includes/Html.php

   1 <?php
   2 # Copyright © 2009 Aryeh Gregor
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 /**
  21  * This class is a collection of static functions that serve two purposes:
  22  *
  23  * 1) Implement any algorithms specified by HTML5, or other HTML
  24  * specifications, in a convenient and self-contained way.
  25  *
  26  * 2) Allow HTML elements to be conveniently and safely generated, like the
  27  * current Xml class but a) less confused (Xml supports HTML-specific things,
  28  * but only sometimes!) and b) not necessarily confined to XML-compatible
  29  * output.
  30  *
  31  * There are two important configuration options this class uses:
  32  *
  33  * $wgHtml5: If this is set to false, then all output should be valid XHTML 1.0
  34  *     Transitional.
  35  * $wgWellFormedXml: If this is set to true, then all output should be
  36  *     well-formed XML (quotes on attributes, self-closing tags, etc.).
  37  *
  38  * This class is meant to be confined to utility functions that are called from
  39  * trusted code paths.  It does not do enforcement of policy like not allowing
  40  * <a> elements.
  41  *
  42  * @since 1.16
  43  */
  44 class Html {
  45         # List of void elements from HTML5, section 9.1.2 as of 2009-08-10
  46         private static $voidElements = array(
  47                 'area',
  48                 'base',
  49                 'br',
  50                 'col',
  51                 'command',
  52                 'embed',
  53                 'hr',
  54                 'img',
  55                 'input',
  56                 'keygen',
  57                 'link',
  58                 'meta',
  59                 'param',
  60                 'source',
  61         );
  62
  63         # Boolean attributes, which may have the value omitted entirely.  Manually
  64         # collected from the HTML5 spec as of 2009-08-10.
  65         private static $boolAttribs = array(
  66                 'async',
  67                 'autobuffer',
  68                 'autofocus',
  69                 'autoplay',
  70                 'checked',
  71                 'controls',
  72                 'defer',
  73                 'disabled',
  74                 'formnovalidate',
  75                 'hidden',
  76                 'ismap',
  77                 'loop',
  78                 'multiple',
  79                 'novalidate',
  80                 'open',
  81                 'readonly',
  82                 'required',
  83                 'reversed',
  84                 'scoped',
  85                 'seamless',
  86         );
  87
  88         /**
  89          * Returns an HTML element in a string.  The major advantage here over
  90          * manually typing out the HTML is that it will escape all attribute
  91          * values.  If you're hardcoding all the attributes, or there are none, you
  92          * should probably type out the string yourself.
  93          *
  94          * This is quite similar to Xml::tags(), but it implements some useful
  95          * HTML-specific logic.  For instance, there is no $allowShortTag
  96          * parameter: the closing tag is magically omitted if $element has an empty
  97          * content model.  If $wgWellFormedXml is false, then a few bytes will be
  98          * shaved off the HTML output as well.  In the future, other HTML-specific
  99          * features might be added, like allowing arrays for the values of
 100          * attributes like class= and media=.
 101          *
 102          * @param $element  string The element's name, e.g., 'a'
 103          * @param $attribs  array  Associative array of attributes, e.g., array(
 104          *   'href' => 'http://www.mediawiki.org/' ).  See expandAttributes() for
 105          *   further documentation.
 106          * @param $contents string The raw HTML contents of the element: *not*
 107          *   escaped!
 108          * @return string Raw HTML
 109          */
 110         public static function rawElement( $element, $attribs = array(), $contents = '' ) {
 111                 global $wgWellFormedXml;
 112                 $start = self::openElement( $element, $attribs );
 113                 if ( in_array( $element, self::$voidElements ) ) {
 114                         if ( $wgWellFormedXml ) {
 115                                 # Silly XML.
 116                                 return substr( $start, 0, -1 ) . ' />';
 117                         }
 118                         return $start;
 119                 } else {
 120                         return "$start$contents" . self::closeElement( $element );
 121                 }
 122         }
 123
 124         /**
 125          * Identical to rawElement(), but HTML-escapes $contents (like
 126          * Xml::element()).
 127          */
 128         public static function element( $element, $attribs = array(), $contents = '' ) {
 129                 return self::rawElement( $element, $attribs, strtr( $contents, array(
 130                         # There's no point in escaping quotes, >, etc. in the contents of
 131                         # elements.
