includes/Xml.php

   1 <?php
   2
   3 /**
   4  * Module of static functions for generating XML
   5  */
   6
   7 class Xml {
   8         /**
   9          * Format an XML element with given attributes and, optionally, text content.
  10          * Element and attribute names are assumed to be ready for literal inclusion.
  11          * Strings are assumed to not contain XML-illegal characters; special
  12          * characters (<, >, &) are escaped but illegals are not touched.
  13          *
  14          * @param $element String:
  15          * @param $attribs Array: Name=>value pairs. Values will be escaped.
  16          * @param $contents String: NULL to make an open tag only; '' for a contentless closed tag (default)
  17          * @return string
  18          */
  19         public static function element( $element, $attribs = null, $contents = '') {
  20                 $out = '<' . $element;
  21                 if( !is_null( $attribs ) ) {
  22                         $out .=  self::expandAttributes( $attribs );
  23                 }
  24                 if( is_null( $contents ) ) {
  25                         $out .= '>';
  26                 } else {
  27                         if( $contents === '' ) {
  28                                 $out .= ' />';
  29                         } else {
  30                                 $out .= '>' . htmlspecialchars( $contents ) . "</$element>";
  31                         }
  32                 }
  33                 return $out;
  34         }
  35
  36         /**
  37          * Given an array of ('attributename' => 'value'), it generates the code
  38          * to set the XML attributes : attributename="value".
  39          * The values are passed to Sanitizer::encodeAttribute.
  40          * Return null if no attributes given.
  41          * @param $attribs Array of attributes for an XML element
  42          */
  43         private static function expandAttributes( $attribs ) {
  44                 if( is_null( $attribs ) ) {
  45                         return null;
  46                 } else {
  47                         $out = '';
  48                         foreach( $attribs as $name => $val ) {
  49                                 $out .= ' ' . $name . '="' . Sanitizer::encodeAttribute( $val ) . '"';
  50                         }
  51                         return $out;
  52                 }
  53         }
  54
  55         /**
  56          * Format an XML element as with self::element(), but run text through the
  57          * UtfNormal::cleanUp() validator first to ensure that no invalid UTF-8
  58          * is passed.
  59          *
  60          * @param $element String:
  61          * @param $attribs Array: Name=>value pairs. Values will be escaped.
  62          * @param $contents String: NULL to make an open tag only; '' for a contentless closed tag (default)
  63          * @return string
  64          */
  65         public static function elementClean( $element, $attribs = array(), $contents = '') {
  66                 if( $attribs ) {
  67                         $attribs = array_map( array( 'UtfNormal', 'cleanUp' ), $attribs );
  68                 }
  69                 if( $contents ) {
  70                         wfProfileIn( __METHOD__ . '-norm' );
  71                         $contents = UtfNormal::cleanUp( $contents );
  72                         wfProfileOut( __METHOD__ . '-norm' );
  73                 }
  74                 return self::element( $element, $attribs, $contents );
  75         }
  76
  77         /** This open an XML element */
  78         public static function openElement( $element, $attribs = null ) {
  79                 return '<' . $element . self::expandAttributes( $attribs ) . '>';
  80         }
  81
  82         // Shortcut
  83         public static function closeElement( $element ) { return "</$element>"; }
  84
  85         /**
  86          * Same as <link>element</link>, but does not escape contents. Handy when the
  87          * content you have is already valid xml.
  88          */
  89         public static function tags( $element, $attribs = null, $contents ) {
  90                 return self::openElement( $element, $attribs ) . $contents . "</$element>";
  91         }
  92
  93         /**
  94          * Create a namespace selector
  95          *
  96          * @param $selected Mixed: the namespace which should be selected, default ''
  97          * @param $allnamespaces String: value of a special item denoting all namespaces. Null to not include (default)
  98          * @param $includehidden Bool: include hidden namespaces?
