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$contents</$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;
 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( $element );
 147
 148                 # Remove HTML5-only attributes if we aren't doing HTML5
 149                 if ( !$wgHtml5 ) {
 150                         if ( $element == 'input' ) {
 151                                 # Whitelist of valid XHTML1 types
 152                                 $validTypes = array(
 153                                         'hidden',
 154                                         'text',
 155                                         'password',
 156                                         'checkbox',
 157                                         'radio',
 158                                         'file',
 159                                         'submit',
 160                                         'image',
 161                                         'reset',
 162                                         'button',
 163                                 );
 164                                 if ( isset( $attribs['type'] )
 165                                 && !in_array( $attribs['type'], $validTypes ) ) {
 166                                         # Fall back to type=text, the default
 167                                         unset( $attribs['type'] );
 168                                 }
 169                         }
 170                         if ( $element == 'textarea' && isset( $attribs['maxlength'] ) ) {
 171                                 unset( $attribs['maxlength'] );
 172                         }
 173                         # Here we're blacklisting some HTML5-only attributes...
 174                         $html5attribs = array(
 175                                 'autocomplete',
 176                                 'autofocus',
 177                                 'max',
 178                                 'min',
 179                                 'multiple',
 180                                 'pattern',
 181                                 'placeholder',
 182                                 'required',
 183                                 'step',
 184                                 'spellcheck',
 185                         );
 186                         foreach ( $html5attribs as $badAttr ) {
 187                                 unset( $attribs[$badAttr] );
 188                         }
 189                 }
 190
 191                 return "<$element" . self::expandAttributes(
 192                         self::dropDefaults( $element, $attribs ) ) . '>';
 193         }
 194
 195         /**
 196          * Given an element name and an associative array of element attributes,
 197          * return an array that is functionally identical to the input array, but
 198          * possibly smaller.  In particular, attributes might be stripped if they
 199          * are given their default values.
 200          *
 201          * This method is not guaranteed to remove all redundant attributes, only
 202          * some common ones and some others selected arbitrarily at random.  It
 203          * only guarantees that the output array should be functionally identical
 204          * to the input array (currently per the HTML 5 draft as of 2009-09-06).
 205          *
 206          * @param $element string Name of the element, e.g., 'a'
 207          * @param $attribs array  Associative array of attributes, e.g., array(
 208          *   'href' => 'http://www.mediawiki.org/' ).  See expandAttributes() for
 209          *   further documentation.
 210          * @return array An array of attributes functionally identical to $attribs
 211          */
 212         private static function dropDefaults( $element, $attribs ) {
 213                 # Don't bother doing anything if we aren't outputting HTML5; it's too
 214                 # much of a pain to maintain two sets of defaults.
 215                 global $wgHtml5;
 216                 if ( !$wgHtml5 ) {
 217                         return $attribs;
 218                 }
 219
 220                 static $attribDefaults = array(
 221                         'area' => array( 'shape' => 'rect' ),
 222                         'button' => array(
 223                                 'formaction' => 'GET',
 224                                 'formenctype' => 'application/x-www-form-urlencoded',
 225                                 'type' => 'submit',
 226                         ),
 227                         'canvas' => array(
 228                                 'height' => '150',
 229                                 'width' => '300',
 230                         ),
 231                         'command' => array( 'type' => 'command' ),
 232                         'form' => array(
 233                                 'action' => 'GET',
 234                                 'autocomplete' => 'on',
 235                                 'enctype' => 'application/x-www-form-urlencoded',
 236                         ),
 237                         'input' => array(
 238                                 'formaction' => 'GET',
 239                                 'type' => 'text',
 240                                 'value' => '',
 241                         ),
 242                         'keygen' => array( 'keytype' => 'rsa' ),
 243                         'link' => array( 'media' => 'all' ),
 244                         'menu' => array( 'type' => 'list' ),
 245                         # Note: the use of text/javascript here instead of other JavaScript
 246                         # MIME types follows the HTML5 spec.
