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          * @var NamespaceInfo
  61          */
  62         protected $nsInfo;
  63
  64         /**
  65          * @param Language $language The language object to use for localizing namespace names,
  66          *   capitalization, etc.
  67          * @param GenderCache $genderCache The gender cache for generating gendered namespace names
  68          * @param string[]|string $localInterwikis
  69          * @param InterwikiLookup|null $interwikiLookup
  70          * @param NamespaceInfo|null $nsInfo
  71          */
  72         public function __construct( Language $language, GenderCache $genderCache,
  73                 $localInterwikis = [], InterwikiLookup $interwikiLookup = null,
  74                 NamespaceInfo $nsInfo = null
  75         ) {
  76                 if ( !$interwikiLookup ) {
  77                         wfDeprecated( __METHOD__ . ' with no InterwikiLookup argument', '1.34' );
  78                         $interwikiLookup = MediaWikiServices::getInstance()->getInterwikiLookup();
  79                 }
  80                 if ( !$nsInfo ) {
  81                         wfDeprecated( __METHOD__ . ' with no NamespaceInfo argument', '1.34' );
  82                         $nsInfo = MediaWikiServices::getInstance()->getNamespaceInfo();
  83                 }
  84                 $this->language = $language;
  85                 $this->genderCache = $genderCache;
  86                 $this->localInterwikis = (array)$localInterwikis;
  87                 $this->interwikiLookup = $interwikiLookup;
  88                 $this->nsInfo = $nsInfo;
  89         }
  90
  91         /**
  92          * @see TitleFormatter::getNamespaceName()
  93          *
  94          * @param int $namespace
  95          * @param string $text
  96          *
  97          * @throws InvalidArgumentException If the namespace is invalid
  98          * @return string Namespace name with underscores (not spaces)
  99          */
 100         public function getNamespaceName( $namespace, $text ) {
 101                 if ( $this->language->needsGenderDistinction() &&
 102                         $this->nsInfo->hasGenderDistinction( $namespace )
 103                 ) {
 104                         // NOTE: we are assuming here that the title text is a user name!
 105                         $gender = $this->genderCache->getGenderOf( $text, __METHOD__ );
 106                         $name = $this->language->getGenderNsText( $namespace, $gender );
 107                 } else {
 108                         $name = $this->language->getNsText( $namespace );
 109                 }
 110
 111                 if ( $name === false ) {
 112                         throw new InvalidArgumentException( 'Unknown namespace ID: ' . $namespace );
 113                 }
 114
 115                 return $name;
 116         }
 117
 118         /**
 119          * @see TitleFormatter::formatTitle()
 120          *
 121          * @param int|bool $namespace The namespace ID (or false, if the namespace should be ignored)
 122          * @param string $text The page title. Should be valid. Only minimal normalization is applied.
 123          *        Underscores will be replaced.
 124          * @param string $fragment The fragment name (may be empty).
 125          * @param string $interwiki The interwiki name (may be empty).
 126          *
 127          * @throws InvalidArgumentException If the namespace is invalid
 128          * @return string
 129          */
 130         public function formatTitle( $namespace, $text, $fragment = '', $interwiki = '' ) {
 131                 $out = '';
 132                 if ( $interwiki !== '' ) {
 133                         $out = $interwiki . ':';
 134                 }
 135
 136                 if ( $namespace != 0 ) {
 137                         try {
 138                                 $nsName = $this->getNamespaceName( $namespace, $text );
 139                         } catch ( InvalidArgumentException $e ) {
 140                                 // See T165149. Awkward, but better than erroneously linking to the main namespace.
 141                                 $nsName = $this->language->getNsText( NS_SPECIAL ) . ":Badtitle/NS{$namespace}";
 142                         }
 143
 144                         $out .= $nsName . ':';
 145                 }
 146                 $out .= $text;
 147
 148                 if ( $fragment !== '' ) {
 149                         $out .= '#' . $fragment;
 150                 }
 151
 152                 $out = str_replace( '_', ' ', $out );
 153
 154                 return $out;
 155         }
 156
 157         /**
 158          * Parses the given text and constructs a TitleValue.
