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\Linker\LinkTarget;
  25
  26 /**
  27  * A codec for %MediaWiki page titles.
  28  *
  29  * @note Normalization and validation is applied while parsing, not when formatting.
  30  * It's possible to construct a TitleValue with an invalid title, and use MediaWikiTitleCodec
  31  * to generate an (invalid) title string from it. TitleValues should be constructed only
  32  * via parseTitle() or from a (semi)trusted source, such as the database.
  33  *
  34  * @see https://www.mediawiki.org/wiki/Requests_for_comment/TitleValue
  35  * @since 1.23
  36  */
  37 class MediaWikiTitleCodec implements TitleFormatter, TitleParser {
  38         /**
  39          * @var Language
  40          */
  41         protected $language;
  42
  43         /**
  44          * @var GenderCache
  45          */
  46         protected $genderCache;
  47
  48         /**
  49          * @var string[]
  50          */
  51         protected $localInterwikis;
  52
  53         /**
  54          * @param Language $language The language object to use for localizing namespace names.
  55          * @param GenderCache $genderCache The gender cache for generating gendered namespace names
  56          * @param string[]|string $localInterwikis
  57          */
  58         public function __construct( Language $language, GenderCache $genderCache,
  59                 $localInterwikis = []
  60         ) {
  61                 $this->language = $language;
  62                 $this->genderCache = $genderCache;
  63                 $this->localInterwikis = (array)$localInterwikis;
  64         }
  65
  66         /**
  67          * @see TitleFormatter::getNamespaceName()
  68          *
  69          * @param int $namespace
  70          * @param string $text
  71          *
  72          * @throws InvalidArgumentException If the namespace is invalid
  73          * @return string
  74          */
  75         public function getNamespaceName( $namespace, $text ) {
  76                 if ( $this->language->needsGenderDistinction() &&
  77                         MWNamespace::hasGenderDistinction( $namespace )
  78                 ) {
  79
  80                         // NOTE: we are assuming here that the title text is a user name!
  81                         $gender = $this->genderCache->getGenderOf( $text, __METHOD__ );
  82                         $name = $this->language->getGenderNsText( $namespace, $gender );
  83                 } else {
  84                         $name = $this->language->getNsText( $namespace );
  85                 }
  86
  87                 if ( $name === false ) {
  88                         throw new InvalidArgumentException( 'Unknown namespace ID: ' . $namespace );
  89                 }
  90
  91                 return $name;
  92         }
  93
  94         /**
  95          * @see TitleFormatter::formatTitle()
  96          *
  97          * @param int|bool $namespace The namespace ID (or false, if the namespace should be ignored)
  98          * @param string $text The page title. Should be valid. Only minimal normalization is applied.
  99          *        Underscores will be replaced.
 100          * @param string $fragment The fragment name (may be empty).
 101          * @param string $interwiki The interwiki name (may be empty).
 102          *
 103          * @throws InvalidArgumentException If the namespace is invalid
 104          * @return string
 105          */
 106         public function formatTitle( $namespace, $text, $fragment = '', $interwiki = '' ) {
 107                 if ( $namespace !== false ) {
 108                         $namespace = $this->getNamespaceName( $namespace, $text );
 109
 110                         if ( $namespace !== '' ) {
 111                                 $text = $namespace . ':' . $text;
 112                         }
 113                 }
 114
 115                 if ( $fragment !== '' ) {
 116                         $text = $text . '#' . $fragment;
 117                 }
 118
 119                 if ( $interwiki !== '' ) {
 120                         $text = $interwiki . ':' . $text;
 121                 }
 122
 123                 $text = str_replace( '_', ' ', $text );
 124
 125                 return $text;
 126         }
 127
 128         /**
 129          * Parses the given text and constructs a TitleValue. Normalization
 130          * is applied according to the rules appropriate for the form specified by $form.
 131          *
 132          * @param string $text The text to parse
 133          * @param int $defaultNamespace Namespace to assume per default (usually NS_MAIN)
 134          *
 135          * @throws MalformedTitleException
 136          * @return TitleValue
 137          */
 138         public function parseTitle( $text, $defaultNamespace ) {
 139                 // NOTE: this is an ugly cludge that allows this class to share the
 140                 // code for parsing with the old Title class. The parser code should
 141                 // be refactored to avoid this.