 247                         'script' => array( 'type' => 'text/javascript' ),
 248                         'style' => array(
 249                                 'media' => 'all',
 250                                 'type' => 'text/css',
 251                         ),
 252                         'textarea' => array( 'wrap' => 'soft' ),
 253                 );
 254
 255                 $element = strtolower( $element );
 256
 257                 foreach ( $attribs as $attrib => $value ) {
 258                         $lcattrib = strtolower( $attrib );
 259                         $value = strval( $value );
 260
 261                         # Simple checks using $attribDefaults
 262                         if ( isset( $attribDefaults[$element][$lcattrib] ) &&
 263                         $attribDefaults[$element][$lcattrib] == $value ) {
 264                                 unset( $attribs[$attrib] );
 265                         }
 266
 267                         if ( $lcattrib == 'class' && $value == '' ) {
 268                                 unset( $attribs[$attrib] );
 269                         }
 270                 }
 271
 272                 # More subtle checks
 273                 if ( $element === 'link' && isset( $attribs['type'] )
 274                 && strval( $attribs['type'] ) == 'text/css' ) {
 275                         unset( $attribs['type'] );
 276                 }
 277                 if ( $element === 'select' && isset( $attribs['size'] ) ) {
 278                         if ( in_array( 'multiple', $attribs )
 279                                 || ( isset( $attribs['multiple'] ) && $attribs['multiple'] !== false )
 280                         ) {
 281                                 # A multi-select
 282                                 if ( strval( $attribs['size'] ) == '4' ) {
 283                                         unset( $attribs['size'] );
 284                                 }
 285                         } else {
 286                                 # Single select
 287                                 if ( strval( $attribs['size'] ) == '1' ) {
 288                                         unset( $attribs['size'] );
 289                                 }
 290                         }
 291                 }
 292
 293                 return $attribs;
 294         }
 295
 296         /**
 297          * Given an associative array of element attributes, generate a string
 298          * to stick after the element name in HTML output.  Like array( 'href' =>
 299          * 'http://www.mediawiki.org/' ) becomes something like
 300          * ' href="http://www.mediawiki.org"'.  Again, this is like
 301          * Xml::expandAttributes(), but it implements some HTML-specific logic.
 302          * For instance, it will omit quotation marks if $wgWellFormedXml is false,
 303          * and will treat boolean attributes specially.
 304          *
 305          * @param $attribs array Associative array of attributes, e.g., array(
 306          *   'href' => 'http://www.mediawiki.org/' ).  Values will be HTML-escaped.
 307          *   A value of false means to omit the attribute.  For boolean attributes,
 308          *   you can omit the key, e.g., array( 'checked' ) instead of
 309          *   array( 'checked' => 'checked' ) or such.
 310          * @return string HTML fragment that goes between element name and '>'
 311          *   (starting with a space if at least one attribute is output)
 312          */
 313         public static function expandAttributes( $attribs ) {
 314                 global $wgHtml5, $wgWellFormedXml;
 315
 316                 $ret = '';
 317                 $attribs = (array)$attribs;
 318                 foreach ( $attribs as $key => $value ) {
 319                         if ( $value === false ) {
 320                                 continue;
 321                         }
 322
 323                         # For boolean attributes, support array( 'foo' ) instead of
 324                         # requiring array( 'foo' => 'meaningless' ).
 325                         if ( is_int( $key )
 326                         && in_array( strtolower( $value ), self::$boolAttribs ) ) {
 327                                 $key = $value;
 328                         }
 329
 330                         # Not technically required in HTML5, but required in XHTML 1.0,
 331                         # and we'd like consistency and better compression anyway.
 332                         $key = strtolower( $key );
 333
 334                         # See the "Attributes" section in the HTML syntax part of HTML5,
 335                         # 9.1.2.3 as of 2009-08-10.  Most attributes can have quotation
 336                         # marks omitted, but not all.  (Although a literal " is not
 337                         # permitted, we don't check for that, since it will be escaped
 338                         # anyway.)
