesheets.
841 # https://msdn.microsoft.com/en-us/library/ms537634.aspx
842 if ( $attribute == 'style' ) {
843 $value = self::checkCss( $value );
844 }
845
846 # Escape HTML id attributes
847 if ( $attribute === 'id' ) {
848 $value = self::escapeIdForAttribute( $value, self::ID_PRIMARY );
849 }
850
851 # Escape HTML id reference lists
852 if ( $attribute === 'aria-describedby'
853 || $attribute === 'aria-flowto'
854 || $attribute === 'aria-labelledby'
855 || $attribute === 'aria-owns'
856 ) {
857 $value = self::escapeIdReferenceList( $value );
858 }
859
860 // RDFa and microdata properties allow URLs, URIs and/or CURIs.
861 // Check them for sanity.
862 if ( $attribute === 'rel' || $attribute === 'rev'
863 # RDFa
864 || $attribute === 'about' || $attribute === 'property'
865 || $attribute === 'resource' || $attribute === 'datatype'
866 || $attribute === 'typeof'
867 # HTML5 microdata
868 || $attribute === 'itemid' || $attribute === 'itemprop'
869 || $attribute === 'itemref' || $attribute === 'itemscope'
870 || $attribute === 'itemtype'
871 ) {
872 // Paranoia. Allow "simple" values but suppress javascript
873 if ( preg_match( self::EVIL_URI_PATTERN, $value ) ) {
874 continue;
875 }
876 }
877
878 # NOTE: even though elements using href/src are not allowed directly, supply
879 # validation code that can be used by tag hook handlers, etc
880 if ( $attribute === 'href' || $attribute === 'src' || $attribute === 'poster' ) {
881 if ( !preg_match( $hrefExp, $value ) ) {
882 continue; // drop any href or src attributes not using an allowed protocol.
883 // NOTE: this also drops all relative URLs
884 }
885 }
886
887 // If this attribute was previously set, override it.
888 // Output should only have one attribute of each name.
889 $out[$attribute] = $value;
890 }
891
892 # itemtype, itemid, itemref don't make sense without itemscope
893 if ( !array_key_exists( 'itemscope', $out ) ) {
894 unset( $out['itemtype'] );
895 unset( $out['itemid'] );
896 unset( $out['itemref'] );
897 }
898 # TODO: Strip itemprop if we aren't descendants of an itemscope or pointed to by an itemref.
899
900 return $out;
901 }
902
903 /**
904 * Given an attribute name, checks whether it is a reserved data attribute
905 * (such as data-mw-foo) which is unavailable to user-generated HTML so MediaWiki
906 * core and extension code can safely use it to communicate with frontend code.
907 * @param string $attr Attribute name.
908 * @return bool
909 */
910 public static function isReservedDataAttribute( $attr ) {
911 // data-ooui is reserved for ooui.
912 // data-mw and data-parsoid are reserved for parsoid.
913 // data-mw-<name here> is reserved for extensions (or core) if
914 // they need to communicate some data to the client and want to be
915 // sure that it isn't coming from an untrusted user.
916 // We ignore the possibility of namespaces since user-generated HTML
917 // can't use them anymore.
918 return (bool)preg_match( '/^data-(ooui|mw|parsoid)/i', $attr );
919 }
920
921 /**
922 * Merge two sets of HTML attributes. Conflicting items in the second set
923 * will override those in the first, except for 'class' attributes which
924 * will be combined (if they're both strings).
925 *
926 * @todo implement merging for other attributes such as style
927 * @param array $a
928 * @param array $b
929 * @return array
930 */
931 static function mergeAttributes( $a, $b ) {
932 $out = array_merge( $a, $b );
933 if ( isset( $a['class'] ) && isset( $b['class'] )
934 && is_string( $a['class'] ) && is_string( $b['class'] )
935 && $a['class'] !== $b['class']
936 ) {
937 $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
938 -1, PREG_SPLIT_NO_EMPTY );
939 $out['class'] = implode( ' ', array_unique( $classes ) );
940 }
941 return $out;
942 }
943
944 /**
945 * Normalize CSS into a format we can easily search for hostile input
946 * - decode character references
947 * - decode escape sequences
948 * - convert characters that IE6 interprets into ascii
949 * - remove comments, unless the entire value is one single comment
950 * @param string $value the css string
951 * @return string normalized css
952 */
953 public static function normalizeCss( $value ) {
954 // Decode character references like &#123;
955 $value = self::decodeCharReferences( $value );
956
957 // Decode escape sequences and line continuation
958 // See the grammar in the CSS 2 spec, appendix D.
959 // This has to be done AFTER decoding character references.
960 // This means it isn't possible for this function to return
961 // unsanitized escape sequences. It is possible to manufacture
962 // input that contains character references that decode to
963 // escape sequences that decode to character references, but
964 // it's OK for the return value to contain character references
965 // because the caller is supposed to escape those anyway.
966 static $decodeRegex;
967 if ( !$decodeRegex ) {
968 $space = '[\\x20\\t\\r\\n\\f]';
969 $nl = '(?:\\n|\\r\\n|\\r|\\f)';
970 $backslash = '\\\\';
971 $decodeRegex = "/ $backslash
972 (?:
973 ($nl) | # 1. Line continuation
974 ([0-9A-Fa-f]{1,6})$space? | # 2. character number
975 (.) | # 3. backslash cancelling special meaning
976 () | # 4. backslash at end of string
977 )/xu";
978 }
979 $value = preg_replace_callback( $decodeRegex,
980 [ __CLASS__, 'cssDecodeCallback' ], $value );
981
982 // Normalize Halfwidth and Fullwidth Unicode block that IE6 might treat as ascii
983 $value = preg_replace_callback(
984 '/[！-［］-ｚ]/u', // U+FF01 to U+FF5A, excluding U+FF3C (T60088)
985 function ( $matches ) {
986 $cp = UtfNormal\Utils::utf8ToCodepoint( $matches[0] );
987 if ( $cp === false ) {
988 return '';
989 }
990 return chr( $cp - 65248 ); // ASCII range \x21-\x7A
991 },
992 $value
993 );
994
995 // Convert more characters IE6 might treat as ascii
996 // U+0280, U+0274, U+207F, U+029F, U+026A, U+207D, U+208D
997 $value = str_replace(
998 [ 'ʀ', 'ɴ', 'ⁿ', 'ʟ', 'ɪ', '⁽', '₍' ],
999 [ 'r', 'n', 'n', 'l', 'i', '(', '(' ],
1000 $value
1001 );
1002
1003 // Let the value through if it's nothing but a single comment, to
1004 // allow other functions which may reject it to pass some error
1005 // message through.
