includes/Html.php

   1 <?php
   2 /**
   3  * Collection of methods to generate HTML content
   4  *
   5  * Copyright © 2009 Aryeh Gregor
   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  */
  25 use MediaWiki\MediaWikiServices;
  26
  27 /**
  28  * This class is a collection of static functions that serve two purposes:
  29  *
  30  * 1) Implement any algorithms specified by HTML5, or other HTML
  31  * specifications, in a convenient and self-contained way.
  32  *
  33  * 2) Allow HTML elements to be conveniently and safely generated, like the
  34  * current Xml class but a) less confused (Xml supports HTML-specific things,
  35  * but only sometimes!) and b) not necessarily confined to XML-compatible
  36  * output.
  37  *
  38  * There are two important configuration options this class uses:
  39  *
  40  * $wgMimeType: If this is set to an xml MIME type then output should be
  41  *     valid XHTML5.
  42  *
  43  * This class is meant to be confined to utility functions that are called from
  44  * trusted code paths.  It does not do enforcement of policy like not allowing
  45  * <a> elements.
  46  *
  47  * @since 1.16
  48  */
  49 class Html {
  50         // List of void elements from HTML5, section 8.1.2 as of 2016-09-19
  51         private static $voidElements = [
  52                 'area',
  53                 'base',
  54                 'br',
  55                 'col',
  56                 'embed',
  57                 'hr',
  58                 'img',
  59                 'input',
  60                 'keygen',
  61                 'link',
  62                 'meta',
  63                 'param',
  64                 'source',
  65                 'track',
  66                 'wbr',
  67         ];
  68
  69         // Boolean attributes, which may have the value omitted entirely.  Manually
  70         // collected from the HTML5 spec as of 2011-08-12.
  71         private static $boolAttribs = [
  72                 'async',
  73                 'autofocus',
  74                 'autoplay',
  75                 'checked',
  76                 'controls',
  77                 'default',
  78                 'defer',
  79                 'disabled',
  80                 'formnovalidate',
  81                 'hidden',
  82                 'ismap',
  83                 'itemscope',
  84                 'loop',
  85                 'multiple',
  86                 'muted',
  87                 'novalidate',
  88                 'open',
  89                 'pubdate',
  90                 'readonly',
  91                 'required',
  92                 'reversed',
  93                 'scoped',
  94                 'seamless',
  95                 'selected',
  96                 'truespeed',
  97                 'typemustmatch',
  98                 // HTML5 Microdata
  99                 'itemscope',
 100         ];
 101
 102         /**
 103          * Modifies a set of attributes meant for button elements
 104          * and apply a set of default attributes when $wgUseMediaWikiUIEverywhere enabled.
 105          * @param array $attrs HTML attributes in an associative array
 106          * @param string[] $modifiers classes to add to the button
 107          * @see https://tools.wmflabs.org/styleguide/desktop/index.html for guidance on available modifiers
 108          * @return array $attrs A modified attribute array
 109          */
 110         public static function buttonAttributes( array $attrs, array $modifiers = [] ) {
 111                 global $wgUseMediaWikiUIEverywhere;
 112                 if ( $wgUseMediaWikiUIEverywhere ) {
 113                         if ( isset( $attrs['class'] ) ) {
 114                                 if ( is_array( $attrs['class'] ) ) {
 115                                         $attrs['class'][] = 'mw-ui-button';
 116                                         $attrs['class'] = array_merge( $attrs['class'], $modifiers );
 117                                         // ensure compatibility with Xml
 118                                         $attrs['class'] = implode( ' ', $attrs['class'] );
 119                                 } else {
 120                                         $attrs['class'] .= ' mw-ui-button ' . implode( ' ', $modifiers );
 121                                 }
 122                         } else {
 123                                 // ensure compatibility with Xml
 124                                 $attrs['class'] = 'mw-ui-button ' . implode( ' ', $modifiers );
 125                         }
 126                 }
 127                 return $attrs;
 128         }
 129
 130         /**
 131          * Modifies a set of attributes meant for text input elements
 132          * and apply a set of default attributes.
 133          * Removes size attribute when $wgUseMediaWikiUIEverywhere enabled.
 134          * @param array $attrs An attribute array.
 135          * @return array $attrs A modified attribute array
 136          */
 137         public static function getTextInputAttributes( array $attrs ) {
 138                 global $wgUseMediaWikiUIEverywhere;
 139                 if ( $wgUseMediaWikiUIEverywhere ) {
 140                         if ( isset( $attrs['class'] ) ) {
 141                                 if ( is_array( $attrs['class'] ) ) {
 142                                         $attrs['class'][] = 'mw-ui-input';
 143                                 } else {
 144                                         $attrs['class'] .= ' mw-ui-input';
 145                                 }
 146                         } else {
 147                                 $attrs['class'] = 'mw-ui-input';
 148                         }
 149                 }
 150                 return $attrs;
 151         }
 152
 153         /**
 154          * Returns an HTML link element in a string styled as a button
 155          * (when $wgUseMediaWikiUIEverywhere is enabled).
 156          *
 157          * @param string $text The text of the element. Will be escaped (not raw HTML)
 158          * @param array $attrs Associative array of attributes, e.g., [
 159          *   'href' => 'https://www.mediawiki.org/' ]. See expandAttributes() for
 160          *   further documentation.
