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  * @license GPL 2+
  22  * @author Daniel Kinzler
  23  */
  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          * @param Language $language The language object to use for localizing namespace names.
  56          * @param GenderCache $genderCache The gender cache for generating gendered namespace names
  57          * @param string[]|string $localInterwikis
  58          */
  59         public function __construct( Language $language, GenderCache $genderCache,
  60                 $localInterwikis = []
  61         ) {
  62                 $this->language = $language;
  63                 $this->genderCache = $genderCache;
  64                 $this->localInterwikis = (array)$localInterwikis;
  65         }
  66
  67         /**
  68          * @see TitleFormatter::getNamespaceName()
  69          *
  70          * @param int $namespace
  71          * @param string $text
  72          *
  73          * @throws InvalidArgumentException If the namespace is invalid
  74          * @return string
  75          */
  76         public function getNamespaceName( $namespace, $text ) {
  77                 if ( $this->language->needsGenderDistinction() &&
  78                         MWNamespace::hasGenderDistinction( $namespace )
  79                 ) {
  80
  81                         // NOTE: we are assuming here that the title text is a user name!
  82                         $gender = $this->genderCache->getGenderOf( $text, __METHOD__ );
  83                         $name = $this->language->getGenderNsText( $namespace, $gender );
  84                 } else {
  85                         $name = $this->language->getNsText( $namespace );
  86                 }
  87
  88                 if ( $name === false ) {
  89                         throw new InvalidArgumentException( 'Unknown namespace ID: ' . $namespace );
  90                 }
  91
  92                 return $name;
  93         }
  94
  95         /**
  96          * @see TitleFormatter::formatTitle()
  97          *
  98          * @param int|bool $namespace The namespace ID (or false, if the namespace should be ignored)
  99          * @param string $text The page title. Should be valid. Only minimal normalization is applied.
 100          *        Underscores will be replaced.
 101          * @param string $fragment The fragment name (may be empty).
 102          * @param string $interwiki The interwiki name (may be empty).
 103          *
 104          * @throws InvalidArgumentException If the namespace is invalid
 105          * @return string
 106          */
 107         public function formatTitle( $namespace, $text, $fragment = '', $interwiki = '' ) {
 108                 if ( $namespace !== false ) {
 109                         // Try to get a namespace name, but fallback
 110                         // to empty string if it doesn't exist
 111                         try {
 112                                 $nsName = $this->getNamespaceName( $namespace, $text );
 113                         } catch ( InvalidArgumentException $e ) {
 114                                 $nsName = '';
 115                         }
 116
 117                         if ( $namespace !== 0 ) {
 118                                 $text = $nsName . ':' . $text;
 119                         }
 120                 }
 121
 122                 if ( $fragment !== '' ) {
 123                         $text = $text . '#' . $fragment;
 124                 }
 125
 126                 if ( $interwiki !== '' ) {
 127                         $text = $interwiki . ':' . $text;
 128                 }
 129
 130                 $text = str_replace( '_', ' ', $text );
 131
 132                 return $text;
 133         }
 134
 135         /**
 136          * Parses the given text and constructs a TitleValue. Normalization
 137          * is applied according to the rules appropriate for the form specified by $form.
 138          *
 139          * @param string $text The text to parse
 140          * @param int $defaultNamespace Namespace to assume per default (usually NS_MAIN)
 141          *
 142          * @throws MalformedTitleException
 143          * @return TitleValue
 144          */
 145         public function parseTitle( $text, $defaultNamespace ) {
 146                 // NOTE: this is an ugly cludge that allows this class to share the
 147                 // code for parsing with the old Title class. The parser code should
 148                 // be refactored to avoid this.
