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
  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                 if ( $namespace !== 0 && $namespace !== false ) {
 116                         // Try to get a namespace name, but fallback
 117                         // to empty string if it doesn't exist. And
 118                         // assume that ns 0 is the empty string.
 119                         try {
 120                                 $nsName = $this->getNamespaceName( $namespace, $text );
 121                         } catch ( InvalidArgumentException $e ) {
 122                                 $nsName = '';
 123                         }
 124                         $text = $nsName . ':' . $text;
 125                 }
 126
 127                 if ( $fragment !== '' ) {
 128                         $text = $text . '#' . $fragment;
 129                 }
 130
 131                 if ( $interwiki !== '' ) {
 132                         $text = $interwiki . ':' . $text;
 133                 }
 134
 135                 $text = str_replace( '_', ' ', $text );
 136
 137                 return $text;
 138         }
 139
 140         /**
 141          * Parses the given text and constructs a TitleValue. Normalization
 142          * is applied according to the rules appropriate for the form specified by $form.
 143          *
 144          * @param string $text The text to parse
 145          * @param int $defaultNamespace Namespace to assume per default (usually NS_MAIN)
 146          *
 147          * @throws MalformedTitleException
 148          * @return TitleValue
 149          */
 150         public function parseTitle( $text, $defaultNamespace = NS_MAIN ) {
 151                 // NOTE: this is an ugly cludge that allows this class to share the
 152                 // code for parsing with the old Title class. The parser code should
 153                 // be refactored to avoid this.
 154                 $parts = $this->splitTitleString( $text, $defaultNamespace );
 155
 156                 // Relative fragment links are not supported by TitleValue
 157                 if ( $parts['dbkey'] === '' ) {
 158                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 159                 }
 160
 161                 return new TitleValue(
 162                         $parts['namespace'],
 163                         $parts['dbkey'],
 164                         $parts['fragment'],
 165                         $parts['interwiki']
 166                 );
 167         }
 168
 169         /**
 170          * @see TitleFormatter::getText()
 171          *
 172          * @param LinkTarget $title
 173          *
 174          * @return string $title->getText()
 175          */
 176         public function getText( LinkTarget $title ) {
 177                 return $title->getText();
 178         }
 179
 180         /**
 181          * @see TitleFormatter::getText()
 182          *
 183          * @param LinkTarget $title
 184          *
 185          * @return string
 186          */
 187         public function getPrefixedText( LinkTarget $title ) {
 188                 return $this->formatTitle(
 189                         $title->getNamespace(),
 190                         $title->getText(),
 191                         '',
 192                         $title->getInterwiki()
 193                 );
 194         }
 195
 196         /**
 197          * @since 1.27
 198          * @see TitleFormatter::getPrefixedDBkey()
 199          * @param LinkTarget $target
 200          * @return string
 201          */
 202         public function getPrefixedDBkey( LinkTarget $target ) {
 203                 $key = '';
 204                 if ( $target->isExternal() ) {
 205                         $key .= $target->getInterwiki() . ':';
 206                 }
 207                 // Try to get a namespace name, but fallback
 208                 // to empty string if it doesn't exist
 209                 try {
 210                         $nsName = $this->getNamespaceName(
 211                                 $target->getNamespace(),
 212                                 $target->getText()
 213                         );
 214                 } catch ( InvalidArgumentException $e ) {
 215                         $nsName = '';
 216                 }
 217
 218                 if ( $target->getNamespace() !== 0 ) {
 219                         $key .= $nsName . ':';
 220                 }
 221
 222                 $key .= $target->getText();
 223
 224                 return strtr( $key, ' ', '_' );
 225         }
 226
 227         /**
 228          * @see TitleFormatter::getText()
 229          *
 230          * @param LinkTarget $title
 231          *
 232          * @return string
 233          */
 234         public function getFullText( LinkTarget $title ) {
 235                 return $this->formatTitle(
 236                         $title->getNamespace(),
 237                         $title->getText(),
 238                         $title->getFragment(),
 239                         $title->getInterwiki()
 240                 );
 241         }
 242
 243         /**
 244          * Normalizes and splits a title string.