 161          * @param string[] $modifiers classes to add to the button
 162          * @see https://tools.wmflabs.org/styleguide/desktop/index.html for guidance on available modifiers
 163          * @return string Raw HTML
 164          */
 165         public static function linkButton( $text, array $attrs, array $modifiers = [] ) {
 166                 return self::element( 'a',
 167                         self::buttonAttributes( $attrs, $modifiers ),
 168                         $text
 169                 );
 170         }
 171
 172         /**
 173          * Returns an HTML link element in a string styled as a button
 174          * (when $wgUseMediaWikiUIEverywhere is enabled).
 175          *
 176          * @param string $contents The raw HTML contents of the element: *not*
 177          *   escaped!
 178          * @param array $attrs Associative array of attributes, e.g., [
 179          *   'href' => 'https://www.mediawiki.org/' ]. See expandAttributes() for
 180          *   further documentation.
 181          * @param string[] $modifiers classes to add to the button
 182          * @see https://tools.wmflabs.org/styleguide/desktop/index.html for guidance on available modifiers
 183          * @return string Raw HTML
 184          */
 185         public static function submitButton( $contents, array $attrs, array $modifiers = [] ) {
 186                 $attrs['type'] = 'submit';
 187                 $attrs['value'] = $contents;
 188                 return self::element( 'input', self::buttonAttributes( $attrs, $modifiers ) );
 189         }
 190
 191         /**
 192          * Returns an HTML element in a string.  The major advantage here over
 193          * manually typing out the HTML is that it will escape all attribute
 194          * values.
 195          *
 196          * This is quite similar to Xml::tags(), but it implements some useful
 197          * HTML-specific logic.  For instance, there is no $allowShortTag
 198          * parameter: the closing tag is magically omitted if $element has an empty
 199          * content model.
 200          *
 201          * @param string $element The element's name, e.g., 'a'
 202          * @param array $attribs Associative array of attributes, e.g., [
 203          *   'href' => 'https://www.mediawiki.org/' ]. See expandAttributes() for
 204          *   further documentation.
 205          * @param string $contents The raw HTML contents of the element: *not*
 206          *   escaped!
 207          * @return string Raw HTML
 208          */
 209         public static function rawElement( $element, $attribs = [], $contents = '' ) {
 210                 $start = self::openElement( $element, $attribs );
 211                 if ( in_array( $element, self::$voidElements ) ) {
 212                         // Silly XML.
 213                         return substr( $start, 0, -1 ) . '/>';
 214                 } else {
 215                         return $start . $contents . self::closeElement( $element );
 216                 }
 217         }
 218
 219         /**
 220          * Identical to rawElement(), but HTML-escapes $contents (like
 221          * Xml::element()).
 222          *
 223          * @param string $element Name of the element, e.g., 'a'
 224          * @param array $attribs Associative array of attributes, e.g., [
 225          *   'href' => 'https://www.mediawiki.org/' ]. See expandAttributes() for
 226          *   further documentation.
 227          * @param string $contents
 228          *
 229          * @return string
 230          */
 231         public static function element( $element, $attribs = [], $contents = '' ) {
 232                 return self::rawElement( $element, $attribs, strtr( $contents, [
 233                         // There's no point in escaping quotes, >, etc. in the contents of
 234                         // elements.
 235                         '&' => '&amp;',
 236                         '<' => '&lt;'
 237                 ] ) );
 238         }
 239
 240         /**
 241          * Identical to rawElement(), but has no third parameter and omits the end
 242          * tag (and the self-closing '/' in XML mode for empty elements).
 243          *
 244          * @param string $element Name of the element, e.g., 'a'
 245          * @param array $attribs Associative array of attributes, e.g., [
 246          *   'href' => 'https://www.mediawiki.org/' ]. See expandAttributes() for
 247          *   further documentation.
 248          *
 249          * @return string
 250          */
 251         public static function openElement( $element, $attribs = [] ) {
 252                 $attribs = (array)$attribs;
 253                 // This is not required in HTML5, but let's do it anyway, for
 254                 // consistency and better compression.
 255                 $element = strtolower( $element );
 256
 257                 // Some people were abusing this by passing things like
 258                 // 'h1 id="foo" to $element, which we don't want.
 259                 if ( strpos( $element, ' ' ) !== false ) {
 260                         wfWarn( __METHOD__ . " given element name with space '$element'" );
 261                 }
 262
 263                 // Remove invalid input types
 264                 if ( $element == 'input' ) {
 265                         $validTypes = [
 266                                 'hidden',
 267                                 'text',
 268                                 'password',
 269                                 'checkbox',
 270                                 'radio',
 271                                 'file',
 272                                 'submit',
 273                                 'image',
 274                                 'reset',
 275                                 'button',
 276
 277                                 // HTML input types
 278                                 'datetime',
 279                                 'datetime-local',
 280                                 'date',
 281                                 'month',
 282                                 'time',
 283                                 'week',
 284                                 'number',
 285                                 'range',
 286                                 'email',
 287                                 'url',
 288                                 'search',
 289                                 'tel',
 290                                 'color',
 291                         ];
 292                         if ( isset( $attribs['type'] ) && !in_array( $attribs['type'], $validTypes ) ) {
 293                                 unset( $attribs['type'] );
 294                         }
 295                 }
 296
 297                 // According to standard the default type for <button> elements is "submit".