 339                         #
 340                         # See also research done on further characters that need to be
 341                         # escaped: http://code.google.com/p/html5lib/issues/detail?id=93
 342                         $badChars = "\\x00- '=<>`/\x{00a0}\x{1680}\x{180e}\x{180F}\x{2000}\x{2001}"
 343                                 . "\x{2002}\x{2003}\x{2004}\x{2005}\x{2006}\x{2007}\x{2008}\x{2009}"
 344                                 . "\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}";
 345                         if ( $wgWellFormedXml || $value === ''
 346                         || preg_match( "![$badChars]!u", $value ) ) {
 347                                 $quote = '"';
 348                         } else {
 349                                 $quote = '';
 350                         }
 351
 352                         if ( in_array( $key, self::$boolAttribs ) ) {
 353                                 # In XHTML 1.0 Transitional, the value needs to be equal to the
 354                                 # key.  In HTML5, we can leave the value empty instead.  If we
 355                                 # don't need well-formed XML, we can omit the = entirely.
 356                                 if ( !$wgWellFormedXml ) {
 357                                         $ret .= " $key";
 358                                 } elseif ( $wgHtml5 ) {
 359                                         $ret .= " $key=\"\"";
 360                                 } else {
 361                                         $ret .= " $key=\"$key\"";
 362                                 }
 363                         } else {
 364                                 # Apparently we need to entity-encode \n, \r, \t, although the
 365                                 # spec doesn't mention that.  Since we're doing strtr() anyway,
 366                                 # and we don't need <> escaped here, we may as well not call
 367                                 # htmlspecialchars().  FIXME: verify that we actually need to
 368                                 # escape \n\r\t here, and explain why, exactly.
 369                                 #
 370                                 # We could call Sanitizer::encodeAttribute() for this, but we
 371                                 # don't because we're stubborn and like our marginal savings on
 372                                 # byte size from not having to encode unnecessary quotes.
 373                                 $map = array(
 374                                         '&' => '&amp;',
 375                                         '"' => '&quot;',
 376                                         "\n" => '&#10;',
 377                                         "\r" => '&#13;',
 378                                         "\t" => '&#9;'
 379                                 );
 380                                 if ( $wgWellFormedXml ) {
 381                                         # This is allowed per spec: <http://www.w3.org/TR/xml/#NT-AttValue>
 382                                         # But reportedly it breaks some XML tools?  FIXME: is this
 383                                         # really true?
 384                                         $map['<'] = '&lt;';
 385                                 }
 386                                 $ret .= " $key=$quote" . strtr( $value, $map ) . $quote;
 387                         }
 388                 }
 389                 return $ret;
 390         }
 391
 392         /**
 393          * Output a <script> tag with the given contents.  TODO: do some useful
 394          * escaping as well, like if $contents contains literal '</script>' or (for
 395          * XML) literal "]]>".
 396          *
 397          * @param $contents string JavaScript
 398          * @return string Raw HTML
 399          */
 400         public static function inlineScript( $contents ) {
 401                 global $wgHtml5, $wgJsMimeType, $wgWellFormedXml;
 402
 403                 $attrs = array();
 404                 if ( !$wgHtml5 ) {
 405                         $attrs['type'] = $wgJsMimeType;
 406                 }
 407                 if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) {
 408                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 409                 }
 410                 return self::rawElement( 'script', $attrs, $contents );
 411         }
 412
 413         /**
 414          * Output a <script> tag linking to the given URL, e.g.,
 415          * <script src=foo.js></script>.
