includes/Html.php

   1 <?php
   2 /**
   3  * Collection of methods to generate HTML content
   4  *
   5  * Copyright © 2009 Aryeh Gregor
   6  * http://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
  26 /**
  27  * This class is a collection of static functions that serve two purposes:
  28  *
  29  * 1) Implement any algorithms specified by HTML5, or other HTML
  30  * specifications, in a convenient and self-contained way.
  31  *
  32  * 2) Allow HTML elements to be conveniently and safely generated, like the
  33  * current Xml class but a) less confused (Xml supports HTML-specific things,
  34  * but only sometimes!) and b) not necessarily confined to XML-compatible
  35  * output.
  36  *
  37  * There are two important configuration options this class uses:
  38  *
  39  * $wgMimeType: If this is set to an xml MIME type then output should be
  40  *     valid XHTML5.
  41  * $wgWellFormedXml: If this is set to true, then all output should be
  42  *     well-formed XML (quotes on attributes, self-closing tags, etc.).
  43  *
  44  * This class is meant to be confined to utility functions that are called from
  45  * trusted code paths.  It does not do enforcement of policy like not allowing
  46  * <a> elements.
  47  *
  48  * @since 1.16
  49  */
  50 class Html {
  51         // List of void elements from HTML5, section 8.1.2 as of 2011-08-12
  52         private static $voidElements = array(
  53                 'area',
  54                 'base',
  55                 'br',
  56                 'col',
  57                 'command',
  58                 'embed',
  59                 'hr',
  60                 'img',
  61                 'input',
  62                 'keygen',
  63                 'link',
  64                 'meta',
  65                 'param',
  66                 'source',
  67                 'track',
  68                 'wbr',
  69         );
  70
  71         // Boolean attributes, which may have the value omitted entirely.  Manually
  72         // collected from the HTML5 spec as of 2011-08-12.
  73         private static $boolAttribs = array(
  74                 'async',
  75                 'autofocus',
  76                 'autoplay',
  77                 'checked',
  78                 'controls',
  79                 'default',
  80                 'defer',
  81                 'disabled',
  82                 'formnovalidate',
  83                 'hidden',
  84                 'ismap',
  85                 'itemscope',
  86                 'loop',
  87                 'multiple',
  88                 'muted',
  89                 'novalidate',
  90                 'open',
  91                 'pubdate',
  92                 'readonly',
  93                 'required',
  94                 'reversed',
  95                 'scoped',
  96                 'seamless',
  97                 'selected',
  98                 'truespeed',
  99                 'typemustmatch',
 100                 // HTML5 Microdata
 101                 'itemscope',
 102         );
 103
 104         /**
 105          * Modifies a set of attributes meant for button elements
 106          * and apply a set of default attributes when $wgUseMediaWikiUIEverywhere enabled.
 107          * @param array $attrs
 108          * @param string[] $modifiers to add to the button
 109          * @see https://tools.wmflabs.org/styleguide/desktop/index.html for guidance on available modifiers
 110          * @return array $attrs A modified attribute array
 111          */
 112         public static function buttonAttributes( $attrs, $modifiers = array() ) {
 113                 global $wgUseMediaWikiUIEverywhere;
 114                 if ( $wgUseMediaWikiUIEverywhere ) {
 115                         if ( isset( $attrs['class'] ) ) {
 116                                 if ( is_array( $attrs['class'] ) ) {
 117                                         $attrs['class'][] = 'mw-ui-button';
 118                                         $attrs = array_merge( $attrs, $modifiers );
 119                                         // ensure compatibility with Xml
 120                                         $attrs['class'] = implode( ' ', $attrs['class'] );
 121                                 } else {
 122                                         $attrs['class'] .= ' mw-ui-button ' . implode( ' ', $modifiers );
 123                                 }
 124                         } else {
 125                                 $attrs['class'] = array( 'mw-ui-button' );
 126                                 // ensure compatibility with Xml
 127                                 $attrs['class'] = implode( ' ', array_merge( $attrs['class'], $modifiers ) );
 128                         }
 129                 }
 130                 return $attrs;
 131         }
 132
 133         /**
 134          * Modifies a set of attributes meant for text input elements
 135          * and apply a set of default attributes.
 136          * Removes size attribute when $wgUseMediaWikiUIEverywhere enabled.
 137          * @param array $attrs An attribute array.
 138          * @return array $attrs A modified attribute array
 139          */
 140         public static function getTextInputAttributes( $attrs ) {
 141                 global $wgUseMediaWikiUIEverywhere;
 142                 if ( !$attrs ) {
 143                         $attrs = array();
 144                 }
 145                 if ( $wgUseMediaWikiUIEverywhere ) {
 146                         if ( isset( $attrs['class'] ) ) {
 147                                 if ( is_array( $attrs['class'] ) ) {
 148                                         $attrs['class'][] = 'mw-ui-input';
 149                                 } else {
 150                                         $attrs['class'] .= ' mw-ui-input';
 151                                 }
 152                         } else {
 153                                 $attrs['class'] = 'mw-ui-input';
 154                         }
 155                 }
 156                 return $attrs;
 157         }
 158
 159         /**
 160          * Returns an HTML link element in a string styled as a button
 161          * (when $wgUseMediaWikiUIEverywhere is enabled).
