includes/Title.php

   1 <?php
   2 # See title.doc
   3
   4 /* private static */ $title_interwiki_cache = array();
   5
   6 # Title class
   7 #
   8 # * Represents a title, which may contain an interwiki designation or namespace
   9 # * Can fetch various kinds of data from the database, albeit inefficiently.
  10 #
  11 class Title {
  12         # All member variables should be considered private
  13         # Please use the accessor functions
  14
  15         var $mTextform;           # Text form (spaces not underscores) of the main part
  16         var $mUrlform;            # URL-encoded form of the main part
  17         var $mDbkeyform;          # Main part with underscores
  18         var $mNamespace;          # Namespace index, i.e. one of the NS_xxxx constants
  19         var $mInterwiki;          # Interwiki prefix (or null string)
  20         var $mFragment;           # Title fragment (i.e. the bit after the #)
  21         var $mArticleID;          # Article ID, fetched from the link cache on demand
  22         var $mRestrictions;       # Array of groups allowed to edit this article
  23                               # Only null or "sysop" are supported
  24         var $mRestrictionsLoaded; # Boolean for initialisation on demand
  25         var $mPrefixedText;       # Text form including namespace/interwiki, initialised on demand
  26
  27 #----------------------------------------------------------------------------
  28 #   Construction
  29 #----------------------------------------------------------------------------
  30
  31         /* private */ function Title()
  32         {
  33                 $this->mInterwiki = $this->mUrlform =
  34                 $this->mTextform = $this->mDbkeyform = "";
  35                 $this->mArticleID = -1;
  36                 $this->mNamespace = 0;
  37                 $this->mRestrictionsLoaded = false;
  38                 $this->mRestrictions = array();
  39         }
  40
  41         # From a prefixed DB key
  42         /* static */ function newFromDBkey( $key )
  43         {
  44                 $t = new Title();
  45                 $t->mDbkeyform = $key;
  46                 if( $t->secureAndSplit() )
  47                         return $t;
  48                 else
  49                         return NULL;
  50         }
  51
  52         # From text, such as what you would find in a link
  53         /* static */ function newFromText( $text )
  54         {
  55                 static $trans;
  56                 $fname = "Title::newFromText";
  57                 wfProfileIn( $fname );
  58
  59                 # Note - mixing latin1 named entities and unicode numbered
  60                 # ones will result in a bad link.
  61                 if( !isset( $trans ) ) {
  62                         global $wgInputEncoding;
  63                         $trans = array_flip( get_html_translation_table( HTML_ENTITIES ) );
  64                         if( strcasecmp( "utf-8", $wgInputEncoding ) == 0 ) {
  65                             $trans = array_map( "utf8_encode", $trans );
  66                         }
  67                 }
  68
  69                 $text = strtr( $text, $trans );
  70
  71                 $text = wfMungeToUtf8( $text );
  72
  73
  74                 # What was this for? TS 2004-03-03
  75                 # $text = urldecode( $text );
  76
  77                 $t = new Title();
  78                 $t->mDbkeyform = str_replace( " ", "_", $text );
  79                 wfProfileOut( $fname );
  80                 if( $t->secureAndSplit() ) {
  81                         return $t;
  82                 } else {
  83                         return NULL;
  84                 }
  85         }
  86
  87         # From a URL-encoded title
  88         /* static */ function newFromURL( $url )
  89         {
  90                 global $wgLang, $wgServer;
  91                 $t = new Title();
  92                 $s = urldecode( $url ); # This is technically wrong, as anything
  93                                                                 # we've gotten is already decoded by PHP.
  94                                                                 # Kept for backwards compatibility with
  95                                                                 # buggy URLs we had for a while...
  96                 $s = $url;
  97
  98                 # For links that came from outside, check for alternate/legacy
  99                 # character encoding.