 298                 // Depending on compatibility mode IE might use "button", instead.
 299                 // We enforce the standard "submit".
 300                 if ( $element == 'button' && !isset( $attribs['type'] ) ) {
 301                         $attribs['type'] = 'submit';
 302                 }
 303
 304                 return "<$element" . self::expandAttributes(
 305                         self::dropDefaults( $element, $attribs ) ) . '>';
 306         }
 307
 308         /**
 309          * Returns "</$element>"
 310          *
 311          * @since 1.17
 312          * @param string $element Name of the element, e.g., 'a'
 313          * @return string A closing tag
 314          */
 315         public static function closeElement( $element ) {
 316                 $element = strtolower( $element );
 317
 318                 return "</$element>";
 319         }
 320
 321         /**
 322          * Given an element name and an associative array of element attributes,
 323          * return an array that is functionally identical to the input array, but
 324          * possibly smaller.  In particular, attributes might be stripped if they
 325          * are given their default values.
 326          *
 327          * This method is not guaranteed to remove all redundant attributes, only
 328          * some common ones and some others selected arbitrarily at random.  It
 329          * only guarantees that the output array should be functionally identical
 330          * to the input array (currently per the HTML 5 draft as of 2009-09-06).
 331          *
 332          * @param string $element Name of the element, e.g., 'a'
 333          * @param array $attribs Associative array of attributes, e.g., [
 334          *   'href' => 'https://www.mediawiki.org/' ].  See expandAttributes() for
 335          *   further documentation.
 336          * @return array An array of attributes functionally identical to $attribs
 337          */
 338         private static function dropDefaults( $element, array $attribs ) {
 339                 // Whenever altering this array, please provide a covering test case
 340                 // in HtmlTest::provideElementsWithAttributesHavingDefaultValues
 341                 static $attribDefaults = [
 342                         'area' => [ 'shape' => 'rect' ],
 343                         'button' => [
 344                                 'formaction' => 'GET',
 345                                 'formenctype' => 'application/x-www-form-urlencoded',
 346                         ],
 347                         'canvas' => [
 348                                 'height' => '150',
 349                                 'width' => '300',
 350                         ],
 351                         'form' => [
 352                                 'action' => 'GET',
 353                                 'autocomplete' => 'on',
 354                                 'enctype' => 'application/x-www-form-urlencoded',
 355                         ],
 356                         'input' => [
 357                                 'formaction' => 'GET',
 358                                 'type' => 'text',
 359                         ],
 360                         'keygen' => [ 'keytype' => 'rsa' ],
 361                         'link' => [ 'media' => 'all' ],
 362                         'menu' => [ 'type' => 'list' ],
 363                         'script' => [ 'type' => 'text/javascript' ],
 364                         'style' => [
 365                                 'media' => 'all',
 366                                 'type' => 'text/css',
 367                         ],
 368                         'textarea' => [ 'wrap' => 'soft' ],
 369                 ];
 370
 371                 $element = strtolower( $element );
 372
 373                 foreach ( $attribs as $attrib => $value ) {
 374                         $lcattrib = strtolower( $attrib );
 375                         if ( is_array( $value ) ) {
 376                                 $value = implode( ' ', $value );
 377                         } else {
 378                                 $value = strval( $value );
 379                         }
 380
 381                         // Simple checks using $attribDefaults
 382                         if ( isset( $attribDefaults[$element][$lcattrib] )
 383                                 && $attribDefaults[$element][$lcattrib] == $value
 384                         ) {
 385                                 unset( $attribs[$attrib] );
 386                         }
 387
 388                         if ( $lcattrib == 'class' && $value == '' ) {
 389                                 unset( $attribs[$attrib] );
 390                         }
 391                 }
 392
 393                 // More subtle checks
 394                 if ( $element === 'link'
 395                         && isset( $attribs['type'] ) && strval( $attribs['type'] ) == 'text/css'
 396                 ) {
 397                         unset( $attribs['type'] );
 398                 }
 399                 if ( $element === 'input' ) {
 400                         $type = $attribs['type'] ?? null;
 401                         $value = $attribs['value'] ?? null;
 402                         if ( $type === 'checkbox' || $type === 'radio' ) {
 403                                 // The default value for checkboxes and radio buttons is 'on'
 404                                 // not ''. By stripping value="" we break radio boxes that
 405                                 // actually wants empty values.
 406                                 if ( $value === 'on' ) {
 407                                         unset( $attribs['value'] );
 408                                 }
 409                         } elseif ( $type === 'submit' ) {
 410                                 // The default value for submit appears to be "Submit" but
 411                                 // let's not bother stripping out localized text that matches
 412                                 // that.
 413                         } else {
 414                                 // The default value for nearly every other field type is ''
 415                                 // The 'range' and 'color' types use different defaults but
 416                                 // stripping a value="" does not hurt them.