 162          *
 163          * @param string $contents The raw HTML contents of the element: *not*
 164          *   escaped!
 165          * @param array $attrs Associative array of attributes, e.g., array(
 166          *   'href' => 'http://www.mediawiki.org/' ). See expandAttributes() for
 167          *   further documentation.
 168          * @param string[] $modifiers to add to the button
 169          * @see http://tools.wmflabs.org/styleguide/desktop/index.html for guidance on available modifiers
 170          * @return string Raw HTML
 171          */
 172         public static function linkButton( $contents, $attrs, $modifiers = array() ) {
 173                 return self::element( 'a',
 174                         self::buttonAttributes( $attrs, $modifiers ),
 175                         $contents
 176                 );
 177         }
 178
 179         /**
 180          * Returns an HTML link element in a string styled as a button
 181          * (when $wgUseMediaWikiUIEverywhere is enabled).
 182          *
 183          * @param string $contents The raw HTML contents of the element: *not*
 184          *   escaped!
 185          * @param array $attrs Associative array of attributes, e.g., array(
 186          *   'href' => 'http://www.mediawiki.org/' ). See expandAttributes() for
 187          *   further documentation.
 188          * @param string[] $modifiers to add to the button
 189          * @see http://tools.wmflabs.org/styleguide/desktop/index.html for guidance on available modifiers
 190          * @return string Raw HTML
 191          */
 192         public static function submitButton( $contents, $attrs, $modifiers = array() ) {
 193                 $attrs['type'] = 'submit';
 194                 $attrs['value'] = $contents;
 195                 return self::element( 'input', self::buttonAttributes( $attrs, $modifiers ) );
 196         }
 197
 198         /**
 199          * Returns an HTML element in a string.  The major advantage here over
 200          * manually typing out the HTML is that it will escape all attribute
 201          * values.
 202          *
 203          * This is quite similar to Xml::tags(), but it implements some useful
 204          * HTML-specific logic.  For instance, there is no $allowShortTag
 205          * parameter: the closing tag is magically omitted if $element has an empty
 206          * content model.  If $wgWellFormedXml is false, then a few bytes will be
 207          * shaved off the HTML output as well.
 208          *
 209          * @param string $element The element's name, e.g., 'a'
 210          * @param array $attribs Associative array of attributes, e.g., array(
 211          *   'href' => 'http://www.mediawiki.org/' ). See expandAttributes() for
 212          *   further documentation.
 213          * @param string $contents The raw HTML contents of the element: *not*
 214          *   escaped!
 215          * @return string Raw HTML
 216          */
 217         public static function rawElement( $element, $attribs = array(), $contents = '' ) {
 218                 global $wgWellFormedXml;
 219                 $start = self::openElement( $element, $attribs );
 220                 if ( in_array( $element, self::$voidElements ) ) {
 221                         if ( $wgWellFormedXml ) {
 222                                 // Silly XML.
 223                                 return substr( $start, 0, -1 ) . ' />';
 224                         }
 225                         return $start;
 226                 } else {
 227                         return "$start$contents" . self::closeElement( $element );
 228                 }
 229         }
 230
 231         /**
 232          * Identical to rawElement(), but HTML-escapes $contents (like
 233          * Xml::element()).
 234          *
 235          * @param string $element
 236          * @param array $attribs
 237          * @param string $contents
 238          *
 239          * @return string
 240          */
 241         public static function element( $element, $attribs = array(), $contents = '' ) {
 242                 return self::rawElement( $element, $attribs, strtr( $contents, array(
 243                         // There's no point in escaping quotes, >, etc. in the contents of
 244                         // elements.
 245                         '&' => '&amp;',
 246                         '<' => '&lt;'
 247                 ) ) );
 248         }
 249
 250         /**
 251          * Identical to rawElement(), but has no third parameter and omits the end
 252          * tag (and the self-closing '/' in XML mode for empty elements).
 253          *
 254          * @param string $element
 255          * @param array $attribs
 256          *
 257          * @return string
 258          */
 259         public static function openElement( $element, $attribs = array() ) {
 260                 $attribs = (array)$attribs;
 261                 // This is not required in HTML5, but let's do it anyway, for
 262                 // consistency and better compression.
 263                 $element = strtolower( $element );
 264
 265                 // Remove invalid input types
 266                 if ( $element == 'input' ) {
 267                         $validTypes = array(
 268                                 'hidden',
 269                                 'text',
 270                                 'password',
 271                                 'checkbox',
 272                                 'radio',
 273                                 'file',
 274                                 'submit',
 275                                 'image',
 276                                 'reset',
 277                                 'button',
 278
 279                                 // HTML input types
 280                                 'datetime',
 281                                 'datetime-local',
 282                                 'date',
 283                                 'month',
 284                                 'time',
 285                                 'week',
 286                                 'number',
 287                                 'range',
 288                                 'email',
 289                                 'url',
 290                                 'search',
 291                                 'tel',
 292                                 'color',
 293                         );
 294                         if ( isset( $attribs['type'] ) && !in_array( $attribs['type'], $validTypes ) ) {
 295                                 unset( $attribs['type'] );
 296                         }
 297                 }
 298
 299                 // According to standard the default type for <button> elements is "submit".
