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 !== false ) {
 116                         // Try to get a namespace name, but fallback
 117                         // to empty string if it doesn't exist
 118                         try {
 119                                 $nsName = $this->getNamespaceName( $namespace, $text );
 120                         } catch ( InvalidArgumentException $e ) {
 121                                 $nsName = '';
 122                         }
 123
 124                         if ( $namespace !== 0 ) {
 125                                 $text = $nsName . ':' . $text;
 126                         }
 127                 }
 128
 129                 if ( $fragment !== '' ) {
 130                         $text = $text . '#' . $fragment;
 131                 }
 132
 133                 if ( $interwiki !== '' ) {
 134                         $text = $interwiki . ':' . $text;
 135                 }
 136
 137                 $text = str_replace( '_', ' ', $text );
 138
 139                 return $text;
 140         }
 141
 142         /**
 143          * Parses the given text and constructs a TitleValue. Normalization
 144          * is applied according to the rules appropriate for the form specified by $form.
 145          *
 146          * @param string $text The text to parse
 147          * @param int $defaultNamespace Namespace to assume per default (usually NS_MAIN)
 148          *
 149          * @throws MalformedTitleException
 150          * @return TitleValue
 151          */
 152         public function parseTitle( $text, $defaultNamespace ) {
 153                 // NOTE: this is an ugly cludge that allows this class to share the
 154                 // code for parsing with the old Title class. The parser code should
 155                 // be refactored to avoid this.
 156                 $parts = $this->splitTitleString( $text, $defaultNamespace );
 157
 158                 // Relative fragment links are not supported by TitleValue
 159                 if ( $parts['dbkey'] === '' ) {
 160                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 161                 }
 162
 163                 return new TitleValue(
 164                         $parts['namespace'],
 165                         $parts['dbkey'],
 166                         $parts['fragment'],
 167                         $parts['interwiki']
 168                 );
 169         }
 170
 171         /**
 172          * @see TitleFormatter::getText()
 173          *
 174          * @param LinkTarget $title
 175          *
 176          * @return string $title->getText()
 177          */
 178         public function getText( LinkTarget $title ) {
 179                 return $this->formatTitle( false, $title->getText(), '' );
 180         }
 181
 182         /**
 183          * @see TitleFormatter::getText()
 184          *
 185          * @param LinkTarget $title
 186          *
 187          * @return string
 188          */
 189         public function getPrefixedText( LinkTarget $title ) {
 190                 return $this->formatTitle(
 191                         $title->getNamespace(),
 192                         $title->getText(),
 193                         '',
 194                         $title->getInterwiki()
 195                 );
 196         }
 197
 198         /**
 199          * @since 1.27
 200          * @see TitleFormatter::getPrefixedDBkey()
 201          * @param LinkTarget $target
 202          * @return string
 203          */
 204         public function getPrefixedDBkey( LinkTarget $target ) {
 205                 $key = '';
 206                 if ( $target->isExternal() ) {
 207                         $key .= $target->getInterwiki() . ':';
 208                 }
 209                 // Try to get a namespace name, but fallback
 210                 // to empty string if it doesn't exist
 211                 try {
 212                         $nsName = $this->getNamespaceName(
 213                                 $target->getNamespace(),
 214                                 $target->getText()
 215                         );
 216                 } catch ( InvalidArgumentException $e ) {
 217                         $nsName = '';
 218                 }
 219
 220                 if ( $target->getNamespace() !== 0 ) {
 221                         $key .= $nsName . ':';
 222                 }
 223
 224                 $key .= $target->getText();
 225
 226                 return strtr( $key, ' ', '_' );
 227         }
 228
 229         /**
 230          * @see TitleFormatter::getText()
 231          *
 232          * @param LinkTarget $title
 233          *
 234          * @return string
 235          */
 236         public function getFullText( LinkTarget $title ) {
 237                 return $this->formatTitle(
 238                         $title->getNamespace(),
 239                         $title->getText(),
 240                         $title->getFragment(),
 241                         $title->getInterwiki()
 242                 );
 243         }
 244
 245         /**
 246          * Normalizes and splits a title string.