  99          * @return String: Html string containing the namespace selector
 100          */
 101         public static function namespaceSelector($selected = '', $allnamespaces = null, $includehidden=false) {
 102                 global $wgContLang;
 103                 if( is_null( $selected ) )
 104                         $selected = '';
 105                 $s = "\n<select id='namespace' name='namespace' class='namespaceselector'>\n";
 106                 $arr = $wgContLang->getFormattedNamespaces();
 107                 if( !is_null($allnamespaces) ) {
 108                         $arr = array($allnamespaces => wfMsg('namespacesall')) + $arr;
 109                 }
 110                 foreach ($arr as $index => $name) {
 111                         if ($index < NS_MAIN) continue;
 112
 113                         $name = $index !== 0 ? $name : wfMsg('blanknamespace');
 114
 115                         if ($index === $selected) {
 116                                 $s .= "\t" . self::element("option",
 117                                                 array("value" => $index, "selected" => "selected"),
 118                                                 $name) . "\n";
 119                         } else {
 120                                 $s .= "\t" . self::element("option", array("value" => $index), $name) . "\n";
 121                         }
 122                 }
 123                 $s .= "</select>\n";
 124                 return $s;
 125         }
 126
 127         /**
 128         * Create a date selector
 129         *
 130         * @param $selected Mixed: the month which should be selected, default ''
 131         * @param $allmonths String: value of a special item denoting all month. Null to not include (default)
 132         * @param string $id Element identifier
 133         * @return String: Html string containing the month selector
 134         */
 135         public static function monthSelector( $selected = '', $allmonths = null, $id = 'month' ) {
 136                 global $wgLang;
 137                 $options = array();
 138             if( is_null( $selected ) )
 139                         $selected = '';
 140             if( !is_null( $allmonths ) )
 141                         $options[] = self::option( wfMsg( 'monthsall' ), $allmonths, $selected === $allmonths );
 142                 for( $i = 1; $i < 13; $i++ )
 143                                 $options[] = self::option( $wgLang->getMonthName( $i ), $i, $selected === $i );
 144                 return self::openElement( 'select', array( 'id' => $id, 'name' => 'month' ) )
 145                         . implode( "\n", $options )
 146                         . self::closeElement( 'select' );
 147         }
 148
 149         /**
 150          *
 151          * @param $language The language code of the selected language
 152          * @param $customisedOnly If true only languages which have some content are listed
 153          * @return array of label and select
 154          */
 155         public static function languageSelector( $selected, $customisedOnly = true ) {
 156                 global $wgContLanguageCode;
 157                 /**
 158                  * Make sure the site language is in the list; a custom language code
 159                  * might not have a defined name...
 160                  */
 161                 $languages = Language::getLanguageNames( $customisedOnly );
 162                 if( !array_key_exists( $wgContLanguageCode, $languages ) ) {
 163                         $languages[$wgContLanguageCode] = $wgContLanguageCode;
 164                 }
 165                 ksort( $languages );
 166
 167                 /**
 168                  * If a bogus value is set, default to the content language.
 169                  * Otherwise, no default is selected and the user ends up
 170                  * with an Afrikaans interface since it's first in the list.
 171                  */
 172                 $selected = isset( $languages[$selected] ) ? $selected : $wgContLanguageCode;
 173                 $options = "\n";
 174                 foreach( $languages as $code => $name ) {
 175                         $options .= Xml::option( "$code - $name", $code, ($code == $selected) ) . "\n";
 176                 }
 177
 178                 return array(
 179                         Xml::label( wfMsg('yourlanguage'), 'wpUserLanguage' ),
 180                         Xml::tags( 'select',
 181                                 array( 'id' => 'wpUserLanguage', 'name' => 'wpUserLanguage' ),
 182                                 $options
 183                         )
 184                 );
 185
 186         }
 187
 188         public static function span( $text, $class, $attribs=array() ) {
 189                 return self::element( 'span', array( 'class' => $class ) + $attribs, $text );
 190         }
 191
 192         /**
 193          * Convenience function to build an HTML text input field
 194          * @return string HTML
 195          */
 196         public static function input( $name, $size=false, $value=false, $attribs=array() ) {
 197                 return self::element( 'input', array(
 198                         'name' => $name,
 199                         'size' => $size,
 200                         'value' => $value ) + $attribs );
 201         }
 202
 203         /**
 204          * Convenience function to build an HTML password input field
 205          * @return string HTML
 206          */
 207         public static function password( $name, $size=false, $value=false, $attribs=array() ) {
 208                 return self::input( $name, $size, $value, array_merge($attribs, array('type' => 'password')));
 209         }
 210
 211         /**
 212          * Internal function for use in checkboxes and radio buttons and such.