 159          *
 160          * @param string $text The text to parse
 161          * @param int $defaultNamespace Namespace to assume per default (usually NS_MAIN)
 162          *
 163          * @throws MalformedTitleException
 164          * @return TitleValue
 165          */
 166         public function parseTitle( $text, $defaultNamespace = NS_MAIN ) {
 167                 // Convert things like &eacute; &#257; or &#x3017; into normalized (T16952) text
 168                 $filteredText = Sanitizer::decodeCharReferencesAndNormalize( $text );
 169
 170                 // NOTE: this is an ugly cludge that allows this class to share the
 171                 // code for parsing with the old Title class. The parser code should
 172                 // be refactored to avoid this.
 173                 $parts = $this->splitTitleString( $filteredText, $defaultNamespace );
 174
 175                 // Fragment-only is okay, but only with no namespace
 176                 if ( $parts['dbkey'] === '' &&
 177                 ( $parts['fragment'] === '' || $parts['namespace'] !== NS_MAIN ) ) {
 178                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 179                 }
 180
 181                 return new TitleValue(
 182                         $parts['namespace'],
 183                         $parts['dbkey'],
 184                         $parts['fragment'],
 185                         $parts['interwiki']
 186                 );
 187         }
 188
 189         /**
 190          * Given a namespace and title, return a TitleValue if valid, or null if invalid.
 191          *
 192          * @param int $namespace
 193          * @param string $text
 194          * @param string $fragment
 195          * @param string $interwiki
 196          *
 197          * @return TitleValue|null
 198          */
 199         public function makeTitleValueSafe( $namespace, $text, $fragment = '', $interwiki = '' ) {
 200                 if ( !$this->nsInfo->exists( $namespace ) ) {
 201                         return null;
 202                 }
 203
 204                 $canonicalNs = $this->nsInfo->getCanonicalName( $namespace );
 205                 $fullText = $canonicalNs == '' ? $text : "$canonicalNs:$text";
 206                 if ( strval( $interwiki ) != '' ) {
 207                         $fullText = "$interwiki:$fullText";
 208                 }
 209                 if ( strval( $fragment ) != '' ) {
 210                         $fullText .= '#' . $fragment;
 211                 }
 212
 213                 try {
 214                         $parts = $this->splitTitleString( $fullText );
 215                 } catch ( MalformedTitleException $e ) {
 216                         return null;
 217                 }
 218
 219                 return new TitleValue(
 220                         $parts['namespace'], $parts['dbkey'], $parts['fragment'], $parts['interwiki'] );
 221         }
 222
 223         /**
 224          * @see TitleFormatter::getText()
 225          *
 226          * @param LinkTarget $title
 227          *
 228          * @return string $title->getText()
 229          */
 230         public function getText( LinkTarget $title ) {
 231                 return $title->getText();
 232         }
 233
 234         /**
 235          * @see TitleFormatter::getText()
 236          *
 237          * @param LinkTarget $title
 238          *
 239          * @return string
 240          */
 241         public function getPrefixedText( LinkTarget $title ) {
 242                 if ( !isset( $title->prefixedText ) ) {
 243                         $title->prefixedText = $this->formatTitle(
 244                                 $title->getNamespace(),
 245                                 $title->getText(),
 246                                 '',
 247                                 $title->getInterwiki()
 248                         );
 249                 }
 250
 251                 return $title->prefixedText;
 252         }
 253
 254         /**
 255          * @since 1.27
 256          * @see TitleFormatter::getPrefixedDBkey()
 257          * @param LinkTarget $target
 258          * @return string
 259          */
 260         public function getPrefixedDBkey( LinkTarget $target ) {
 261                 return strtr( $this->formatTitle(
 262                         $target->getNamespace(),
 263                         $target->getDBkey(),
 264                         '',
 265                         $target->getInterwiki()
 266                 ), ' ', '_' );
 267         }
 268
 269         /**
 270          * @see TitleFormatter::getText()
 271          *
 272          * @param LinkTarget $title
 273          *
 274          * @return string
 275          */
 276         public function getFullText( LinkTarget $title ) {
 277                 return $this->formatTitle(
 278                         $title->getNamespace(),
 279                         $title->getText(),
 280                         $title->getFragment(),
 281                         $title->getInterwiki()
 282                 );
 283         }
 284
 285         /**
 286          * Normalizes and splits a title string.