 100                 wfDebug( "Refer: {$_SERVER['HTTP_REFERER']}\n" );
 101                 wfDebug( "Servr: $wgServer\n" );
 102                 if( empty( $_SERVER["HTTP_REFERER"] ) ||
 103                         strncmp($wgServer, $_SERVER["HTTP_REFERER"], strlen( $wgServer ) ) )
 104                 {
 105                         $s = $wgLang->checkTitleEncoding( $s );
 106                 }
 107
 108                 $t->mDbkeyform = str_replace( " ", "_", $s );
 109                 if( $t->secureAndSplit() ) {
 110                         return $t;
 111                 } else {
 112                         return NULL;
 113                 }
 114         }
 115
 116         # From a cur_id
 117         # This is inefficiently implemented, the cur row is requested but not
 118         # used for anything else
 119         /* static */ function newFromID( $id )
 120         {
 121                 $fname = "Title::newFromID";
 122                 $row = wfGetArray( "cur", array( "cur_namespace", "cur_title" ),
 123                         array( "cur_id" => $id ), $fname );
 124                 if ( $row !== false ) {
 125                         $title = Title::makeTitle( $row->cur_namespace, $row->cur_title );
 126                 } else {
 127                         $title = NULL;
 128                 }
 129                 return $title;
 130         }
 131
 132         # From a namespace index and a DB key
 133         /* static */ function makeTitle( $ns, $title )
 134         {
 135                 $t = new Title();
 136                 $t->mDbkeyform = Title::makeName( $ns, $title );
 137                 if( $t->secureAndSplit() ) {
 138                         return $t;
 139                 } else {
 140                         return NULL;
 141                 }
 142         }
 143
 144         function newMainPage()
 145         {
 146                 return Title::newFromText( wfMsg( "mainpage" ) );
 147         }
 148
 149 #----------------------------------------------------------------------------
 150 #       Static functions
 151 #----------------------------------------------------------------------------
 152
 153         # Get the prefixed DB key associated with an ID
 154         /* static */ function nameOf( $id )
 155         {
 156                 $sql = "SELECT cur_namespace,cur_title FROM cur WHERE " .
 157                   "cur_id={$id}";
 158                 $res = wfQuery( $sql, DB_READ, "Article::nameOf" );
 159                 if ( 0 == wfNumRows( $res ) ) { return NULL; }
 160
 161                 $s = wfFetchObject( $res );
 162                 $n = Title::makeName( $s->cur_namespace, $s->cur_title );
 163                 return $n;
 164         }
 165
 166         # Get a regex character class describing the legal characters in a link
 167         /* static */ function legalChars()
 168         {
 169                 # Missing characters:
 170                 #  * []|# Needed for link syntax
 171                 #  * % and + are corrupted by Apache when they appear in the path
 172                 #
 173                 # Theoretically 0x80-0x9F of ISO 8859-1 should be disallowed, but
 174                 # this breaks interlanguage links
 175
 176                 $set = " !\"$&'()*,\\-.\\/0-9:;<=>?@A-Z\\\\^_`a-z{}~\\x80-\\xFF";
 177                 return $set;
 178         }
 179
 180         # Returns a stripped-down a title string ready for the search index
 181         # Takes a namespace index and a text-form main part
 182         /* static */ function indexTitle( $ns, $title )
 183         {
 184                 global $wgDBminWordLen, $wgLang;
 185
 186                 $lc = SearchEngine::legalSearchChars() . "&#;";
 187                 $t = $wgLang->stripForSearch( $title );
 188                 $t = preg_replace( "/[^{$lc}]+/", " ", $t );
 189                 $t = strtolower( $t );
 190
 191                 # Handle 's, s'
 192                 $t = preg_replace( "/([{$lc}]+)'s( |$)/", "\\1 \\1's ", $t );
 193                 $t = preg_replace( "/([{$lc}]+)s'( |$)/", "\\1s ", $t );
 194
 195                 $t = preg_replace( "/\\s+/", " ", $t );
 196
 197                 if ( $ns == Namespace::getImage() ) {
 198                         $t = preg_replace( "/ (png|gif|jpg|jpeg|ogg)$/", "", $t );
 199                 }
 200                 return trim( $t );
 201         }
 202
 203         # Make a prefixed DB key from a DB key and a namespace index
 204         /* static */ function makeName( $ns, $title )
 205         {
 206                 global $wgLang;
 207
 208                 $n = $wgLang->getNsText( $ns );