 132                         '&' => '&amp;',
 133                         '<' => '&lt;'
 134                 ) ) );
 135         }
 136
 137         /**
 138          * Identical to rawElement(), but has no third parameter and omits the end
 139          * tag (and the self-closing '/' in XML mode for empty elements).
 140          */
 141         public static function openElement( $element, $attribs = array() ) {
 142                 global $wgHtml5, $wgWellFormedXml;
 143                 $attribs = (array)$attribs;
 144                 # This is not required in HTML5, but let's do it anyway, for
 145                 # consistency and better compression.
 146                 $element = strtolower( $element );
 147
 148                 # In text/html, initial <html> and <head> tags can be omitted under
 149                 # pretty much any sane circumstances, if they have no attributes.  See:
 150                 # <http://www.whatwg.org/specs/web-apps/current-work/multipage/syntax.html#optional-tags>
 151                 if ( !$wgWellFormedXml && !$attribs
 152                 && in_array( $element, array( 'html', 'head' ) ) ) {
 153                         return '';
 154                 }
 155
 156                 # Remove HTML5-only attributes if we aren't doing HTML5
 157                 if ( !$wgHtml5 ) {
 158                         if ( $element == 'input' ) {
 159                                 # Whitelist of valid XHTML1 types
 160                                 $validTypes = array(
 161                                         'hidden',
 162                                         'text',
 163                                         'password',
 164                                         'checkbox',
 165                                         'radio',
 166                                         'file',
 167                                         'submit',
 168                                         'image',
 169                                         'reset',
 170                                         'button',
 171                                 );
 172                                 if ( isset( $attribs['type'] )
 173                                 && !in_array( $attribs['type'], $validTypes ) ) {
 174                                         # Fall back to type=text, the default
 175                                         unset( $attribs['type'] );
 176                                 }
 177                         }
 178                         if ( $element == 'textarea' && isset( $attribs['maxlength'] ) ) {
 179                                 unset( $attribs['maxlength'] );
 180                         }
 181                         # Here we're blacklisting some HTML5-only attributes...
 182                         $html5attribs = array(
 183                                 'autocomplete',
 184                                 'autofocus',
 185                                 'max',
 186                                 'min',
 187                                 'multiple',
 188                                 'pattern',
 189                                 'placeholder',
 190                                 'required',
 191                                 'step',
 192                                 'spellcheck',
 193                         );
 194                         foreach ( $html5attribs as $badAttr ) {
 195                                 unset( $attribs[$badAttr] );
 196                         }
 197                 }
 198
 199                 return "<$element" . self::expandAttributes(
 200                         self::dropDefaults( $element, $attribs ) ) . '>';
 201         }
 202
 203         /**
 204          * Returns "</$element>", except if $wgWellFormedXml is off, in which case
 205          * it returns the empty string when that's guaranteed to be safe.
 206          *
 207          * @param $element string Name of the element, e.g., 'a'
 208          * @return string A closing tag, if required
 209          */
 210         public static function closeElement( $element ) {
 211                 global $wgWellFormedXml;
 212
 213                 $element = strtolower( $element );
 214
 215                 # Reference:
 216                 # http://www.whatwg.org/specs/web-apps/current-work/multipage/syntax.html#optional-tags
 217                 if ( !$wgWellFormedXml && in_array( $element, array(
 218                         'html',
 219                         'head',
 220                         'body',
 221                         'li',
 222                         'dt',
 223                         'dd',
 224                         'tr',
 225                         'td',
 226                         'th',
 227                 ) ) ) {
 228                         return '';
 229                 }
 230                 return "</$element>";
 231         }
 232
 233         /**
 234          * Given an element name and an associative array of element attributes,
 235          * return an array that is functionally identical to the input array, but
 236          * possibly smaller.  In particular, attributes might be stripped if they
 237          * are given their default values.
 238          *
 239          * This method is not guaranteed to remove all redundant attributes, only
 240          * some common ones and some others selected arbitrarily at random.  It
 241          * only guarantees that the output array should be functionally identical
 242          * to the input array (currently per the HTML 5 draft as of 2009-09-06).
 243          *
 244          * @param $element string Name of the element, e.g., 'a'
 245          * @param $attribs array  Associative array of attributes, e.g., array(
 246          *   'href' => 'http://www.mediawiki.org/' ).  See expandAttributes() for
 247          *   further documentation.
