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          * @param array $exclude Array of namespace indexes to exclude
 100          * @return String: Html string containing the namespace selector
 101          */
 102         public static function namespaceSelector($selected = '', $allnamespaces = null, $includehidden=false, $exclude = array() ) {
 103                 global $wgContLang;
 104                 if( is_null( $selected ) )
 105                         $selected = '';
 106                 $s = "\n<select id='namespace' name='namespace' class='namespaceselector'>\n";
 107                 $arr = $wgContLang->getFormattedNamespaces();
 108                 if( !is_null($allnamespaces) ) {
 109                         $arr = array($allnamespaces => wfMsg('namespacesall')) + $arr;
 110                 }
 111                 foreach ($arr as $index => $name) {
 112                         if( $index < NS_MAIN || in_array( $index, $exclude ) )
 113                                 continue;
 114
 115                         $name = $index !== 0 ? $name : wfMsg('blanknamespace');
 116
 117                         if ($index === $selected) {
 118                                 $s .= "\t" . self::element("option",
 119                                                 array("value" => $index, "selected" => "selected"),
 120                                                 $name) . "\n";
 121                         } else {
 122                                 $s .= "\t" . self::element("option", array("value" => $index), $name) . "\n";
 123                         }
 124                 }
 125                 $s .= "</select>\n";
 126                 return $s;
 127         }
 128
 129         /**
 130         * Create a date selector
 131         *
 132         * @param $selected Mixed: the month which should be selected, default ''
 133         * @param $allmonths String: value of a special item denoting all month. Null to not include (default)
 134         * @param string $id Element identifier
 135         * @return String: Html string containing the month selector
 136         */
 137         public static function monthSelector( $selected = '', $allmonths = null, $id = 'month' ) {
 138                 global $wgLang;
 139                 $options = array();
 140             if( is_null( $selected ) )
 141                         $selected = '';
 142             if( !is_null( $allmonths ) )
 143                         $options[] = self::option( wfMsg( 'monthsall' ), $allmonths, $selected === $allmonths );
 144                 for( $i = 1; $i < 13; $i++ )
 145                                 $options[] = self::option( $wgLang->getMonthName( $i ), $i, $selected === $i );
 146                 return self::openElement( 'select', array( 'id' => $id, 'name' => 'month' ) )
 147                         . implode( "\n", $options )
 148                         . self::closeElement( 'select' );
 149         }
 150
 151         /**
 152          *
 153          * @param $language The language code of the selected language
 154          * @param $customisedOnly If true only languages which have some content are listed
 155          * @return array of label and select
 156          */
 157         public static function languageSelector( $selected, $customisedOnly = true ) {
 158                 global $wgContLanguageCode;
 159                 /**
 160                  * Make sure the site language is in the list; a custom language code
 161                  * might not have a defined name...
 162                  */
 163                 $languages = Language::getLanguageNames( $customisedOnly );
 164                 if( !array_key_exists( $wgContLanguageCode, $languages ) ) {
 165                         $languages[$wgContLanguageCode] = $wgContLanguageCode;
 166                 }
 167                 ksort( $languages );
 168
 169                 /**
 170                  * If a bogus value is set, default to the content language.
 171                  * Otherwise, no default is selected and the user ends up
 172                  * with an Afrikaans interface since it's first in the list.
 173                  */
 174                 $selected = isset( $languages[$selected] ) ? $selected : $wgContLanguageCode;
 175                 $options = "\n";
 176                 foreach( $languages as $code => $name ) {
 177                         $options .= Xml::option( "$code - $name", $code, ($code == $selected) ) . "\n";
 178                 }
 179
 180                 return array(
 181                         Xml::label( wfMsg('yourlanguage'), 'wpUserLanguage' ),
 182                         Xml::tags( 'select',
 183                                 array( 'id' => 'wpUserLanguage', 'name' => 'wpUserLanguage' ),
 184                                 $options
 185                         )
 186                 );
 187
 188         }
 189
 190         public static function span( $text, $class, $attribs=array() ) {
 191                 return self::element( 'span', array( 'class' => $class ) + $attribs, $text );
 192         }
 193
 194         /**
 195          * Convenience function to build an HTML text input field
 196          * @return string HTML
 197          */
 198         public static function input( $name, $size=false, $value=false, $attribs=array() ) {
 199                 return self::element( 'input', array(
 200                         'name' => $name,
 201                         'size' => $size,
 202                         'value' => $value ) + $attribs );
 203         }
 204
 205         /**
 206          * Convenience function to build an HTML password input field
 207          * @return string HTML
 208          */
 209         public static function password( $name, $size=false, $value=false, $attribs=array() ) {
 210                 return self::input( $name, $size, $value, array_merge($attribs, array('type' => 'password')));
 211         }
 212
 213         /**
 214          * Internal function for use in checkboxes and radio buttons and such.