 209                 if ( "" == $n ) { return $title; }
 210                 else { return "{$n}:{$title}"; }
 211         }
 212
 213         # Arguably static
 214         # Returns the URL associated with an interwiki prefix
 215         # The URL contains $1, which is replaced by the title
 216         function getInterwikiLink( $key )
 217         {
 218                 global $wgMemc, $wgDBname, $title_interwiki_cache;
 219                 $k = "$wgDBname:interwiki:$key";
 220
 221                 if( array_key_exists( $k, $title_interwiki_cache ) )
 222                         return $title_interwiki_cache[$k]->iw_url;
 223
 224                 $s = $wgMemc->get( $k );
 225                 if( $s ) {
 226                         $title_interwiki_cache[$k] = $s;
 227                         return $s->iw_url;
 228                 }
 229                 $dkey = wfStrencode( $key );
 230                 $query = "SELECT iw_url FROM interwiki WHERE iw_prefix='$dkey'";
 231                 $res = wfQuery( $query, DB_READ, "Title::getInterwikiLink" );
 232                 if(!$res) return "";
 233
 234                 $s = wfFetchObject( $res );
 235                 if(!$s) {
 236                         $s = (object)false;
 237                         $s->iw_url = "";
 238                 }
 239                 $wgMemc->set( $k, $s );
 240                 $title_interwiki_cache[$k] = $s;
 241                 return $s->iw_url;
 242         }
 243
 244 #----------------------------------------------------------------------------
 245 #       Other stuff
 246 #----------------------------------------------------------------------------
 247
 248         # Simple accessors
 249         # See the definitions at the top of this file
 250
 251         function getText() { return $this->mTextform; }
 252         function getPartialURL() { return $this->mUrlform; }
 253         function getDBkey() { return $this->mDbkeyform; }
 254         function getNamespace() { return $this->mNamespace; }
 255         function setNamespace( $n ) { $this->mNamespace = $n; }
 256         function getInterwiki() { return $this->mInterwiki; }
 257         function getFragment() { return $this->mFragment; }
 258
 259         # Get title for search index
 260         function getIndexTitle()
 261         {
 262                 return Title::indexTitle( $this->mNamespace, $this->mTextform );
 263         }
 264
 265         # Get prefixed title with underscores
 266         function getPrefixedDBkey()
 267         {
 268                 $s = $this->prefix( $this->mDbkeyform );
 269                 $s = str_replace( " ", "_", $s );
 270                 return $s;
 271         }
 272
 273         # Get prefixed title with spaces
 274         # This is the form usually used for display
 275         function getPrefixedText()
 276         {
 277                 if ( empty( $this->mPrefixedText ) ) {
 278                         $s = $this->prefix( $this->mTextform );
 279                         $s = str_replace( "_", " ", $s );
 280                         $this->mPrefixedText = $s;
 281                 }
 282                 return $this->mPrefixedText;
 283         }
 284
 285         # Get a URL-encoded title (not an actual URL) including interwiki
 286         function getPrefixedURL()
 287         {
 288                 $s = $this->prefix( $this->mDbkeyform );
 289                 $s = str_replace( " ", "_", $s );
 290
 291                 $s = wfUrlencode ( $s ) ;
 292
 293                 # Cleaning up URL to make it look nice -- is this safe?
 294                 $s = preg_replace( "/%3[Aa]/", ":", $s );
 295                 $s = preg_replace( "/%2[Ff]/", "/", $s );
 296                 $s = str_replace( "%28", "(", $s );
 297                 $s = str_replace( "%29", ")", $s );
 298
 299                 return $s;
 300         }
 301
 302         # Get a real URL referring to this title, with interwiki link and fragment
 303         function getFullURL( $query = "" )
 304         {
 305                 global $wgLang, $wgArticlePath, $wgServer, $wgScript;
 306
 307                 if ( "" == $this->mInterwiki ) {
 308                         $p = $wgArticlePath;
 309                         return $wgServer . $this->getLocalUrl( $query );
 310                 }
 311
 312                 $p = $this->getInterwikiLink( $this->mInterwiki );
 313                 $n = $wgLang->getNsText( $this->mNamespace );
 314                 if ( "" != $n ) { $n .= ":"; }
 315                 $u = str_replace( "$1", $n . $this->mUrlform, $p );
 316                 if ( "" != $this->mFragment ) {