1006 if ( !preg_match( '! ^ \s* /\* [^*\\/]* \*/ \s* $ !x', $value ) ) {
1007 // Remove any comments; IE gets token splitting wrong
1008 // This must be done AFTER decoding character references and
1009 // escape sequences, because those steps can introduce comments
1010 // This step cannot introduce character references or escape
1011 // sequences, because it replaces comments with spaces rather
1012 // than removing them completely.
1013 $value = StringUtils::delimiterReplace( '/*', '*/', ' ', $value );
1014
1015 // Remove anything after a comment-start token, to guard against
1016 // incorrect client implementations.
1017 $commentPos = strpos( $value, '/*' );
1018 if ( $commentPos !== false ) {
1019 $value = substr( $value, 0, $commentPos );
1020 }
1021 }
1022
1023 // S followed by repeat, iteration, or prolonged sound marks,
1024 // which IE will treat as "ss"
1025 $value = preg_replace(
1026 '/s(?:
1027 \xE3\x80\xB1 | # U+3031
1028 \xE3\x82\x9D | # U+309D
1029 \xE3\x83\xBC | # U+30FC
1030 \xE3\x83\xBD | # U+30FD
1031 \xEF\xB9\xBC | # U+FE7C
1032 \xEF\xB9\xBD | # U+FE7D
1033 \xEF\xBD\xB0 # U+FF70
1034 )/ix',
1035 'ss',
1036 $value
1037 );
1038
1039 return $value;
1040 }
1041
1042 /**
1043 * Pick apart some CSS and check it for forbidden or unsafe structures.
1044 * Returns a sanitized string. This sanitized string will have
1045 * character references and escape sequences decoded and comments
1046 * stripped (unless it is itself one valid comment, in which case the value
1047 * will be passed through). If the input is just too evil, only a comment
1048 * complaining about evilness will be returned.
1049 *
1050 * Currently URL references, 'expression', 'tps' are forbidden.
1051 *
1052 * NOTE: Despite the fact that character references are decoded, the
1053 * returned string may contain character references given certain
1054 * clever input strings. These character references must
1055 * be escaped before the return value is embedded in HTML.
1056 *
1057 * @param string $value
1058 * @return string
1059 */
1060 static function checkCss( $value ) {
1061 $value = self::normalizeCss( $value );
1062
1063 // Reject problematic keywords and control characters
1064 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ||
1065 strpos( $value, UtfNormal\Constants::UTF8_REPLACEMENT ) !== false ) {
1066 return '/* invalid control char */';
1067 } elseif ( preg_match(
1068 '! expression
1069 | filter\s*:
1070 | accelerator\s*:
1071 | -o-link\s*:
1072 | -o-link-source\s*:
1073 | -o-replace\s*:
1074 | url\s*\(
1075 | image\s*\(
1076 | image-set\s*\(
1077 | attr\s*\([^)]+[\s,]+url
1078 !ix', $value ) ) {
1079 return '/* insecure input */';
1080 }
1081 return $value;
1082 }
1083
1084 /**
1085 * @param array $matches
1086 * @return string
1087 */
1088 static function cssDecodeCallback( $matches ) {
1089 if ( $matches[1] !== '' ) {
1090 // Line continuation
1091 return '';
1092 } elseif ( $matches[2] !== '' ) {
1093 $char = UtfNormal\Utils::codepointToUtf8( hexdec( $matches[2] ) );
1094 } elseif ( $matches[3] !== '' ) {
1095 $char = $matches[3];
1096 } else {
1097 $char = '\\';
1098 }
1099 if ( $char == "\n" || $char == '"' || $char == "'" || $char == '\\' ) {
1100 // These characters need to be escaped in strings
1101 // Clean up the escape sequence to avoid parsing errors by clients
1102 return '\\' . dechex( ord( $char ) ) . ' ';
1103 } else {
1104 // Decode unnecessary escape
1105 return $char;
1106 }
1107 }
1108
1109 /**
1110 * Take a tag soup fragment listing an HTML element's attributes
1111 * and normalize it to well-formed XML, discarding unwanted attributes.
1112 * Output is safe for further wikitext processing, with escaping of
1113 * values that could trigger problems.
1114 *
1115 * - Normalizes attribute names to lowercase
1116 * - Discards attributes not on a whitelist for the given element
1117 * - Turns broken or invalid entities into plaintext
1118 * - Double-quotes all attribute values
1119 * - Attributes without values are given the name as attribute
1120 * - Double attributes are discarded
1121 * - Unsafe style attributes are discarded
1122 * - Prepends space if there are attributes.
1123 * - (Optionally) Sorts attributes by name.
1124 *
1125 * @param string $text
1126 * @param string $element
1127 * @param bool $sorted Whether to sort the attributes (default: false)
1128 * @return string
1129 */
1130 static function fixTagAttributes( $text, $element, $sorted = false ) {
1131 if ( trim( $text ) == '' ) {
1132 return '';
1133 }
1134
1135 $decoded = self::decodeTagAttributes( $text );
1136 $stripped = self::validateTagAttributes( $decoded, $element );
1137
1138 if ( $sorted ) {
1139 ksort( $stripped );
1140 }
1141
1142 return self::safeEncodeTagAttributes( $stripped );
1143 }
1144
1145 /**
1146 * Encode an attribute value for HTML output.