 215          * @return array
 216          */
 217         public static function attrib( $name, $present = true ) {
 218                 return $present ? array( $name => $name ) : array();
 219         }
 220
 221         /**
 222          * Convenience function to build an HTML checkbox
 223          * @return string HTML
 224          */
 225         public static function check( $name, $checked=false, $attribs=array() ) {
 226                 return self::element( 'input', array_merge(
 227                         array(
 228                                 'name' => $name,
 229                                 'type' => 'checkbox',
 230                                 'value' => 1 ),
 231                         self::attrib( 'checked', $checked ),
 232                         $attribs ) );
 233         }
 234
 235         /**
 236          * Convenience function to build an HTML radio button
 237          * @return string HTML
 238          */
 239         public static function radio( $name, $value, $checked=false, $attribs=array() ) {
 240                 return self::element( 'input', array(
 241                         'name' => $name,
 242                         'type' => 'radio',
 243                         'value' => $value ) + self::attrib( 'checked', $checked ) + $attribs );
 244         }
 245
 246         /**
 247          * Convenience function to build an HTML form label
 248          * @return string HTML
 249          */
 250         public static function label( $label, $id ) {
 251                 return self::element( 'label', array( 'for' => $id ), $label );
 252         }
 253
 254         /**
 255          * Convenience function to build an HTML text input field with a label
 256          * @return string HTML
 257          */
 258         public static function inputLabel( $label, $name, $id, $size=false, $value=false, $attribs=array() ) {
 259                 return Xml::label( $label, $id ) .
 260                         '&nbsp;' .
 261                         self::input( $name, $size, $value, array( 'id' => $id ) + $attribs );
 262         }
 263
 264         /**
 265          * Convenience function to build an HTML checkbox with a label
 266          * @return string HTML
 267          */
 268         public static function checkLabel( $label, $name, $id, $checked=false, $attribs=array() ) {
 269                 return self::check( $name, $checked, array( 'id' => $id ) + $attribs ) .
 270                         '&nbsp;' .
 271                         self::label( $label, $id );
 272         }
 273
 274         /**
 275          * Convenience function to build an HTML radio button with a label
 276          * @return string HTML
 277          */
 278         public static function radioLabel( $label, $name, $value, $id, $checked=false, $attribs=array() ) {
 279                 return self::radio( $name, $value, $checked, array( 'id' => $id ) + $attribs ) .
 280                         '&nbsp;' .
 281                         self::label( $label, $id );
 282         }
 283
 284         /**
 285          * Convenience function to build an HTML submit button
 286          * @param $value String: label text for the button
 287          * @param $attribs Array: optional custom attributes
 288          * @return string HTML
 289          */
 290         public static function submitButton( $value, $attribs=array() ) {
 291                 return self::element( 'input', array( 'type' => 'submit', 'value' => $value ) + $attribs );
 292         }
 293
 294         /**
 295          * Convenience function to build an HTML hidden form field.
 296          * @todo Document $name parameter.
 297          * @param $name FIXME
 298          * @param $value String: label text for the button
 299          * @param $attribs Array: optional custom attributes
 300          * @return string HTML
 301          */
 302         public static function hidden( $name, $value, $attribs=array() ) {
 303                 return self::element( 'input', array(
 304                         'name' => $name,
 305                         'type' => 'hidden',
 306                         'value' => $value ) + $attribs );
 307         }
 308
 309         /**
 310          * Convenience function to build an HTML drop-down list item.
 311          * @param $text String: text for this item
 312          * @param $value String: form submission value; if empty, use text
 313          * @param $selected boolean: if true, will be the default selected item
 314          * @param $attribs array: optional additional HTML attributes
 315          * @return string HTML
 316          */
 317         public static function option( $text, $value=null, $selected=false,
 318                         $attribs=array() ) {
 319                 if( !is_null( $value ) ) {
 320                         $attribs['value'] = $value;
 321                 }
 322                 if( $selected ) {
 323                         $attribs['selected'] = 'selected';
 324                 }
 325                 return self::element( 'option', $attribs, $text );
 326         }
 327
 328         /**
 329          * Returns an escaped string suitable for inclusion in a string literal
 330          * for JavaScript source code.
 331          * Illegal control characters are assumed not to be present.