 245          *
 246          * This function removes illegal characters, splits off the interwiki and
 247          * namespace prefixes, sets the other forms, and canonicalizes
 248          * everything.
 249          *
 250          * @todo this method is only exposed as a temporary measure to ease refactoring.
 251          * It was copied with minimal changes from Title::secureAndSplit().
 252          *
 253          * @todo This method should be split up and an appropriate interface
 254          * defined for use by the Title class.
 255          *
 256          * @param string $text
 257          * @param int $defaultNamespace
 258          *
 259          * @throws MalformedTitleException If $text is not a valid title string.
 260          * @return array A map with the fields 'interwiki', 'fragment', 'namespace',
 261          *         'user_case_dbkey', and 'dbkey'.
 262          */
 263         public function splitTitleString( $text, $defaultNamespace = NS_MAIN ) {
 264                 $dbkey = str_replace( ' ', '_', $text );
 265
 266                 # Initialisation
 267                 $parts = [
 268                         'interwiki' => '',
 269                         'local_interwiki' => false,
 270                         'fragment' => '',
 271                         'namespace' => $defaultNamespace,
 272                         'dbkey' => $dbkey,
 273                         'user_case_dbkey' => $dbkey,
 274                 ];
 275
 276                 # Strip Unicode bidi override characters.
 277                 # Sometimes they slip into cut-n-pasted page titles, where the
 278                 # override chars get included in list displays.
 279                 $dbkey = preg_replace( '/\xE2\x80[\x8E\x8F\xAA-\xAE]/S', '', $dbkey );
 280
 281                 # Clean up whitespace
 282                 # Note: use of the /u option on preg_replace here will cause
 283                 # input with invalid UTF-8 sequences to be nullified out in PHP 5.2.x,
 284                 # conveniently disabling them.
 285                 $dbkey = preg_replace(
 286                         '/[ _\xA0\x{1680}\x{180E}\x{2000}-\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}]+/u',
 287                         '_',
 288                         $dbkey
 289                 );
 290                 $dbkey = trim( $dbkey, '_' );
 291
 292                 if ( strpos( $dbkey, UtfNormal\Constants::UTF8_REPLACEMENT ) !== false ) {
 293                         # Contained illegal UTF-8 sequences or forbidden Unicode chars.
 294                         throw new MalformedTitleException( 'title-invalid-utf8', $text );
 295                 }
 296
 297                 $parts['dbkey'] = $dbkey;
 298
 299                 # Initial colon indicates main namespace rather than specified default
 300                 # but should not create invalid {ns,title} pairs such as {0,Project:Foo}
 301                 if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 302                         $parts['namespace'] = NS_MAIN;
 303                         $dbkey = substr( $dbkey, 1 ); # remove the colon but continue processing
 304                         $dbkey = trim( $dbkey, '_' ); # remove any subsequent whitespace
 305                 }
 306
 307                 if ( $dbkey == '' ) {
 308                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 309                 }
 310
 311                 # Namespace or interwiki prefix
 312                 $prefixRegexp = "/^(.+?)_*:_*(.*)$/S";
 313                 do {
 314                         $m = [];
 315                         if ( preg_match( $prefixRegexp, $dbkey, $m ) ) {
 316                                 $p = $m[1];
 317                                 $ns = $this->language->getNsIndex( $p );
 318                                 if ( $ns !== false ) {
 319                                         # Ordinary namespace
 320                                         $dbkey = $m[2];
 321                                         $parts['namespace'] = $ns;
 322                                         # For Talk:X pages, check if X has a "namespace" prefix
 323                                         if ( $ns == NS_TALK && preg_match( $prefixRegexp, $dbkey, $x ) ) {
 324                                                 if ( $this->language->getNsIndex( $x[1] ) ) {
 325                                                         # Disallow Talk:File:x type titles...