 142                 $parts = $this->splitTitleString( $text, $defaultNamespace );
 143
 144                 // Relative fragment links are not supported by TitleValue
 145                 if ( $parts['dbkey'] === '' ) {
 146                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 147                 }
 148
 149                 return new TitleValue(
 150                         $parts['namespace'],
 151                         $parts['dbkey'],
 152                         $parts['fragment'],
 153                         $parts['interwiki']
 154                 );
 155         }
 156
 157         /**
 158          * @see TitleFormatter::getText()
 159          *
 160          * @param LinkTarget $title
 161          *
 162          * @return string $title->getText()
 163          */
 164         public function getText( LinkTarget $title ) {
 165                 return $this->formatTitle( false, $title->getText(), '' );
 166         }
 167
 168         /**
 169          * @see TitleFormatter::getText()
 170          *
 171          * @param LinkTarget $title
 172          *
 173          * @return string
 174          */
 175         public function getPrefixedText( LinkTarget $title ) {
 176                 return $this->formatTitle(
 177                         $title->getNamespace(),
 178                         $title->getText(),
 179                         '',
 180                         $title->getInterwiki()
 181                 );
 182         }
 183
 184         /**
 185          * @since 1.27
 186          * @see TitleFormatter::getPrefixedDBkey()
 187          * @param LinkTarget $target
 188          * @return string
 189          */
 190         public function getPrefixedDBkey( LinkTarget $target ) {
 191                 $key = '';
 192                 if ( $target->isExternal() ) {
 193                         $key .= $target->getInterwiki() . ':';
 194                 }
 195                 $nsName = $this->getNamespaceName(
 196                         $target->getNamespace(),
 197                         $target->getText()
 198                 );
 199
 200                 if ( $nsName !== '' ) {
 201                         $key .= $nsName . ':';
 202                 }
 203
 204                 $key .= $target->getText();
 205
 206                 return strtr( $key, ' ', '_' );
 207         }
 208
 209         /**
 210          * @see TitleFormatter::getText()
 211          *
 212          * @param LinkTarget $title
 213          *
 214          * @return string
 215          */
 216         public function getFullText( LinkTarget $title ) {
 217                 return $this->formatTitle(
 218                         $title->getNamespace(),
 219                         $title->getText(),
 220                         $title->getFragment(),
 221                         $title->getInterwiki()
 222                 );
 223         }
 224
 225         /**
 226          * Normalizes and splits a title string.
 227          *
 228          * This function removes illegal characters, splits off the interwiki and
 229          * namespace prefixes, sets the other forms, and canonicalizes
 230          * everything.
 231          *
 232          * @todo this method is only exposed as a temporary measure to ease refactoring.
 233          * It was copied with minimal changes from Title::secureAndSplit().
 234          *
 235          * @todo This method should be split up and an appropriate interface
 236          * defined for use by the Title class.
 237          *
 238          * @param string $text
 239          * @param int $defaultNamespace
 240          *
 241          * @throws MalformedTitleException If $text is not a valid title string.
 242          * @return array A map with the fields 'interwiki', 'fragment', 'namespace',
 243          *         'user_case_dbkey', and 'dbkey'.
 244          */
 245         public function splitTitleString( $text, $defaultNamespace = NS_MAIN ) {
 246                 $dbkey = str_replace( ' ', '_', $text );
 247
 248                 # Initialisation
 249                 $parts = [
 250                         'interwiki' => '',
 251                         'local_interwiki' => false,
 252                         'fragment' => '',
 253                         'namespace' => $defaultNamespace,
 254                         'dbkey' => $dbkey,
 255                         'user_case_dbkey' => $dbkey,
 256                 ];
 257
 258                 # Strip Unicode bidi override characters.
 259                 # Sometimes they slip into cut-n-pasted page titles, where the
 260                 # override chars get included in list displays.
 261                 $dbkey = preg_replace( '/\xE2\x80[\x8E\x8F\xAA-\xAE]/S', '', $dbkey );
 262
 263                 # Clean up whitespace
 264                 # Note: use of the /u option on preg_replace here will cause
 265                 # input with invalid UTF-8 sequences to be nullified out in PHP 5.2.x,
 266                 # conveniently disabling them.
 267                 $dbkey = preg_replace(
 268                         '/[ _\xA0\x{1680}\x{180E}\x{2000}-\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}]+/u',
 269                         '_',
 270                         $dbkey
 271                 );
 272                 $dbkey = trim( $dbkey, '_' );
 273
 274                 if ( strpos( $dbkey, UtfNormal\Constants::UTF8_REPLACEMENT ) !== false ) {
 275                         # Contained illegal UTF-8 sequences or forbidden Unicode chars.