 317                         $u .= "#" . wfUrlencode( $this->mFragment );
 318                 }
 319                 return $u;
 320         }
 321
 322         # Get a URL with an optional query string, no fragment
 323         # * If $query=="", it will use $wgArticlePath
 324         # * Returns a full for an interwiki link, loses any query string
 325         # * Optionally adds the server and escapes for HTML
 326         # * Setting $query to "-" makes an old-style URL with nothing in the
 327         #   query except a title
 328
 329         function getURL() {
 330                 die( "Call to obsolete obsolete function Title::getURL()" );
 331         }
 332
 333         function getLocalURL( $query = "" )
 334         {
 335                 global $wgLang, $wgArticlePath, $wgScript;
 336
 337                 if ( $this->isExternal() ) {
 338                         return $this->getFullURL();
 339                 }
 340
 341                 $dbkey = wfUrlencode( $this->getPrefixedDBkey() );
 342                 if ( $query == "" ) {
 343                         $url = str_replace( "$1", $dbkey, $wgArticlePath );
 344                 } else {
 345                         if ( $query == "-" ) {
 346                                 $query = "";
 347                         }
 348                         if ( $wgScript != "" ) {
 349                                 $url = "{$wgScript}?title={$dbkey}&{$query}";
 350                         } else {
 351                                 # Top level wiki
 352                                 $url = "/{$dbkey}?{$query}";
 353                         }
 354                 }
 355                 return $url;
 356         }
 357
 358         function escapeLocalURL( $query = "" ) {
 359                 return wfEscapeHTML( $this->getLocalURL( $query ) );
 360         }
 361
 362         function escapeFullURL( $query = "" ) {
 363                 return wfEscapeHTML( $this->getFullURL( $query ) );
 364         }
 365
 366         function getInternalURL( $query = "" ) {
 367                 # Used in various Squid-related code, in case we have a different
 368                 # internal hostname for the server than the exposed one.
 369                 global $wgInternalServer;
 370                 return $wgInternalServer . $this->getLocalURL( $query );
 371         }
 372
 373         # Get the edit URL, or a null string if it is an interwiki link
 374         function getEditURL()
 375         {
 376                 global $wgServer, $wgScript;
 377
 378                 if ( "" != $this->mInterwiki ) { return ""; }
 379                 $s = $this->getLocalURL( "action=edit" );
 380
 381                 return $s;
 382         }
 383
 384         # Get HTML-escaped displayable text
 385         # For the title field in <a> tags
 386         function getEscapedText()
 387         {
 388                 return wfEscapeHTML( $this->getPrefixedText() );
 389         }
 390
 391         # Is the title interwiki?
 392         function isExternal() { return ( "" != $this->mInterwiki ); }
 393
 394         # Does the title correspond to a protected article?
 395         function isProtected()
 396         {
 397                 if ( -1 == $this->mNamespace ) { return true; }
 398                 $a = $this->getRestrictions();
 399                 if ( in_array( "sysop", $a ) ) { return true; }
 400                 return false;
 401         }
 402
 403         # Is the page a log page, i.e. one where the history is messed up by
 404         # LogPage.php? This used to be used for suppressing diff links in recent
 405         # changes, but now that's done by setting a flag in the recentchanges
 406         # table. Hence, this probably is no longer used.
 407         function isLog()
 408         {
 409                 if ( $this->mNamespace != Namespace::getWikipedia() ) {
 410                         return false;
 411                 }
 412                 if ( ( 0 == strcmp( wfMsg( "uploadlogpage" ), $this->mDbkeyform ) ) ||
 413                   ( 0 == strcmp( wfMsg( "dellogpage" ), $this->mDbkeyform ) ) ) {
 414                         return true;
 415                 }
 416                 return false;
 417         }
 418
 419         # Is $wgUser is watching this page?
 420         function userIsWatching()
 421         {
 422                 global $wgUser;
 423
 424                 if ( -1 == $this->mNamespace ) { return false; }
 425                 if ( 0 == $wgUser->getID() ) { return false; }
 426
 427                 return $wgUser->isWatched( $this );
 428         }
 429
 430         # Can $wgUser edit this page?
