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