 300                 // Depending on compatibility mode IE might use "button", instead.
 301                 // We enforce the standard "submit".
 302                 if ( $element == 'button' && !isset( $attribs['type'] ) ) {
 303                         $attribs['type'] = 'submit';
 304                 }
 305
 306                 return "<$element" . self::expandAttributes(
 307                         self::dropDefaults( $element, $attribs ) ) . '>';
 308         }
 309
 310         /**
 311          * Returns "</$element>"
 312          *
 313          * @since 1.17
 314          * @param string $element Name of the element, e.g., 'a'
 315          * @return string A closing tag
 316          */
 317         public static function closeElement( $element ) {
 318                 $element = strtolower( $element );
 319
 320                 return "</$element>";
 321         }
 322
 323         /**
 324          * Given an element name and an associative array of element attributes,
 325          * return an array that is functionally identical to the input array, but
 326          * possibly smaller.  In particular, attributes might be stripped if they
 327          * are given their default values.
 328          *
 329          * This method is not guaranteed to remove all redundant attributes, only
 330          * some common ones and some others selected arbitrarily at random.  It
 331          * only guarantees that the output array should be functionally identical
 332          * to the input array (currently per the HTML 5 draft as of 2009-09-06).
 333          *
 334          * @param string $element Name of the element, e.g., 'a'
 335          * @param array $attribs Associative array of attributes, e.g., array(
 336          *   'href' => 'http://www.mediawiki.org/' ).  See expandAttributes() for
 337          *   further documentation.
 338          * @return array An array of attributes functionally identical to $attribs
 339          */
 340         private static function dropDefaults( $element, $attribs ) {
 341
 342                 // Whenever altering this array, please provide a covering test case
 343                 // in HtmlTest::provideElementsWithAttributesHavingDefaultValues
 344                 static $attribDefaults = array(
 345                         'area' => array( 'shape' => 'rect' ),
 346                         'button' => array(
 347                                 'formaction' => 'GET',
 348                                 'formenctype' => 'application/x-www-form-urlencoded',
 349                         ),
 350                         'canvas' => array(
 351                                 'height' => '150',
 352                                 'width' => '300',
 353                         ),
 354                         'command' => array( 'type' => 'command' ),
 355                         'form' => array(
 356                                 'action' => 'GET',
 357                                 'autocomplete' => 'on',
 358                                 'enctype' => 'application/x-www-form-urlencoded',
 359                         ),
 360                         'input' => array(
 361                                 'formaction' => 'GET',
 362                                 'type' => 'text',
 363                         ),
 364                         'keygen' => array( 'keytype' => 'rsa' ),
 365                         'link' => array( 'media' => 'all' ),
 366                         'menu' => array( 'type' => 'list' ),
 367                         // Note: the use of text/javascript here instead of other JavaScript
 368                         // MIME types follows the HTML5 spec.
 369                         'script' => array( 'type' => 'text/javascript' ),
 370                         'style' => array(
 371                                 'media' => 'all',
 372                                 'type' => 'text/css',
 373                         ),
 374                         'textarea' => array( 'wrap' => 'soft' ),
 375                 );
 376
 377                 $element = strtolower( $element );
 378
 379                 foreach ( $attribs as $attrib => $value ) {
 380                         $lcattrib = strtolower( $attrib );
 381                         if ( is_array( $value ) ) {
 382                                 $value = implode( ' ', $value );
 383                         } else {
 384                                 $value = strval( $value );
 385                         }
 386
 387                         // Simple checks using $attribDefaults
 388                         if ( isset( $attribDefaults[$element][$lcattrib] )
 389                                 && $attribDefaults[$element][$lcattrib] == $value
 390                         ) {
 391                                 unset( $attribs[$attrib] );
 392                         }
 393
 394                         if ( $lcattrib == 'class' && $value == '' ) {
 395                                 unset( $attribs[$attrib] );
 396                         }
 397                 }
 398
 399                 // More subtle checks
 400                 if ( $element === 'link'
 401                         && isset( $attribs['type'] ) && strval( $attribs['type'] ) == 'text/css'
 402                 ) {
 403                         unset( $attribs['type'] );
 404                 }
 405                 if ( $element === 'input' ) {
 406                         $type = isset( $attribs['type'] ) ? $attribs['type'] : null;
 407                         $value = isset( $attribs['value'] ) ? $attribs['value'] : null;
 408                         if ( $type === 'checkbox' || $type === 'radio' ) {
 409                                 // The default value for checkboxes and radio buttons is 'on'
 410                                 // not ''. By stripping value="" we break radio boxes that
 411                                 // actually wants empty values.
 412                                 if ( $value === 'on' ) {
 413                                         unset( $attribs['value'] );
 414                                 }
 415                         } elseif ( $type === 'submit' ) {
 416                                 // The default value for submit appears to be "Submit" but
 417                                 // let's not bother stripping out localized text that matches
 418                                 // that.
 419                         } else {
 420                                 // The default value for nearly every other field type is ''
 421                                 // The 'range' and 'color' types use different defaults but
 422                                 // stripping a value="" does not hurt them.