1147 * @param string $text
1148 * @return string HTML-encoded text fragment
1149 */
1150 static function encodeAttribute( $text ) {
1151 $encValue = htmlspecialchars( $text, ENT_QUOTES );
1152
1153 // Whitespace is normalized during attribute decoding,
1154 // so if we've been passed non-spaces we must encode them
1155 // ahead of time or they won't be preserved.
1156 $encValue = strtr( $encValue, [
1157 "\n" => '&#10;',
1158 "\r" => '&#13;',
1159 "\t" => '&#9;',
1160 ] );
1161
1162 return $encValue;
1163 }
1164
1165 /**
1166 * Armor French spaces with a replacement character
1167 *
1168 * @since 1.32
1169 * @param string $text Text to armor
1170 * @param string $space Space character for the French spaces, defaults to '&#160;'
1171 * @return string Armored text
1172 */
1173 public static function armorFrenchSpaces( $text, $space = '&#160;' ) {
1174 // Replace $ with \$ and \ with \\
1175 $space = preg_replace( '#(?<!\\\\)(\\$|\\\\)#', '\\\\$1', $space );
1176 $fixtags = [
1177 # French spaces, last one Guillemet-left
1178 # only if there is something before the space
1179 # and a non-word character after the punctuation.
1180 '/(\S) (?=[?:;!%»›](?!\w))/u' => "\\1$space",
1181 # French spaces, Guillemet-right
1182 '/([«‹]) /u' => "\\1$space",
1183 ];
1184 return preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1185 }
1186
1187 /**
1188 * Encode an attribute value for HTML tags, with extra armoring
1189 * against further wiki processing.
1190 * @param string $text
1191 * @return string HTML-encoded text fragment
1192 */
1193 static function safeEncodeAttribute( $text ) {
1194 $encValue = self::encodeAttribute( $text );
1195
1196 # Templates and links may be expanded in later parsing,
1197 # creating invalid or dangerous output. Suppress this.
1198 $encValue = strtr( $encValue, [
1199 '<' => '&lt;', // This should never happen,
1200 '>' => '&gt;', // we've received invalid input
1201 '"' => '&quot;', // which should have been escaped.
1202 '{' => '&#123;',
1203 '}' => '&#125;', // prevent unpaired language conversion syntax
1204 '[' => '&#91;',
1205 ']' => '&#93;',
1206 "''" => '&#39;&#39;',
1207 'ISBN' => '&#73;SBN',
1208 'RFC' => '&#82;FC',
1209 'PMID' => '&#80;MID',
1210 '|' => '&#124;',
1211 '__' => '&#95;_',
1212 ] );
1213
1214 # Armor against French spaces detection (T5158)
1215 $encValue = self::armorFrenchSpaces( $encValue, '&#32;' );
1216
1217 # Stupid hack
1218 $encValue = preg_replace_callback(
1219 '/((?i)' . wfUrlProtocols() . ')/',
1220 function ( $matches ) {
1221 return str_replace( ':', '&#58;', $matches[1] );
1222 },
1223 $encValue );
1224 return $encValue;
1225 }
1226
1227 /**
1228 * Given a value, escape it so that it can be used in an id attribute and
1229 * return it. This will use HTML5 validation, allowing anything but ASCII
1230 * whitespace.
1231 *
1232 * To ensure we don't have to bother escaping anything, we also strip ', ".
1233 * TODO: Is this the best tactic?
1234 *
1235 * We also strip # because it upsets IE, and % because it could be
1236 * ambiguous if it's part of something that looks like a percent escape
1237 * (which don't work reliably in fragments cross-browser).
1238 *
1239 * @deprecated since 1.30, use one of this class' escapeIdFor*() functions
1240 *
1241 * @see https://www.w3.org/TR/html401/types.html#type-name Valid characters
1242 * in the id and name attributes
1243 * @see https://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with
1244 * the id attribute
1245 * @see https://www.w3.org/TR/html5/dom.html#the-id-attribute
1246 * HTML5 definition of id attribute
1247 *
1248 * @param string $id Id to escape
1249 * @param string|array $options String or array of strings (default is array()):
1250 * 'noninitial': This is a non-initial fragment of an id, not a full id,
1251 * so don't pay attention if the first character isn't valid at the
1252 * beginning of an id.
1253 * @return string
1254 */
1255 static function escapeId( $id, $options = [] ) {
1256 $options = (array)$options;
1257
1258 // HTML4-style escaping
1259 static $replace = [
1260 '%3A' => ':',
1261 '%' => '.'
1262 ];
1263
1264 $id = urlencode( strtr( $id, ' ', '_' ) );
1265 $id = strtr( $id, $replace );
1266
1267 if ( !preg_match( '/^[a-zA-Z]/', $id ) && !in_array( 'noninitial', $options ) ) {
1268 // Initial character must be a letter!
1269 $id = "x$id";
1270 }
1271 return $id;
1272 }
1273
1274 /**
1275 * Given a section name or other user-generated or otherwise unsafe string, escapes it to be
1276 * a valid HTML id attribute.
1277 *
1278 * WARNING: unlike escapeId(), the output of this function is not guaranteed to be HTML safe,
1279 * be sure to use proper escaping.
1280 *
1281 * @param string $id String to escape
1282 * @param int $mode One of ID_* constants, specifying whether the primary or fallback encoding
1283 * should be used.
1284 * @return string|bool Escaped ID or false if fallback encoding is requested but it's not
1285 * configured.