 149                 $parts = $this->splitTitleString( $text, $defaultNamespace );
 150
 151                 // Relative fragment links are not supported by TitleValue
 152                 if ( $parts['dbkey'] === '' ) {
 153                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 154                 }
 155
 156                 return new TitleValue(
 157                         $parts['namespace'],
 158                         $parts['dbkey'],
 159                         $parts['fragment'],
 160                         $parts['interwiki']
 161                 );
 162         }
 163
 164         /**
 165          * @see TitleFormatter::getText()
 166          *
 167          * @param LinkTarget $title
 168          *
 169          * @return string $title->getText()
 170          */
 171         public function getText( LinkTarget $title ) {
 172                 return $this->formatTitle( false, $title->getText(), '' );
 173         }
 174
 175         /**
 176          * @see TitleFormatter::getText()
 177          *
 178          * @param LinkTarget $title
 179          *
 180          * @return string
 181          */
 182         public function getPrefixedText( LinkTarget $title ) {
 183                 return $this->formatTitle(
 184                         $title->getNamespace(),
 185                         $title->getText(),
 186                         '',
 187                         $title->getInterwiki()
 188                 );
 189         }
 190
 191         /**
 192          * @since 1.27
 193          * @see TitleFormatter::getPrefixedDBkey()
 194          * @param LinkTarget $target
 195          * @return string
 196          */
 197         public function getPrefixedDBkey( LinkTarget $target ) {
 198                 $key = '';
 199                 if ( $target->isExternal() ) {
 200                         $key .= $target->getInterwiki() . ':';
 201                 }
 202                 // Try to get a namespace name, but fallback
 203                 // to empty string if it doesn't exist
 204                 try {
 205                         $nsName = $this->getNamespaceName(
 206                                 $target->getNamespace(),
 207                                 $target->getText()
 208                         );
 209                 } catch ( InvalidArgumentException $e ) {
 210                         $nsName = '';
 211                 }
 212
 213                 if ( $target->getNamespace() !== 0 ) {
 214                         $key .= $nsName . ':';
 215                 }
 216
 217                 $key .= $target->getText();
 218
 219                 return strtr( $key, ' ', '_' );
 220         }
 221
 222         /**
 223          * @see TitleFormatter::getText()
 224          *
 225          * @param LinkTarget $title
 226          *
 227          * @return string
 228          */
 229         public function getFullText( LinkTarget $title ) {
 230                 return $this->formatTitle(
 231                         $title->getNamespace(),
 232                         $title->getText(),
 233                         $title->getFragment(),
 234                         $title->getInterwiki()
 235                 );
 236         }
 237
 238         /**
 239          * Normalizes and splits a title string.
 240          *
 241          * This function removes illegal characters, splits off the interwiki and
 242          * namespace prefixes, sets the other forms, and canonicalizes
 243          * everything.
 244          *
 245          * @todo this method is only exposed as a temporary measure to ease refactoring.
 246          * It was copied with minimal changes from Title::secureAndSplit().
 247          *
 248          * @todo This method should be split up and an appropriate interface
 249          * defined for use by the Title class.
 250          *
 251          * @param string $text
 252          * @param int $defaultNamespace
 253          *
 254          * @throws MalformedTitleException If $text is not a valid title string.
 255          * @return array A map with the fields 'interwiki', 'fragment', 'namespace',
 256          *         'user_case_dbkey', and 'dbkey'.
 257          */
 258         public function splitTitleString( $text, $defaultNamespace = NS_MAIN ) {
 259                 $dbkey = str_replace( ' ', '_', $text );
 260
 261                 # Initialisation
 262                 $parts = [
 263                         'interwiki' => '',
 264                         'local_interwiki' => false,
 265                         'fragment' => '',
 266                         'namespace' => $defaultNamespace,
 267                         'dbkey' => $dbkey,
 268                         'user_case_dbkey' => $dbkey,
 269                 ];
 270
 271                 # Strip Unicode bidi override characters.
 272                 # Sometimes they slip into cut-n-pasted page titles, where the
 273                 # override chars get included in list displays.
 274                 $dbkey = preg_replace( '/\xE2\x80[\x8E\x8F\xAA-\xAE]/S', '', $dbkey );
 275
 276                 # Clean up whitespace
 277                 # Note: use of the /u option on preg_replace here will cause
 278                 # input with invalid UTF-8 sequences to be nullified out in PHP 5.2.x,
 279                 # conveniently disabling them.
 280                 $dbkey = preg_replace(
 281                         '/[ _\xA0\x{1680}\x{180E}\x{2000}-\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}]+/u',
 282                         '_',
 283                         $dbkey
 284                 );
 285                 $dbkey = trim( $dbkey, '_' );
 286
 287                 if ( strpos( $dbkey, UtfNormal\Constants::UTF8_REPLACEMENT ) !== false ) {
 288                         # Contained illegal UTF-8 sequences or forbidden Unicode chars.
