includes/title/MediaWikiTitleCodec.php

   1 <?php
   2 /**
   3  * A codec for MediaWiki page titles.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @author Daniel Kinzler
  22  */
  23 use MediaWiki\Interwiki\InterwikiLookup;
  24 use MediaWiki\MediaWikiServices;
  25 use MediaWiki\Linker\LinkTarget;
  26
  27 /**
  28  * A codec for MediaWiki page titles.
  29  *
  30  * @note Normalization and validation is applied while parsing, not when formatting.
  31  * It's possible to construct a TitleValue with an invalid title, and use MediaWikiTitleCodec
  32  * to generate an (invalid) title string from it. TitleValues should be constructed only
  33  * via parseTitle() or from a (semi)trusted source, such as the database.
  34  *
  35  * @see https://www.mediawiki.org/wiki/Requests_for_comment/TitleValue
  36  * @since 1.23
  37  */
  38 class MediaWikiTitleCodec implements TitleFormatter, TitleParser {
  39         /**
  40          * @var Language
  41          */
  42         protected $language;
  43
  44         /**
  45          * @var GenderCache
  46          */
  47         protected $genderCache;
  48
  49         /**
  50          * @var string[]
  51          */
  52         protected $localInterwikis;
  53
  54         /**
  55          * @var InterwikiLookup
  56          */
  57         protected $interwikiLookup;
  58
  59         /**
  60          * @param Language $language The language object to use for localizing namespace names.
  61          * @param GenderCache $genderCache The gender cache for generating gendered namespace names
  62          * @param string[]|string $localInterwikis
  63          * @param InterwikiLookup|null $interwikiLookup
  64          */
  65         public function __construct( Language $language, GenderCache $genderCache,
  66                 $localInterwikis = [], $interwikiLookup = null
  67         ) {
  68                 $this->language = $language;
  69                 $this->genderCache = $genderCache;
  70                 $this->localInterwikis = (array)$localInterwikis;
  71                 $this->interwikiLookup = $interwikiLookup ?:
  72                         MediaWikiServices::getInstance()->getInterwikiLookup();
  73         }
  74
  75         /**
  76          * @see TitleFormatter::getNamespaceName()
  77          *
  78          * @param int $namespace
  79          * @param string $text
  80          *
  81          * @throws InvalidArgumentException If the namespace is invalid
  82          * @return string Namespace name with underscores (not spaces)
  83          */
  84         public function getNamespaceName( $namespace, $text ) {
  85                 if ( $this->language->needsGenderDistinction() &&
  86                         MWNamespace::hasGenderDistinction( $namespace )
  87                 ) {
  88                         // NOTE: we are assuming here that the title text is a user name!
  89                         $gender = $this->genderCache->getGenderOf( $text, __METHOD__ );
  90                         $name = $this->language->getGenderNsText( $namespace, $gender );
  91                 } else {
  92                         $name = $this->language->getNsText( $namespace );
  93                 }
  94
  95                 if ( $name === false ) {
  96                         throw new InvalidArgumentException( 'Unknown namespace ID: ' . $namespace );
  97                 }
  98
  99                 return $name;
 100         }
 101
 102         /**
 103          * @see TitleFormatter::formatTitle()
 104          *
 105          * @param int|bool $namespace The namespace ID (or false, if the namespace should be ignored)
 106          * @param string $text The page title. Should be valid. Only minimal normalization is applied.
 107          *        Underscores will be replaced.
 108          * @param string $fragment The fragment name (may be empty).
 109          * @param string $interwiki The interwiki name (may be empty).
 110          *
 111          * @throws InvalidArgumentException If the namespace is invalid
 112          * @return string
 113          */
 114         public function formatTitle( $namespace, $text, $fragment = '', $interwiki = '' ) {
 115                 $out = '';
 116                 if ( $interwiki !== '' ) {
 117                         $out = $interwiki . ':';
 118                 }
 119
 120                 if ( $namespace != 0 ) {
 121                         try {
 122                                 $nsName = $this->getNamespaceName( $namespace, $text );
 123                         } catch ( InvalidArgumentException $e ) {
 124                                 // See T165149. Awkward, but better than erroneously linking to the main namespace.