 326                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 327                                                 } elseif ( $this->interwikiLookup->isValidInterwiki( $x[1] ) ) {
 328                                                         # Disallow Talk:Interwiki:x type titles...
 329                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 330                                                 }
 331                                         }
 332                                 } elseif ( $this->interwikiLookup->isValidInterwiki( $p ) ) {
 333                                         # Interwiki link
 334                                         $dbkey = $m[2];
 335                                         $parts['interwiki'] = $this->language->lc( $p );
 336
 337                                         # Redundant interwiki prefix to the local wiki
 338                                         foreach ( $this->localInterwikis as $localIW ) {
 339                                                 if ( 0 == strcasecmp( $parts['interwiki'], $localIW ) ) {
 340                                                         if ( $dbkey == '' ) {
 341                                                                 # Empty self-links should point to the Main Page, to ensure
 342                                                                 # compatibility with cross-wiki transclusions and the like.
 343                                                                 $mainPage = Title::newMainPage();
 344                                                                 return [
 345                                                                         'interwiki' => $mainPage->getInterwiki(),
 346                                                                         'local_interwiki' => true,
 347                                                                         'fragment' => $mainPage->getFragment(),
 348                                                                         'namespace' => $mainPage->getNamespace(),
 349                                                                         'dbkey' => $mainPage->getDBkey(),
 350                                                                         'user_case_dbkey' => $mainPage->getUserCaseDBKey()
 351                                                                 ];
 352                                                         }
 353                                                         $parts['interwiki'] = '';
 354                                                         # local interwikis should behave like initial-colon links
 355                                                         $parts['local_interwiki'] = true;
 356
 357                                                         # Do another namespace split...
 358                                                         continue 2;
 359                                                 }
 360                                         }
 361
 362                                         # If there's an initial colon after the interwiki, that also
 363                                         # resets the default namespace
 364                                         if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 365                                                 $parts['namespace'] = NS_MAIN;
 366                                                 $dbkey = substr( $dbkey, 1 );
 367                                                 $dbkey = trim( $dbkey, '_' );
 368                                         }
 369                                 }
 370                                 # If there's no recognized interwiki or namespace,
 371                                 # then let the colon expression be part of the title.
 372                         }
 373                         break;
 374                 } while ( true );
 375
 376                 $fragment = strstr( $dbkey, '#' );
 377                 if ( false !== $fragment ) {
 378                         $parts['fragment'] = str_replace( '_', ' ', substr( $fragment, 1 ) );
 379                         $dbkey = substr( $dbkey, 0, strlen( $dbkey ) - strlen( $fragment ) );
 380                         # remove whitespace again: prevents "Foo_bar_#"
 381                         # becoming "Foo_bar_"
 382                         $dbkey = preg_replace( '/_*$/', '', $dbkey );
 383                 }
 384
 385                 # Reject illegal characters.
 386                 $rxTc = self::getTitleInvalidRegex();
 387                 $matches = [];
 388                 if ( preg_match( $rxTc, $dbkey, $matches ) ) {
 389                         throw new MalformedTitleException( 'title-invalid-characters', $text, [ $matches[0] ] );
 390                 }
 391
 392                 # Pages with "/./" or "/../" appearing in the URLs will often be un-
 393                 # reachable due to the way web browsers deal with 'relative' URLs.
 394                 # Also, they conflict with subpage syntax.  Forbid them explicitly.
 395                 if (
 396                         strpos( $dbkey, '.' ) !== false &&
 397                         (
 398                                 $dbkey === '.' || $dbkey === '..' ||
 399                                 strpos( $dbkey, './' ) === 0 ||
 400                                 strpos( $dbkey, '../' ) === 0 ||
 401                                 strpos( $dbkey, '/./' ) !== false ||
 402                                 strpos( $dbkey, '/../' ) !== false ||
 403                                 substr( $dbkey, -2 ) == '/.' ||
 404                                 substr( $dbkey, -3 ) == '/..'
 405                         )
 406                 ) {
 407                         throw new MalformedTitleException( 'title-invalid-relative', $text );
 408                 }
 409
 410                 # Magic tilde sequences? Nu-uh!