 287          *
 288          * This function removes illegal characters, splits off the interwiki and
 289          * namespace prefixes, sets the other forms, and canonicalizes
 290          * everything.
 291          *
 292          * @todo this method is only exposed as a temporary measure to ease refactoring.
 293          * It was copied with minimal changes from Title::secureAndSplit().
 294          *
 295          * @todo This method should be split up and an appropriate interface
 296          * defined for use by the Title class.
 297          *
 298          * @param string $text
 299          * @param int $defaultNamespace
 300          *
 301          * @throws MalformedTitleException If $text is not a valid title string.
 302          * @return array A map with the fields 'interwiki', 'fragment', 'namespace',
 303          *         'user_case_dbkey', and 'dbkey'.
 304          */
 305         public function splitTitleString( $text, $defaultNamespace = NS_MAIN ) {
 306                 $dbkey = str_replace( ' ', '_', $text );
 307
 308                 # Initialisation
 309                 $parts = [
 310                         'interwiki' => '',
 311                         'local_interwiki' => false,
 312                         'fragment' => '',
 313                         'namespace' => $defaultNamespace,
 314                         'dbkey' => $dbkey,
 315                         'user_case_dbkey' => $dbkey,
 316                 ];
 317
 318                 # Strip Unicode bidi override characters.
 319                 # Sometimes they slip into cut-n-pasted page titles, where the
 320                 # override chars get included in list displays.
 321                 $dbkey = preg_replace( '/[\x{200E}\x{200F}\x{202A}-\x{202E}]+/u', '', $dbkey );
 322
 323                 # Clean up whitespace
 324                 # Note: use of the /u option on preg_replace here will cause
 325                 # input with invalid UTF-8 sequences to be nullified out in PHP 5.2.x,
 326                 # conveniently disabling them.
 327                 $dbkey = preg_replace(
 328                         '/[ _\xA0\x{1680}\x{180E}\x{2000}-\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}]+/u',
 329                         '_',
 330                         $dbkey
 331                 );
 332                 $dbkey = trim( $dbkey, '_' );
 333
 334                 if ( strpos( $dbkey, UtfNormal\Constants::UTF8_REPLACEMENT ) !== false ) {
 335                         # Contained illegal UTF-8 sequences or forbidden Unicode chars.
 336                         throw new MalformedTitleException( 'title-invalid-utf8', $text );
 337                 }
 338
 339                 $parts['dbkey'] = $dbkey;
 340
 341                 # Initial colon indicates main namespace rather than specified default
 342                 # but should not create invalid {ns,title} pairs such as {0,Project:Foo}
 343                 if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 344                         $parts['namespace'] = NS_MAIN;
 345                         $dbkey = substr( $dbkey, 1 ); # remove the colon but continue processing
 346                         $dbkey = trim( $dbkey, '_' ); # remove any subsequent whitespace
 347                 }
 348
 349                 if ( $dbkey == '' ) {
 350                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 351                 }
 352
 353                 # Namespace or interwiki prefix
 354                 $prefixRegexp = "/^(.+?)_*:_*(.*)$/S";
 355                 do {
 356                         $m = [];
 357                         if ( preg_match( $prefixRegexp, $dbkey, $m ) ) {
 358                                 $p = $m[1];
 359                                 $ns = $this->language->getNsIndex( $p );
 360                                 if ( $ns !== false ) {
 361                                         # Ordinary namespace
 362                                         $dbkey = $m[2];
 363                                         $parts['namespace'] = $ns;
 364                                         # For Talk:X pages, check if X has a "namespace" prefix
 365                                         if ( $ns == NS_TALK && preg_match( $prefixRegexp, $dbkey, $x ) ) {
 366                                                 if ( $this->language->getNsIndex( $x[1] ) ) {
 367                                                         # Disallow Talk:File:x type titles...
