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 $modifiers to add to the button
 108          * @see https://tools.wmflabs.org/styleguide/desktop/index.html for guidance on available modifiers
 109          * @return array $attrs A modified attribute array
 110          */
 111         public static function buttonAttributes( $attrs, $modifiers = array() ) {
 112                 global $wgUseMediaWikiUIEverywhere;
 113                 if ( $wgUseMediaWikiUIEverywhere ) {
 114                         if ( isset( $attrs['class'] ) ) {
 115                                 if ( is_array( $attrs['class'] ) ) {
 116                                         $attrs['class'][] = 'mw-ui-button';
 117                                         $attrs = array_merge( $attrs, $modifiers );
 118                                         // ensure compatibility with Xml
 119                                         $attrs['class'] = implode( ' ', $attrs['class'] );
 120                                 } else {
 121                                         $attrs['class'] .= ' mw-ui-button ' . implode( ' ', $modifiers );
 122                                 }
 123                         } else {
 124                                 $attrs['class'] = array( 'mw-ui-button' );
 125                                 // ensure compatibility with Xml
 126                                 $attrs['class'] = implode( ' ', array_merge( $attrs['class'], $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( $attrs ) {
 140                 global $wgUseMediaWikiUIEverywhere;
 141                 if ( !$attrs ) {
 142                         $attrs = array();
 143                 }
 144                 if ( isset( $attrs['class'] ) ) {
 145                         if ( is_array( $attrs['class'] ) ) {
 146                                 $attrs['class'][] = 'mw-ui-input';
 147                         } else {
 148                                 $attrs['class'] .= ' mw-ui-input';
 149                         }
 150                 } else {
 151                         $attrs['class'] = 'mw-ui-input';
 152                 }
 153                 if ( $wgUseMediaWikiUIEverywhere ) {
 154                         // Note that size can effect the desired width rendering of mw-ui-input elements
 155                         // so it is removed. Left intact when mediawiki ui not enabled.
 156                         unset( $attrs['size'] );
 157                 }
 158                 return $attrs;
 159         }
 160
 161         /**
 162          * Returns an HTML link element in a string styled as a button (when $wgUseMediaWikiUIEverywhere is enabled).
 163          *
 164          * @param string $contents The raw HTML contents of the element: *not*
 165          *   escaped!
 166          * @param array $attrs Associative array of attributes, e.g., array(
 167          *   'href' => 'http://www.mediawiki.org/' ). See expandAttributes() for
 168          *   further documentation.
 169          * @param array $modifiers to add to the button
 170          * @see http://tools.wmflabs.org/styleguide/desktop/index.html for guidance on available modifiers
 171          * @return string Raw HTML
 172          */
 173         public static function linkButton( $contents, $attrs, $modifiers = array() ) {
 174                 return Html::element( 'a',
 175                         self::buttonAttributes( $attrs, $modifiers ),
 176                         $contents
 177                 );
 178         }
 179
 180         /**
 181          * Returns an HTML link element in a string styled as a button (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 array $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 Html::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.  If you're hardcoding all the attributes, or there are none, you
 202          * should probably just type out the html element yourself.
 203          *
 204          * This is quite similar to Xml::tags(), but it implements some useful
 205          * HTML-specific logic.  For instance, there is no $allowShortTag
 206          * parameter: the closing tag is magically omitted if $element has an empty
 207          * content model.  If $wgWellFormedXml is false, then a few bytes will be
 208          * shaved off the HTML output as well.
 209          *
 210          * @param string $element The element's name, e.g., 'a'
 211          * @param array $attribs Associative array of attributes, e.g., array(
 212          *   'href' => 'http://www.mediawiki.org/' ). See expandAttributes() for
 213          *   further documentation.
 214          * @param string $contents The raw HTML contents of the element: *not*
 215          *   escaped!
 216          * @return string Raw HTML
 217          */
 218         public static function rawElement( $element, $attribs = array(), $contents = '' ) {
 219                 global $wgWellFormedXml;
 220                 $start = self::openElement( $element, $attribs );
 221                 if ( in_array( $element, self::$voidElements ) ) {
 222                         if ( $wgWellFormedXml ) {
 223                                 // Silly XML.
 224                                 return substr( $start, 0, -1 ) . ' />';
 225                         }
 226                         return $start;
 227                 } else {
 228                         return "$start$contents" . self::closeElement( $element );
 229                 }
 230         }
 231
 232         /**
 233          * Identical to rawElement(), but HTML-escapes $contents (like
 234          * Xml::element()).