 423                                 if ( $value === '' ) {
 424                                         unset( $attribs['value'] );
 425                                 }
 426                         }
 427                 }
 428                 if ( $element === 'select' && isset( $attribs['size'] ) ) {
 429                         if ( in_array( 'multiple', $attribs )
 430                                 || ( isset( $attribs['multiple'] ) && $attribs['multiple'] !== false )
 431                         ) {
 432                                 // A multi-select
 433                                 if ( strval( $attribs['size'] ) == '4' ) {
 434                                         unset( $attribs['size'] );
 435                                 }
 436                         } else {
 437                                 // Single select
 438                                 if ( strval( $attribs['size'] ) == '1' ) {
 439                                         unset( $attribs['size'] );
 440                                 }
 441                         }
 442                 }
 443
 444                 return $attribs;
 445         }
 446
 447         /**
 448          * Given an associative array of element attributes, generate a string
 449          * to stick after the element name in HTML output.  Like array( 'href' =>
 450          * 'http://www.mediawiki.org/' ) becomes something like
 451          * ' href="http://www.mediawiki.org"'.  Again, this is like
 452          * Xml::expandAttributes(), but it implements some HTML-specific logic.
 453          * For instance, it will omit quotation marks if $wgWellFormedXml is false,
 454          * and will treat boolean attributes specially.
 455          *
 456          * Attributes that can contain space-separated lists ('class', 'accesskey' and 'rel') array
 457          * values are allowed as well, which will automagically be normalized
 458          * and converted to a space-separated string. In addition to a numerical
 459          * array, the attribute value may also be an associative array. See the
 460          * example below for how that works.
 461          *
 462          * @par Numerical array
 463          * @code
 464          *     Html::element( 'em', array(
 465          *         'class' => array( 'foo', 'bar' )
 466          *     ) );
 467          *     // gives '<em class="foo bar"></em>'
 468          * @endcode
 469          *
 470          * @par Associative array
 471          * @code
 472          *     Html::element( 'em', array(
 473          *         'class' => array( 'foo', 'bar', 'foo' => false, 'quux' => true )
 474          *     ) );
 475          *     // gives '<em class="bar quux"></em>'
 476          * @endcode
 477          *
 478          * @param array $attribs Associative array of attributes, e.g., array(
 479          *   'href' => 'http://www.mediawiki.org/' ).  Values will be HTML-escaped.
 480          *   A value of false means to omit the attribute.  For boolean attributes,
 481          *   you can omit the key, e.g., array( 'checked' ) instead of
 482          *   array( 'checked' => 'checked' ) or such.
 483          *
 484          * @throws MWException If an attribute that doesn't allow lists is set to an array
 485          * @return string HTML fragment that goes between element name and '>'
 486          *   (starting with a space if at least one attribute is output)
 487          */
 488         public static function expandAttributes( $attribs ) {
 489                 global $wgWellFormedXml;
 490
 491                 $ret = '';
 492                 $attribs = (array)$attribs;
 493                 foreach ( $attribs as $key => $value ) {
 494                         // Support intuitive array( 'checked' => true/false ) form
 495                         if ( $value === false || is_null( $value ) ) {
 496                                 continue;
 497                         }
 498
 499                         // For boolean attributes, support array( 'foo' ) instead of
 500                         // requiring array( 'foo' => 'meaningless' ).
 501                         if ( is_int( $key ) && in_array( strtolower( $value ), self::$boolAttribs ) ) {
 502                                 $key = $value;
 503                         }
 504
 505                         // Not technically required in HTML5 but we'd like consistency
 506                         // and better compression anyway.
 507                         $key = strtolower( $key );
 508
 509                         // Bug 23769: Blacklist all form validation attributes for now.  Current
 510                         // (June 2010) WebKit has no UI, so the form just refuses to submit
 511                         // without telling the user why, which is much worse than failing
 512                         // server-side validation.  Opera is the only other implementation at
 513                         // this time, and has ugly UI, so just kill the feature entirely until
 514                         // we have at least one good implementation.
 515
 516                         // As the default value of "1" for "step" rejects decimal
 517                         // numbers to be entered in 'type="number"' fields, allow
 518                         // the special case 'step="any"'.
 519
 520                         if ( in_array( $key, array( 'max', 'min', 'pattern', 'required' ) )
 521                                 || $key === 'step' && $value !== 'any' ) {
 522                                 continue;
 523                         }
 524
 525                         // http://www.w3.org/TR/html401/index/attributes.html ("space-separated")
 526                         // http://www.w3.org/TR/html5/index.html#attributes-1 ("space-separated")
 527                         $spaceSeparatedListAttributes = array(
 528                                 'class', // html4, html5
 529                                 'accesskey', // as of html5, multiple space-separated values allowed
 530                                 // html4-spec doesn't document rel= as space-separated
 531                                 // but has been used like that and is now documented as such
 532                                 // in the html5-spec.
 533                                 'rel',
 534                         );
 535
 536                         // Specific features for attributes that allow a list of space-separated values
 537                         if ( in_array( $key, $spaceSeparatedListAttributes ) ) {
 538                                 // Apply some normalization and remove duplicates
 539
 540                                 // Convert into correct array. Array can contain space-separated
 541                                 // values. Implode/explode to get those into the main array as well.