 368                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 369                                                 } elseif ( $this->interwikiLookup->isValidInterwiki( $x[1] ) ) {
 370                                                         # Disallow Talk:Interwiki:x type titles...
 371                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 372                                                 }
 373                                         }
 374                                 } elseif ( $this->interwikiLookup->isValidInterwiki( $p ) ) {
 375                                         # Interwiki link
 376                                         $dbkey = $m[2];
 377                                         $parts['interwiki'] = $this->language->lc( $p );
 378
 379                                         # Redundant interwiki prefix to the local wiki
 380                                         foreach ( $this->localInterwikis as $localIW ) {
 381                                                 if ( strcasecmp( $parts['interwiki'], $localIW ) == 0 ) {
 382                                                         if ( $dbkey == '' ) {
 383                                                                 # Empty self-links should point to the Main Page, to ensure
 384                                                                 # compatibility with cross-wiki transclusions and the like.
 385                                                                 $mainPage = Title::newMainPage();
 386                                                                 return [
 387                                                                         'interwiki' => $mainPage->getInterwiki(),
 388                                                                         'local_interwiki' => true,
 389                                                                         'fragment' => $mainPage->getFragment(),
 390                                                                         'namespace' => $mainPage->getNamespace(),
 391                                                                         'dbkey' => $mainPage->getDBkey(),
 392                                                                         'user_case_dbkey' => $mainPage->getUserCaseDBKey()
 393                                                                 ];
 394                                                         }
 395                                                         $parts['interwiki'] = '';
 396                                                         # local interwikis should behave like initial-colon links
 397                                                         $parts['local_interwiki'] = true;
 398
 399                                                         # Do another namespace split...
 400                                                         continue 2;
 401                                                 }
 402                                         }
 403
 404                                         # If there's an initial colon after the interwiki, that also
 405                                         # resets the default namespace
 406                                         if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 407                                                 $parts['namespace'] = NS_MAIN;
 408                                                 $dbkey = substr( $dbkey, 1 );
 409                                                 $dbkey = trim( $dbkey, '_' );
 410                                         }
 411                                 }
 412                                 # If there's no recognized interwiki or namespace,
 413                                 # then let the colon expression be part of the title.
 414                         }
 415                         break;
 416                 } while ( true );
 417
 418                 $fragment = strstr( $dbkey, '#' );
 419                 if ( $fragment !== false ) {
 420                         $parts['fragment'] = str_replace( '_', ' ', substr( $fragment, 1 ) );
 421                         $dbkey = substr( $dbkey, 0, strlen( $dbkey ) - strlen( $fragment ) );
 422                         # remove whitespace again: prevents "Foo_bar_#"
 423                         # becoming "Foo_bar_"
 424                         $dbkey = preg_replace( '/_*$/', '', $dbkey );
 425                 }
 426
 427                 # Reject illegal characters.
 428                 $rxTc = self::getTitleInvalidRegex();
 429                 $matches = [];
 430                 if ( preg_match( $rxTc, $dbkey, $matches ) ) {
 431                         throw new MalformedTitleException( 'title-invalid-characters', $text, [ $matches[0] ] );
 432                 }
 433
 434                 # Pages with "/./" or "/../" appearing in the URLs will often be un-
 435                 # reachable due to the way web browsers deal with 'relative' URLs.
 436                 # Also, they conflict with subpage syntax.  Forbid them explicitly.
 437                 if (
 438                         strpos( $dbkey, '.' ) !== false &&
 439                         (
 440                                 $dbkey === '.' || $dbkey === '..' ||
 441                                 strpos( $dbkey, './' ) === 0 ||
 442                                 strpos( $dbkey, '../' ) === 0 ||
 443                                 strpos( $dbkey, '/./' ) !== false ||
 444                                 strpos( $dbkey, '/../' ) !== false ||
 445                                 substr( $dbkey, -2 ) == '/.' ||
 446                                 substr( $dbkey, -3 ) == '/..'
