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