1286 *
1287 * @since 1.30
1288 */
1289 public static function escapeIdForAttribute( $id, $mode = self::ID_PRIMARY ) {
1290 global $wgFragmentMode;
1291
1292 if ( !isset( $wgFragmentMode[$mode] ) ) {
1293 if ( $mode === self::ID_PRIMARY ) {
1294 throw new UnexpectedValueException( '$wgFragmentMode is configured with no primary mode' );
1295 }
1296 return false;
1297 }
1298
1299 $internalMode = $wgFragmentMode[$mode];
1300
1301 return self::escapeIdInternal( $id, $internalMode );
1302 }
1303
1304 /**
1305 * Given a section name or other user-generated or otherwise unsafe string, escapes it to be
1306 * a valid URL fragment.
1307 *
1308 * WARNING: unlike escapeId(), the output of this function is not guaranteed to be HTML safe,
1309 * be sure to use proper escaping.
1310 *
1311 * @param string $id String to escape
1312 * @return string Escaped ID
1313 *
1314 * @since 1.30
1315 */
1316 public static function escapeIdForLink( $id ) {
1317 global $wgFragmentMode;
1318
1319 if ( !isset( $wgFragmentMode[self::ID_PRIMARY] ) ) {
1320 throw new UnexpectedValueException( '$wgFragmentMode is configured with no primary mode' );
1321 }
1322
1323 $mode = $wgFragmentMode[self::ID_PRIMARY];
1324
1325 $id = self::escapeIdInternal( $id, $mode );
1326
1327 return $id;
1328 }
1329
1330 /**
1331 * Given a section name or other user-generated or otherwise unsafe string, escapes it to be
1332 * a valid URL fragment for external interwikis.
1333 *
1334 * @param string $id String to escape
1335 * @return string Escaped ID
1336 *
1337 * @since 1.30
1338 */
1339 public static function escapeIdForExternalInterwiki( $id ) {
1340 global $wgExternalInterwikiFragmentMode;
1341
1342 $id = self::escapeIdInternal( $id, $wgExternalInterwikiFragmentMode );
1343
1344 return $id;
1345 }
1346
1347 /**
1348 * Helper for escapeIdFor*() functions. Performs most of the actual escaping.
1349 *
1350 * @param string $id String to escape
1351 * @param string $mode One of modes from $wgFragmentMode
1352 * @return string
1353 */
1354 private static function escapeIdInternal( $id, $mode ) {
1355 switch ( $mode ) {
1356 case 'html5':
1357 $id = str_replace( ' ', '_', $id );
1358 break;
1359 case 'legacy':
1360 // This corresponds to 'noninitial' mode of the old escapeId()
1361 static $replace = [
1362 '%3A' => ':',
1363 '%' => '.'
1364 ];
1365
1366 $id = urlencode( str_replace( ' ', '_', $id ) );
1367 $id = strtr( $id, $replace );
1368 break;
1369 default:
1370 throw new InvalidArgumentException( "Invalid mode '$mode' passed to '" . __METHOD__ );
1371 }
1372
1373 return $id;
1374 }
1375
1376 /**
1377 * Given a string containing a space delimited list of ids, escape each id
1378 * to match ids escaped by the escapeIdForAttribute() function.
1379 *
1380 * @since 1.27
1381 *
1382 * @param string $referenceString Space delimited list of ids
1383 * @return string
1384 */
1385 public static function escapeIdReferenceList( $referenceString ) {
1386 # Explode the space delimited list string into an array of tokens
1387 $references = preg_split( '/\s+/', "{$referenceString}", -1, PREG_SPLIT_NO_EMPTY );
1388
1389 # Escape each token as an id
1390 foreach ( $references as &$ref ) {
1391 $ref = self::escapeIdForAttribute( $ref );
1392 }
1393
1394 # Merge the array back to a space delimited list string
1395 # If the array is empty, the result will be an empty string ('')
1396 $referenceString = implode( ' ', $references );
1397
1398 return $referenceString;
1399 }
1400
1401 /**
1402 * Given a value, escape it so that it can be used as a CSS class and
1403 * return it.
1404 *
1405 * @todo For extra validity, input should be validated UTF-8.
1406 *
1407 * @see https://www.w3.org/TR/CSS21/syndata.html Valid characters/format
1408 *
1409 * @param string $class
1410 * @return string
1411 */
1412 static function escapeClass( $class ) {
1413 // Convert ugly stuff to underscores and kill underscores in ugly places
1414 return rtrim( preg_replace(
1415 [ '/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/', '/_+/' ],
1416 '_',
1417 $class ), '_' );
1418 }
1419
1420 /**
1421 * Given HTML input, escape with htmlspecialchars but un-escape entities.
1422 * This allows (generally harmless) entities like &#160; to survive.
1423 *
1424 * @param string $html HTML to escape
1425 * @return string Escaped input
1426 */
1427 static function escapeHtmlAllowEntities( $html ) {
1428 $html = self::decodeCharReferences( $html );
1429 # It seems wise to escape ' as well as ", as a matter of course. Can't
1430 # hurt. Use ENT_SUBSTITUTE so that incorrectly truncated multibyte characters
1431 # don't cause the entire string to disappear.
1432 $html = htmlspecialchars( $html, ENT_QUOTES | ENT_SUBSTITUTE );
1433 return $html;
1434 }
1435
1436 /**
1437 * Return an associative array of attribute names and values from
1438 * a partial tag string. Attribute names are forced to lowercase,
1439 * character references are decoded to UTF-8 text.
1440 *
1441 * @param string $text
1442 * @return array
1443 */
1444 public static function decodeTagAttributes( $text ) {
1445 if ( trim( $text ) == '' ) {
1446 return [];
1447 }
1448
1449 $pairs = [];
1450 if ( !preg_match_all(
1451 self::getAttribsRegex(),
1452 $text,
1453 $pairs,
1454 PREG_SET_ORDER ) ) {
1455 return [];
1456 }
1457
1458 $attribs = [];
1459 foreach ( $pairs as $set ) {
1460 $attribute = strtolower( $set[1] );
1461
1462 // Filter attribute names with unacceptable characters
1463 if ( !preg_match( self::getAttribNameRegex(), $attribute ) ) {
1464 continue;
1465 }
1466
1467 $value = self::getTagAttributeCallback( $set );
1468
1469 // Normalize whitespace
1470 $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
1471 $value = trim( $value );
1472
1473 // Decode character references
1474 $attribs[$attribute] = self::decodeCharReferences( $value );
1475 }
1476 return $attribs;
1477 }
1478
1479 /**
1480 * Build a partial tag string from an associative array of attribute
1481 * names and values as returned by decodeTagAttributes.