 235          *
 236          * @param string $element
 237          * @param array $attribs
 238          * @param string $contents
 239          *
 240          * @return string
 241          */
 242         public static function element( $element, $attribs = array(), $contents = '' ) {
 243                 return self::rawElement( $element, $attribs, strtr( $contents, array(
 244                         // There's no point in escaping quotes, >, etc. in the contents of
 245                         // elements.
 246                         '&' => '&amp;',
 247                         '<' => '&lt;'
 248                 ) ) );
 249         }
 250
 251         /**
 252          * Identical to rawElement(), but has no third parameter and omits the end
 253          * tag (and the self-closing '/' in XML mode for empty elements).
 254          *
 255          * @param string $element
 256          * @param array $attribs
 257          *
 258          * @return string
 259          */
 260         public static function openElement( $element, $attribs = array() ) {
 261                 global $wgWellFormedXml;
 262                 $attribs = (array)$attribs;
 263                 // This is not required in HTML5, but let's do it anyway, for
 264                 // consistency and better compression.
 265                 $element = strtolower( $element );
 266
 267                 // In text/html, initial <html> and <head> tags can be omitted under
 268                 // pretty much any sane circumstances, if they have no attributes.  See:
 269                 // <http://www.whatwg.org/html/syntax.html#optional-tags>
 270                 if ( !$wgWellFormedXml && !$attribs
 271                 && in_array( $element, array( 'html', 'head' ) ) ) {
 272                         return '';
 273                 }
 274
 275                 // Remove invalid input types
 276                 if ( $element == 'input' ) {
 277                         $validTypes = array(
 278                                 'hidden',
 279                                 'text',
 280                                 'password',
 281                                 'checkbox',
 282                                 'radio',
 283                                 'file',
 284                                 'submit',
 285                                 'image',
 286                                 'reset',
 287                                 'button',
 288
 289                                 // HTML input types
 290                                 'datetime',
 291                                 'datetime-local',
 292                                 'date',
 293                                 'month',
 294                                 'time',
 295                                 'week',
 296                                 'number',
 297                                 'range',
 298                                 'email',
 299                                 'url',
 300                                 'search',
 301                                 'tel',
 302                                 'color',
 303                         );
 304                         if ( isset( $attribs['type'] )
 305                         && !in_array( $attribs['type'], $validTypes ) ) {
 306                                 unset( $attribs['type'] );
 307                         }
 308                 }
 309
 310                 // According to standard the default type for <button> elements is "submit".
 311                 // Depending on compatibility mode IE might use "button", instead.
 312                 // We enforce the standard "submit".
 313                 if ( $element == 'button' && !isset( $attribs['type'] ) ) {
 314                         $attribs['type'] = 'submit';
 315                 }
 316
 317                 return "<$element" . self::expandAttributes(
 318                         self::dropDefaults( $element, $attribs ) ) . '>';
 319         }
 320
 321         /**
 322          * Returns "</$element>"
 323          *
 324          * @since 1.17
 325          * @param string $element Name of the element, e.g., 'a'
 326          * @return string A closing tag
 327          */
 328         public static function closeElement( $element ) {
 329                 $element = strtolower( $element );
 330
 331                 return "</$element>";
 332         }
 333
 334         /**
 335          * Given an element name and an associative array of element attributes,
 336          * return an array that is functionally identical to the input array, but
 337          * possibly smaller.  In particular, attributes might be stripped if they
 338          * are given their default values.
 339          *
 340          * This method is not guaranteed to remove all redundant attributes, only
 341          * some common ones and some others selected arbitrarily at random.  It
 342          * only guarantees that the output array should be functionally identical
 343          * to the input array (currently per the HTML 5 draft as of 2009-09-06).
 344          *
 345          * @param string $element Name of the element, e.g., 'a'
 346          * @param array $attribs Associative array of attributes, e.g., array(
 347          *   'href' => 'http://www.mediawiki.org/' ).  See expandAttributes() for
 348          *   further documentation.
 349          * @return array An array of attributes functionally identical to $attribs
 350          */
 351         private static function dropDefaults( $element, $attribs ) {
 352
 353                 // Whenever altering this array, please provide a covering test case
 354                 // in HtmlTest::provideElementsWithAttributesHavingDefaultValues
 355                 static $attribDefaults = array(
 356                         'area' => array( 'shape' => 'rect' ),
 357                         'button' => array(
 358                                 'formaction' => 'GET',
 359                                 'formenctype' => 'application/x-www-form-urlencoded',
 360                         ),
 361                         'canvas' => array(
 362                                 'height' => '150',
 363                                 'width' => '300',
 364                         ),
 365                         'command' => array( 'type' => 'command' ),
 366                         'form' => array(
 367                                 'action' => 'GET',
 368                                 'autocomplete' => 'on',
 369                                 'enctype' => 'application/x-www-form-urlencoded',
 370                         ),
 371                         'input' => array(
 372                                 'formaction' => 'GET',
 373                                 'type' => 'text',
 374                         ),
 375                         'keygen' => array( 'keytype' => 'rsa' ),
 376                         'link' => array( 'media' => 'all' ),
 377                         'menu' => array( 'type' => 'list' ),
 378                         // Note: the use of text/javascript here instead of other JavaScript
 379                         // MIME types follows the HTML5 spec.