 125                                 $nsName = $this->language->getNsText( NS_SPECIAL ) . ":Badtitle/NS{$namespace}";
 126                         }
 127
 128                         $out .= $nsName . ':';
 129                 }
 130                 $out .= $text;
 131
 132                 if ( $fragment !== '' ) {
 133                         $out .= '#' . $fragment;
 134                 }
 135
 136                 $out = str_replace( '_', ' ', $out );
 137
 138                 return $out;
 139         }
 140
 141         /**
 142          * Parses the given text and constructs a TitleValue. Normalization
 143          * is applied according to the rules appropriate for the form specified by $form.
 144          *
 145          * @param string $text The text to parse
 146          * @param int $defaultNamespace Namespace to assume per default (usually NS_MAIN)
 147          *
 148          * @throws MalformedTitleException
 149          * @return TitleValue
 150          */
 151         public function parseTitle( $text, $defaultNamespace = NS_MAIN ) {
 152                 // Convert things like &eacute; &#257; or &#x3017; into normalized (T16952) text
 153                 $filteredText = Sanitizer::decodeCharReferencesAndNormalize( $text );
 154
 155                 // NOTE: this is an ugly cludge that allows this class to share the
 156                 // code for parsing with the old Title class. The parser code should
 157                 // be refactored to avoid this.
 158                 $parts = $this->splitTitleString( $filteredText, $defaultNamespace );
 159
 160                 // Fragment-only is okay, but only with no namespace
 161                 if ( $parts['dbkey'] === '' &&
 162                 ( $parts['fragment'] === '' || $parts['namespace'] !== NS_MAIN ) ) {
 163                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 164                 }
 165
 166                 return new TitleValue(
 167                         $parts['namespace'],
 168                         $parts['dbkey'],
 169                         $parts['fragment'],
 170                         $parts['interwiki']
 171                 );
 172         }
 173
 174         /**
 175          * @see TitleFormatter::getText()
 176          *
 177          * @param LinkTarget $title
 178          *
 179          * @return string $title->getText()
 180          */
 181         public function getText( LinkTarget $title ) {
 182                 return $title->getText();
 183         }
 184
 185         /**
 186          * @see TitleFormatter::getText()
 187          *
 188          * @param LinkTarget $title
 189          *
 190          * @return string
 191          */
 192         public function getPrefixedText( LinkTarget $title ) {
 193                 if ( !isset( $title->prefixedText ) ) {
 194                         $title->prefixedText = $this->formatTitle(
 195                                 $title->getNamespace(),
 196                                 $title->getText(),
 197                                 '',
 198                                 $title->getInterwiki()
 199                         );
 200                 }
 201
 202                 return $title->prefixedText;
 203         }
 204
 205         /**
 206          * @since 1.27
 207          * @see TitleFormatter::getPrefixedDBkey()
 208          * @param LinkTarget $target
 209          * @return string
 210          */
 211         public function getPrefixedDBkey( LinkTarget $target ) {
 212                 return strtr( $this->formatTitle(
 213                         $target->getNamespace(),
 214                         $target->getDBkey(),
 215                         '',
 216                         $target->getInterwiki()
 217                 ), ' ', '_' );
 218         }
 219
 220         /**
 221          * @see TitleFormatter::getText()
 222          *
 223          * @param LinkTarget $title
 224          *
 225          * @return string
 226          */
 227         public function getFullText( LinkTarget $title ) {
 228                 return $this->formatTitle(
 229                         $title->getNamespace(),
 230                         $title->getText(),
 231                         $title->getFragment(),
 232                         $title->getInterwiki()
 233                 );
 234         }
 235
 236         /**
 237          * Normalizes and splits a title string.