1482 *
1483 * @param array $assoc_array
1484 * @return string
1485 */
1486 public static function safeEncodeTagAttributes( $assoc_array ) {
1487 $attribs = [];
1488 foreach ( $assoc_array as $attribute => $value ) {
1489 $encAttribute = htmlspecialchars( $attribute );
1490 $encValue = self::safeEncodeAttribute( $value );
1491
1492 $attribs[] = "$encAttribute=\"$encValue\"";
1493 }
1494 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
1495 }
1496
1497 /**
1498 * Pick the appropriate attribute value from a match set from the
1499 * attribs regex matches.
1500 *
1501 * @param array $set
1502 * @throws MWException When tag conditions are not met.
1503 * @return string
1504 */
1505 private static function getTagAttributeCallback( $set ) {
1506 if ( isset( $set[5] ) ) {
1507 # No quotes.
1508 return $set[5];
1509 } elseif ( isset( $set[4] ) ) {
1510 # Single-quoted
1511 return $set[4];
1512 } elseif ( isset( $set[3] ) ) {
1513 # Double-quoted
1514 return $set[3];
1515 } elseif ( !isset( $set[2] ) ) {
1516 # In XHTML, attributes must have a value so return an empty string.
1517 # See "Empty attribute syntax",
1518 # https://www.w3.org/TR/html5/syntax.html#syntax-attribute-name
1519 return "";
1520 } else {
1521 throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1522 }
1523 }
1524
1525 /**
1526 * @param string $text
1527 * @return string
1528 */
1529 private static function normalizeWhitespace( $text ) {
1530 return trim( preg_replace(
1531 '/(?:\r\n|[\x20\x0d\x0a\x09])+/',
1532 ' ',
1533 $text ) );
1534 }
1535
1536 /**
1537 * Normalizes whitespace in a section name, such as might be returned
1538 * by Parser::stripSectionName(), for use in the id's that are used for
1539 * section links.
1540 *
1541 * @param string $section
1542 * @return string
1543 */
1544 static function normalizeSectionNameWhitespace( $section ) {
1545 return trim( preg_replace( '/[ _]+/', ' ', $section ) );
1546 }
1547
1548 /**
1549 * Ensure that any entities and character references are legal
1550 * for XML and XHTML specifically. Any stray bits will be
1551 * &amp;-escaped to result in a valid text fragment.
1552 *
1553 * a. named char refs can only be &lt; &gt; &amp; &quot;, others are
1554 * numericized (this way we're well-formed even without a DTD)
1555 * b. any numeric char refs must be legal chars, not invalid or forbidden
1556 * c. use lower cased "&#x", not "&#X"
1557 * d. fix or reject non-valid attributes
1558 *
1559 * @param string $text
1560 * @return string
1561 * @private
1562 */
1563 static function normalizeCharReferences( $text ) {
1564 return preg_replace_callback(
1565 self::CHAR_REFS_REGEX,
1566 [ self::class, 'normalizeCharReferencesCallback' ],
1567 $text );
1568 }
1569
1570 /**
1571 * @param string $matches
1572 * @return string
1573 */
1574 static function normalizeCharReferencesCallback( $matches ) {
1575 $ret = null;
1576 if ( $matches[1] != '' ) {
1577 $ret = self::normalizeEntity( $matches[1] );
1578 } elseif ( $matches[2] != '' ) {
1579 $ret = self::decCharReference( $matches[2] );
1580 } elseif ( $matches[3] != '' ) {
1581 $ret = self::hexCharReference( $matches[3] );
1582 }
1583 if ( is_null( $ret ) ) {
1584 return htmlspecialchars( $matches[0] );
1585 } else {
1586 return $ret;
1587 }
1588 }
1589
1590 /**
1591 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1592 * return the equivalent numeric entity reference (except for the core &lt;
1593 * &gt; &amp; &quot;). If the entity is a MediaWiki-specific alias, returns
1594 * the HTML equivalent. Otherwise, returns HTML-escaped text of
1595 * pseudo-entity source (eg &amp;foo;)
1596 *
1597 * @param string $name
1598 * @return string
1599 */
1600 static function normalizeEntity( $name ) {
1601 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1602 return '&' . self::$htmlEntityAliases[$name] . ';';
1603 } elseif ( in_array( $name, [ 'lt', 'gt', 'amp', 'quot' ] ) ) {
1604 return "&$name;";
1605 } elseif ( isset( self::$htmlEntities[$name] ) ) {
1606 return '&#' . self::$htmlEntities[$name] . ';';
1607 } else {
1608 return "&amp;$name;";
1609 }
1610 }
1611
1612 /**
1613 * @param int $codepoint
1614 * @return null|string
1615 */
1616 static function decCharReference( $codepoint ) {
1617 $point = intval( $codepoint );
1618 if ( self::validateCodepoint( $point ) ) {
1619 return sprintf( '&#%d;', $point );
1620 } else {
1621 return null;
1622 }
1623 }
1624
1625 /**
1626 * @param int $codepoint
1627 * @return null|string
1628 */
1629 static function hexCharReference( $codepoint ) {
1630 $point = hexdec( $codepoint );
1631 if ( self::validateCodepoint( $point ) ) {
1632 return sprintf( '&#x%x;', $point );
1633 } else {
1634 return null;
1635 }
1636 }
1637
1638 /**
1639 * Returns true if a given Unicode codepoint is a valid character in
1640 * both HTML5 and XML.
1641 * @param int $codepoint
1642 * @return bool
1643 */
1644 private static function validateCodepoint( $codepoint ) {
1645 # U+000C is valid in HTML5 but not allowed in XML.