 542                                 if ( is_array( $value ) ) {
 543                                         // If input wasn't an array, we can skip this step
 544                                         $newValue = array();
 545                                         foreach ( $value as $k => $v ) {
 546                                                 if ( is_string( $v ) ) {
 547                                                         // String values should be normal `array( 'foo' )`
 548                                                         // Just append them
 549                                                         if ( !isset( $value[$v] ) ) {
 550                                                                 // As a special case don't set 'foo' if a
 551                                                                 // separate 'foo' => true/false exists in the array
 552                                                                 // keys should be authoritative
 553                                                                 $newValue[] = $v;
 554                                                         }
 555                                                 } elseif ( $v ) {
 556                                                         // If the value is truthy but not a string this is likely
 557                                                         // an array( 'foo' => true ), falsy values don't add strings
 558                                                         $newValue[] = $k;
 559                                                 }
 560                                         }
 561                                         $value = implode( ' ', $newValue );
 562                                 }
 563                                 $value = explode( ' ', $value );
 564
 565                                 // Normalize spacing by fixing up cases where people used
 566                                 // more than 1 space and/or a trailing/leading space
 567                                 $value = array_diff( $value, array( '', ' ' ) );
 568
 569                                 // Remove duplicates and create the string
 570                                 $value = implode( ' ', array_unique( $value ) );
 571                         } elseif ( is_array( $value ) ) {
 572                                 throw new MWException( "HTML attribute $key can not contain a list of values" );
 573                         }
 574
 575                         // See the "Attributes" section in the HTML syntax part of HTML5,
 576                         // 9.1.2.3 as of 2009-08-10.  Most attributes can have quotation
 577                         // marks omitted, but not all.  (Although a literal " is not
 578                         // permitted, we don't check for that, since it will be escaped
 579                         // anyway.)
 580
 581                         // See also research done on further characters that need to be
 582                         // escaped: http://code.google.com/p/html5lib/issues/detail?id=93
 583                         $badChars = "\\x00- '=<>`/\x{00a0}\x{1680}\x{180e}\x{180F}\x{2000}\x{2001}"
 584                                 . "\x{2002}\x{2003}\x{2004}\x{2005}\x{2006}\x{2007}\x{2008}\x{2009}"
 585                                 . "\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}";
 586                         if ( $wgWellFormedXml || $value === '' || preg_match( "![$badChars]!u", $value ) ) {
 587                                 $quote = '"';
 588                         } else {
 589                                 $quote = '';
 590                         }
 591
 592                         if ( in_array( $key, self::$boolAttribs ) ) {
 593                                 // In HTML5, we can leave the value empty. If we don't need
 594                                 // well-formed XML, we can omit the = entirely.
 595                                 if ( !$wgWellFormedXml ) {
 596                                         $ret .= " $key";
 597                                 } else {
 598                                         $ret .= " $key=\"\"";
 599                                 }
 600                         } else {
 601                                 // Apparently we need to entity-encode \n, \r, \t, although the
 602                                 // spec doesn't mention that.  Since we're doing strtr() anyway,
 603                                 // we may as well not call htmlspecialchars().
 604                                 // @todo FIXME: Verify that we actually need to
 605                                 // escape \n\r\t here, and explain why, exactly.
 606                                 #
 607                                 // We could call Sanitizer::encodeAttribute() for this, but we
 608                                 // don't because we're stubborn and like our marginal savings on
 609                                 // byte size from not having to encode unnecessary quotes.
 610                                 // The only difference between this transform and the one by
 611                                 // Sanitizer::encodeAttribute() is '<' is only encoded here if
 612                                 // $wgWellFormedXml is set, and ' is not encoded.
 613                                 $map = array(
 614                                         '&' => '&amp;',
 615                                         '"' => '&quot;',
 616                                         '>' => '&gt;',
 617                                         "\n" => '&#10;',
 618                                         "\r" => '&#13;',
 619                                         "\t" => '&#9;'
 620                                 );
 621                                 if ( $wgWellFormedXml ) {
 622                                         // This is allowed per spec: <http://www.w3.org/TR/xml/#NT-AttValue>
 623                                         // But reportedly it breaks some XML tools?
 624                                         // @todo FIXME: Is this really true?
 625                                         $map['<'] = '&lt;';
 626                                 }
 627                                 $ret .= " $key=$quote" . strtr( $value, $map ) . $quote;
 628                         }
 629                 }
 630                 return $ret;
 631         }
 632
 633         /**
 634          * Output a "<script>" tag with the given contents.
 635          *
 636          * @todo do some useful escaping as well, like if $contents contains
 637          * literal "</script>" or (for XML) literal "]]>".
 638          *
 639          * @param string $contents JavaScript
 640          * @return string Raw HTML
 641          */
 642         public static function inlineScript( $contents ) {
 643                 global $wgWellFormedXml;
 644
 645                 $attrs = array();
 646
 647                 if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) {
 648                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 649                 }
 650
 651                 return self::rawElement( 'script', $attrs, $contents );
 652         }
 653
 654         /**
 655          * Output a "<script>" tag linking to the given URL, e.g.,
 656          * "<script src=foo.js></script>".