 411                 if ( strpos( $dbkey, '~~~' ) !== false ) {
 412                         throw new MalformedTitleException( 'title-invalid-magic-tilde', $text );
 413                 }
 414
 415                 # Limit the size of titles to 255 bytes. This is typically the size of the
 416                 # underlying database field. We make an exception for special pages, which
 417                 # don't need to be stored in the database, and may edge over 255 bytes due
 418                 # to subpage syntax for long titles, e.g. [[Special:Block/Long name]]
 419                 $maxLength = ( $parts['namespace'] != NS_SPECIAL ) ? 255 : 512;
 420                 if ( strlen( $dbkey ) > $maxLength ) {
 421                         throw new MalformedTitleException( 'title-invalid-too-long', $text,
 422                                 [ Message::numParam( $maxLength ) ] );
 423                 }
 424
 425                 # Normally, all wiki links are forced to have an initial capital letter so [[foo]]
 426                 # and [[Foo]] point to the same place.  Don't force it for interwikis, since the
 427                 # other site might be case-sensitive.
 428                 $parts['user_case_dbkey'] = $dbkey;
 429                 if ( $parts['interwiki'] === '' ) {
 430                         $dbkey = Title::capitalize( $dbkey, $parts['namespace'] );
 431                 }
 432
 433                 # Can't make a link to a namespace alone... "empty" local links can only be
 434                 # self-links with a fragment identifier.
 435                 if ( $dbkey == '' && $parts['interwiki'] === '' ) {
 436                         if ( $parts['namespace'] != NS_MAIN ) {
 437                                 throw new MalformedTitleException( 'title-invalid-empty', $text );
 438                         }
 439                 }
 440
 441                 // Allow IPv6 usernames to start with '::' by canonicalizing IPv6 titles.
 442                 // IP names are not allowed for accounts, and can only be referring to
 443                 // edits from the IP. Given '::' abbreviations and caps/lowercaps,
 444                 // there are numerous ways to present the same IP. Having sp:contribs scan
 445                 // them all is silly and having some show the edits and others not is
 446                 // inconsistent. Same for talk/userpages. Keep them normalized instead.
 447                 if ( $parts['namespace'] == NS_USER || $parts['namespace'] == NS_USER_TALK ) {
 448                         $dbkey = IP::sanitizeIP( $dbkey );
 449                 }
 450
 451                 // Any remaining initial :s are illegal.
 452                 if ( $dbkey !== '' && ':' == $dbkey[0] ) {
 453                         throw new MalformedTitleException( 'title-invalid-leading-colon', $text );
 454                 }
 455
 456                 # Fill fields
 457                 $parts['dbkey'] = $dbkey;
 458
 459                 return $parts;
 460         }
 461
 462         /**
 463          * Returns a simple regex that will match on characters and sequences invalid in titles.
 464          * Note that this doesn't pick up many things that could be wrong with titles, but that
 465          * replacing this regex with something valid will make many titles valid.
 466          * Previously Title::getTitleInvalidRegex()
 467          *
 468          * @return string Regex string
 469          * @since 1.25
 470          */
 471         public static function getTitleInvalidRegex() {
 472                 static $rxTc = false;
 473                 if ( !$rxTc ) {
 474                         # Matching titles will be held as illegal.
 475                         $rxTc = '/' .
 476                                 # Any character not allowed is forbidden...
 477                                 '[^' . Title::legalChars() . ']' .
 478                                 # URL percent encoding sequences interfere with the ability
 479                                 # to round-trip titles -- you can't link to them consistently.
 480                                 '|%[0-9A-Fa-f]{2}' .
 481                                 # XML/HTML character references produce similar issues.
 482                                 '|&[A-Za-z0-9\x80-\xff]+;' .
 483                                 '|&#[0-9]+;' .
 484                                 '|&#x[0-9A-Fa-f]+;' .
 485                                 '/S';
 486                 }
 487
 488                 return $rxTc;
 489         }
 490 }