 247          *
 248          * This function removes illegal characters, splits off the interwiki and
 249          * namespace prefixes, sets the other forms, and canonicalizes
 250          * everything.
 251          *
 252          * @todo this method is only exposed as a temporary measure to ease refactoring.
 253          * It was copied with minimal changes from Title::secureAndSplit().
 254          *
 255          * @todo This method should be split up and an appropriate interface
 256          * defined for use by the Title class.
 257          *
 258          * @param string $text
 259          * @param int $defaultNamespace
 260          *
 261          * @throws MalformedTitleException If $text is not a valid title string.
 262          * @return array A map with the fields 'interwiki', 'fragment', 'namespace',
 263          *         'user_case_dbkey', and 'dbkey'.
 264          */
 265         public function splitTitleString( $text, $defaultNamespace = NS_MAIN ) {
 266                 $dbkey = str_replace( ' ', '_', $text );
 267
 268                 # Initialisation
 269                 $parts = [
 270                         'interwiki' => '',
 271                         'local_interwiki' => false,
 272                         'fragment' => '',
 273                         'namespace' => $defaultNamespace,
 274                         'dbkey' => $dbkey,
 275                         'user_case_dbkey' => $dbkey,
 276                 ];
 277
 278                 # Strip Unicode bidi override characters.
 279                 # Sometimes they slip into cut-n-pasted page titles, where the
 280                 # override chars get included in list displays.
 281                 $dbkey = preg_replace( '/\xE2\x80[\x8E\x8F\xAA-\xAE]/S', '', $dbkey );
 282
 283                 # Clean up whitespace
 284                 # Note: use of the /u option on preg_replace here will cause
 285                 # input with invalid UTF-8 sequences to be nullified out in PHP 5.2.x,
 286                 # conveniently disabling them.
 287                 $dbkey = preg_replace(
 288                         '/[ _\xA0\x{1680}\x{180E}\x{2000}-\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}]+/u',
 289                         '_',
 290                         $dbkey
 291                 );
 292                 $dbkey = trim( $dbkey, '_' );
 293
 294                 if ( strpos( $dbkey, UtfNormal\Constants::UTF8_REPLACEMENT ) !== false ) {
 295                         # Contained illegal UTF-8 sequences or forbidden Unicode chars.
 296                         throw new MalformedTitleException( 'title-invalid-utf8', $text );
 297                 }
 298
 299                 $parts['dbkey'] = $dbkey;
 300
 301                 # Initial colon indicates main namespace rather than specified default
 302                 # but should not create invalid {ns,title} pairs such as {0,Project:Foo}
 303                 if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 304                         $parts['namespace'] = NS_MAIN;
 305                         $dbkey = substr( $dbkey, 1 ); # remove the colon but continue processing
 306                         $dbkey = trim( $dbkey, '_' ); # remove any subsequent whitespace
 307                 }
 308
 309                 if ( $dbkey == '' ) {
 310                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 311                 }
 312
 313                 # Namespace or interwiki prefix
 314                 $prefixRegexp = "/^(.+?)_*:_*(.*)$/S";
 315                 do {
 316                         $m = [];
 317                         if ( preg_match( $prefixRegexp, $dbkey, $m ) ) {
 318                                 $p = $m[1];
 319                                 $ns = $this->language->getNsIndex( $p );
 320                                 if ( $ns !== false ) {
 321                                         # Ordinary namespace
 322                                         $dbkey = $m[2];
 323                                         $parts['namespace'] = $ns;
 324                                         # For Talk:X pages, check if X has a "namespace" prefix
 325                                         if ( $ns == NS_TALK && preg_match( $prefixRegexp, $dbkey, $x ) ) {
 326                                                 if ( $this->language->getNsIndex( $x[1] ) ) {
 327                                                         # Disallow Talk:File:x type titles...