 380                         'script' => array( 'type' => 'text/javascript' ),
 381                         'style' => array(
 382                                 'media' => 'all',
 383                                 'type' => 'text/css',
 384                         ),
 385                         'textarea' => array( 'wrap' => 'soft' ),
 386                 );
 387
 388                 $element = strtolower( $element );
 389
 390                 foreach ( $attribs as $attrib => $value ) {
 391                         $lcattrib = strtolower( $attrib );
 392                         if ( is_array( $value ) ) {
 393                                 $value = implode( ' ', $value );
 394                         } else {
 395                                 $value = strval( $value );
 396                         }
 397
 398                         // Simple checks using $attribDefaults
 399                         if ( isset( $attribDefaults[$element][$lcattrib] ) &&
 400                         $attribDefaults[$element][$lcattrib] == $value ) {
 401                                 unset( $attribs[$attrib] );
 402                         }
 403
 404                         if ( $lcattrib == 'class' && $value == '' ) {
 405                                 unset( $attribs[$attrib] );
 406                         }
 407                 }
 408
 409                 // More subtle checks
 410                 if ( $element === 'link' && isset( $attribs['type'] )
 411                 && strval( $attribs['type'] ) == 'text/css' ) {
 412                         unset( $attribs['type'] );
 413                 }
 414                 if ( $element === 'input' ) {
 415                         $type = isset( $attribs['type'] ) ? $attribs['type'] : null;
 416                         $value = isset( $attribs['value'] ) ? $attribs['value'] : null;
 417                         if ( $type === 'checkbox' || $type === 'radio' ) {
 418                                 // The default value for checkboxes and radio buttons is 'on'
 419                                 // not ''. By stripping value="" we break radio boxes that
 420                                 // actually wants empty values.
 421                                 if ( $value === 'on' ) {
 422                                         unset( $attribs['value'] );
 423                                 }
 424                         } elseif ( $type === 'submit' ) {
 425                                 // The default value for submit appears to be "Submit" but
 426                                 // let's not bother stripping out localized text that matches
 427                                 // that.
 428                         } else {
 429                                 // The default value for nearly every other field type is ''
 430                                 // The 'range' and 'color' types use different defaults but
 431                                 // stripping a value="" does not hurt them.
 432                                 if ( $value === '' ) {
 433                                         unset( $attribs['value'] );
 434                                 }
 435                         }
 436                 }
 437                 if ( $element === 'select' && isset( $attribs['size'] ) ) {
 438                         if ( in_array( 'multiple', $attribs )
 439                                 || ( isset( $attribs['multiple'] ) && $attribs['multiple'] !== false )
 440                         ) {
 441                                 // A multi-select
 442                                 if ( strval( $attribs['size'] ) == '4' ) {
 443                                         unset( $attribs['size'] );
 444                                 }
 445                         } else {
 446                                 // Single select
 447                                 if ( strval( $attribs['size'] ) == '1' ) {
 448                                         unset( $attribs['size'] );
 449                                 }
 450                         }
 451                 }
 452
 453                 return $attribs;
 454         }
 455
 456         /**
 457          * Given an associative array of element attributes, generate a string
 458          * to stick after the element name in HTML output.  Like array( 'href' =>
 459          * 'http://www.mediawiki.org/' ) becomes something like
 460          * ' href="http://www.mediawiki.org"'.  Again, this is like
 461          * Xml::expandAttributes(), but it implements some HTML-specific logic.
 462          * For instance, it will omit quotation marks if $wgWellFormedXml is false,
 463          * and will treat boolean attributes specially.
 464          *
 465          * Attributes that can contain space-separated lists ('class', 'accesskey' and 'rel') array
 466          * values are allowed as well, which will automagically be normalized
 467          * and converted to a space-separated string. In addition to a numerical
 468          * array, the attribute value may also be an associative array. See the
 469          * example below for how that works.