 248          * @return array An array of attributes functionally identical to $attribs
 249          */
 250         private static function dropDefaults( $element, $attribs ) {
 251                 # Don't bother doing anything if we aren't outputting HTML5; it's too
 252                 # much of a pain to maintain two sets of defaults.
 253                 global $wgHtml5;
 254                 if ( !$wgHtml5 ) {
 255                         return $attribs;
 256                 }
 257
 258                 static $attribDefaults = array(
 259                         'area' => array( 'shape' => 'rect' ),
 260                         'button' => array(
 261                                 'formaction' => 'GET',
 262                                 'formenctype' => 'application/x-www-form-urlencoded',
 263                                 'type' => 'submit',
 264                         ),
 265                         'canvas' => array(
 266                                 'height' => '150',
 267                                 'width' => '300',
 268                         ),
 269                         'command' => array( 'type' => 'command' ),
 270                         'form' => array(
 271                                 'action' => 'GET',
 272                                 'autocomplete' => 'on',
 273                                 'enctype' => 'application/x-www-form-urlencoded',
 274                         ),
 275                         'input' => array(
 276                                 'formaction' => 'GET',
 277                                 'type' => 'text',
 278                                 'value' => '',
 279                         ),
 280                         'keygen' => array( 'keytype' => 'rsa' ),
 281                         'link' => array( 'media' => 'all' ),
 282                         'menu' => array( 'type' => 'list' ),
 283                         # Note: the use of text/javascript here instead of other JavaScript
 284                         # MIME types follows the HTML5 spec.
 285                         'script' => array( 'type' => 'text/javascript' ),
 286                         'style' => array(
 287                                 'media' => 'all',
 288                                 'type' => 'text/css',
 289                         ),
 290                         'textarea' => array( 'wrap' => 'soft' ),
 291                 );
 292
 293                 $element = strtolower( $element );
 294
 295                 foreach ( $attribs as $attrib => $value ) {
 296                         $lcattrib = strtolower( $attrib );
 297                         $value = strval( $value );
 298
 299                         # Simple checks using $attribDefaults
 300                         if ( isset( $attribDefaults[$element][$lcattrib] ) &&
 301                         $attribDefaults[$element][$lcattrib] == $value ) {
 302                                 unset( $attribs[$attrib] );
 303                         }
 304
 305                         if ( $lcattrib == 'class' && $value == '' ) {
 306                                 unset( $attribs[$attrib] );
 307                         }
 308                 }
 309
 310                 # More subtle checks
 311                 if ( $element === 'link' && isset( $attribs['type'] )
 312                 && strval( $attribs['type'] ) == 'text/css' ) {
 313                         unset( $attribs['type'] );
 314                 }
 315                 if ( $element === 'select' && isset( $attribs['size'] ) ) {
 316                         if ( in_array( 'multiple', $attribs )
 317                                 || ( isset( $attribs['multiple'] ) && $attribs['multiple'] !== false )
 318                         ) {
 319                                 # A multi-select
 320                                 if ( strval( $attribs['size'] ) == '4' ) {
 321                                         unset( $attribs['size'] );
 322                                 }
 323                         } else {
 324                                 # Single select
 325                                 if ( strval( $attribs['size'] ) == '1' ) {
 326                                         unset( $attribs['size'] );
 327                                 }
 328                         }
 329                 }
 330
 331                 return $attribs;
 332         }
 333
 334         /**
 335          * Given an associative array of element attributes, generate a string
 336          * to stick after the element name in HTML output.  Like array( 'href' =>
 337          * 'http://www.mediawiki.org/' ) becomes something like
 338          * ' href="http://www.mediawiki.org"'.  Again, this is like
 339          * Xml::expandAttributes(), but it implements some HTML-specific logic.
 340          * For instance, it will omit quotation marks if $wgWellFormedXml is false,
 341          * and will treat boolean attributes specially.
 342          *
 343          * @param $attribs array Associative array of attributes, e.g., array(
 344          *   'href' => 'http://www.mediawiki.org/' ).  Values will be HTML-escaped.
 345          *   A value of false means to omit the attribute.  For boolean attributes,
 346          *   you can omit the key, e.g., array( 'checked' ) instead of
 347          *   array( 'checked' => 'checked' ) or such.