1646 # U+000D is valid in XML but not allowed in HTML5.
1647 # U+007F - U+009F are disallowed in HTML5 (control characters).
1648 return $codepoint == 0x09
1649 || $codepoint == 0x0a
1650 || ( $codepoint >= 0x20 && $codepoint <= 0x7e )
1651 || ( $codepoint >= 0xa0 && $codepoint <= 0xd7ff )
1652 || ( $codepoint >= 0xe000 && $codepoint <= 0xfffd )
1653 || ( $codepoint >= 0x10000 && $codepoint <= 0x10ffff );
1654 }
1655
1656 /**
1657 * Decode any character references, numeric or named entities,
1658 * in the text and return a UTF-8 string.
1659 *
1660 * @param string $text
1661 * @return string
1662 */
1663 public static function decodeCharReferences( $text ) {
1664 return preg_replace_callback(
1665 self::CHAR_REFS_REGEX,
1666 [ self::class, 'decodeCharReferencesCallback' ],
1667 $text );
1668 }
1669
1670 /**
1671 * Decode any character references, numeric or named entities,
1672 * in the next and normalize the resulting string. (T16952)
1673 *
1674 * This is useful for page titles, not for text to be displayed,
1675 * MediaWiki allows HTML entities to escape normalization as a feature.
1676 *
1677 * @param string $text Already normalized, containing entities
1678 * @return string Still normalized, without entities
1679 */
1680 public static function decodeCharReferencesAndNormalize( $text ) {
1681 $text = preg_replace_callback(
1682 self::CHAR_REFS_REGEX,
1683 [ self::class, 'decodeCharReferencesCallback' ],
1684 $text,
1685 -1, //limit
1686 $count
1687 );
1688
1689 if ( $count ) {
1690 return MediaWikiServices::getInstance()->getContentLanguage()->normalize( $text );
1691 } else {
1692 return $text;
1693 }
1694 }
1695
1696 /**
1697 * @param string $matches
1698 * @return string
1699 */
1700 static function decodeCharReferencesCallback( $matches ) {
1701 if ( $matches[1] != '' ) {
1702 return self::decodeEntity( $matches[1] );
1703 } elseif ( $matches[2] != '' ) {
1704 return self::decodeChar( intval( $matches[2] ) );
1705 } elseif ( $matches[3] != '' ) {
1706 return self::decodeChar( hexdec( $matches[3] ) );
1707 }
1708 # Last case should be an ampersand by itself
1709 return $matches[0];
1710 }
1711
1712 /**
1713 * Return UTF-8 string for a codepoint if that is a valid
1714 * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1715 * @param int $codepoint
1716 * @return string
1717 * @private
1718 */
1719 static function decodeChar( $codepoint ) {
1720 if ( self::validateCodepoint( $codepoint ) ) {
1721 return UtfNormal\Utils::codepointToUtf8( $codepoint );
1722 } else {
1723 return UtfNormal\Constants::UTF8_REPLACEMENT;
1724 }
1725 }
1726
1727 /**
1728 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1729 * return the UTF-8 encoding of that character. Otherwise, returns
1730 * pseudo-entity source (eg "&foo;")
1731 *
1732 * @param string $name
1733 * @return string
1734 */
1735 static function decodeEntity( $name ) {
1736 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1737 $name = self::$htmlEntityAliases[$name];
1738 }
1739 if ( isset( self::$htmlEntities[$name] ) ) {
1740 return UtfNormal\Utils::codepointToUtf8( self::$htmlEntities[$name] );
1741 } else {
1742 return "&$name;";
1743 }
1744 }
1745
1746 /**
1747 * Fetch the whitelist of acceptable attributes for a given element name.
1748 *
1749 * @param string $element
1750 * @return array
1751 */
1752 static function attributeWhitelist( $element ) {
1753 $list = self::setupAttributeWhitelist();
1754 return $list[$element] ?? [];
1755 }
1756
1757 /**
1758 * Foreach array key (an allowed HTML element), return an array
1759 * of allowed attributes
1760 * @return array
1761 */
1762 static function setupAttributeWhitelist() {
1763 static $whitelist;
1764
1765 if ( $whitelist !== null ) {
1766 return $whitelist;
1767 }
1768
1769 $common = [
1770 # HTML
1771 'id',
1772 'class',
1773 'style',
1774 'lang',
1775 'dir',
1776 'title',
1777
1778 # WAI-ARIA
1779 'aria-describedby',
1780 'aria-flowto',
1781 'aria-label',
1782 'aria-labelledby',
1783 'aria-owns',
1784 'role',
1785
1786 # RDFa
1787 # These attributes are specified in section 9 of
1788 # https://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1789 'about',
1790 'property',
1791 'resource',
1792 'datatype',
1793 'typeof',
1794
1795 # Microdata. These are specified by
1796 # https://html.spec.whatwg.org/multipage/microdata.html#the-microdata-model
1797 'itemid',
1798 'itemprop',
1799 'itemref',
1800 'itemscope',
1801 'itemtype',
1802 ];
1803
1804 $block = array_merge( $common, [ 'align' ] );
1805 $tablealign = [ 'align', 'valign' ];
1806 $tablecell = [
1807 'abbr',
1808 'axis',
1809 'headers',
1810 'scope',
1811 'rowspan',
1812 'colspan',
1813 'nowrap', # deprecated
1814 'width', # deprecated
1815 'height', # deprecated
1816 'bgcolor', # deprecated
1817 ];
1818
1819 # Numbers refer to sections in HTML 4.01 standard describing the element.