 470          *
 471          * @par Numerical array
 472          * @code
 473          *     Html::element( 'em', array(
 474          *         'class' => array( 'foo', 'bar' )
 475          *     ) );
 476          *     // gives '<em class="foo bar"></em>'
 477          * @endcode
 478          *
 479          * @par Associative array
 480          * @code
 481          *     Html::element( 'em', array(
 482          *         'class' => array( 'foo', 'bar', 'foo' => false, 'quux' => true )
 483          *     ) );
 484          *     // gives '<em class="bar quux"></em>'
 485          * @endcode
 486          *
 487          * @param array $attribs Associative array of attributes, e.g., array(
 488          *   'href' => 'http://www.mediawiki.org/' ).  Values will be HTML-escaped.
 489          *   A value of false means to omit the attribute.  For boolean attributes,
 490          *   you can omit the key, e.g., array( 'checked' ) instead of
 491          *   array( 'checked' => 'checked' ) or such.
 492          *
 493          * @throws MWException If an attribute that doesn't allow lists is set to an array
 494          * @return string HTML fragment that goes between element name and '>'
 495          *   (starting with a space if at least one attribute is output)
 496          */
 497         public static function expandAttributes( $attribs ) {
 498                 global $wgWellFormedXml;
 499
 500                 $ret = '';
 501                 $attribs = (array)$attribs;
 502                 foreach ( $attribs as $key => $value ) {
 503                         // Support intuitive array( 'checked' => true/false ) form
 504                         if ( $value === false || is_null( $value ) ) {
 505                                 continue;
 506                         }
 507
 508                         // For boolean attributes, support array( 'foo' ) instead of
 509                         // requiring array( 'foo' => 'meaningless' ).
 510                         if ( is_int( $key )
 511                         && in_array( strtolower( $value ), self::$boolAttribs ) ) {
 512                                 $key = $value;
 513                         }
 514
 515                         // Not technically required in HTML5 but we'd like consistency
 516                         // and better compression anyway.
 517                         $key = strtolower( $key );
 518
 519                         // Bug 23769: Blacklist all form validation attributes for now.  Current
 520                         // (June 2010) WebKit has no UI, so the form just refuses to submit
 521                         // without telling the user why, which is much worse than failing
 522                         // server-side validation.  Opera is the only other implementation at
 523                         // this time, and has ugly UI, so just kill the feature entirely until
 524                         // we have at least one good implementation.
 525
 526                         // As the default value of "1" for "step" rejects decimal
 527                         // numbers to be entered in 'type="number"' fields, allow
 528                         // the special case 'step="any"'.
 529
 530                         if ( in_array( $key, array( 'max', 'min', 'pattern', 'required' ) )
 531                                 || $key === 'step' && $value !== 'any' ) {
 532                                 continue;
 533                         }
 534
 535                         // http://www.w3.org/TR/html401/index/attributes.html ("space-separated")
 536                         // http://www.w3.org/TR/html5/index.html#attributes-1 ("space-separated")
 537                         $spaceSeparatedListAttributes = array(
 538                                 'class', // html4, html5
 539                                 'accesskey', // as of html5, multiple space-separated values allowed
 540                                 // html4-spec doesn't document rel= as space-separated
 541                                 // but has been used like that and is now documented as such
 542                                 // in the html5-spec.
 543                                 'rel',
 544                         );
 545
 546                         // Specific features for attributes that allow a list of space-separated values
 547                         if ( in_array( $key, $spaceSeparatedListAttributes ) ) {
 548                                 // Apply some normalization and remove duplicates
 549
 550                                 // Convert into correct array. Array can contain space-separated
 551                                 // values. Implode/explode to get those into the main array as well.
 552                                 if ( is_array( $value ) ) {
 553                                         // If input wasn't an array, we can skip this step
 554                                         $newValue = array();
 555                                         foreach ( $value as $k => $v ) {
 556                                                 if ( is_string( $v ) ) {
 557                                                         // String values should be normal `array( 'foo' )`
 558                                                         // Just append them
 559                                                         if ( !isset( $value[$v] ) ) {
 560                                                                 // As a special case don't set 'foo' if a
 561                                                                 // separate 'foo' => true/false exists in the array
 562                                                                 // keys should be authoritative
 563                                                                 $newValue[] = $v;
 564                                                         }
 565                                                 } elseif ( $v ) {
 566                                                         // If the value is truthy but not a string this is likely
 567                                                         // an array( 'foo' => true ), falsy values don't add strings
 568                                                         $newValue[] = $k;
 569                                                 }
 570                                         }
 571                                         $value = implode( ' ', $newValue );
 572                                 }
 573                                 $value = explode( ' ', $value );
 574
 575                                 // Normalize spacing by fixing up cases where people used
 576                                 // more than 1 space and/or a trailing/leading space
 577                                 $value = array_diff( $value, array( '', ' ' ) );
 578
 579                                 // Remove duplicates and create the string
 580                                 $value = implode( ' ', array_unique( $value ) );
 581                         } elseif ( is_array( $value ) ) {
 582                                 throw new MWException( "HTML attribute $key can not contain a list of values" );
 583                         }
 584
 585                         // See the "Attributes" section in the HTML syntax part of HTML5,
 586                         // 9.1.2.3 as of 2009-08-10.  Most attributes can have quotation
 587                         // marks omitted, but not all.  (Although a literal " is not
 588                         // permitted, we don't check for that, since it will be escaped
 589                         // anyway.)