 328                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 329                                                 } elseif ( $this->interwikiLookup->isValidInterwiki( $x[1] ) ) {
 330                                                         # Disallow Talk:Interwiki:x type titles...
 331                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 332                                                 }
 333                                         }
 334                                 } elseif ( $this->interwikiLookup->isValidInterwiki( $p ) ) {
 335                                         # Interwiki link
 336                                         $dbkey = $m[2];
 337                                         $parts['interwiki'] = $this->language->lc( $p );
 338
 339                                         # Redundant interwiki prefix to the local wiki
 340                                         foreach ( $this->localInterwikis as $localIW ) {
 341                                                 if ( 0 == strcasecmp( $parts['interwiki'], $localIW ) ) {
 342                                                         if ( $dbkey == '' ) {
 343                                                                 # Empty self-links should point to the Main Page, to ensure
 344                                                                 # compatibility with cross-wiki transclusions and the like.
 345                                                                 $mainPage = Title::newMainPage();
 346                                                                 return [
 347                                                                         'interwiki' => $mainPage->getInterwiki(),
 348                                                                         'local_interwiki' => true,
 349                                                                         'fragment' => $mainPage->getFragment(),
 350                                                                         'namespace' => $mainPage->getNamespace(),
 351                                                                         'dbkey' => $mainPage->getDBkey(),
 352                                                                         'user_case_dbkey' => $mainPage->getUserCaseDBKey()
 353                                                                 ];
 354                                                         }
 355                                                         $parts['interwiki'] = '';
 356                                                         # local interwikis should behave like initial-colon links
 357                                                         $parts['local_interwiki'] = true;
 358
 359                                                         # Do another namespace split...
 360                                                         continue 2;
 361                                                 }
 362                                         }
 363
 364                                         # If there's an initial colon after the interwiki, that also
 365                                         # resets the default namespace
 366                                         if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 367                                                 $parts['namespace'] = NS_MAIN;
 368                                                 $dbkey = substr( $dbkey, 1 );
 369                                                 $dbkey = trim( $dbkey, '_' );
 370                                         }
 371                                 }
 372                                 # If there's no recognized interwiki or namespace,
 373                                 # then let the colon expression be part of the title.
 374                         }
 375                         break;
 376                 } while ( true );
 377
 378                 $fragment = strstr( $dbkey, '#' );
 379                 if ( false !== $fragment ) {
 380                         $parts['fragment'] = str_replace( '_', ' ', substr( $fragment, 1 ) );
 381                         $dbkey = substr( $dbkey, 0, strlen( $dbkey ) - strlen( $fragment ) );
 382                         # remove whitespace again: prevents "Foo_bar_#"
 383                         # becoming "Foo_bar_"
 384                         $dbkey = preg_replace( '/_*$/', '', $dbkey );
 385                 }
 386
 387                 # Reject illegal characters.
 388                 $rxTc = self::getTitleInvalidRegex();
 389                 $matches = [];
 390                 if ( preg_match( $rxTc, $dbkey, $matches ) ) {
 391                         throw new MalformedTitleException( 'title-invalid-characters', $text, [ $matches[0] ] );
 392                 }
 393
 394                 # Pages with "/./" or "/../" appearing in the URLs will often be un-
 395                 # reachable due to the way web browsers deal with 'relative' URLs.
 396                 # Also, they conflict with subpage syntax.  Forbid them explicitly.
 397                 if (
 398                         strpos( $dbkey, '.' ) !== false &&
 399                         (
 400                                 $dbkey === '.' || $dbkey === '..' ||
 401                                 strpos( $dbkey, './' ) === 0 ||
 402                                 strpos( $dbkey, '../' ) === 0 ||
 403                                 strpos( $dbkey, '/./' ) !== false ||
 404                                 strpos( $dbkey, '/../' ) !== false ||
 405                                 substr( $dbkey, -2 ) == '/.' ||
 406                                 substr( $dbkey, -3 ) == '/..'
 407                         )
 408                 ) {
 409                         throw new MalformedTitleException( 'title-invalid-relative', $text );
 410                 }
 411
 412                 # Magic tilde sequences? Nu-uh!