 348          * @return string HTML fragment that goes between element name and '>'
 349          *   (starting with a space if at least one attribute is output)
 350          */
 351         public static function expandAttributes( $attribs ) {
 352                 global $wgHtml5, $wgWellFormedXml;
 353
 354                 $ret = '';
 355                 $attribs = (array)$attribs;
 356                 foreach ( $attribs as $key => $value ) {
 357                         if ( $value === false ) {
 358                                 continue;
 359                         }
 360
 361                         # For boolean attributes, support array( 'foo' ) instead of
 362                         # requiring array( 'foo' => 'meaningless' ).
 363                         if ( is_int( $key )
 364                         && in_array( strtolower( $value ), self::$boolAttribs ) ) {
 365                                 $key = $value;
 366                         }
 367
 368                         # Not technically required in HTML5, but required in XHTML 1.0,
 369                         # and we'd like consistency and better compression anyway.
 370                         $key = strtolower( $key );
 371
 372                         # See the "Attributes" section in the HTML syntax part of HTML5,
 373                         # 9.1.2.3 as of 2009-08-10.  Most attributes can have quotation
 374                         # marks omitted, but not all.  (Although a literal " is not
 375                         # permitted, we don't check for that, since it will be escaped
 376                         # anyway.)
 377                         #
 378                         # See also research done on further characters that need to be
 379                         # escaped: http://code.google.com/p/html5lib/issues/detail?id=93
 380                         $badChars = "\\x00- '=<>`/\x{00a0}\x{1680}\x{180e}\x{180F}\x{2000}\x{2001}"
 381                                 . "\x{2002}\x{2003}\x{2004}\x{2005}\x{2006}\x{2007}\x{2008}\x{2009}"
 382                                 . "\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}";
 383                         if ( $wgWellFormedXml || $value === ''
 384                         || preg_match( "![$badChars]!u", $value ) ) {
 385                                 $quote = '"';
 386                         } else {
 387                                 $quote = '';
 388                         }
 389
 390                         if ( in_array( $key, self::$boolAttribs ) ) {
 391                                 # In XHTML 1.0 Transitional, the value needs to be equal to the
 392                                 # key.  In HTML5, we can leave the value empty instead.  If we
 393                                 # don't need well-formed XML, we can omit the = entirely.
 394                                 if ( !$wgWellFormedXml ) {
 395                                         $ret .= " $key";
 396                                 } elseif ( $wgHtml5 ) {
 397                                         $ret .= " $key=\"\"";
 398                                 } else {
 399                                         $ret .= " $key=\"$key\"";
 400                                 }
 401                         } else {
 402                                 # Apparently we need to entity-encode \n, \r, \t, although the
 403                                 # spec doesn't mention that.  Since we're doing strtr() anyway,
 404                                 # and we don't need <> escaped here, we may as well not call
 405                                 # htmlspecialchars().  FIXME: verify that we actually need to
 406                                 # escape \n\r\t here, and explain why, exactly.
 407                                 #
 408                                 # We could call Sanitizer::encodeAttribute() for this, but we
 409                                 # don't because we're stubborn and like our marginal savings on
 410                                 # byte size from not having to encode unnecessary quotes.
 411                                 $map = array(
 412                                         '&' => '&amp;',
 413                                         '"' => '&quot;',
 414                                         "\n" => '&#10;',
 415                                         "\r" => '&#13;',
 416                                         "\t" => '&#9;'
 417                                 );
 418                                 if ( $wgWellFormedXml ) {
 419                                         # This is allowed per spec: <http://www.w3.org/TR/xml/#NT-AttValue>
 420                                         # But reportedly it breaks some XML tools?  FIXME: is this
 421                                         # really true?
 422                                         $map['<'] = '&lt;';
 423                                 }
 424                                 $ret .= " $key=$quote" . strtr( $value, $map ) . $quote;
 425                         }
 426                 }
 427                 return $ret;
 428         }
 429
 430         /**
 431          * Output a <script> tag with the given contents.  TODO: do some useful
 432          * escaping as well, like if $contents contains literal '</script>' or (for
 433          * XML) literal "]]>".
 434          *
 435          * @param $contents string JavaScript
 436          * @return string Raw HTML
 437          */
 438         public static function inlineScript( $contents ) {
 439                 global $wgHtml5, $wgJsMimeType, $wgWellFormedXml;
 440
 441                 $attrs = array();
 442                 if ( !$wgHtml5 ) {
 443                         $attrs['type'] = $wgJsMimeType;
 444                 }
 445                 if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) {
 446                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 447                 }
 448                 return self::rawElement( 'script', $attrs, $contents );
 449         }
 450
 451         /**
 452          * Output a <script> tag linking to the given URL, e.g.,
 453          * <script src=foo.js></script>.
