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