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                 $options = array();
 106
 107                 // Godawful hack... we'll be frequently passed selected namespaces
 108                 // as strings since PHP is such a shithole.
 109                 // But we also don't want blanks and nulls and "all"s matching 0,
 110                 // so let's convert *just* string ints to clean ints.
 111                 if( preg_match( '/^\d+$/', $selected ) ) {
 112                         $selected = intval( $selected );
 113                 }
 114
 115                 if( !is_null( $all ) )
 116                         $namespaces = array( $all => wfMsg( 'namespacesall' ) ) + $namespaces;
 117                 foreach( $namespaces as $index => $name ) {
 118                         if( $index < NS_MAIN )
 119                                 continue;
 120                         if( $index === 0 )
 121                                 $name = wfMsg( 'blanknamespace' );
 122                         $options[] = self::option( $name, $index, $index === $selected );
 123                 }
 124
 125                 return Xml::openElement( 'select', array( 'id' => 'namespace', 'name' => $element_name,
 126                         'class' => 'namespaceselector' ) )
 127                         . "\n"
 128                         . implode( "\n", $options )
 129                         . "\n"
 130                         . Xml::closeElement( 'select' );
 131         }
 132
 133         /**
 134         * Create a date selector
 135         *
 136         * @param $selected Mixed: the month which should be selected, default ''
 137         * @param $allmonths String: value of a special item denoting all month. Null to not include (default)
 138         * @param string $id Element identifier
 139         * @return String: Html string containing the month selector
 140         */
 141         public static function monthSelector( $selected = '', $allmonths = null, $id = 'month' ) {
 142                 global $wgLang;
 143                 $options = array();
 144             if( is_null( $selected ) )
 145                         $selected = '';
 146             if( !is_null( $allmonths ) )
 147                         $options[] = self::option( wfMsg( 'monthsall' ), $allmonths, $selected === $allmonths );
 148                 for( $i = 1; $i < 13; $i++ )
 149                                 $options[] = self::option( $wgLang->getMonthName( $i ), $i, $selected === $i );
 150                 return self::openElement( 'select', array( 'id' => $id, 'name' => 'month' ) )
 151                         . implode( "\n", $options )
 152                         . self::closeElement( 'select' );
 153         }
 154
 155         /**
 156          *
 157          * @param $language The language code of the selected language
 158          * @param $customisedOnly If true only languages which have some content are listed
 159          * @return array of label and select
 160          */
 161         public static function languageSelector( $selected, $customisedOnly = true ) {
 162                 global $wgContLanguageCode;
 163                 /**
 164                  * Make sure the site language is in the list; a custom language code
 165                  * might not have a defined name...
 166                  */
 167                 $languages = Language::getLanguageNames( $customisedOnly );
 168                 if( !array_key_exists( $wgContLanguageCode, $languages ) ) {
 169                         $languages[$wgContLanguageCode] = $wgContLanguageCode;
 170                 }
 171                 ksort( $languages );
 172
 173                 /**
 174                  * If a bogus value is set, default to the content language.
 175                  * Otherwise, no default is selected and the user ends up
 176                  * with an Afrikaans interface since it's first in the list.
 177                  */
 178                 $selected = isset( $languages[$selected] ) ? $selected : $wgContLanguageCode;
 179                 $options = "\n";
 180                 foreach( $languages as $code => $name ) {
 181                         $options .= Xml::option( "$code - $name", $code, ($code == $selected) ) . "\n";
 182                 }
 183
 184                 return array(
 185                         Xml::label( wfMsg('yourlanguage'), 'wpUserLanguage' ),
 186                         Xml::tags( 'select',
 187                                 array( 'id' => 'wpUserLanguage', 'name' => 'wpUserLanguage' ),
 188                                 $options
 189                         )
 190                 );
 191
 192         }
 193
 194         public static function span( $text, $class, $attribs=array() ) {
 195                 return self::element( 'span', array( 'class' => $class ) + $attribs, $text );
 196         }
 197
 198         /**
 199          * Convenience function to build an HTML text input field
 200          * @return string HTML
 201          */
 202         public static function input( $name, $size=false, $value=false, $attribs=array() ) {
 203                 return self::element( 'input', array(
 204                         'name' => $name,
 205                         'size' => $size,
 206                         'value' => $value ) + $attribs );
 207         }
 208
 209         /**
 210          * Convenience function to build an HTML password input field
 211          * @return string HTML
 212          */
 213         public static function password( $name, $size=false, $value=false, $attribs=array() ) {
 214                 return self::input( $name, $size, $value, array_merge($attribs, array('type' => 'password')));
 215         }
 216
 217         /**
 218          * Internal function for use in checkboxes and radio buttons and such.