 289                         throw new MalformedTitleException( 'title-invalid-utf8', $text );
 290                 }
 291
 292                 $parts['dbkey'] = $dbkey;
 293
 294                 # Initial colon indicates main namespace rather than specified default
 295                 # but should not create invalid {ns,title} pairs such as {0,Project:Foo}
 296                 if ( $dbkey !== '' && ':' == $dbkey[0] ) {
 297                         $parts['namespace'] = NS_MAIN;
 298                         $dbkey = substr( $dbkey, 1 ); # remove the colon but continue processing
 299                         $dbkey = trim( $dbkey, '_' ); # remove any subsequent whitespace
 300                 }
 301
 302                 if ( $dbkey == '' ) {
 303                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 304                 }
 305
 306                 # Namespace or interwiki prefix
 307                 $prefixRegexp = "/^(.+?)_*:_*(.*)$/S";
 308                 do {
 309                         $m = [];
 310                         if ( preg_match( $prefixRegexp, $dbkey, $m ) ) {
 311                                 $p = $m[1];
 312                                 $ns = $this->language->getNsIndex( $p );
 313                                 $interwikiLookup = MediaWikiServices::getInstance()->getInterwikiLookup();
 314                                 if ( $ns !== false ) {
 315                                         # Ordinary namespace
 316                                         $dbkey = $m[2];
 317                                         $parts['namespace'] = $ns;
 318                                         # For Talk:X pages, check if X has a "namespace" prefix
 319                                         if ( $ns == NS_TALK && preg_match( $prefixRegexp, $dbkey, $x ) ) {
 320                                                 if ( $this->language->getNsIndex( $x[1] ) ) {
 321                                                         # Disallow Talk:File:x type titles...
 322                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 323                                                 } elseif ( $interwikiLookup->isValidInterwiki( $x[1] ) ) {
 324                                                         // TODO: get rid of global state!
 325                                                         # Disallow Talk:Interwiki:x type titles...
 326                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 327                                                 }
 328                                         }
 329                                 } elseif ( $interwikiLookup->isValidInterwiki( $p ) ) {
 330                                         # Interwiki link
 331                                         $dbkey = $m[2];
 332                                         $parts['interwiki'] = $this->language->lc( $p );
 333
 334                                         # Redundant interwiki prefix to the local wiki
 335                                         foreach ( $this->localInterwikis as $localIW ) {
 336                                                 if ( 0 == strcasecmp( $parts['interwiki'], $localIW ) ) {
 337                                                         if ( $dbkey == '' ) {
 338                                                                 # Empty self-links should point to the Main Page, to ensure
 339                                                                 # compatibility with cross-wiki transclusions and the like.
 340                                                                 $mainPage = Title::newMainPage();
 341                                                                 return [
 342                                                                         'interwiki' => $mainPage->getInterwiki(),
 343                                                                         'local_interwiki' => true,
 344                                                                         'fragment' => $mainPage->getFragment(),
 345                                                                         'namespace' => $mainPage->getNamespace(),
 346                                                                         'dbkey' => $mainPage->getDBkey(),
 347                                                                         'user_case_dbkey' => $mainPage->getUserCaseDBKey()
 348                                                                 ];
 349                                                         }
 350                                                         $parts['interwiki'] = '';
 351                                                         # local interwikis should behave like initial-colon links
 352                                                         $parts['local_interwiki'] = true;
 353
 354                                                         # Do another namespace split...
 355                                                         continue 2;
 356                                                 }
 357                                         }
 358
 359                                         # If there's an initial colon after the interwiki, that also
 360                                         # resets the default namespace
 361                                         if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 362                                                 $parts['namespace'] = NS_MAIN;
 363                                                 $dbkey = substr( $dbkey, 1 );
 364                                         }
 365                                 }
 366                                 # If there's no recognized interwiki or namespace,
 367                                 # then let the colon expression be part of the title.
 368                         }
 369                         break;
 370                 } while ( true );
 371
 372                 $fragment = strstr( $dbkey, '#' );
 373                 if ( false !== $fragment ) {
 374                         $parts['fragment'] = str_replace( '_', ' ', substr( $fragment, 1 ) );
 375                         $dbkey = substr( $dbkey, 0, strlen( $dbkey ) - strlen( $fragment ) );
 376                         # remove whitespace again: prevents "Foo_bar_#"
 377                         # becoming "Foo_bar_"
 378                         $dbkey = preg_replace( '/_*$/', '', $dbkey );
 379                 }
 380
 381                 # Reject illegal characters.
 382                 $rxTc = self::getTitleInvalidRegex();
 383                 $matches = [];
 384                 if ( preg_match( $rxTc, $dbkey, $matches ) ) {
 385                         throw new MalformedTitleException( 'title-invalid-characters', $text, [ $matches[0] ] );
 386                 }
 387
 388                 # Pages with "/./" or "/../" appearing in the URLs will often be un-
 389                 # reachable due to the way web browsers deal with 'relative' URLs.
 390                 # Also, they conflict with subpage syntax.  Forbid them explicitly.