 213          * @return array
 214          */
 215         public static function attrib( $name, $present = true ) {
 216                 return $present ? array( $name => $name ) : array();
 217         }
 218
 219         /**
 220          * Convenience function to build an HTML checkbox
 221          * @return string HTML
 222          */
 223         public static function check( $name, $checked=false, $attribs=array() ) {
 224                 return self::element( 'input', array_merge(
 225                         array(
 226                                 'name' => $name,
 227                                 'type' => 'checkbox',
 228                                 'value' => 1 ),
 229                         self::attrib( 'checked', $checked ),
 230                         $attribs ) );
 231         }
 232
 233         /**
 234          * Convenience function to build an HTML radio button
 235          * @return string HTML
 236          */
 237         public static function radio( $name, $value, $checked=false, $attribs=array() ) {
 238                 return self::element( 'input', array(
 239                         'name' => $name,
 240                         'type' => 'radio',
 241                         'value' => $value ) + self::attrib( 'checked', $checked ) + $attribs );
 242         }
 243
 244         /**
 245          * Convenience function to build an HTML form label
 246          * @return string HTML
 247          */
 248         public static function label( $label, $id ) {
 249                 return self::element( 'label', array( 'for' => $id ), $label );
 250         }
 251
 252         /**
 253          * Convenience function to build an HTML text input field with a label
 254          * @return string HTML
 255          */
 256         public static function inputLabel( $label, $name, $id, $size=false, $value=false, $attribs=array() ) {
 257                 return Xml::label( $label, $id ) .
 258                         '&nbsp;' .
 259                         self::input( $name, $size, $value, array( 'id' => $id ) + $attribs );
 260         }
 261
 262         /**
 263          * Convenience function to build an HTML checkbox with a label
 264          * @return string HTML
 265          */
 266         public static function checkLabel( $label, $name, $id, $checked=false, $attribs=array() ) {
 267                 return self::check( $name, $checked, array( 'id' => $id ) + $attribs ) .
 268                         '&nbsp;' .
 269                         self::label( $label, $id );
 270         }
 271
 272         /**
 273          * Convenience function to build an HTML radio button with a label
 274          * @return string HTML
 275          */
 276         public static function radioLabel( $label, $name, $value, $id, $checked=false, $attribs=array() ) {
 277                 return self::radio( $name, $value, $checked, array( 'id' => $id ) + $attribs ) .
 278                         '&nbsp;' .
 279                         self::label( $label, $id );
 280         }
 281
 282         /**
 283          * Convenience function to build an HTML submit button
 284          * @param $value String: label text for the button
 285          * @param $attribs Array: optional custom attributes
 286          * @return string HTML
 287          */
 288         public static function submitButton( $value, $attribs=array() ) {
 289                 return self::element( 'input', array( 'type' => 'submit', 'value' => $value ) + $attribs );
 290         }
 291
 292         /**
 293          * Convenience function to build an HTML hidden form field.
 294          * @todo Document $name parameter.
 295          * @param $name FIXME
 296          * @param $value String: label text for the button
 297          * @param $attribs Array: optional custom attributes
 298          * @return string HTML
 299          */
 300         public static function hidden( $name, $value, $attribs=array() ) {
 301                 return self::element( 'input', array(
 302                         'name' => $name,
 303                         'type' => 'hidden',
 304                         'value' => $value ) + $attribs );
 305         }
 306
 307         /**
 308          * Convenience function to build an HTML drop-down list item.
 309          * @param $text String: text for this item
 310          * @param $value String: form submission value; if empty, use text
 311          * @param $selected boolean: if true, will be the default selected item
 312          * @param $attribs array: optional additional HTML attributes
 313          * @return string HTML
 314          */
 315         public static function option( $text, $value=null, $selected=false,
 316                         $attribs=array() ) {
 317                 if( !is_null( $value ) ) {
 318                         $attribs['value'] = $value;
 319                 }
 320                 if( $selected ) {
 321                         $attribs['selected'] = 'selected';
 322                 }
 323                 return self::element( 'option', $attribs, $text );
 324         }
 325
 326         /**
 327          * Returns an escaped string suitable for inclusion in a string literal
 328          * for JavaScript source code.
 329          * Illegal control characters are assumed not to be present.