 657          *
 658          * @param string $url
 659          * @return string Raw HTML
 660          */
 661         public static function linkedScript( $url ) {
 662                 $attrs = array( 'src' => $url );
 663
 664                 return self::element( 'script', $attrs );
 665         }
 666
 667         /**
 668          * Output a "<style>" tag with the given contents for the given media type
 669          * (if any).  TODO: do some useful escaping as well, like if $contents
 670          * contains literal "</style>" (admittedly unlikely).
 671          *
 672          * @param string $contents CSS
 673          * @param string $media A media type string, like 'screen'
 674          * @return string Raw HTML
 675          */
 676         public static function inlineStyle( $contents, $media = 'all' ) {
 677                 global $wgWellFormedXml;
 678
 679                 if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) {
 680                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 681                 }
 682
 683                 return self::rawElement( 'style', array(
 684                         'type' => 'text/css',
 685                         'media' => $media,
 686                 ), $contents );
 687         }
 688
 689         /**
 690          * Output a "<link rel=stylesheet>" linking to the given URL for the given
 691          * media type (if any).
 692          *
 693          * @param string $url
 694          * @param string $media A media type string, like 'screen'
 695          * @return string Raw HTML
 696          */
 697         public static function linkedStyle( $url, $media = 'all' ) {
 698                 return self::element( 'link', array(
 699                         'rel' => 'stylesheet',
 700                         'href' => $url,
 701                         'type' => 'text/css',
 702                         'media' => $media,
 703                 ) );
 704         }
 705
 706         /**
 707          * Convenience function to produce an "<input>" element.  This supports the
 708          * new HTML5 input types and attributes.
 709          *
 710          * @param string $name Name attribute
 711          * @param string $value Value attribute
 712          * @param string $type Type attribute
 713          * @param array $attribs Associative array of miscellaneous extra
 714          *   attributes, passed to Html::element()
 715          * @return string Raw HTML
 716          */
 717         public static function input( $name, $value = '', $type = 'text', $attribs = array() ) {
 718                 $attribs['type'] = $type;
 719                 $attribs['value'] = $value;
 720                 $attribs['name'] = $name;
 721                 if ( in_array( $type, array( 'text', 'search', 'email', 'password', 'number' ) ) ) {
 722                         $attribs = self::getTextInputAttributes( $attribs );
 723                 }
 724                 return self::element( 'input', $attribs );
 725         }
 726
 727         /**
 728          * Convenience function to produce a checkbox (input element with type=checkbox)
 729          *
 730          * @param string $name Name attribute
 731          * @param bool $checked Whether the checkbox is checked or not
 732          * @param array $attribs Array of additional attributes
 733          * @return string Raw HTML
 734          */
 735         public static function check( $name, $checked = false, array $attribs = array() ) {
 736                 if ( isset( $attribs['value'] ) ) {
 737                         $value = $attribs['value'];
 738                         unset( $attribs['value'] );
 739                 } else {
 740                         $value = 1;
 741                 }
 742
 743                 if ( $checked ) {
 744                         $attribs[] = 'checked';
 745                 }
 746
 747                 return self::input( $name, $value, 'checkbox', $attribs );
 748         }
 749
 750         /**
 751          * Convenience function to produce a checkbox (input element with type=checkbox)
 752          *
 753          * @param string $name Name attribute
 754          * @param bool $checked Whether the checkbox is checked or not
 755          * @param array $attribs Array of additional attributes
 756          * @return string Raw HTML
 757          */
 758         public static function radio( $name, $checked = false, array $attribs = array() ) {
 759                 if ( isset( $attribs['value'] ) ) {
 760                         $value = $attribs['value'];
 761                         unset( $attribs['value'] );
 762                 } else {
 763                         $value = 1;
 764                 }
 765
 766                 if ( $checked ) {
 767                         $attribs[] = 'checked';
 768                 }
 769
 770                 return self::input( $name, $value, 'radio', $attribs );
 771         }
 772
 773         /**
 774          * Convenience function for generating a label for inputs.
 775          *
 776          * @param string $label Contents of the label
 777          * @param string $id ID of the element being labeled
 778          * @param array $attribs Additional attributes
 779          * @return string Raw HTML
 780          */
 781         public static function label( $label, $id, array $attribs = array() ) {
 782                 $attribs += array(
 783                         'for' => $id
 784                 );
 785                 return self::element( 'label', $attribs, $label );
 786         }
 787
 788         /**
 789          * Convenience function to produce an input element with type=hidden
 790          *
 791          * @param string $name Name attribute
 792          * @param string $value Value attribute
 793          * @param array $attribs Associative array of miscellaneous extra
 794          *   attributes, passed to Html::element()
 795          * @return string Raw HTML
 796          */
 797         public static function hidden( $name, $value, $attribs = array() ) {
 798                 return self::input( $name, $value, 'hidden', $attribs );
 799         }
 800
 801         /**
 802          * Convenience function to produce a <textarea> element.
 803          *
 804          * This supports leaving out the cols= and rows= which Xml requires and are
 805          * required by HTML4/XHTML but not required by HTML5.
