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