 417                                 if ( $value === '' ) {
 418                                         unset( $attribs['value'] );
 419                                 }
 420                         }
 421                 }
 422                 if ( $element === 'select' && isset( $attribs['size'] ) ) {
 423                         if ( in_array( 'multiple', $attribs )
 424                                 || ( isset( $attribs['multiple'] ) && $attribs['multiple'] !== false )
 425                         ) {
 426                                 // A multi-select
 427                                 if ( strval( $attribs['size'] ) == '4' ) {
 428                                         unset( $attribs['size'] );
 429                                 }
 430                         } else {
 431                                 // Single select
 432                                 if ( strval( $attribs['size'] ) == '1' ) {
 433                                         unset( $attribs['size'] );
 434                                 }
 435                         }
 436                 }
 437
 438                 return $attribs;
 439         }
 440
 441         /**
 442          * Given an associative array of element attributes, generate a string
 443          * to stick after the element name in HTML output.  Like [ 'href' =>
 444          * 'https://www.mediawiki.org/' ] becomes something like
 445          * ' href="https://www.mediawiki.org"'.  Again, this is like
 446          * Xml::expandAttributes(), but it implements some HTML-specific logic.
 447          *
 448          * Attributes that can contain space-separated lists ('class', 'accesskey' and 'rel') array
 449          * values are allowed as well, which will automagically be normalized
 450          * and converted to a space-separated string. In addition to a numerical
 451          * array, the attribute value may also be an associative array. See the
 452          * example below for how that works.
 453          *
 454          * @par Numerical array
 455          * @code
 456          *     Html::element( 'em', [
 457          *         'class' => [ 'foo', 'bar' ]
 458          *     ] );
 459          *     // gives '<em class="foo bar"></em>'
 460          * @endcode
 461          *
 462          * @par Associative array
 463          * @code
 464          *     Html::element( 'em', [
 465          *         'class' => [ 'foo', 'bar', 'foo' => false, 'quux' => true ]
 466          *     ] );
 467          *     // gives '<em class="bar quux"></em>'
 468          * @endcode
 469          *
 470          * @param array $attribs Associative array of attributes, e.g., [
 471          *   'href' => 'https://www.mediawiki.org/' ].  Values will be HTML-escaped.
 472          *   A value of false or null means to omit the attribute.  For boolean attributes,
 473          *   you can omit the key, e.g., [ 'checked' ] instead of
 474          *   [ 'checked' => 'checked' ] or such.
 475          *
 476          * @throws MWException If an attribute that doesn't allow lists is set to an array
 477          * @return string HTML fragment that goes between element name and '>'
 478          *   (starting with a space if at least one attribute is output)
 479          */
 480         public static function expandAttributes( array $attribs ) {
 481                 $ret = '';
 482                 foreach ( $attribs as $key => $value ) {
 483                         // Support intuitive [ 'checked' => true/false ] form
 484                         if ( $value === false || is_null( $value ) ) {
 485                                 continue;
 486                         }
 487
 488                         // For boolean attributes, support [ 'foo' ] instead of
 489                         // requiring [ 'foo' => 'meaningless' ].
 490                         if ( is_int( $key ) && in_array( strtolower( $value ), self::$boolAttribs ) ) {
 491                                 $key = $value;
 492                         }
 493
 494                         // Not technically required in HTML5 but we'd like consistency
 495                         // and better compression anyway.
 496                         $key = strtolower( $key );
 497
 498                         // https://www.w3.org/TR/html401/index/attributes.html ("space-separated")
 499                         // https://www.w3.org/TR/html5/index.html#attributes-1 ("space-separated")
 500                         $spaceSeparatedListAttributes = [
 501                                 'class', // html4, html5
 502                                 'accesskey', // as of html5, multiple space-separated values allowed
 503                                 // html4-spec doesn't document rel= as space-separated
 504                                 // but has been used like that and is now documented as such
 505                                 // in the html5-spec.
 506                                 'rel',
 507                         ];
 508
 509                         // Specific features for attributes that allow a list of space-separated values
 510                         if ( in_array( $key, $spaceSeparatedListAttributes ) ) {
 511                                 // Apply some normalization and remove duplicates
 512
 513                                 // Convert into correct array. Array can contain space-separated
 514                                 // values. Implode/explode to get those into the main array as well.
 515                                 if ( is_array( $value ) ) {
 516                                         // If input wasn't an array, we can skip this step
 517                                         $newValue = [];
 518                                         foreach ( $value as $k => $v ) {
 519                                                 if ( is_string( $v ) ) {
 520                                                         // String values should be normal `[ 'foo' ]`
 521                                                         // Just append them
 522                                                         if ( !isset( $value[$v] ) ) {
 523                                                                 // As a special case don't set 'foo' if a
 524                                                                 // separate 'foo' => true/false exists in the array
 525                                                                 // keys should be authoritative
 526                                                                 $newValue[] = $v;
 527                                                         }
 528                                                 } elseif ( $v ) {
 529                                                         // If the value is truthy but not a string this is likely
 530                                                         // an [ 'foo' => true ], falsy values don't add strings
 531                                                         $newValue[] = $k;
 532                                                 }
 533                                         }
 534                                         $value = implode( ' ', $newValue );
 535                                 }
 536                                 $value = explode( ' ', $value );
 537
 538                                 // Normalize spacing by fixing up cases where people used
 539                                 // more than 1 space and/or a trailing/leading space
 540                                 $value = array_diff( $value, [ '', ' ' ] );
 541
 542                                 // Remove duplicates and create the string
 543                                 $value = implode( ' ', array_unique( $value ) );
 544                         } elseif ( is_array( $value ) ) {
 545                                 throw new MWException( "HTML attribute $key can not contain a list of values" );
 546                         }
 547
 548                         $quote = '"';
 549
 550                         if ( in_array( $key, self::$boolAttribs ) ) {
 551                                 $ret .= " $key=\"\"";
 552                         } else {
 553                                 $ret .= " $key=$quote" . Sanitizer::encodeAttribute( $value ) . $quote;
 554                         }
 555                 }
 556                 return $ret;
 557         }
 558
 559         /**
 560          * Output an HTML script tag with the given contents.