 413                 if ( strpos( $dbkey, '~~~' ) !== false ) {
 414                         throw new MalformedTitleException( 'title-invalid-magic-tilde', $text );
 415                 }
 416
 417                 # Limit the size of titles to 255 bytes. This is typically the size of the
 418                 # underlying database field. We make an exception for special pages, which
 419                 # don't need to be stored in the database, and may edge over 255 bytes due
 420                 # to subpage syntax for long titles, e.g. [[Special:Block/Long name]]
 421                 $maxLength = ( $parts['namespace'] != NS_SPECIAL ) ? 255 : 512;
 422                 if ( strlen( $dbkey ) > $maxLength ) {
 423                         throw new MalformedTitleException( 'title-invalid-too-long', $text,
 424                                 [ Message::numParam( $maxLength ) ] );
 425                 }
 426
 427                 # Normally, all wiki links are forced to have an initial capital letter so [[foo]]
 428                 # and [[Foo]] point to the same place.  Don't force it for interwikis, since the
 429                 # other site might be case-sensitive.
 430                 $parts['user_case_dbkey'] = $dbkey;
 431                 if ( $parts['interwiki'] === '' ) {
 432                         $dbkey = Title::capitalize( $dbkey, $parts['namespace'] );
 433                 }
 434
 435                 # Can't make a link to a namespace alone... "empty" local links can only be
 436                 # self-links with a fragment identifier.
 437                 if ( $dbkey == '' && $parts['interwiki'] === '' ) {
 438                         if ( $parts['namespace'] != NS_MAIN ) {
 439                                 throw new MalformedTitleException( 'title-invalid-empty', $text );
 440                         }
 441                 }
 442
 443                 // Allow IPv6 usernames to start with '::' by canonicalizing IPv6 titles.
 444                 // IP names are not allowed for accounts, and can only be referring to
 445                 // edits from the IP. Given '::' abbreviations and caps/lowercaps,
 446                 // there are numerous ways to present the same IP. Having sp:contribs scan
 447                 // them all is silly and having some show the edits and others not is
 448                 // inconsistent. Same for talk/userpages. Keep them normalized instead.
 449                 if ( $parts['namespace'] == NS_USER || $parts['namespace'] == NS_USER_TALK ) {
 450                         $dbkey = IP::sanitizeIP( $dbkey );
 451                 }
 452
 453                 // Any remaining initial :s are illegal.
 454                 if ( $dbkey !== '' && ':' == $dbkey[0] ) {
 455                         throw new MalformedTitleException( 'title-invalid-leading-colon', $text );
 456                 }
 457
 458                 # Fill fields
 459                 $parts['dbkey'] = $dbkey;
 460
 461                 return $parts;
 462         }
 463
 464         /**
 465          * Returns a simple regex that will match on characters and sequences invalid in titles.
 466          * Note that this doesn't pick up many things that could be wrong with titles, but that
 467          * replacing this regex with something valid will make many titles valid.
 468          * Previously Title::getTitleInvalidRegex()
 469          *
 470          * @return string Regex string
 471          * @since 1.25
 472          */
 473         public static function getTitleInvalidRegex() {
 474                 static $rxTc = false;
 475                 if ( !$rxTc ) {
 476                         # Matching titles will be held as illegal.
 477                         $rxTc = '/' .
 478                                 # Any character not allowed is forbidden...
 479                                 '[^' . Title::legalChars() . ']' .
 480                                 # URL percent encoding sequences interfere with the ability
 481                                 # to round-trip titles -- you can't link to them consistently.
 482                                 '|%[0-9A-Fa-f]{2}' .
 483                                 # XML/HTML character references produce similar issues.
 484                                 '|&[A-Za-z0-9\x80-\xff]+;' .
 485                                 '|&#[0-9]+;' .
 486                                 '|&#x[0-9A-Fa-f]+;' .
 487                                 '/S';
 488                 }
 489
 490                 return $rxTc;
 491         }
 492 }