 806          *
 807          * @param string $name Name attribute
 808          * @param string $value Value attribute
 809          * @param array $attribs Associative array of miscellaneous extra
 810          *   attributes, passed to Html::element()
 811          * @return string Raw HTML
 812          */
 813         public static function textarea( $name, $value = '', $attribs = array() ) {
 814                 $attribs['name'] = $name;
 815
 816                 if ( substr( $value, 0, 1 ) == "\n" ) {
 817                         // Workaround for bug 12130: browsers eat the initial newline
 818                         // assuming that it's just for show, but they do keep the later
 819                         // newlines, which we may want to preserve during editing.
 820                         // Prepending a single newline
 821                         $spacedValue = "\n" . $value;
 822                 } else {
 823                         $spacedValue = $value;
 824                 }
 825                 return self::element( 'textarea', self::getTextInputAttributes( $attribs ), $spacedValue );
 826         }
 827
 828         /**
 829          * Build a drop-down box for selecting a namespace
 830          *
 831          * @param array $params Params to set.
 832          * - selected: [optional] Id of namespace which should be pre-selected
 833          * - all: [optional] Value of item for "all namespaces". If null or unset,
 834          *   no "<option>" is generated to select all namespaces.
 835          * - label: text for label to add before the field.
 836          * - exclude: [optional] Array of namespace ids to exclude.
 837          * - disable: [optional] Array of namespace ids for which the option should
 838          *   be disabled in the selector.
 839          * @param array $selectAttribs HTML attributes for the generated select element.
 840          * - id:   [optional], default: 'namespace'.
 841          * - name: [optional], default: 'namespace'.
 842          * @return string HTML code to select a namespace.
 843          */
 844         public static function namespaceSelector( array $params = array(),
 845                 array $selectAttribs = array()
 846         ) {
 847                 global $wgContLang;
 848
 849                 ksort( $selectAttribs );
 850
 851                 // Is a namespace selected?
 852                 if ( isset( $params['selected'] ) ) {
 853                         // If string only contains digits, convert to clean int. Selected could also
 854                         // be "all" or "" etc. which needs to be left untouched.
 855                         // PHP is_numeric() has issues with large strings, PHP ctype_digit has other issues
 856                         // and returns false for already clean ints. Use regex instead..
 857                         if ( preg_match( '/^\d+$/', $params['selected'] ) ) {
 858                                 $params['selected'] = intval( $params['selected'] );
 859                         }
 860                         // else: leaves it untouched for later processing
 861                 } else {
 862                         $params['selected'] = '';
 863                 }
 864
 865                 if ( !isset( $params['exclude'] ) || !is_array( $params['exclude'] ) ) {
 866                         $params['exclude'] = array();
 867                 }
 868                 if ( !isset( $params['disable'] ) || !is_array( $params['disable'] ) ) {
 869                         $params['disable'] = array();
 870                 }
 871
 872                 // Associative array between option-values and option-labels
 873                 $options = array();
 874
 875                 if ( isset( $params['all'] ) ) {
 876                         // add an option that would let the user select all namespaces.
 877                         // Value is provided by user, the name shown is localized for the user.
 878                         $options[$params['all']] = wfMessage( 'namespacesall' )->text();
 879                 }
 880                 // Add all namespaces as options (in the content language)
 881                 $options += $wgContLang->getFormattedNamespaces();
 882
 883                 // Convert $options to HTML and filter out namespaces below 0
 884                 $optionsHtml = array();
 885                 foreach ( $options as $nsId => $nsName ) {
 886                         if ( $nsId < NS_MAIN || in_array( $nsId, $params['exclude'] ) ) {
 887                                 continue;
 888                         }
 889                         if ( $nsId === NS_MAIN ) {
 890                                 // For other namespaces use the namespace prefix as label, but for
 891                                 // main we don't use "" but the user message describing it (e.g. "(Main)" or "(Article)")
 892                                 $nsName = wfMessage( 'blanknamespace' )->text();
 893                         } elseif ( is_int( $nsId ) ) {
 894                                 $nsName = $wgContLang->convertNamespace( $nsId );
 895                         }
 896                         $optionsHtml[] = self::element(
 897                                 'option', array(
 898                                         'disabled' => in_array( $nsId, $params['disable'] ),
 899                                         'value' => $nsId,
 900                                         'selected' => $nsId === $params['selected'],
 901                                 ), $nsName
 902                         );
 903                 }
 904
 905                 if ( !array_key_exists( 'id', $selectAttribs ) ) {
 906                         $selectAttribs['id'] = 'namespace';
 907                 }
 908
 909                 if ( !array_key_exists( 'name', $selectAttribs ) ) {
 910                         $selectAttribs['name'] = 'namespace';
 911                 }
 912
 913                 $ret = '';
 914                 if ( isset( $params['label'] ) ) {
 915                         $ret .= self::element(
 916                                 'label', array(
 917                                         'for' => isset( $selectAttribs['id'] ) ? $selectAttribs['id'] : null,
 918                                 ), $params['label']
 919                         ) . '&#160;';
 920                 }
 921
 922                 // Wrap options in a <select>
 923                 $ret .= self::openElement( 'select', $selectAttribs )
 924                         . "\n"
 925                         . implode( "\n", $optionsHtml )
 926                         . "\n"
 927                         . self::closeElement( 'select' );
 928
 929                 return $ret;
 930         }
 931
 932         /**
 933          * Constructs the opening html-tag with necessary doctypes depending on
 934          * global variables.