 454          *
 455          * @param $url string
 456          * @return string Raw HTML
 457          */
 458         public static function linkedScript( $url ) {
 459                 global $wgHtml5, $wgJsMimeType;
 460
 461                 $attrs = array( 'src' => $url );
 462                 if ( !$wgHtml5 ) {
 463                         $attrs['type'] = $wgJsMimeType;
 464                 }
 465                 return self::element( 'script', $attrs );
 466         }
 467
 468         /**
 469          * Output a <style> tag with the given contents for the given media type
 470          * (if any).  TODO: do some useful escaping as well, like if $contents
 471          * contains literal '</style>' (admittedly unlikely).
 472          *
 473          * @param $contents string CSS
 474          * @param $media mixed A media type string, like 'screen'
 475          * @return string Raw HTML
 476          */
 477         public static function inlineStyle( $contents, $media = 'all' ) {
 478                 global $wgWellFormedXml;
 479
 480                 if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) {
 481                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 482                 }
 483                 return self::rawElement( 'style', array(
 484                         'type' => 'text/css',
 485                         'media' => $media,
 486                 ), $contents );
 487         }
 488
 489         /**
 490          * Output a <link rel=stylesheet> linking to the given URL for the given
 491          * media type (if any).
 492          *
 493          * @param $url string
 494          * @param $media mixed A media type string, like 'screen'
 495          * @return string Raw HTML
 496          */
 497         public static function linkedStyle( $url, $media = 'all' ) {
 498                 return self::element( 'link', array(
 499                         'rel' => 'stylesheet',
 500                         'href' => $url,
 501                         'type' => 'text/css',
 502                         'media' => $media,
 503                 ) );
 504         }
 505
 506         /**
 507          * Convenience function to produce an <input> element.  This supports the
 508          * new HTML5 input types and attributes, and will silently strip them if
 509          * $wgHtml5 is false.
 510          *
 511          * @param $name    string name attribute
 512          * @param $value   mixed  value attribute
 513          * @param $type    string type attribute
 514          * @param $attribs array  Associative array of miscellaneous extra
 515          *   attributes, passed to Html::element()
 516          * @return string Raw HTML
 517          */
 518         public static function input( $name, $value = '', $type = 'text', $attribs = array() ) {
 519                 $attribs['type'] = $type;
 520                 $attribs['value'] = $value;
 521                 $attribs['name'] = $name;
 522
 523                 return self::element( 'input', $attribs );
 524         }
 525
 526         /**
 527          * Convenience function to produce an input element with type=hidden, like
 528          * Xml::hidden.
 529          *
 530          * @param $name    string name attribute
 531          * @param $value   string value attribute
 532          * @param $attribs array  Associative array of miscellaneous extra
 533          *   attributes, passed to Html::element()
 534          * @return string Raw HTML
 535          */
 536         public static function hidden( $name, $value, $attribs = array() ) {
 537                 return self::input( $name, $value, 'hidden', $attribs );
 538         }
 539
 540         /**
 541          * Convenience function to produce an <input> element.  This supports leaving
 542          * out the cols= and rows= which Xml requires and are required by HTML4/XHTML
 543          * but not required by HTML5 and will silently set cols="" and rows="" if
 544          * $wgHtml5 is false and cols and rows are omitted (HTML4 validates present
 545          * but empty cols="" and rows="" as valid).
 546          *
 547          * @param $name    string name attribute
 548          * @param $value   string value attribute
 549          * @param $attribs array  Associative array of miscellaneous extra
 550          *   attributes, passed to Html::element()
 551          * @return string Raw HTML
 552          */
 553         public static function textarea( $name, $value = '', $attribs = array() ) {
 554                 global $wgHtml5;
 555                 $attribs['name'] = $name;
 556                 if ( !$wgHtml5 ) {
 557                         if ( !isset( $attribs['cols'] ) )
 558                                 $attribs['cols'] = "";
 559                         if ( !isset( $attribs['rows'] ) )
 560                                 $attribs['rows'] = "";
 561                 }
 562                 return self::element( 'textarea', $attribs, $value );
 563         }
 564 }