 219          * @return array
 220          */
 221         public static function attrib( $name, $present = true ) {
 222                 return $present ? array( $name => $name ) : array();
 223         }
 224
 225         /**
 226          * Convenience function to build an HTML checkbox
 227          * @return string HTML
 228          */
 229         public static function check( $name, $checked=false, $attribs=array() ) {
 230                 return self::element( 'input', array_merge(
 231                         array(
 232                                 'name' => $name,
 233                                 'type' => 'checkbox',
 234                                 'value' => 1 ),
 235                         self::attrib( 'checked', $checked ),
 236                         $attribs ) );
 237         }
 238
 239         /**
 240          * Convenience function to build an HTML radio button
 241          * @return string HTML
 242          */
 243         public static function radio( $name, $value, $checked=false, $attribs=array() ) {
 244                 return self::element( 'input', array(
 245                         'name' => $name,
 246                         'type' => 'radio',
 247                         'value' => $value ) + self::attrib( 'checked', $checked ) + $attribs );
 248         }
 249
 250         /**
 251          * Convenience function to build an HTML form label
 252          * @return string HTML
 253          */
 254         public static function label( $label, $id ) {
 255                 return self::element( 'label', array( 'for' => $id ), $label );
 256         }
 257
 258         /**
 259          * Convenience function to build an HTML text input field with a label
 260          * @return string HTML
 261          */
 262         public static function inputLabel( $label, $name, $id, $size=false, $value=false, $attribs=array() ) {
 263                 return Xml::label( $label, $id ) .
 264                         '&nbsp;' .
 265                         self::input( $name, $size, $value, array( 'id' => $id ) + $attribs );
 266         }
 267
 268         /**
 269          * Convenience function to build an HTML checkbox with a label
 270          * @return string HTML
 271          */
 272         public static function checkLabel( $label, $name, $id, $checked=false, $attribs=array() ) {
 273                 return self::check( $name, $checked, array( 'id' => $id ) + $attribs ) .
 274                         '&nbsp;' .
 275                         self::label( $label, $id );
 276         }
 277
 278         /**
 279          * Convenience function to build an HTML radio button with a label
 280          * @return string HTML
 281          */
 282         public static function radioLabel( $label, $name, $value, $id, $checked=false, $attribs=array() ) {
 283                 return self::radio( $name, $value, $checked, array( 'id' => $id ) + $attribs ) .
 284                         '&nbsp;' .
 285                         self::label( $label, $id );
 286         }
 287
 288         /**
 289          * Convenience function to build an HTML submit button
 290          * @param $value String: label text for the button
 291          * @param $attribs Array: optional custom attributes
 292          * @return string HTML
 293          */
 294         public static function submitButton( $value, $attribs=array() ) {
 295                 return self::element( 'input', array( 'type' => 'submit', 'value' => $value ) + $attribs );
 296         }
 297
 298         /**
 299          * Convenience function to build an HTML hidden form field.
 300          * @todo Document $name parameter.
 301          * @param $name FIXME
 302          * @param $value String: label text for the button
 303          * @param $attribs Array: optional custom attributes
 304          * @return string HTML
 305          */
 306         public static function hidden( $name, $value, $attribs=array() ) {
 307                 return self::element( 'input', array(
 308                         'name' => $name,
 309                         'type' => 'hidden',
 310                         'value' => $value ) + $attribs );
 311         }
 312
 313         /**
 314          * Convenience function to build an HTML drop-down list item.
 315          * @param $text String: text for this item
 316          * @param $value String: form submission value; if empty, use text
 317          * @param $selected boolean: if true, will be the default selected item
 318          * @param $attribs array: optional additional HTML attributes
 319          * @return string HTML
 320          */
 321         public static function option( $text, $value=null, $selected=false,
 322                         $attribs=array() ) {
 323                 if( !is_null( $value ) ) {
 324                         $attribs['value'] = $value;
 325                 }
 326                 if( $selected ) {
 327                         $attribs['selected'] = 'selected';
 328                 }
 329                 return self::element( 'option', $attribs, $text );
 330         }
 331
 332         /**
 333          * Returns an escaped string suitable for inclusion in a string literal
 334          * for JavaScript source code.
 335          * Illegal control characters are assumed not to be present.