1820 # See: https://www.w3.org/TR/html4/
1821 $whitelist = [
1822 # 7.5.4
1823 'div' => $block,
1824 'center' => $common, # deprecated
1825 'span' => $common,
1826
1827 # 7.5.5
1828 'h1' => $block,
1829 'h2' => $block,
1830 'h3' => $block,
1831 'h4' => $block,
1832 'h5' => $block,
1833 'h6' => $block,
1834
1835 # 7.5.6
1836 # address
1837
1838 # 8.2.4
1839 'bdo' => $common,
1840
1841 # 9.2.1
1842 'em' => $common,
1843 'strong' => $common,
1844 'cite' => $common,
1845 'dfn' => $common,
1846 'code' => $common,
1847 'samp' => $common,
1848 'kbd' => $common,
1849 'var' => $common,
1850 'abbr' => $common,
1851 # acronym
1852
1853 # 9.2.2
1854 'blockquote' => array_merge( $common, [ 'cite' ] ),
1855 'q' => array_merge( $common, [ 'cite' ] ),
1856
1857 # 9.2.3
1858 'sub' => $common,
1859 'sup' => $common,
1860
1861 # 9.3.1
1862 'p' => $block,
1863
1864 # 9.3.2
1865 'br' => array_merge( $common, [ 'clear' ] ),
1866
1867 # https://www.w3.org/TR/html5/text-level-semantics.html#the-wbr-element
1868 'wbr' => $common,
1869
1870 # 9.3.4
1871 'pre' => array_merge( $common, [ 'width' ] ),
1872
1873 # 9.4
1874 'ins' => array_merge( $common, [ 'cite', 'datetime' ] ),
1875 'del' => array_merge( $common, [ 'cite', 'datetime' ] ),
1876
1877 # 10.2
1878 'ul' => array_merge( $common, [ 'type' ] ),
1879 'ol' => array_merge( $common, [ 'type', 'start', 'reversed' ] ),
1880 'li' => array_merge( $common, [ 'type', 'value' ] ),
1881
1882 # 10.3
1883 'dl' => $common,
1884 'dd' => $common,
1885 'dt' => $common,
1886
1887 # 11.2.1
1888 'table' => array_merge( $common,
1889 [ 'summary', 'width', 'border', 'frame',
1890 'rules', 'cellspacing', 'cellpadding',
1891 'align', 'bgcolor',
1892 ] ),
1893
1894 # 11.2.2
1895 'caption' => $block,
1896
1897 # 11.2.3
1898 'thead' => $common,
1899 'tfoot' => $common,
1900 'tbody' => $common,
1901
1902 # 11.2.4
1903 'colgroup' => array_merge( $common, [ 'span' ] ),
1904 'col' => array_merge( $common, [ 'span' ] ),
1905
1906 # 11.2.5
1907 'tr' => array_merge( $common, [ 'bgcolor' ], $tablealign ),
1908
1909 # 11.2.6
1910 'td' => array_merge( $common, $tablecell, $tablealign ),
1911 'th' => array_merge( $common, $tablecell, $tablealign ),
1912
1913 # 12.2
1914 # NOTE: <a> is not allowed directly, but the attrib
1915 # whitelist is used from the Parser object
1916 'a' => array_merge( $common, [ 'href', 'rel', 'rev' ] ), # rel/rev esp. for RDFa
1917
1918 # 13.2
1919 # Not usually allowed, but may be used for extension-style hooks
1920 # such as <math> when it is rasterized, or if $wgAllowImageTag is
1921 # true
1922 'img' => array_merge( $common, [ 'alt', 'src', 'width', 'height', 'srcset' ] ),
1923
1924 'video' => array_merge( $common, [ 'poster', 'controls', 'preload', 'width', 'height' ] ),
1925 'source' => array_merge( $common, [ 'type', 'src' ] ),
1926 'track' => array_merge( $common, [ 'type', 'src', 'srclang', 'kind', 'label' ] ),
1927
1928 # 15.2.1
1929 'tt' => $common,
1930 'b' => $common,
1931 'i' => $common,
1932 'big' => $common,
1933 'small' => $common,
1934 'strike' => $common,
1935 's' => $common,
1936 'u' => $common,
1937
1938 # 15.2.2
1939 'font' => array_merge( $common, [ 'size', 'color', 'face' ] ),
1940 # basefont
1941
1942 # 15.3
1943 'hr' => array_merge( $common, [ 'width' ] ),
1944
1945 # HTML Ruby annotation text module, simple ruby only.
1946 # https://www.w3.org/TR/html5/text-level-semantics.html#the-ruby-element
1947 'ruby' => $common,
1948 # rbc
1949 'rb' => $common,
1950 'rp' => $common,
1951 'rt' => $common, # array_merge( $common, array( 'rbspan' ) ),
1952 'rtc' => $common,
1953
1954 # MathML root element, where used for extensions
1955 # 'title' may not be 100% valid here; it's XHTML
1956 # https://www.w3.org/TR/REC-MathML/
1957 'math' => [ 'class', 'style', 'id', 'title' ],
1958
1959 // HTML 5 section 4.5
1960 'figure' => $common,
1961 'figcaption' => $common,
1962
1963 # HTML 5 section 4.6
1964 'bdi' => $common,
1965
1966 # HTML5 elements, defined by:
1967 # https://html.spec.whatwg.org/multipage/semantics.html#the-data-element
1968 'data' => array_merge( $common, [ 'value' ] ),
1969 'time' => array_merge( $common, [ 'datetime' ] ),
1970 'mark' => $common,
1971
1972 // meta and link are only permitted by removeHTMLtags when Microdata
1973 // is enabled so we don't bother adding a conditional to hide these
1974 // Also meta and link are only valid in WikiText as Microdata elements
1975 // (ie: validateTag rejects tags missing the attributes needed for Microdata)
1976 // So we don't bother including $common attributes that have no purpose.
1977 'meta' => [ 'itemprop', 'content' ],
1978 'link' => [ 'itemprop', 'href', 'title' ],
1979 ];
1980
1981 return $whitelist;
1982 }
1983
1984 /**
1985 * Take a fragment of (potentially invalid) HTML and return
1986 * a version with any tags removed, encoded as plain text.