 561          *
 562          * It is unsupported for the contents to contain the sequence `<script` or `</script`
 563          * (case-insensitive). This ensures the script can be terminated easily and consistently.
 564          * It is the responsibility of the caller to avoid such character sequence by escaping
 565          * or avoiding it. If found at run-time, the contents are replaced with a comment, and
 566          * a warning is logged server-side.
 567          *
 568          * @param string $contents JavaScript
 569          * @param string|null $nonce Nonce for CSP header, from OutputPage::getCSPNonce()
 570          * @return string Raw HTML
 571          */
 572         public static function inlineScript( $contents, $nonce = null ) {
 573                 $attrs = [];
 574                 if ( $nonce !== null ) {
 575                         $attrs['nonce'] = $nonce;
 576                 } elseif ( ContentSecurityPolicy::isNonceRequired( RequestContext::getMain()->getConfig() ) ) {
 577                         wfWarn( "no nonce set on script. CSP will break it" );
 578                 }
 579
 580                 if ( preg_match( '/<\/?script/i', $contents ) ) {
 581                         wfLogWarning( __METHOD__ . ': Illegal character sequence found in inline script.' );
 582                         $contents = '/* ERROR: Invalid script */';
 583                 }
 584
 585                 return self::rawElement( 'script', $attrs, $contents );
 586         }
 587
 588         /**
 589          * Output a "<script>" tag linking to the given URL, e.g.,
 590          * "<script src=foo.js></script>".
 591          *
 592          * @param string $url
 593          * @param string|null $nonce Nonce for CSP header, from OutputPage::getCSPNonce()
 594          * @return string Raw HTML
 595          */
 596         public static function linkedScript( $url, $nonce = null ) {
 597                 $attrs = [ 'src' => $url ];
 598                 if ( $nonce !== null ) {
 599                         $attrs['nonce'] = $nonce;
 600                 } elseif ( ContentSecurityPolicy::isNonceRequired( RequestContext::getMain()->getConfig() ) ) {
 601                         wfWarn( "no nonce set on script. CSP will break it" );
 602                 }
 603
 604                 return self::element( 'script', $attrs );
 605         }
 606
 607         /**
 608          * Output a "<style>" tag with the given contents for the given media type
 609          * (if any).  TODO: do some useful escaping as well, like if $contents
 610          * contains literal "</style>" (admittedly unlikely).
 611          *
 612          * @param string $contents CSS
 613          * @param string $media A media type string, like 'screen'
 614          * @param array $attribs (since 1.31) Associative array of attributes, e.g., [
 615          *   'href' => 'https://www.mediawiki.org/' ]. See expandAttributes() for
 616          *   further documentation.
 617          * @return string Raw HTML
 618          */
 619         public static function inlineStyle( $contents, $media = 'all', $attribs = [] ) {
 620                 // Don't escape '>' since that is used
 621                 // as direct child selector.
 622                 // Remember, in css, there is no "x" for hexadecimal escapes, and
 623                 // the space immediately after an escape sequence is swallowed.
 624                 $contents = strtr( $contents, [
 625                         '<' => '\3C ',
 626                         // CDATA end tag for good measure, but the main security
 627                         // is from escaping the '<'.
 628                         ']]>' => '\5D\5D\3E '
 629                 ] );
 630
 631                 if ( preg_match( '/[<&]/', $contents ) ) {
 632                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 633                 }
 634
 635                 return self::rawElement( 'style', [
 636                         'media' => $media,
 637                 ] + $attribs, $contents );
 638         }
 639
 640         /**
 641          * Output a "<link rel=stylesheet>" linking to the given URL for the given
 642          * media type (if any).
 643          *
 644          * @param string $url
 645          * @param string $media A media type string, like 'screen'
 646          * @return string Raw HTML
 647          */
 648         public static function linkedStyle( $url, $media = 'all' ) {
 649                 return self::element( 'link', [
 650                         'rel' => 'stylesheet',
 651                         'href' => $url,
 652                         'media' => $media,
 653                 ] );
 654         }
 655
 656         /**
 657          * Convenience function to produce an "<input>" element.  This supports the
 658          * new HTML5 input types and attributes.