 590                         #
 591                         // See also research done on further characters that need to be
 592                         // escaped: http://code.google.com/p/html5lib/issues/detail?id=93
 593                         $badChars = "\\x00- '=<>`/\x{00a0}\x{1680}\x{180e}\x{180F}\x{2000}\x{2001}"
 594                                 . "\x{2002}\x{2003}\x{2004}\x{2005}\x{2006}\x{2007}\x{2008}\x{2009}"
 595                                 . "\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}";
 596                         if ( $wgWellFormedXml || $value === ''
 597                         || preg_match( "![$badChars]!u", $value ) ) {
 598                                 $quote = '"';
 599                         } else {
 600                                 $quote = '';
 601                         }
 602
 603                         if ( in_array( $key, self::$boolAttribs ) ) {
 604                                 // In HTML5, we can leave the value empty. If we don't need
 605                                 // well-formed XML, we can omit the = entirely.
 606                                 if ( !$wgWellFormedXml ) {
 607                                         $ret .= " $key";
 608                                 } else {
 609                                         $ret .= " $key=\"\"";
 610                                 }
 611                         } else {
 612                                 // Apparently we need to entity-encode \n, \r, \t, although the
 613                                 // spec doesn't mention that.  Since we're doing strtr() anyway,
 614                                 // and we don't need <> escaped here, we may as well not call
 615                                 // htmlspecialchars().
 616                                 // @todo FIXME: Verify that we actually need to
 617                                 // escape \n\r\t here, and explain why, exactly.
 618                                 #
 619                                 // We could call Sanitizer::encodeAttribute() for this, but we
 620                                 // don't because we're stubborn and like our marginal savings on
 621                                 // byte size from not having to encode unnecessary quotes.
 622                                 $map = array(
 623                                         '&' => '&amp;',
 624                                         '"' => '&quot;',
 625                                         "\n" => '&#10;',
 626                                         "\r" => '&#13;',
 627                                         "\t" => '&#9;'
 628                                 );
 629                                 if ( $wgWellFormedXml ) {
 630                                         // This is allowed per spec: <http://www.w3.org/TR/xml/#NT-AttValue>
 631                                         // But reportedly it breaks some XML tools?
 632                                         // @todo FIXME: Is this really true?
 633                                         $map['<'] = '&lt;';
 634                                 }
 635                                 $ret .= " $key=$quote" . strtr( $value, $map ) . $quote;
 636                         }
 637                 }
 638                 return $ret;
 639         }
 640
 641         /**
 642          * Output a "<script>" tag with the given contents.
 643          *
 644          * @todo do some useful escaping as well, like if $contents contains
 645          * literal "</script>" or (for XML) literal "]]>".
 646          *
 647          * @param string $contents JavaScript
 648          * @return string Raw HTML
 649          */
 650         public static function inlineScript( $contents ) {
 651                 global $wgWellFormedXml;
 652
 653                 $attrs = array();
 654
 655                 if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) {
 656                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 657                 }
 658
 659                 return self::rawElement( 'script', $attrs, $contents );
 660         }
 661
 662         /**
 663          * Output a "<script>" tag linking to the given URL, e.g.,
 664          * "<script src=foo.js></script>".
 665          *
 666          * @param string $url
 667          * @return string Raw HTML
 668          */
 669         public static function linkedScript( $url ) {
 670                 $attrs = array( 'src' => $url );
 671
 672                 return self::element( 'script', $attrs );
 673         }
 674
 675         /**
 676          * Output a "<style>" tag with the given contents for the given media type
 677          * (if any).  TODO: do some useful escaping as well, like if $contents
 678          * contains literal "</style>" (admittedly unlikely).
 679          *
 680          * @param string $contents CSS
 681          * @param string $media A media type string, like 'screen'
 682          * @return string Raw HTML
 683          */
 684         public static function inlineStyle( $contents, $media = 'all' ) {
 685                 global $wgWellFormedXml;
 686
 687                 if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) {
 688                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 689                 }
 690
 691                 return self::rawElement( 'style', array(
 692                         'type' => 'text/css',
 693                         'media' => $media,
 694                 ), $contents );
 695         }
 696
 697         /**
 698          * Output a "<link rel=stylesheet>" linking to the given URL for the given
 699          * media type (if any).
