includes/Html.php

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