 659          *
 660          * @param string $name Name attribute
 661          * @param string $value Value attribute
 662          * @param string $type Type attribute
 663          * @param array $attribs Associative array of miscellaneous extra
 664          *   attributes, passed to Html::element()
 665          * @return string Raw HTML
 666          */
 667         public static function input( $name, $value = '', $type = 'text', array $attribs = [] ) {
 668                 $attribs['type'] = $type;
 669                 $attribs['value'] = $value;
 670                 $attribs['name'] = $name;
 671                 if ( in_array( $type, [ 'text', 'search', 'email', 'password', 'number' ] ) ) {
 672                         $attribs = self::getTextInputAttributes( $attribs );
 673                 }
 674                 if ( in_array( $type, [ 'button', 'reset', 'submit' ] ) ) {
 675                         $attribs = self::buttonAttributes( $attribs );
 676                 }
 677                 return self::element( 'input', $attribs );
 678         }
 679
 680         /**
 681          * Convenience function to produce a checkbox (input element with type=checkbox)
 682          *
 683          * @param string $name Name attribute
 684          * @param bool $checked Whether the checkbox is checked or not
 685          * @param array $attribs Array of additional attributes
 686          * @return string Raw HTML
 687          */
 688         public static function check( $name, $checked = false, array $attribs = [] ) {
 689                 if ( isset( $attribs['value'] ) ) {
 690                         $value = $attribs['value'];
 691                         unset( $attribs['value'] );
 692                 } else {
 693                         $value = 1;
 694                 }
 695
 696                 if ( $checked ) {
 697                         $attribs[] = 'checked';
 698                 }
 699
 700                 return self::input( $name, $value, 'checkbox', $attribs );
 701         }
 702
 703         /**
 704          * Return the HTML for a message box.
 705          * @since 1.31
 706          * @param string $html of contents of box
 707          * @param string $className corresponding to box
 708          * @param string $heading (optional)
 709          * @return string of HTML representing a box.
 710          */
 711         private static function messageBox( $html, $className, $heading = '' ) {
 712                 if ( $heading !== '' ) {
 713                         $html = self::element( 'h2', [], $heading ) . $html;
 714                 }
 715                 return self::rawElement( 'div', [ 'class' => $className ], $html );
 716         }
 717
 718         /**
 719          * Return a warning box.
 720          * @since 1.31
 721          * @param string $html of contents of box
 722          * @return string of HTML representing a warning box.
 723          */
 724         public static function warningBox( $html ) {
 725                 return self::messageBox( $html, 'warningbox' );
 726         }
 727
 728         /**
 729          * Return an error box.
 730          * @since 1.31
 731          * @param string $html of contents of error box
 732          * @param string $heading (optional)
 733          * @return string of HTML representing an error box.
 734          */
 735         public static function errorBox( $html, $heading = '' ) {
 736                 return self::messageBox( $html, 'errorbox', $heading );
 737         }
 738
 739         /**
 740          * Return a success box.
 741          * @since 1.31
 742          * @param string $html of contents of box
 743          * @return string of HTML representing a success box.
 744          */
 745         public static function successBox( $html ) {
 746                 return self::messageBox( $html, 'successbox' );
 747         }
 748
 749         /**
 750          * Convenience function to produce a radio button (input element with type=radio)
 751          *
 752          * @param string $name Name attribute
 753          * @param bool $checked Whether the radio button is checked or not
 754          * @param array $attribs Array of additional attributes
 755          * @return string Raw HTML
 756          */
 757         public static function radio( $name, $checked = false, array $attribs = [] ) {
 758                 if ( isset( $attribs['value'] ) ) {
 759                         $value = $attribs['value'];
 760                         unset( $attribs['value'] );
 761                 } else {
 762                         $value = 1;
 763                 }
 764
 765                 if ( $checked ) {
 766                         $attribs[] = 'checked';
 767                 }
 768
 769                 return self::input( $name, $value, 'radio', $attribs );
 770         }
 771
 772         /**
 773          * Convenience function for generating a label for inputs.
 774          *
 775          * @param string $label Contents of the label
 776          * @param string $id ID of the element being labeled
 777          * @param array $attribs Additional attributes
 778          * @return string Raw HTML
 779          */
 780         public static function label( $label, $id, array $attribs = [] ) {
 781                 $attribs += [
 782                         'for' => $id
 783                 ];
 784                 return self::element( 'label', $attribs, $label );
 785         }
 786
 787         /**
 788          * Convenience function to produce an input element with type=hidden
 789          *
 790          * @param string $name Name attribute
 791          * @param string $value Value attribute
 792          * @param array $attribs Associative array of miscellaneous extra
 793          *   attributes, passed to Html::element()
 794          * @return string Raw HTML
 795          */
 796         public static function hidden( $name, $value, array $attribs = [] ) {
 797                 return self::input( $name, $value, 'hidden', $attribs );
 798         }
 799
 800         /**
 801          * Convenience function to produce a <textarea> element.
 802          *
 803          * This supports leaving out the cols= and rows= which Xml requires and are
 804          * required by HTML4/XHTML but not required by HTML5.
 805          *
 806          * @param string $name Name attribute
 807          * @param string $value Value attribute
 808          * @param array $attribs Associative array of miscellaneous extra
 809          *   attributes, passed to Html::element()
 810          * @return string Raw HTML
 811          */
 812         public static function textarea( $name, $value = '', array $attribs = [] ) {
 813                 $attribs['name'] = $name;
 814
 815                 if ( substr( $value, 0, 1 ) == "\n" ) {
 816                         // Workaround for T14130: browsers eat the initial newline
 817                         // assuming that it's just for show, but they do keep the later
 818                         // newlines, which we may want to preserve during editing.