 700          *
 701          * @param string $url
 702          * @param string $media A media type string, like 'screen'
 703          * @return string Raw HTML
 704          */
 705         public static function linkedStyle( $url, $media = 'all' ) {
 706                 return self::element( 'link', array(
 707                         'rel' => 'stylesheet',
 708                         'href' => $url,
 709                         'type' => 'text/css',
 710                         'media' => $media,
 711                 ) );
 712         }
 713
 714         /**
 715          * Convenience function to produce an "<input>" element.  This supports the
 716          * new HTML5 input types and attributes.
 717          *
 718          * @param string $name Name attribute
 719          * @param array $value Value attribute
 720          * @param string $type Type attribute
 721          * @param array $attribs Associative array of miscellaneous extra
 722          *   attributes, passed to Html::element()
 723          * @return string Raw HTML
 724          */
 725         public static function input( $name, $value = '', $type = 'text', $attribs = array() ) {
 726                 $attribs['type'] = $type;
 727                 $attribs['value'] = $value;
 728                 $attribs['name'] = $name;
 729                 if ( in_array( $type, array( 'text', 'search', 'email', 'password', 'number' ) ) ) {
 730                         $attribs = Html::getTextInputAttributes( $attribs );
 731                 }
 732                 return self::element( 'input', $attribs );
 733         }
 734
 735         /**
 736          * Convenience function to produce a checkbox (input element with type=checkbox)
 737          *
 738          * @param string $name Name attribute
 739          * @param bool $checked Whether the checkbox is checked or not
 740          * @param array $attribs Array of additional attributes
 741          * @return string
 742          */
 743         public static function check( $name, $checked = false, array $attribs = array() ) {
 744                 if ( isset( $attribs['value'] ) ) {
 745                         $value = $attribs['value'];
 746                         unset( $attribs['value'] );
 747                 } else {
 748                         $value = 1;
 749                 }
 750
 751                 if ( $checked ) {
 752                         $attribs[] = 'checked';
 753                 }
 754
 755                 return self::input( $name, $value, 'checkbox', $attribs );
 756         }
 757
 758         /**
 759          * Convenience function to produce a checkbox (input element with type=checkbox)
 760          *
 761          * @param string $name Name attribute
 762          * @param bool $checked Whether the checkbox is checked or not
 763          * @param array $attribs Array of additional attributes
 764          * @return string
 765          */
 766         public static function radio( $name, $checked = false, array $attribs = array() ) {
 767                 if ( isset( $attribs['value'] ) ) {
 768                         $value = $attribs['value'];
 769                         unset( $attribs['value'] );
 770                 } else {
 771                         $value = 1;
 772                 }
 773
 774                 if ( $checked ) {
 775                         $attribs[] = 'checked';
 776                 }
 777
 778                 return self::input( $name, $value, 'radio', $attribs );
 779         }
 780
 781         /**
 782          * Convenience function for generating a label for inputs.
 783          *
 784          * @param string $label Contents of the label
 785          * @param string $id ID of the element being labeled
 786          * @param array $attribs Additional attributes
 787          * @return string
 788          */
 789         public static function label( $label, $id, array $attribs = array() ) {
 790                 $attribs += array(
 791                         'for' => $id
 792                 );
 793                 return self::element( 'label', $attribs, $label );
 794         }
 795
 796         /**
 797          * Convenience function to produce an input element with type=hidden
 798          *
 799          * @param string $name Name attribute
 800          * @param string $value Value attribute
 801          * @param array $attribs Associative array of miscellaneous extra
 802          *   attributes, passed to Html::element()
 803          * @return string Raw HTML
 804          */
 805         public static function hidden( $name, $value, $attribs = array() ) {
 806                 return self::input( $name, $value, 'hidden', $attribs );
 807         }
 808
 809         /**
 810          * Convenience function to produce a <textarea> element.
 811          *
 812          * This supports leaving out the cols= and rows= which Xml requires and are
 813          * required by HTML4/XHTML but not required by HTML5.
 814          *
 815          * @param string $name Name attribute
 816          * @param string $value Value attribute
 817          * @param array $attribs Associative array of miscellaneous extra
 818          *   attributes, passed to Html::element()
 819          * @return string Raw HTML
 820          */
 821         public static function textarea( $name, $value = '', $attribs = array() ) {
 822                 $attribs['name'] = $name;
 823
 824                 if ( substr( $value, 0, 1 ) == "\n" ) {
 825                         // Workaround for bug 12130: browsers eat the initial newline
 826                         // assuming that it's just for show, but they do keep the later
 827                         // newlines, which we may want to preserve during editing.