 276                         throw new MalformedTitleException( 'title-invalid-utf8', $text );
 277                 }
 278
 279                 $parts['dbkey'] = $dbkey;
 280
 281                 # Initial colon indicates main namespace rather than specified default
 282                 # but should not create invalid {ns,title} pairs such as {0,Project:Foo}
 283                 if ( $dbkey !== '' && ':' == $dbkey[0] ) {
 284                         $parts['namespace'] = NS_MAIN;
 285                         $dbkey = substr( $dbkey, 1 ); # remove the colon but continue processing
 286                         $dbkey = trim( $dbkey, '_' ); # remove any subsequent whitespace
 287                 }
 288
 289                 if ( $dbkey == '' ) {
 290                         throw new MalformedTitleException( 'title-invalid-empty', $text );
 291                 }
 292
 293                 # Namespace or interwiki prefix
 294                 $prefixRegexp = "/^(.+?)_*:_*(.*)$/S";
 295                 do {
 296                         $m = [];
 297                         if ( preg_match( $prefixRegexp, $dbkey, $m ) ) {
 298                                 $p = $m[1];
 299                                 $ns = $this->language->getNsIndex( $p );
 300                                 if ( $ns !== false ) {
 301                                         # Ordinary namespace
 302                                         $dbkey = $m[2];
 303                                         $parts['namespace'] = $ns;
 304                                         # For Talk:X pages, check if X has a "namespace" prefix
 305                                         if ( $ns == NS_TALK && preg_match( $prefixRegexp, $dbkey, $x ) ) {
 306                                                 if ( $this->language->getNsIndex( $x[1] ) ) {
 307                                                         # Disallow Talk:File:x type titles...
 308                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 309                                                 } elseif ( Interwiki::isValidInterwiki( $x[1] ) ) {
 310                                                         // TODO: get rid of global state!
 311                                                         # Disallow Talk:Interwiki:x type titles...
 312                                                         throw new MalformedTitleException( 'title-invalid-talk-namespace', $text );
 313                                                 }
 314                                         }
 315                                 } elseif ( Interwiki::isValidInterwiki( $p ) ) {
 316                                         # Interwiki link
 317                                         $dbkey = $m[2];
 318                                         $parts['interwiki'] = $this->language->lc( $p );
 319
 320                                         # Redundant interwiki prefix to the local wiki
 321                                         foreach ( $this->localInterwikis as $localIW ) {
 322                                                 if ( 0 == strcasecmp( $parts['interwiki'], $localIW ) ) {
 323                                                         if ( $dbkey == '' ) {
 324                                                                 # Empty self-links should point to the Main Page, to ensure
 325                                                                 # compatibility with cross-wiki transclusions and the like.
 326                                                                 $mainPage = Title::newMainPage();
 327                                                                 return [
 328                                                                         'interwiki' => $mainPage->getInterwiki(),
 329                                                                         'local_interwiki' => true,
 330                                                                         'fragment' => $mainPage->getFragment(),
 331                                                                         'namespace' => $mainPage->getNamespace(),
 332                                                                         'dbkey' => $mainPage->getDBkey(),
 333                                                                         'user_case_dbkey' => $mainPage->getUserCaseDBKey()
 334                                                                 ];
 335                                                         }
 336                                                         $parts['interwiki'] = '';
 337                                                         # local interwikis should behave like initial-colon links
 338                                                         $parts['local_interwiki'] = true;
 339
 340                                                         # Do another namespace split...
 341                                                         continue 2;
 342                                                 }
 343                                         }
 344
 345                                         # If there's an initial colon after the interwiki, that also
 346                                         # resets the default namespace
 347                                         if ( $dbkey !== '' && $dbkey[0] == ':' ) {
 348                                                 $parts['namespace'] = NS_MAIN;
 349                                                 $dbkey = substr( $dbkey, 1 );
 350                                         }
 351                                 }
 352                                 # If there's no recognized interwiki or namespace,
 353                                 # then let the colon expression be part of the title.
 354                         }
 355                         break;
 356                 } while ( true );
 357
 358                 $fragment = strstr( $dbkey, '#' );
 359                 if ( false !== $fragment ) {
 360                         $parts['fragment'] = str_replace( '_', ' ', substr( $fragment, 1 ) );
 361                         $dbkey = substr( $dbkey, 0, strlen( $dbkey ) - strlen( $fragment ) );
 362                         # remove whitespace again: prevents "Foo_bar_#"
 363                         # becoming "Foo_bar_"
 364                         $dbkey = preg_replace( '/_*$/', '', $dbkey );
 365                 }
 366
 367                 # Reject illegal characters.
 368                 $rxTc = self::getTitleInvalidRegex();
 369                 $matches = [];
 370                 if ( preg_match( $rxTc, $dbkey, $matches ) ) {
 371                         throw new MalformedTitleException( 'title-invalid-characters', $text, [ $matches[0] ] );
 372                 }
 373
 374                 # Pages with "/./" or "/../" appearing in the URLs will often be un-
 375                 # reachable due to the way web browsers deal with 'relative' URLs.
 376                 # Also, they conflict with subpage syntax.  Forbid them explicitly.