 238          *
 239          * This function removes illegal characters, splits off the interwiki and
 240          * namespace prefixes, sets the other forms, and canonicalizes
 241          * everything.
 242          *
 243          * @todo this method is only exposed as a temporary measure to ease refactoring.
 244          * It was copied with minimal changes from Title::secureAndSplit().
 245          *
 246          * @todo This method should be split up and an appropriate interface
 247          * defined for use by the Title class.
 248          *
 249          * @param string $text
 250          * @param int $defaultNamespace
 251          *
 252          * @throws MalformedTitleException If $text is not a valid title string.
 253          * @return array A map with the fields 'interwiki', 'fragment', 'namespace',
 254          *         'user_case_dbkey', and 'dbkey'.
 255          */
 256         public function splitTitleString( $text, $defaultNamespace = NS_MAIN ) {
 257                 $dbkey = str_replace( ' ', '_', $text );
 258
 259                 # Initialisation
 260                 $parts = [
 261                         'interwiki' => '',
 262                         'local_interwiki' => false,
 263                         'fragment' => '',
 264                         'namespace' => $defaultNamespace,
 265                         'dbkey' => $dbkey,
 266                         'user_case_dbkey' => $dbkey,
 267                 ];
 268
 269                 # Strip Unicode bidi override characters.
 270                 # Sometimes they slip into cut-n-pasted page titles, where the
 271                 # override chars get included in list displays.
 272                 $dbkey = preg_replace( '/\xE2\x80[\x8E\x8F\xAA-\xAE]/S', '', $dbkey );
 273
 274                 # Clean up whitespace
 275                 # Note: use of the /u option on preg_replace here will cause
 276                 # input with invalid UTF-8 sequences to be nullified out in PHP 5.2.x,
 277                 # conveniently disabling them.
 278                 $dbkey = preg_replace(
 279                         '/[ _\xA0\x{1680}\x{180E}\x{2000}-\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}]+/u',
 280                         '_',
 281                         $dbkey
 282                 );
 283                 $dbkey = trim( $dbkey, '_' );
 284
 285                 if ( strpos( $dbkey, UtfNormal\Constants::UTF8_REPLACEMENT ) !== false ) {
 286                         # Contained illegal UTF-8 sequences or forbidden Unicode chars.
 287                         throw new MalformedTitleException( 'title-invalid-utf8', $text );
 288                 }
 289
 290                 $parts['dbkey'] = $dbkey;
 291
 292                 # Initial colon indicates main namespace rather than specified default
 293                 # but should not create invalid {ns,title} pairs such as {0,Project:Foo}
 294                 if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 295                         $parts['namespace'] = NS_MAIN;
 296                         $dbkey = substr( $dbkey, 1 ); # remove the colon but continue processing
 297                         $dbkey = trim( $dbkey, '_' ); # remove any subsequent whitespace
 298                 }
 299
 300                 if ( $dbkey == '' ) {
 301                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 302                 }
 303
 304                 # Namespace or interwiki prefix
 305                 $prefixRegexp = "/^(.+?)_*:_*(.*)$/S";
 306                 do {
 307                         $m = [];
 308                         if ( preg_match( $prefixRegexp, $dbkey, $m ) ) {
 309                                 $p = $m[1];
 310                                 $ns = $this->language->getNsIndex( $p );
 311                                 if ( $ns !== false ) {
 312                                         # Ordinary namespace
 313                                         $dbkey = $m[2];
 314                                         $parts['namespace'] = $ns;
 315                                         # For Talk:X pages, check if X has a "namespace" prefix
 316                                         if ( $ns == NS_TALK && preg_match( $prefixRegexp, $dbkey, $x ) ) {
 317                                                 if ( $this->language->getNsIndex( $x[1] ) ) {
 318                                                         # Disallow Talk:File:x type titles...