 828                         // Prepending a single newline
 829                         $spacedValue = "\n" . $value;
 830                 } else {
 831                         $spacedValue = $value;
 832                 }
 833                 return self::element( 'textarea', Html::getTextInputAttributes( $attribs ), $spacedValue );
 834         }
 835
 836         /**
 837          * Build a drop-down box for selecting a namespace
 838          *
 839          * @param array $params Params to set.
 840          * - selected: [optional] Id of namespace which should be pre-selected
 841          * - all: [optional] Value of item for "all namespaces". If null or unset,
 842          *   no "<option>" is generated to select all namespaces.
 843          * - label: text for label to add before the field.
 844          * - exclude: [optional] Array of namespace ids to exclude.
 845          * - disable: [optional] Array of namespace ids for which the option should
 846          *   be disabled in the selector.
 847          * @param array $selectAttribs HTML attributes for the generated select element.
 848          * - id:   [optional], default: 'namespace'.
 849          * - name: [optional], default: 'namespace'.
 850          * @return string HTML code to select a namespace.
 851          */
 852         public static function namespaceSelector( array $params = array(),
 853                 array $selectAttribs = array()
 854         ) {
 855                 global $wgContLang;
 856
 857                 ksort( $selectAttribs );
 858
 859                 // Is a namespace selected?
 860                 if ( isset( $params['selected'] ) ) {
 861                         // If string only contains digits, convert to clean int. Selected could also
 862                         // be "all" or "" etc. which needs to be left untouched.
 863                         // PHP is_numeric() has issues with large strings, PHP ctype_digit has other issues
 864                         // and returns false for already clean ints. Use regex instead..
 865                         if ( preg_match( '/^\d+$/', $params['selected'] ) ) {
 866                                 $params['selected'] = intval( $params['selected'] );
 867                         }
 868                         // else: leaves it untouched for later processing
 869                 } else {
 870                         $params['selected'] = '';
 871                 }
 872
 873                 if ( !isset( $params['exclude'] ) || !is_array( $params['exclude'] ) ) {
 874                         $params['exclude'] = array();
 875                 }
 876                 if ( !isset( $params['disable'] ) || !is_array( $params['disable'] ) ) {
 877                         $params['disable'] = array();
 878                 }
 879
 880                 // Associative array between option-values and option-labels
 881                 $options = array();
 882
 883                 if ( isset( $params['all'] ) ) {
 884                         // add an option that would let the user select all namespaces.
 885                         // Value is provided by user, the name shown is localized for the user.
 886                         $options[$params['all']] = wfMessage( 'namespacesall' )->text();
 887                 }
 888                 // Add all namespaces as options (in the content language)
 889                 $options += $wgContLang->getFormattedNamespaces();
 890
 891                 // Convert $options to HTML and filter out namespaces below 0
 892                 $optionsHtml = array();
 893                 foreach ( $options as $nsId => $nsName ) {
 894                         if ( $nsId < NS_MAIN || in_array( $nsId, $params['exclude'] ) ) {
 895                                 continue;
 896                         }
 897                         if ( $nsId === NS_MAIN ) {
 898                                 // For other namespaces use use the namespace prefix as label, but for
 899                                 // main we don't use "" but the user message describing it (e.g. "(Main)" or "(Article)")
 900                                 $nsName = wfMessage( 'blanknamespace' )->text();
 901                         } elseif ( is_int( $nsId ) ) {
 902                                 $nsName = $wgContLang->convertNamespace( $nsId );
 903                         }
 904                         $optionsHtml[] = Html::element(
 905                                 'option', array(
 906                                         'disabled' => in_array( $nsId, $params['disable'] ),
 907                                         'value' => $nsId,
 908                                         'selected' => $nsId === $params['selected'],
 909                                 ), $nsName
 910                         );
 911                 }
 912
 913                 if ( !array_key_exists( 'id', $selectAttribs ) ) {
 914                         $selectAttribs['id'] = 'namespace';
 915                 }
 916
 917                 if ( !array_key_exists( 'name', $selectAttribs ) ) {
 918                         $selectAttribs['name'] = 'namespace';
 919                 }
 920
 921                 $ret = '';
 922                 if ( isset( $params['label'] ) ) {
 923                         $ret .= Html::element(
 924                                 'label', array(
 925                                         'for' => isset( $selectAttribs['id'] ) ? $selectAttribs['id'] : null,
 926                                 ), $params['label']
 927                         ) . '&#160;';
 928                 }
 929
 930                 // Wrap options in a <select>
 931                 $ret .= Html::openElement( 'select', $selectAttribs )
 932                         . "\n"
 933                         . implode( "\n", $optionsHtml )
 934                         . "\n"
 935                         . Html::closeElement( 'select' );
 936
 937                 return $ret;
 938         }
 939
 940         /**
 941          * Constructs the opening html-tag with necessary doctypes depending on
 942          * global variables.