 377                 if (
 378                         strpos( $dbkey, '.' ) !== false &&
 379                         (
 380                                 $dbkey === '.' || $dbkey === '..' ||
 381                                 strpos( $dbkey, './' ) === 0 ||
 382                                 strpos( $dbkey, '../' ) === 0 ||
 383                                 strpos( $dbkey, '/./' ) !== false ||
 384                                 strpos( $dbkey, '/../' ) !== false ||
 385                                 substr( $dbkey, -2 ) == '/.' ||
 386                                 substr( $dbkey, -3 ) == '/..'
 387                         )
 388                 ) {
 389                         throw new MalformedTitleException( 'title-invalid-relative', $text );
 390                 }
 391
 392                 # Magic tilde sequences? Nu-uh!
 393                 if ( strpos( $dbkey, '~~~' ) !== false ) {
 394                         throw new MalformedTitleException( 'title-invalid-magic-tilde', $text );
 395                 }
 396
 397                 # Limit the size of titles to 255 bytes. This is typically the size of the
 398                 # underlying database field. We make an exception for special pages, which
 399                 # don't need to be stored in the database, and may edge over 255 bytes due
 400                 # to subpage syntax for long titles, e.g. [[Special:Block/Long name]]
 401                 $maxLength = ( $parts['namespace'] != NS_SPECIAL ) ? 255 : 512;
 402                 if ( strlen( $dbkey ) > $maxLength ) {
 403                         throw new MalformedTitleException( 'title-invalid-too-long', $text,
 404                                 [ Message::numParam( $maxLength ) ] );
 405                 }
 406
 407                 # Normally, all wiki links are forced to have an initial capital letter so [[foo]]
 408                 # and [[Foo]] point to the same place.  Don't force it for interwikis, since the
 409                 # other site might be case-sensitive.
 410                 $parts['user_case_dbkey'] = $dbkey;
 411                 if ( $parts['interwiki'] === '' ) {
 412                         $dbkey = Title::capitalize( $dbkey, $parts['namespace'] );
 413                 }
 414
 415                 # Can't make a link to a namespace alone... "empty" local links can only be
 416                 # self-links with a fragment identifier.
 417                 if ( $dbkey == '' && $parts['interwiki'] === '' ) {
 418                         if ( $parts['namespace'] != NS_MAIN ) {
 419                                 throw new MalformedTitleException( 'title-invalid-empty', $text );
 420                         }
 421                 }
 422
 423                 // Allow IPv6 usernames to start with '::' by canonicalizing IPv6 titles.
 424                 // IP names are not allowed for accounts, and can only be referring to
 425                 // edits from the IP. Given '::' abbreviations and caps/lowercaps,
 426                 // there are numerous ways to present the same IP. Having sp:contribs scan
 427                 // them all is silly and having some show the edits and others not is
 428                 // inconsistent. Same for talk/userpages. Keep them normalized instead.
 429                 if ( $parts['namespace'] == NS_USER || $parts['namespace'] == NS_USER_TALK ) {
 430                         $dbkey = IP::sanitizeIP( $dbkey );
 431                 }
 432
 433                 // Any remaining initial :s are illegal.
 434                 if ( $dbkey !== '' && ':' == $dbkey[0] ) {
 435                         throw new MalformedTitleException( 'title-invalid-leading-colon', $text );
 436                 }
 437
 438                 # Fill fields
 439                 $parts['dbkey'] = $dbkey;
 440
 441                 return $parts;
 442         }
 443
 444         /**
 445          * Returns a simple regex that will match on characters and sequences invalid in titles.
 446          * Note that this doesn't pick up many things that could be wrong with titles, but that
 447          * replacing this regex with something valid will make many titles valid.
 448          * Previously Title::getTitleInvalidRegex()
 449          *
 450          * @return string Regex string
 451          * @since 1.25
 452          */
 453         public static function getTitleInvalidRegex() {
 454                 static $rxTc = false;
 455                 if ( !$rxTc ) {
 456                         # Matching titles will be held as illegal.
 457                         $rxTc = '/' .
 458                                 # Any character not allowed is forbidden...
 459                                 '[^' . Title::legalChars() . ']' .
 460                                 # URL percent encoding sequences interfere with the ability
 461                                 # to round-trip titles -- you can't link to them consistently.
 462                                 '|%[0-9A-Fa-f]{2}' .
 463                                 # XML/HTML character references produce similar issues.
 464                                 '|&[A-Za-z0-9\x80-\xff]+;' .
 465                                 '|&#[0-9]+;' .
 466                                 '|&#x[0-9A-Fa-f]+;' .
 467                                 '/S';
 468                 }
 469
 470                 return $rxTc;
 471         }
 472 }