 819                         // Prepending a single newline
 820                         $spacedValue = "\n" . $value;
 821                 } else {
 822                         $spacedValue = $value;
 823                 }
 824                 return self::element( 'textarea', self::getTextInputAttributes( $attribs ), $spacedValue );
 825         }
 826
 827         /**
 828          * Helper for Html::namespaceSelector().
 829          * @param array $params See Html::namespaceSelector()
 830          * @return array
 831          */
 832         public static function namespaceSelectorOptions( array $params = [] ) {
 833                 if ( !isset( $params['exclude'] ) || !is_array( $params['exclude'] ) ) {
 834                         $params['exclude'] = [];
 835                 }
 836
 837                 if ( $params['in-user-lang'] ?? false ) {
 838                         global $wgLang;
 839                         $lang = $wgLang;
 840                 } else {
 841                         $lang = MediaWikiServices::getInstance()->getContentLanguage();
 842                 }
 843
 844                 $optionsOut = [];
 845                 if ( isset( $params['all'] ) ) {
 846                         // add an option that would let the user select all namespaces.
 847                         // Value is provided by user, the name shown is localized for the user.
 848                         $optionsOut[$params['all']] = wfMessage( 'namespacesall' )->text();
 849                 }
 850                 // Add all namespaces as options
 851                 $options = $lang->getFormattedNamespaces();
 852                 // Filter out namespaces below 0 and massage labels
 853                 foreach ( $options as $nsId => $nsName ) {
 854                         if ( $nsId < NS_MAIN || in_array( $nsId, $params['exclude'] ) ) {
 855                                 continue;
 856                         }
 857                         if ( $nsId === NS_MAIN ) {
 858                                 // For other namespaces use the namespace prefix as label, but for
 859                                 // main we don't use "" but the user message describing it (e.g. "(Main)" or "(Article)")
 860                                 $nsName = wfMessage( 'blanknamespace' )->text();
 861                         } elseif ( is_int( $nsId ) ) {
 862                                 $nsName = $lang->convertNamespace( $nsId );
 863                         }
 864                         $optionsOut[$nsId] = $nsName;
 865                 }
 866
 867                 return $optionsOut;
 868         }
 869
 870         /**
 871          * Build a drop-down box for selecting a namespace
 872          *
 873          * @param array $params Params to set.
 874          * - selected: [optional] Id of namespace which should be pre-selected
 875          * - all: [optional] Value of item for "all namespaces". If null or unset,
 876          *   no "<option>" is generated to select all namespaces.
 877          * - label: text for label to add before the field.
 878          * - exclude: [optional] Array of namespace ids to exclude.
 879          * - disable: [optional] Array of namespace ids for which the option should
 880          *   be disabled in the selector.
 881          * @param array $selectAttribs HTML attributes for the generated select element.
 882          * - id:   [optional], default: 'namespace'.
 883          * - name: [optional], default: 'namespace'.
 884          * @return string HTML code to select a namespace.
 885          */
 886         public static function namespaceSelector( array $params = [],
 887                 array $selectAttribs = []
 888         ) {
 889                 ksort( $selectAttribs );
 890
 891                 // Is a namespace selected?
 892                 if ( isset( $params['selected'] ) ) {
 893                         // If string only contains digits, convert to clean int. Selected could also
 894                         // be "all" or "" etc. which needs to be left untouched.
 895                         // PHP is_numeric() has issues with large strings, PHP ctype_digit has other issues
 896                         // and returns false for already clean ints. Use regex instead..
 897                         if ( preg_match( '/^\d+$/', $params['selected'] ) ) {
 898                                 $params['selected'] = intval( $params['selected'] );
 899                         }
 900                         // else: leaves it untouched for later processing
 901                 } else {
 902                         $params['selected'] = '';
 903                 }
 904
 905                 if ( !isset( $params['disable'] ) || !is_array( $params['disable'] ) ) {
 906                         $params['disable'] = [];
 907                 }
 908
 909                 // Associative array between option-values and option-labels
 910                 $options = self::namespaceSelectorOptions( $params );
 911
 912                 // Convert $options to HTML
 913                 $optionsHtml = [];
 914                 foreach ( $options as $nsId => $nsName ) {
 915                         $optionsHtml[] = self::element(
 916                                 'option', [
 917                                         'disabled' => in_array( $nsId, $params['disable'] ),
 918                                         'value' => $nsId,
 919                                         'selected' => $nsId === $params['selected'],
 920                                 ], $nsName
 921                         );
 922                 }
 923
 924                 if ( !array_key_exists( 'id', $selectAttribs ) ) {
 925                         $selectAttribs['id'] = 'namespace';
 926                 }
 927
 928                 if ( !array_key_exists( 'name', $selectAttribs ) ) {
 929                         $selectAttribs['name'] = 'namespace';
 930                 }
 931
 932                 $ret = '';
 933                 if ( isset( $params['label'] ) ) {
 934                         $ret .= self::element(
 935                                 'label', [
 936                                         'for' => $selectAttribs['id'] ?? null,
 937                                 ], $params['label']
 938                         ) . "\u{00A0}";
 939                 }
 940
 941                 // Wrap options in a <select>
 942                 $ret .= self::openElement( 'select', $selectAttribs )
 943                         . "\n"
 944                         . implode( "\n", $optionsHtml )
 945                         . "\n"
 946                         . self::closeElement( 'select' );
 947
 948                 return $ret;
 949         }
 950
 951         /**
 952          * Constructs the opening html-tag with necessary doctypes depending on
 953          * global variables.