 330          *
 331          * @param string $string
 332          * @return string
 333          */
 334         public static function escapeJsString( $string ) {
 335                 // See ECMA 262 section 7.8.4 for string literal format
 336                 $pairs = array(
 337                         "\\" => "\\\\",
 338                         "\"" => "\\\"",
 339                         '\'' => '\\\'',
 340                         "\n" => "\\n",
 341                         "\r" => "\\r",
 342
 343                         # To avoid closing the element or CDATA section
 344                         "<" => "\\x3c",
 345                         ">" => "\\x3e",
 346
 347                         # To avoid any complaints about bad entity refs
 348                         "&" => "\\x26",
 349
 350                         # Work around https://bugzilla.mozilla.org/show_bug.cgi?id=274152
 351                         # Encode certain Unicode formatting chars so affected
 352                         # versions of Gecko don't misinterpret our strings;
 353                         # this is a common problem with Farsi text.
 354                         "\xe2\x80\x8c" => "\\u200c", // ZERO WIDTH NON-JOINER
 355                         "\xe2\x80\x8d" => "\\u200d", // ZERO WIDTH JOINER
 356                 );
 357                 return strtr( $string, $pairs );
 358         }
 359
 360         /**
 361          * Encode a variable of unknown type to JavaScript.
 362          * Arrays are converted to JS arrays, objects are converted to JS associative
 363          * arrays (objects). So cast your PHP associative arrays to objects before
 364          * passing them to here.
 365          */
 366         public static function encodeJsVar( $value ) {
 367                 if ( is_bool( $value ) ) {
 368                         $s = $value ? 'true' : 'false';
 369                 } elseif ( is_null( $value ) ) {
 370                         $s = 'null';
 371                 } elseif ( is_int( $value ) ) {
 372                         $s = $value;
 373                 } elseif ( is_array( $value ) ) {
 374                         $s = '[';
 375                         foreach ( $value as $elt ) {
 376                                 if ( $s != '[' ) {
 377                                         $s .= ', ';
 378                                 }
 379                                 $s .= self::encodeJsVar( $elt );
 380                         }
 381                         $s .= ']';
 382                 } elseif ( is_object( $value ) ) {
 383                         $s = '{';
 384                         foreach ( (array)$value as $name => $elt ) {
 385                                 if ( $s != '{' ) {
 386                                         $s .= ', ';
 387                                 }
 388                                 $s .= '"' . self::escapeJsString( $name ) . '": ' .
 389                                         self::encodeJsVar( $elt );
 390                         }
 391                         $s .= '}';
 392                 } else {
 393                         $s = '"' . self::escapeJsString( $value ) . '"';
 394                 }
 395                 return $s;
 396         }
 397
 398
 399         /**
 400          * Check if a string is well-formed XML.
 401          * Must include the surrounding tag.
 402          *
 403          * @param $text String: string to test.
 404          * @return bool
 405          *
 406          * @todo Error position reporting return
 407          */
 408         public static function isWellFormed( $text ) {
 409                 $parser = xml_parser_create( "UTF-8" );
 410
 411                 # case folding violates XML standard, turn it off
 412                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 413
 414                 if( !xml_parse( $parser, $text, true ) ) {
 415                         //$err = xml_error_string( xml_get_error_code( $parser ) );
 416                         //$position = xml_get_current_byte_index( $parser );
 417                         //$fragment = $this->extractFragment( $html, $position );
 418                         //$this->mXmlError = "$err at byte $position:\n$fragment";
 419                         xml_parser_free( $parser );
 420                         return false;
 421                 }
 422                 xml_parser_free( $parser );
 423                 return true;
 424         }
 425
 426         /**
 427          * Check if a string is a well-formed XML fragment.
 428          * Wraps fragment in an \<html\> bit and doctype, so it can be a fragment
 429          * and can use HTML named entities.
 430          *
 431          * @param $text String:
 432          * @return bool
 433          */
 434         public static function isWellFormedXmlFragment( $text ) {
 435                 $html =
 436                         Sanitizer::hackDocType() .
 437                         '<html>' .
 438                         $text .
 439                         '</html>';
 440                 return Xml::isWellFormed( $html );
 441         }
 442
 443         /**
 444          * Replace " > and < with their respective HTML entities ( &quot;,
 445          * &gt;, &lt;)
 446          *
 447          * @param $in String: text that might contain HTML tags.
 448          * @return string Escaped string
 449          */
 450         public static function escapeTagsOnly( $in ) {
 451                 return str_replace(
 452                         array( '"', '>', '<' ),
 453                         array( '&quot;', '&gt;', '&lt;' ),
 454                         $in );
 455         }
 456 }
 457