 431         function userCanEdit()
 432         {
 433                 global $wgUser;
 434
 435                 if ( -1 == $this->mNamespace ) { return false; }
 436                 # if ( 0 == $this->getArticleID() ) { return false; }
 437                 if ( $this->mDbkeyform == "_" ) { return false; }
 438
 439                 $ur = $wgUser->getRights();
 440                 foreach ( $this->getRestrictions() as $r ) {
 441                         if ( "" != $r && ( ! in_array( $r, $ur ) ) ) {
 442                                 return false;
 443                         }
 444                 }
 445                 return true;
 446         }
 447
 448         # Accessor/initialisation for mRestrictions
 449         function getRestrictions()
 450         {
 451                 $id = $this->getArticleID();
 452                 if ( 0 == $id ) { return array(); }
 453
 454                 if ( ! $this->mRestrictionsLoaded ) {
 455                         $res = wfGetSQL( "cur", "cur_restrictions", "cur_id=$id" );
 456                         $this->mRestrictions = explode( ",", trim( $res ) );
 457                         $this->mRestrictionsLoaded = true;
 458                 }
 459                 return $this->mRestrictions;
 460         }
 461
 462         # Is there a version of this page in the deletion archive?
 463         function isDeleted() {
 464                 $ns = $this->getNamespace();
 465                 $t = wfStrencode( $this->getDBkey() );
 466                 $sql = "SELECT COUNT(*) AS n FROM archive WHERE ar_namespace=$ns AND ar_title='$t'";
 467                 if( $res = wfQuery( $sql, DB_READ ) ) {
 468                         $s = wfFetchObject( $res );
 469                         return $s->n;
 470                 }
 471                 return 0;
 472         }
 473
 474         # Get the article ID from the link cache
 475         # Used very heavily, e.g. in Parser::replaceInternalLinks()
 476         function getArticleID()
 477         {
 478                 global $wgLinkCache;
 479
 480                 if ( -1 != $this->mArticleID ) { return $this->mArticleID; }
 481                 $this->mArticleID = $wgLinkCache->addLinkObj( $this );
 482                 return $this->mArticleID;
 483         }
 484
 485         # This clears some fields in this object, and clears any associated keys in the
 486         # "bad links" section of $wgLinkCache. This is called from Article::insertNewArticle()
 487         # to allow loading of the new cur_id. It's also called from Article::doDeleteArticle()
 488         function resetArticleID( $newid )
 489         {
 490                 global $wgLinkCache;
 491                 $wgLinkCache->clearBadLink( $this->getPrefixedDBkey() );
 492
 493                 if ( 0 == $newid ) { $this->mArticleID = -1; }
 494                 else { $this->mArticleID = $newid; }
 495                 $this->mRestrictionsLoaded = false;
 496                 $this->mRestrictions = array();
 497         }
 498
 499         # Updates cur_touched
 500         # Called from LinksUpdate.php
 501         function invalidateCache() {
 502                 $now = wfTimestampNow();
 503                 $ns = $this->getNamespace();
 504                 $ti = wfStrencode( $this->getDBkey() );
 505                 $sql = "UPDATE cur SET cur_touched='$now' WHERE cur_namespace=$ns AND cur_title='$ti'";
 506                 return wfQuery( $sql, DB_WRITE, "Title::invalidateCache" );
 507         }
 508
 509         # Prefixes some arbitrary text with the namespace or interwiki prefix of this object
 510         /* private */ function prefix( $name )
 511         {
 512                 global $wgLang;
 513
 514                 $p = "";
 515                 if ( "" != $this->mInterwiki ) {
 516                         $p = $this->mInterwiki . ":";
 517                 }
 518                 if ( 0 != $this->mNamespace ) {
 519                         $p .= $wgLang->getNsText( $this->mNamespace ) . ":";
 520                 }
 521                 return $p . $name;
 522         }
 523
 524         # Secure and split - main initialisation function for this object
 525         #
 526         # Assumes that mDbkeyform has been set, and is urldecoded
 527     # and uses undersocres, but not otherwise munged.  This function
 528     # removes illegal characters, splits off the winterwiki and
 529     # namespace prefixes, sets the other forms, and canonicalizes
 530     # everything.
