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