 416          *
 417          * @param $url string
 418          * @return string Raw HTML
 419          */
 420         public static function linkedScript( $url ) {
 421                 global $wgHtml5, $wgJsMimeType;
 422
 423                 $attrs = array( 'src' => $url );
 424                 if ( !$wgHtml5 ) {
 425                         $attrs['type'] = $wgJsMimeType;
 426                 }
 427                 return self::element( 'script', $attrs );
 428         }
 429
 430         /**
 431          * Output a <style> tag with the given contents for the given media type
 432          * (if any).  TODO: do some useful escaping as well, like if $contents
 433          * contains literal '</style>' (admittedly unlikely).
 434          *
 435          * @param $contents string CSS
 436          * @param $media mixed A media type string, like 'screen'
 437          * @return string Raw HTML
 438          */
 439         public static function inlineStyle( $contents, $media = 'all' ) {
 440                 global $wgWellFormedXml;
 441
 442                 if ( $wgWellFormedXml && preg_match( '/[<&]/', $contents ) ) {
 443                         $contents = "/*<![CDATA[*/$contents/*]]>*/";
 444                 }
 445                 return self::rawElement( 'style', array(
 446                         'type' => 'text/css',
 447                         'media' => $media,
 448                 ), $contents );
 449         }
 450
 451         /**
 452          * Output a <link rel=stylesheet> linking to the given URL for the given
 453          * media type (if any).
 454          *
 455          * @param $url string
 456          * @param $media mixed A media type string, like 'screen'
 457          * @return string Raw HTML
 458          */
 459         public static function linkedStyle( $url, $media = 'all' ) {
 460                 return self::element( 'link', array(
 461                         'rel' => 'stylesheet',
 462                         'href' => $url,
 463                         'type' => 'text/css',
 464                         'media' => $media,
 465                 ) );
 466         }
 467
 468         /**
 469          * Convenience function to produce an <input> element.  This supports the
 470          * new HTML5 input types and attributes, and will silently strip them if
 471          * $wgHtml5 is false.
 472          *
 473          * @param $name    string name attribute
 474          * @param $value   mixed  value attribute
 475          * @param $type    string type attribute
 476          * @param $attribs array  Associative array of miscellaneous extra
 477          *   attributes, passed to Html::element()
 478          * @return string Raw HTML
 479          */
 480         public static function input( $name, $value = '', $type = 'text', $attribs = array() ) {
 481                 $attribs['type'] = $type;
 482                 $attribs['value'] = $value;
 483                 $attribs['name'] = $name;
 484
 485                 return self::element( 'input', $attribs );
 486         }
 487
 488         /**
 489          * Convenience function to produce an input element with type=hidden, like
 490          * Xml::hidden.
 491          *
 492          * @param $name    string name attribute
 493          * @param $value   string value attribute
 494          * @param $attribs array  Associative array of miscellaneous extra
 495          *   attributes, passed to Html::element()
 496          * @return string Raw HTML
 497          */
 498         public static function hidden( $name, $value, $attribs = array() ) {
 499                 return self::input( $name, $value, 'hidden', $attribs );
 500         }
 501
 502         /**
 503          * Convenience function to produce an <input> element.  This supports leaving
 504          * out the cols= and rows= which Xml requires and are required by HTML4/XHTML
 505          * but not required by HTML5 and will silently set cols="" and rows="" if
 506          * $wgHtml5 is false and cols and rows are omitted (HTML4 validates present
 507          * but empty cols="" and rows="" as valid).
 508          *
 509          * @param $name    string name attribute
 510          * @param $value   string value attribute
 511          * @param $attribs array  Associative array of miscellaneous extra
 512          *   attributes, passed to Html::element()
 513          * @return string Raw HTML
 514          */
 515         public static function textarea( $name, $value = '', $attribs = array() ) {
 516                 global $wgHtml5;
 517                 $attribs['name'] = $name;
 518                 if ( !$wgHtml5 ) {
 519                         if ( !isset( $attribs['cols'] ) )
 520                                 $attribs['cols'] = "";
 521                         if ( !isset( $attribs['rows'] ) )
 522                                 $attribs['rows'] = "";
 523                 }
 524                 return self::element( 'textarea', $attribs, $value );
 525         }
 526 }