 319                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 320                                                 } elseif ( $this->interwikiLookup->isValidInterwiki( $x[1] ) ) {
 321                                                         # Disallow Talk:Interwiki:x type titles...
 322                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 323                                                 }
 324                                         }
 325                                 } elseif ( $this->interwikiLookup->isValidInterwiki( $p ) ) {
 326                                         # Interwiki link
 327                                         $dbkey = $m[2];
 328                                         $parts['interwiki'] = $this->language->lc( $p );
 329
 330                                         # Redundant interwiki prefix to the local wiki
 331                                         foreach ( $this->localInterwikis as $localIW ) {
 332                                                 if ( strcasecmp( $parts['interwiki'], $localIW ) == 0 ) {
 333                                                         if ( $dbkey == '' ) {
 334                                                                 # Empty self-links should point to the Main Page, to ensure
 335                                                                 # compatibility with cross-wiki transclusions and the like.
 336                                                                 $mainPage = Title::newMainPage();
 337                                                                 return [
 338                                                                         'interwiki' => $mainPage->getInterwiki(),
 339                                                                         'local_interwiki' => true,
 340                                                                         'fragment' => $mainPage->getFragment(),
 341                                                                         'namespace' => $mainPage->getNamespace(),
 342                                                                         'dbkey' => $mainPage->getDBkey(),
 343                                                                         'user_case_dbkey' => $mainPage->getUserCaseDBKey()
 344                                                                 ];
 345                                                         }
 346                                                         $parts['interwiki'] = '';
 347                                                         # local interwikis should behave like initial-colon links
 348                                                         $parts['local_interwiki'] = true;
 349
 350                                                         # Do another namespace split...
 351                                                         continue 2;
 352                                                 }
 353                                         }
 354
 355                                         # If there's an initial colon after the interwiki, that also
 356                                         # resets the default namespace
 357                                         if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 358                                                 $parts['namespace'] = NS_MAIN;
 359                                                 $dbkey = substr( $dbkey, 1 );
 360                                                 $dbkey = trim( $dbkey, '_' );
 361                                         }
 362                                 }
 363                                 # If there's no recognized interwiki or namespace,
 364                                 # then let the colon expression be part of the title.
 365                         }
 366                         break;
 367                 } while ( true );
 368
 369                 $fragment = strstr( $dbkey, '#' );
 370                 if ( $fragment !== false ) {
 371                         $parts['fragment'] = str_replace( '_', ' ', substr( $fragment, 1 ) );
 372                         $dbkey = substr( $dbkey, 0, strlen( $dbkey ) - strlen( $fragment ) );
 373                         # remove whitespace again: prevents "Foo_bar_#"
 374                         # becoming "Foo_bar_"
 375                         $dbkey = preg_replace( '/_*$/', '', $dbkey );
 376                 }
 377
 378                 # Reject illegal characters.
 379                 $rxTc = self::getTitleInvalidRegex();
 380                 $matches = [];
 381                 if ( preg_match( $rxTc, $dbkey, $matches ) ) {
 382                         throw new MalformedTitleException( 'title-invalid-characters', $text, [ $matches[0] ] );
 383                 }
 384
 385                 # Pages with "/./" or "/../" appearing in the URLs will often be un-
 386                 # reachable due to the way web browsers deal with 'relative' URLs.
 387                 # Also, they conflict with subpage syntax.  Forbid them explicitly.
 388                 if (
 389                         strpos( $dbkey, '.' ) !== false &&
 390                         (
 391                                 $dbkey === '.' || $dbkey === '..' ||
 392                                 strpos( $dbkey, './' ) === 0 ||
 393                                 strpos( $dbkey, '../' ) === 0 ||
 394                                 strpos( $dbkey, '/./' ) !== false ||
 395                                 strpos( $dbkey, '/../' ) !== false ||
 396                                 substr( $dbkey, -2 ) == '/.' ||
 397                                 substr( $dbkey, -3 ) == '/..'