 332          *
 333          * @param string $string
 334          * @return string
 335          */
 336         public static function escapeJsString( $string ) {
 337                 // See ECMA 262 section 7.8.4 for string literal format
 338                 $pairs = array(
 339                         "\\" => "\\\\",
 340                         "\"" => "\\\"",
 341                         '\'' => '\\\'',
 342                         "\n" => "\\n",
 343                         "\r" => "\\r",
 344
 345                         # To avoid closing the element or CDATA section
 346                         "<" => "\\x3c",
 347                         ">" => "\\x3e",
 348
 349                         # To avoid any complaints about bad entity refs
 350                         "&" => "\\x26",
 351
 352                         # Work around https://bugzilla.mozilla.org/show_bug.cgi?id=274152
 353                         # Encode certain Unicode formatting chars so affected
 354                         # versions of Gecko don't misinterpret our strings;
 355                         # this is a common problem with Farsi text.
 356                         "\xe2\x80\x8c" => "\\u200c", // ZERO WIDTH NON-JOINER
 357                         "\xe2\x80\x8d" => "\\u200d", // ZERO WIDTH JOINER
 358                 );
 359                 return strtr( $string, $pairs );
 360         }
 361
 362         /**
 363          * Encode a variable of unknown type to JavaScript.
 364          * Arrays are converted to JS arrays, objects are converted to JS associative
 365          * arrays (objects). So cast your PHP associative arrays to objects before
 366          * passing them to here.
 367          */
 368         public static function encodeJsVar( $value ) {
 369                 if ( is_bool( $value ) ) {
 370                         $s = $value ? 'true' : 'false';
 371                 } elseif ( is_null( $value ) ) {
 372                         $s = 'null';
 373                 } elseif ( is_int( $value ) ) {
 374                         $s = $value;
 375                 } elseif ( is_array( $value ) ) {
 376                         $s = '[';
 377                         foreach ( $value as $elt ) {
 378                                 if ( $s != '[' ) {
 379                                         $s .= ', ';
 380                                 }
 381                                 $s .= self::encodeJsVar( $elt );
 382                         }
 383                         $s .= ']';
 384                 } elseif ( is_object( $value ) ) {
 385                         $s = '{';
 386                         foreach ( (array)$value as $name => $elt ) {
 387                                 if ( $s != '{' ) {
 388                                         $s .= ', ';
 389                                 }
 390                                 $s .= '"' . self::escapeJsString( $name ) . '": ' .
 391                                         self::encodeJsVar( $elt );
 392                         }
 393                         $s .= '}';
 394                 } else {
 395                         $s = '"' . self::escapeJsString( $value ) . '"';
 396                 }
 397                 return $s;
 398         }
 399
 400
 401         /**
 402          * Check if a string is well-formed XML.
 403          * Must include the surrounding tag.
 404          *
 405          * @param $text String: string to test.
 406          * @return bool
 407          *
 408          * @todo Error position reporting return
 409          */
 410         public static function isWellFormed( $text ) {
 411                 $parser = xml_parser_create( "UTF-8" );
 412
 413                 # case folding violates XML standard, turn it off
 414                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 415
 416                 if( !xml_parse( $parser, $text, true ) ) {
 417                         //$err = xml_error_string( xml_get_error_code( $parser ) );
 418                         //$position = xml_get_current_byte_index( $parser );
 419                         //$fragment = $this->extractFragment( $html, $position );
 420                         //$this->mXmlError = "$err at byte $position:\n$fragment";
 421                         xml_parser_free( $parser );
 422                         return false;
 423                 }
 424                 xml_parser_free( $parser );
 425                 return true;
 426         }
 427
 428         /**
 429          * Check if a string is a well-formed XML fragment.
 430          * Wraps fragment in an \<html\> bit and doctype, so it can be a fragment
 431          * and can use HTML named entities.
 432          *
 433          * @param $text String:
 434          * @return bool
 435          */
 436         public static function isWellFormedXmlFragment( $text ) {
 437                 $html =
 438                         Sanitizer::hackDocType() .
 439                         '<html>' .
 440                         $text .
 441                         '</html>';
 442                 return Xml::isWellFormed( $html );
 443         }
 444
 445         /**
 446          * Replace " > and < with their respective HTML entities ( &quot;,
 447          * &gt;, &lt;)
 448          *
 449          * @param $in String: text that might contain HTML tags.
 450          * @return string Escaped string
 451          */
 452         public static function escapeTagsOnly( $in ) {
 453                 return str_replace(
 454                         array( '"', '>', '<' ),
 455                         array( '&quot;', '&gt;', '&lt;' ),
 456                         $in );
 457         }
 458 }
 459