 935          *
 936          * @param array $attribs Associative array of miscellaneous extra
 937          *   attributes, passed to Html::element() of html tag.
 938          * @return string Raw HTML
 939          */
 940         public static function htmlHeader( $attribs = array() ) {
 941                 $ret = '';
 942
 943                 global $wgHtml5Version, $wgMimeType, $wgXhtmlNamespaces;
 944
 945                 $isXHTML = self::isXmlMimeType( $wgMimeType );
 946
 947                 if ( $isXHTML ) { // XHTML5
 948                         // XML MIME-typed markup should have an xml header.
 949                         // However a DOCTYPE is not needed.
 950                         $ret .= "<?xml version=\"1.0\" encoding=\"UTF-8\" ?" . ">\n";
 951
 952                         // Add the standard xmlns
 953                         $attribs['xmlns'] = 'http://www.w3.org/1999/xhtml';
 954
 955                         // And support custom namespaces
 956                         foreach ( $wgXhtmlNamespaces as $tag => $ns ) {
 957                                 $attribs["xmlns:$tag"] = $ns;
 958                         }
 959                 } else { // HTML5
 960                         // DOCTYPE
 961                         $ret .= "<!DOCTYPE html>\n";
 962                 }
 963
 964                 if ( $wgHtml5Version ) {
 965                         $attribs['version'] = $wgHtml5Version;
 966                 }
 967
 968                 $html = self::openElement( 'html', $attribs );
 969
 970                 if ( $html ) {
 971                         $html .= "\n";
 972                 }
 973
 974                 $ret .= $html;
 975
 976                 return $ret;
 977         }
 978
 979         /**
 980          * Determines if the given MIME type is xml.
 981          *
 982          * @param string $mimetype MIME type
 983          * @return bool
 984          */
 985         public static function isXmlMimeType( $mimetype ) {
 986                 # http://www.whatwg.org/html/infrastructure.html#xml-mime-type
 987                 # * text/xml
 988                 # * application/xml
 989                 # * Any MIME type with a subtype ending in +xml (this implicitly includes application/xhtml+xml)
 990                 return (bool)preg_match( '!^(text|application)/xml$|^.+/.+\+xml$!', $mimetype );
 991         }
 992
 993         /**
 994          * Get HTML for an info box with an icon.
 995          *
 996          * @param string $text Wikitext, get this with wfMessage()->plain()
 997          * @param string $icon Path to icon file (used as 'src' attribute)
 998          * @param string $alt Alternate text for the icon
 999          * @param string $class Additional class name to add to the wrapper div
1000          *
1001          * @return string
1002          */
1003         static function infoBox( $text, $icon, $alt, $class = '' ) {
1004                 $s = self::openElement( 'div', array( 'class' => "mw-infobox $class" ) );
1005
1006                 $s .= self::openElement( 'div', array( 'class' => 'mw-infobox-left' ) ) .
1007                                 self::element( 'img',
1008                                         array(
1009                                                 'src' => $icon,
1010                                                 'alt' => $alt,
1011                                         )
1012                                 ) .
1013                                 self::closeElement( 'div' );
1014
1015                 $s .= self::openElement( 'div', array( 'class' => 'mw-infobox-right' ) ) .
1016                                 $text .
1017                                 self::closeElement( 'div' );
1018                 $s .= self::element( 'div', array( 'style' => 'clear: left;' ), ' ' );
1019
1020                 $s .= self::closeElement( 'div' );
1021
1022                 $s .= self::element( 'div', array( 'style' => 'clear: left;' ), ' ' );
1023
1024                 return $s;
1025         }
1026
1027         /**
1028          * Generate a srcset attribute value.
1029          *
1030          * Generates a srcset attribute value from an array mapping pixel densities
1031          * to URLs. A trailing 'x' in pixel density values is optional.
1032          *
1033          * @note srcset width and height values are not supported.
1034          *
1035          * @see http://www.whatwg.org/html/embedded-content-1.html#attr-img-srcset
1036          *
1037          * @par Example:
1038          * @code
1039          *     Html::srcSet( array(
1040          *         '1x'   => 'standard.jpeg',
1041          *         '1.5x' => 'large.jpeg',
1042          *         '3x'   => 'extra-large.jpeg',
1043          *     ) );
1044          *     // gives 'standard.jpeg 1x, large.jpeg 1.5x, extra-large.jpeg 2x'
1045          * @endcode
1046          *
1047          * @param string[] $urls
1048          * @return string
1049          */
1050         static function srcSet( $urls ) {
1051                 $candidates = array();
1052                 foreach ( $urls as $density => $url ) {
1053                         // Cast density to float to strip 'x'.
1054                         $candidates[] = $url . ' ' . (float)$density . 'x';
1055                 }
1056                 return implode( ", ", $candidates );
1057         }
1058 }