 954          *
 955          * @param array $attribs Associative array of miscellaneous extra
 956          *   attributes, passed to Html::element() of html tag.
 957          * @return string Raw HTML
 958          */
 959         public static function htmlHeader( array $attribs = [] ) {
 960                 $ret = '';
 961
 962                 global $wgHtml5Version, $wgMimeType, $wgXhtmlNamespaces;
 963
 964                 $isXHTML = self::isXmlMimeType( $wgMimeType );
 965
 966                 if ( $isXHTML ) { // XHTML5
 967                         // XML MIME-typed markup should have an xml header.
 968                         // However a DOCTYPE is not needed.
 969                         $ret .= "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
 970
 971                         // Add the standard xmlns
 972                         $attribs['xmlns'] = 'http://www.w3.org/1999/xhtml';
 973
 974                         // And support custom namespaces
 975                         foreach ( $wgXhtmlNamespaces as $tag => $ns ) {
 976                                 $attribs["xmlns:$tag"] = $ns;
 977                         }
 978                 } else { // HTML5
 979                         $ret .= "<!DOCTYPE html>\n";
 980                 }
 981
 982                 if ( $wgHtml5Version ) {
 983                         $attribs['version'] = $wgHtml5Version;
 984                 }
 985
 986                 $ret .= self::openElement( 'html', $attribs );
 987
 988                 return $ret;
 989         }
 990
 991         /**
 992          * Determines if the given MIME type is xml.
 993          *
 994          * @param string $mimetype MIME type
 995          * @return bool
 996          */
 997         public static function isXmlMimeType( $mimetype ) {
 998                 # https://html.spec.whatwg.org/multipage/infrastructure.html#xml-mime-type
 999                 # * text/xml
1000                 # * application/xml
1001                 # * Any MIME type with a subtype ending in +xml (this implicitly includes application/xhtml+xml)
1002                 return (bool)preg_match( '!^(text|application)/xml$|^.+/.+\+xml$!', $mimetype );
1003         }
1004
1005         /**
1006          * Get HTML for an information message box with an icon.
1007          *
1008          * @internal For use by the WebInstaller class.
1009          * @param string $rawHtml HTML
1010          * @param string $icon Path to icon file (used as 'src' attribute)
1011          * @param string $alt Alternate text for the icon
1012          * @param string $class Additional class name to add to the wrapper div
1013          * @return string HTML
1014          */
1015         public static function infoBox( $rawHtml, $icon, $alt, $class = '' ) {
1016                 $s = self::openElement( 'div', [ 'class' => "mw-infobox $class" ] );
1017
1018                 $s .= self::openElement( 'div', [ 'class' => 'mw-infobox-left' ] ) .
1019                                 self::element( 'img',
1020                                         [
1021                                                 'src' => $icon,
1022                                                 'alt' => $alt,
1023                                         ]
1024                                 ) .
1025                                 self::closeElement( 'div' );
1026
1027                 $s .= self::openElement( 'div', [ 'class' => 'mw-infobox-right' ] ) .
1028                                 $rawHtml .
1029                                 self::closeElement( 'div' );
1030                 $s .= self::element( 'div', [ 'style' => 'clear: left;' ], ' ' );
1031
1032                 $s .= self::closeElement( 'div' );
1033
1034                 $s .= self::element( 'div', [ 'style' => 'clear: left;' ], ' ' );
1035
1036                 return $s;
1037         }
1038
1039         /**
1040          * Generate a srcset attribute value.
1041          *
1042          * Generates a srcset attribute value from an array mapping pixel densities
1043          * to URLs. A trailing 'x' in pixel density values is optional.
1044          *
1045          * @note srcset width and height values are not supported.
1046          *
1047          * @see https://html.spec.whatwg.org/#attr-img-srcset
1048          *
1049          * @par Example:
1050          * @code
1051          *     Html::srcSet( [
1052          *         '1x'   => 'standard.jpeg',
1053          *         '1.5x' => 'large.jpeg',
1054          *         '3x'   => 'extra-large.jpeg',
1055          *     ] );
1056          *     // gives 'standard.jpeg 1x, large.jpeg 1.5x, extra-large.jpeg 2x'
1057          * @endcode
1058          *
1059          * @param string[] $urls
1060          * @return string
1061          */
1062         static function srcSet( array $urls ) {
1063                 $candidates = [];
1064                 foreach ( $urls as $density => $url ) {
1065                         // Cast density to float to strip 'x', then back to string to serve
1066                         // as array index.
1067                         $density = (string)(float)$density;
1068                         $candidates[$density] = $url;
1069                 }
1070
1071                 // Remove duplicates that are the same as a smaller value
1072                 ksort( $candidates, SORT_NUMERIC );
1073                 $candidates = array_unique( $candidates );
1074
1075                 // Append density info to the url
1076                 foreach ( $candidates as $density => $url ) {
1077                         $candidates[$density] = $url . ' ' . $density . 'x';
1078                 }
1079
1080                 return implode( ", ", $candidates );
1081         }
1082 }