 447                         )
 448                 ) {
 449                         throw new MalformedTitleException( 'title-invalid-relative', $text );
 450                 }
 451
 452                 # Magic tilde sequences? Nu-uh!
 453                 if ( strpos( $dbkey, '~~~' ) !== false ) {
 454                         throw new MalformedTitleException( 'title-invalid-magic-tilde', $text );
 455                 }
 456
 457                 # Limit the size of titles to 255 bytes. This is typically the size of the
 458                 # underlying database field. We make an exception for special pages, which
 459                 # don't need to be stored in the database, and may edge over 255 bytes due
 460                 # to subpage syntax for long titles, e.g. [[Special:Block/Long name]]
 461                 $maxLength = ( $parts['namespace'] != NS_SPECIAL ) ? 255 : 512;
 462                 if ( strlen( $dbkey ) > $maxLength ) {
 463                         throw new MalformedTitleException( 'title-invalid-too-long', $text,
 464                                 [ Message::numParam( $maxLength ) ] );
 465                 }
 466
 467                 # Normally, all wiki links are forced to have an initial capital letter so [[foo]]
 468                 # and [[Foo]] point to the same place.  Don't force it for interwikis, since the
 469                 # other site might be case-sensitive.
 470                 $parts['user_case_dbkey'] = $dbkey;
 471                 if ( $parts['interwiki'] === '' && $this->nsInfo->isCapitalized( $parts['namespace'] ) ) {
 472                         $dbkey = $this->language->ucfirst( $dbkey );
 473                 }
 474
 475                 # Can't make a link to a namespace alone... "empty" local links can only be
 476                 # self-links with a fragment identifier.
 477                 if ( $dbkey == '' && $parts['interwiki'] === '' && $parts['namespace'] != NS_MAIN ) {
 478                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 479                 }
 480
 481                 // Allow IPv6 usernames to start with '::' by canonicalizing IPv6 titles.
 482                 // IP names are not allowed for accounts, and can only be referring to
 483                 // edits from the IP. Given '::' abbreviations and caps/lowercaps,
 484                 // there are numerous ways to present the same IP. Having sp:contribs scan
 485                 // them all is silly and having some show the edits and others not is
 486                 // inconsistent. Same for talk/userpages. Keep them normalized instead.
 487                 if ( $parts['namespace'] == NS_USER || $parts['namespace'] == NS_USER_TALK ) {
 488                         $dbkey = IP::sanitizeIP( $dbkey );
 489                 }
 490
 491                 // Any remaining initial :s are illegal.
 492                 if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 493                         throw new MalformedTitleException( 'title-invalid-leading-colon', $text );
 494                 }
 495
 496                 # Fill fields
 497                 $parts['dbkey'] = $dbkey;
 498
 499                 return $parts;
 500         }
 501
 502         /**
 503          * Returns a simple regex that will match on characters and sequences invalid in titles.
 504          * Note that this doesn't pick up many things that could be wrong with titles, but that
 505          * replacing this regex with something valid will make many titles valid.
 506          * Previously Title::getTitleInvalidRegex()
 507          *
 508          * @return string Regex string
 509          * @since 1.25
 510          */
 511         public static function getTitleInvalidRegex() {
 512                 static $rxTc = false;
 513                 if ( !$rxTc ) {
 514                         # Matching titles will be held as illegal.
 515                         $rxTc = '/' .
 516                                 # Any character not allowed is forbidden...
 517                                 '[^' . Title::legalChars() . ']' .
 518                                 # URL percent encoding sequences interfere with the ability
 519                                 # to round-trip titles -- you can't link to them consistently.
 520                                 '|%[0-9A-Fa-f]{2}' .
 521                                 # XML/HTML character references produce similar issues.
 522                                 '|&[A-Za-z0-9\x80-\xff]+;' .
 523                                 '|&#[0-9]+;' .
 524                                 '|&#x[0-9A-Fa-f]+;' .
 525                                 '/S';
 526                 }
 527
 528                 return $rxTc;
 529         }
 530 }