 943          *
 944          * @param array $attribs Associative array of miscellaneous extra
 945          *   attributes, passed to Html::element() of html tag.
 946          * @return string Raw HTML
 947          */
 948         public static function htmlHeader( $attribs = array() ) {
 949                 $ret = '';
 950
 951                 global $wgHtml5Version, $wgMimeType, $wgXhtmlNamespaces;
 952
 953                 $isXHTML = self::isXmlMimeType( $wgMimeType );
 954
 955                 if ( $isXHTML ) { // XHTML5
 956                         // XML MIME-typed markup should have an xml header.
 957                         // However a DOCTYPE is not needed.
 958                         $ret .= "<?xml version=\"1.0\" encoding=\"UTF-8\" ?" . ">\n";
 959
 960                         // Add the standard xmlns
 961                         $attribs['xmlns'] = 'http://www.w3.org/1999/xhtml';
 962
 963                         // And support custom namespaces
 964                         foreach ( $wgXhtmlNamespaces as $tag => $ns ) {
 965                                 $attribs["xmlns:$tag"] = $ns;
 966                         }
 967                 } else { // HTML5
 968                         // DOCTYPE
 969                         $ret .= "<!DOCTYPE html>\n";
 970                 }
 971
 972                 if ( $wgHtml5Version ) {
 973                         $attribs['version'] = $wgHtml5Version;
 974                 }
 975
 976                 $html = Html::openElement( 'html', $attribs );
 977
 978                 if ( $html ) {
 979                         $html .= "\n";
 980                 }
 981
 982                 $ret .= $html;
 983
 984                 return $ret;
 985         }
 986
 987         /**
 988          * Determines if the given MIME type is xml.
 989          *
 990          * @param string $mimetype MIME type
 991          * @return bool
 992          */
 993         public static function isXmlMimeType( $mimetype ) {
 994                 # http://www.whatwg.org/html/infrastructure.html#xml-mime-type
 995                 # * text/xml
 996                 # * application/xml
 997                 # * Any MIME type with a subtype ending in +xml (this implicitly includes application/xhtml+xml)
 998                 return (bool)preg_match( '!^(text|application)/xml$|^.+/.+\+xml$!', $mimetype );
 999         }
1000
1001         /**
1002          * Get HTML for an info box with an icon.
1003          *
1004          * @param string $text Wikitext, get this with wfMessage()->plain()
1005          * @param string $icon Path to icon file (used as 'src' attribute)
1006          * @param string $alt Alternate text for the icon
1007          * @param string $class Additional class name to add to the wrapper div
1008          *
1009          * @return string
1010          */
1011         static function infoBox( $text, $icon, $alt, $class = false ) {
1012                 $s = Html::openElement( 'div', array( 'class' => "mw-infobox $class" ) );
1013
1014                 $s .= Html::openElement( 'div', array( 'class' => 'mw-infobox-left' ) ) .
1015                                 Html::element( 'img',
1016                                         array(
1017                                                 'src' => $icon,
1018                                                 'alt' => $alt,
1019                                         )
1020                                 ) .
1021                                 Html::closeElement( 'div' );
1022
1023                 $s .= Html::openElement( 'div', array( 'class' => 'mw-infobox-right' ) ) .
1024                                 $text .
1025                                 Html::closeElement( 'div' );
1026                 $s .= Html::element( 'div', array( 'style' => 'clear: left;' ), ' ' );
1027
1028                 $s .= Html::closeElement( 'div' );
1029
1030                 $s .= Html::element( 'div', array( 'style' => 'clear: left;' ), ' ' );
1031
1032                 return $s;
1033         }
1034
1035         /**
1036          * Generate a srcset attribute value from an array mapping pixel densities
1037          * to URLs. Note that srcset supports width and height values as well, which
1038          * are not used here.
1039          *
1040          * @param array $urls
1041          * @return string
1042          */
1043         static function srcSet( $urls ) {
1044                 $candidates = array();
1045                 foreach ( $urls as $density => $url ) {
1046                         // Image candidate syntax per current whatwg live spec, 2012-09-23:
1047                         // http://www.whatwg.org/html/embedded-content-1.html#attr-img-srcset
1048                         $candidates[] = "{$url} {$density}x";
1049                 }
1050                 return implode( ", ", $candidates );
1051         }
1052 }