 398                         )
 399                 ) {
 400                         throw new MalformedTitleException( 'title-invalid-relative', $text );
 401                 }
 402
 403                 # Magic tilde sequences? Nu-uh!
 404                 if ( strpos( $dbkey, '~~~' ) !== false ) {
 405                         throw new MalformedTitleException( 'title-invalid-magic-tilde', $text );
 406                 }
 407
 408                 # Limit the size of titles to 255 bytes. This is typically the size of the
 409                 # underlying database field. We make an exception for special pages, which
 410                 # don't need to be stored in the database, and may edge over 255 bytes due
 411                 # to subpage syntax for long titles, e.g. [[Special:Block/Long name]]
 412                 $maxLength = ( $parts['namespace'] != NS_SPECIAL ) ? 255 : 512;
 413                 if ( strlen( $dbkey ) > $maxLength ) {
 414                         throw new MalformedTitleException( 'title-invalid-too-long', $text,
 415                                 [ Message::numParam( $maxLength ) ] );
 416                 }
 417
 418                 # Normally, all wiki links are forced to have an initial capital letter so [[foo]]
 419                 # and [[Foo]] point to the same place.  Don't force it for interwikis, since the
 420                 # other site might be case-sensitive.
 421                 $parts['user_case_dbkey'] = $dbkey;
 422                 if ( $parts['interwiki'] === '' ) {
 423                         $dbkey = Title::capitalize( $dbkey, $parts['namespace'] );
 424                 }
 425
 426                 # Can't make a link to a namespace alone... "empty" local links can only be
 427                 # self-links with a fragment identifier.
 428                 if ( $dbkey == '' && $parts['interwiki'] === '' && $parts['namespace'] != NS_MAIN ) {
 429                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 430                 }
 431
 432                 // Allow IPv6 usernames to start with '::' by canonicalizing IPv6 titles.
 433                 // IP names are not allowed for accounts, and can only be referring to
 434                 // edits from the IP. Given '::' abbreviations and caps/lowercaps,
 435                 // there are numerous ways to present the same IP. Having sp:contribs scan
 436                 // them all is silly and having some show the edits and others not is
 437                 // inconsistent. Same for talk/userpages. Keep them normalized instead.
 438                 if ( $parts['namespace'] == NS_USER || $parts['namespace'] == NS_USER_TALK ) {
 439                         $dbkey = IP::sanitizeIP( $dbkey );
 440                 }
 441
 442                 // Any remaining initial :s are illegal.
 443                 if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 444                         throw new MalformedTitleException( 'title-invalid-leading-colon', $text );
 445                 }
 446
 447                 # Fill fields
 448                 $parts['dbkey'] = $dbkey;
 449
 450                 return $parts;
 451         }
 452
 453         /**
 454          * Returns a simple regex that will match on characters and sequences invalid in titles.
 455          * Note that this doesn't pick up many things that could be wrong with titles, but that
 456          * replacing this regex with something valid will make many titles valid.
 457          * Previously Title::getTitleInvalidRegex()
 458          *
 459          * @return string Regex string
 460          * @since 1.25
 461          */
 462         public static function getTitleInvalidRegex() {
 463                 static $rxTc = false;
 464                 if ( !$rxTc ) {
 465                         # Matching titles will be held as illegal.
 466                         $rxTc = '/' .
 467                                 # Any character not allowed is forbidden...
 468                                 '[^' . Title::legalChars() . ']' .
 469                                 # URL percent encoding sequences interfere with the ability
 470                                 # to round-trip titles -- you can't link to them consistently.
 471                                 '|%[0-9A-Fa-f]{2}' .
 472                                 # XML/HTML character references produce similar issues.
 473                                 '|&[A-Za-z0-9\x80-\xff]+;' .
 474                                 '|&#[0-9]+;' .
 475                                 '|&#x[0-9A-Fa-f]+;' .
 476                                 '/S';
 477                 }
 478
 479                 return $rxTc;
 480         }
 481 }