 336          *
 337          * @param string $string
 338          * @return string
 339          */
 340         public static function escapeJsString( $string ) {
 341                 // See ECMA 262 section 7.8.4 for string literal format
 342                 $pairs = array(
 343                         "\\" => "\\\\",
 344                         "\"" => "\\\"",
 345                         '\'' => '\\\'',
 346                         "\n" => "\\n",
 347                         "\r" => "\\r",
 348
 349                         # To avoid closing the element or CDATA section
 350                         "<" => "\\x3c",
 351                         ">" => "\\x3e",
 352
 353                         # To avoid any complaints about bad entity refs
 354                         "&" => "\\x26",
 355
 356                         # Work around https://bugzilla.mozilla.org/show_bug.cgi?id=274152
 357                         # Encode certain Unicode formatting chars so affected
 358                         # versions of Gecko don't misinterpret our strings;
 359                         # this is a common problem with Farsi text.
 360                         "\xe2\x80\x8c" => "\\u200c", // ZERO WIDTH NON-JOINER
 361                         "\xe2\x80\x8d" => "\\u200d", // ZERO WIDTH JOINER
 362                 );
 363                 return strtr( $string, $pairs );
 364         }
 365
 366         /**
 367          * Encode a variable of unknown type to JavaScript.
 368          * Arrays are converted to JS arrays, objects are converted to JS associative
 369          * arrays (objects). So cast your PHP associative arrays to objects before
 370          * passing them to here.
 371          */
 372         public static function encodeJsVar( $value ) {
 373                 if ( is_bool( $value ) ) {
 374                         $s = $value ? 'true' : 'false';
 375                 } elseif ( is_null( $value ) ) {
 376                         $s = 'null';
 377                 } elseif ( is_int( $value ) ) {
 378                         $s = $value;
 379                 } elseif ( is_array( $value ) ) {
 380                         $s = '[';
 381                         foreach ( $value as $elt ) {
 382                                 if ( $s != '[' ) {
 383                                         $s .= ', ';
 384                                 }
 385                                 $s .= self::encodeJsVar( $elt );
 386                         }
 387                         $s .= ']';
 388                 } elseif ( is_object( $value ) ) {
 389                         $s = '{';
 390                         foreach ( (array)$value as $name => $elt ) {
 391                                 if ( $s != '{' ) {
 392                                         $s .= ', ';
 393                                 }
 394                                 $s .= '"' . self::escapeJsString( $name ) . '": ' .
 395                                         self::encodeJsVar( $elt );
 396                         }
 397                         $s .= '}';
 398                 } else {
 399                         $s = '"' . self::escapeJsString( $value ) . '"';
 400                 }
 401                 return $s;
 402         }
 403
 404
 405         /**
 406          * Check if a string is well-formed XML.
 407          * Must include the surrounding tag.
 408          *
 409          * @param $text String: string to test.
 410          * @return bool
 411          *
 412          * @todo Error position reporting return
 413          */
 414         public static function isWellFormed( $text ) {
 415                 $parser = xml_parser_create( "UTF-8" );
 416
 417                 # case folding violates XML standard, turn it off
 418                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 419
 420                 if( !xml_parse( $parser, $text, true ) ) {
 421                         //$err = xml_error_string( xml_get_error_code( $parser ) );
 422                         //$position = xml_get_current_byte_index( $parser );
 423                         //$fragment = $this->extractFragment( $html, $position );
 424                         //$this->mXmlError = "$err at byte $position:\n$fragment";
 425                         xml_parser_free( $parser );
 426                         return false;
 427                 }
 428                 xml_parser_free( $parser );
 429                 return true;
 430         }
 431
 432         /**
 433          * Check if a string is a well-formed XML fragment.
 434          * Wraps fragment in an \<html\> bit and doctype, so it can be a fragment
 435          * and can use HTML named entities.
 436          *
 437          * @param $text String:
 438          * @return bool
 439          */
 440         public static function isWellFormedXmlFragment( $text ) {
 441                 $html =
 442                         Sanitizer::hackDocType() .
 443                         '<html>' .
 444                         $text .
 445                         '</html>';
 446                 return Xml::isWellFormed( $html );
 447         }
 448
 449         /**
 450          * Replace " > and < with their respective HTML entities ( &quot;,
 451          * &gt;, &lt;)
 452          *
 453          * @param $in String: text that might contain HTML tags.
 454          * @return string Escaped string
 455          */
 456         public static function escapeTagsOnly( $in ) {
 457                 return str_replace(
 458                         array( '"', '>', '<' ),
 459                         array( '&quot;', '&gt;', '&lt;' ),
 460                         $in );
 461         }
 462 }
 463