 531         #
 532         /* private */ function secureAndSplit()
 533         {
 534                 global $wgLang, $wgLocalInterwiki;
 535                 $fname = "Title::secureAndSplit";
 536                 wfProfileIn( $fname );
 537
 538                 static $imgpre = false;
 539                 static $rxTc = false;
 540
 541                 # Initialisation
 542                 if ( $imgpre === false ) {
 543                         $imgpre = ":" . $wgLang->getNsText( Namespace::getImage() ) . ":";
 544                         $rxTc = "/[^" . Title::legalChars() . "]/";
 545                 }
 546
 547
 548                 $this->mInterwiki = $this->mFragment = "";
 549                 $this->mNamespace = 0;
 550
 551                 # Clean up whitespace
 552                 #
 553                 $t = preg_replace( "/[\\s_]+/", "_", $this->mDbkeyform );
 554                 if ( "_" == $t{0} ) {
 555                         $t = substr( $t, 1 );
 556                 }
 557                 $l = strlen( $t );
 558                 if ( $l && ( "_" == $t{$l-1} ) ) {
 559                         $t = substr( $t, 0, $l-1 );
 560                 }
 561                 if ( "" == $t ) {
 562                         wfProfileOut( $fname );
 563                         return false;
 564                 }
 565
 566                 $this->mDbkeyform = $t;
 567                 $done = false;
 568
 569                 # :Image: namespace
 570                 if ( 0 == strncasecmp( $imgpre, $t, strlen( $imgpre ) ) ) {
 571                         $t = substr( $t, 1 );
 572                 }
 573
 574                 # Redundant initial colon
 575                 if ( ":" == $t{0} ) {
 576                         $r = substr( $t, 1 );
 577                 } else {
 578                         # Namespace or interwiki prefix
 579                         if ( preg_match( "/^((?:i|x|[a-z]{2,3})(?:-[a-z0-9]+)?|[A-Za-z0-9_\\x80-\\xff]+):_*(.*)$/", $t, $m ) ) {
 580                                 #$p = strtolower( $m[1] );
 581                                 $p = $m[1];
 582                                 if ( $ns = $wgLang->getNsIndex( strtolower( $p ) )) {
 583                                         # Ordinary namespace
 584                                         $t = $m[2];
 585                                         $this->mNamespace = $ns;
 586                                 } elseif ( $this->getInterwikiLink( $p ) ) {
 587                                         # Interwiki link
 588                                         $t = $m[2];
 589                                         $this->mInterwiki = $p;
 590
 591                                         if ( !preg_match( "/^([A-Za-z0-9_\\x80-\\xff]+):(.*)$/", $t, $m ) ) {
 592                                                 $done = true;
 593                                         } elseif($this->mInterwiki != $wgLocalInterwiki) {
 594                                                 $done = true;
 595                                         }
 596                                 }
 597                         }
 598                         $r = $t;
 599                 }
 600
 601                 # Redundant interwiki prefix to the local wiki
 602                 if ( 0 == strcmp( $this->mInterwiki, $wgLocalInterwiki ) ) {
 603                         $this->mInterwiki = "";
 604                 }
 605                 # We already know that some pages won't be in the database!
 606                 #
 607                 if ( "" != $this->mInterwiki || -1 == $this->mNamespace ) {
 608                         $this->mArticleID = 0;
 609                 }
 610                 $f = strstr( $r, "#" );
 611                 if ( false !== $f ) {
 612                         $this->mFragment = substr( $f, 1 );
 613                         $r = substr( $r, 0, strlen( $r ) - strlen( $f ) );
 614                 }
 615
 616                 # Reject illegal characters.
 617                 #
 618                 if( preg_match( $rxTc, $r ) ) {
 619                         return false;
 620                 }
 621
 622                 # "." and ".." conflict with the directories of those names
 623                 if ( $r === "." || $r === ".." ) {
 624                         return false;
 625                 }
 626
 627                 # Initial capital letter
 628                 if( $this->mInterwiki == "") $t = $wgLang->ucfirst( $r );
 629
 630                 # Fill fields
 631                 $this->mDbkeyform = $t;
 632                 $this->mUrlform = wfUrlencode( $t );
 633
 634                 $this->mTextform = str_replace( "_", " ", $t );
 635
 636                 wfProfileOut( $fname );
 637                 return true;
 638         }
 639
 640         # Get a title object associated with the talk page of this article
 641         function getTalkPage() {
 642                 return Title::makeTitle( Namespace::getTalk( $this->getNamespace() ), $this->getDBkey() );
 643         }
 644
 645         # Get a title object associated with the subject page of this talk page
 646         function getSubjectPage() {
 647                 return Title::makeTitle( Namespace::getSubject( $this->getNamespace() ), $this->getDBkey() );
 648         }
 649 }
 650 ?>