1987 *
1988 * Warning: this return value must be further escaped for literal
1989 * inclusion in HTML output as of 1.10!
1990 *
1991 * @param string $html HTML fragment
1992 * @return string
1993 */
1994 static function stripAllTags( $html ) {
1995 // Use RemexHtml to tokenize $html and extract the text
1996 $handler = new RemexStripTagHandler;
1997 $tokenizer = new RemexHtml\Tokenizer\Tokenizer( $handler, $html, [
1998 'ignoreErrors' => true,
1999 // don't ignore char refs, we want them to be decoded
2000 'ignoreNulls' => true,
2001 'skipPreprocess' => true,
2002 ] );
2003 $tokenizer->execute();
2004 $text = $handler->getResult();
2005
2006 $text = self::normalizeWhitespace( $text );
2007 return $text;
2008 }
2009
2010 /**
2011 * Hack up a private DOCTYPE with HTML's standard entity declarations.
2012 * PHP 4 seemed to know these if you gave it an HTML doctype, but
2013 * PHP 5.1 doesn't.
2014 *
2015 * Use for passing XHTML fragments to PHP's XML parsing functions
2016 *
2017 * @return string
2018 */
2019 static function hackDocType() {
2020 $out = "<!DOCTYPE html [\n";
2021 foreach ( self::$htmlEntities as $entity => $codepoint ) {
2022 $out .= "<!ENTITY $entity \"&#$codepoint;\">";
2023 }
2024 $out .= "]>\n";
2025 return $out;
2026 }
2027
2028 /**
2029 * @param string $url
2030 * @return mixed|string
2031 */
2032 static function cleanUrl( $url ) {
2033 # Normalize any HTML entities in input. They will be
2034 # re-escaped by makeExternalLink().
2035 $url = self::decodeCharReferences( $url );
2036
2037 # Escape any control characters introduced by the above step
2038 $url = preg_replace_callback( '/[\][<>"\\x00-\\x20\\x7F\|]/',
2039 [ __CLASS__, 'cleanUrlCallback' ], $url );
2040
2041 # Validate hostname portion
2042 $matches = [];
2043 if ( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
2044 list( /* $whole */, $protocol, $host, $rest ) = $matches;
2045
2046 // Characters that will be ignored in IDNs.
2047 // https://tools.ietf.org/html/rfc3454#section-3.1
2048 // Strip them before further processing so blacklists and such work.
2049 $strip = "/
2050 \\s| # general whitespace
2051 \xc2\xad| # 00ad SOFT HYPHEN
2052 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
2053 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
2054 \xe2\x81\xa0| # 2060 WORD JOINER
2055 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
2056 \xcd\x8f| # 034f COMBINING GRAPHEME JOINER
2057 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
2058 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
2059 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
2060 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
2061 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
2062 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe0f VARIATION SELECTOR-1-16
2063 /xuD";
2064
2065 $host = preg_replace( $strip, '', $host );
2066
2067 // IPv6 host names are bracketed with []. Url-decode these.
2068 if ( substr_compare( "//%5B", $host, 0, 5 ) === 0 &&
2069 preg_match( '!^//%5B([0-9A-Fa-f:.]+)%5D((:\d+)?)$!', $host, $matches )
2070 ) {
2071 $host = '//[' . $matches[1] . ']' . $matches[2];
2072 }
2073
2074 // @todo FIXME: Validate hostnames here
2075
2076 return $protocol . $host . $rest;
2077 } else {
2078 return $url;
2079 }
2080 }
2081
2082 /**
2083 * @param array $matches
2084 * @return string
2085 */
2086 static function cleanUrlCallback( $matches ) {
2087 return urlencode( $matches[0] );
2088 }
2089
2090 /**
2091 * Does a string look like an e-mail address?
2092 *
2093 * This validates an email address using an HTML5 specification found at:
2094 * http://www.whatwg.org/html/states-of-the-type-attribute.html#valid-e-mail-address
2095 * Which as of 2011-01-24 says:
2096 *
2097 * A valid e-mail address is a string that matches the ABNF production
2098 * 1*( atext / "." ) "@" ldh-str *( "." ldh-str ) where atext is defined
2099 * in RFC 5322 section 3.2.3, and ldh-str is defined in RFC 1034 section
2100 * 3.5.
2101 *
2102 * This function is an implementation of the specification as requested in
2103 * T24449.
2104 *
2105 * Client-side forms will use the same standard validation rules via JS or
2106 * HTML 5 validation; additional restrictions can be enforced server-side
2107 * by extensions via the 'isValidEmailAddr' hook.
2108 *
2109 * Note that this validation doesn't 100% match RFC 2822, but is believed
2110 * to be liberal enough for wide use. Some invalid addresses will still
2111 * pass validation here.
2112 *
2113 * @since 1.18
2114 *
2115 * @param string $addr E-mail address
2116 * @return bool
2117 */
2118 public static function validateEmail( $addr ) {
2119 $result = null;
2120 if ( !Hooks::run( 'isValidEmailAddr', [ $addr, &$result ] ) ) {
2121 return $result;
2122 }
2123
2124 // Please note strings below are enclosed in brackets [], this make the
2125 // hyphen "-" a range indicator. Hence it is double backslashed below.
2126 // See T28948
2127 $rfc5322_atext = "a-z0-9!#$%&'*+\\-\/=?^_`{|}~";
2128 $rfc1034_ldh_str = "a-z0-9\\-";
2129
2130 $html5_email_regexp = "/
2131 ^ # start of string
2132 [$rfc5322_atext\\.]+ # user part which is liberal :p
2133 @ # 'apostrophe'
2134 [$rfc1034_ldh_str]+ # First domain part
2135 (\\.[$rfc1034_ldh_str]+)* # Following part prefixed with a dot
2136 $ # End of string
2137 /ix"; // case Insensitive, eXtended
2138
2139 return (bool)preg_match( $html5_email_regexp, $addr );
2140 }
2141 }
Wiklou, le Wiki du Wiklou