 391                 if (
 392                         strpos( $dbkey, '.' ) !== false &&
 393                         (
 394                                 $dbkey === '.' || $dbkey === '..' ||
 395                                 strpos( $dbkey, './' ) === 0 ||
 396                                 strpos( $dbkey, '../' ) === 0 ||
 397                                 strpos( $dbkey, '/./' ) !== false ||
 398                                 strpos( $dbkey, '/../' ) !== false ||
 399                                 substr( $dbkey, -2 ) == '/.' ||
 400                                 substr( $dbkey, -3 ) == '/..'
 401                         )
 402                 ) {
 403                         throw new MalformedTitleException( 'title-invalid-relative', $text );
 404                 }
 405
 406                 # Magic tilde sequences? Nu-uh!
 407                 if ( strpos( $dbkey, '~~~' ) !== false ) {
 408                         throw new MalformedTitleException( 'title-invalid-magic-tilde', $text );
 409                 }
 410
 411                 # Limit the size of titles to 255 bytes. This is typically the size of the
 412                 # underlying database field. We make an exception for special pages, which
 413                 # don't need to be stored in the database, and may edge over 255 bytes due
 414                 # to subpage syntax for long titles, e.g. [[Special:Block/Long name]]
 415                 $maxLength = ( $parts['namespace'] != NS_SPECIAL ) ? 255 : 512;
 416                 if ( strlen( $dbkey ) > $maxLength ) {
 417                         throw new MalformedTitleException( 'title-invalid-too-long', $text,
 418                                 [ Message::numParam( $maxLength ) ] );
 419                 }
 420
 421                 # Normally, all wiki links are forced to have an initial capital letter so [[foo]]
 422                 # and [[Foo]] point to the same place.  Don't force it for interwikis, since the
 423                 # other site might be case-sensitive.
 424                 $parts['user_case_dbkey'] = $dbkey;
 425                 if ( $parts['interwiki'] === '' ) {
 426                         $dbkey = Title::capitalize( $dbkey, $parts['namespace'] );
 427                 }
 428
 429                 # Can't make a link to a namespace alone... "empty" local links can only be
 430                 # self-links with a fragment identifier.
 431                 if ( $dbkey == '' && $parts['interwiki'] === '' ) {
 432                         if ( $parts['namespace'] != NS_MAIN ) {
 433                                 throw new MalformedTitleException( 'title-invalid-empty', $text );
 434                         }
 435                 }
 436
 437                 // Allow IPv6 usernames to start with '::' by canonicalizing IPv6 titles.
 438                 // IP names are not allowed for accounts, and can only be referring to
 439                 // edits from the IP. Given '::' abbreviations and caps/lowercaps,
 440                 // there are numerous ways to present the same IP. Having sp:contribs scan
 441                 // them all is silly and having some show the edits and others not is
 442                 // inconsistent. Same for talk/userpages. Keep them normalized instead.
 443                 if ( $parts['namespace'] == NS_USER || $parts['namespace'] == NS_USER_TALK ) {
 444                         $dbkey = IP::sanitizeIP( $dbkey );
 445                 }
 446
 447                 // Any remaining initial :s are illegal.
 448                 if ( $dbkey !== '' && ':' == $dbkey[0] ) {
 449                         throw new MalformedTitleException( 'title-invalid-leading-colon', $text );
 450                 }
 451
 452                 # Fill fields
 453                 $parts['dbkey'] = $dbkey;
 454
 455                 return $parts;
 456         }
 457
 458         /**
 459          * Returns a simple regex that will match on characters and sequences invalid in titles.
 460          * Note that this doesn't pick up many things that could be wrong with titles, but that
 461          * replacing this regex with something valid will make many titles valid.
 462          * Previously Title::getTitleInvalidRegex()
 463          *
 464          * @return string Regex string
 465          * @since 1.25
 466          */
 467         public static function getTitleInvalidRegex() {
 468                 static $rxTc = false;
 469                 if ( !$rxTc ) {
 470                         # Matching titles will be held as illegal.
 471                         $rxTc = '/' .
 472                                 # Any character not allowed is forbidden...
 473                                 '[^' . Title::legalChars() . ']' .
 474                                 # URL percent encoding sequences interfere with the ability
 475                                 # to round-trip titles -- you can't link to them consistently.
 476                                 '|%[0-9A-Fa-f]{2}' .
 477                                 # XML/HTML character references produce similar issues.
 478                                 '|&[A-Za-z0-9\x80-\xff]+;' .
 479                                 '|&#[0-9]+;' .
 480                                 '|&#x[0-9A-Fa-f]+;' .
 481                                 '/S';
 482                 }
 483
 484                 return $rxTc;
 485         }
 486 }