includes/parser/LinkHolderArray.php

   1 <?php
   2 /**
   3  * Holder of replacement pairs for wiki links
   4  *
   5  * @file
   6  */
   7
   8 /**
   9  * @ingroup Parser
  10  */
  11 class LinkHolderArray {
  12         var $internals = array(), $interwikis = array();
  13         var $size = 0;
  14         var $parent;
  15         protected $tempIdOffset;
  16
  17         function __construct( $parent ) {
  18                 $this->parent = $parent;
  19         }
  20
  21         /**
  22          * Reduce memory usage to reduce the impact of circular references
  23          */
  24         function __destruct() {
  25                 foreach ( $this as $name => $value ) {
  26                         unset( $this->$name );
  27                 }
  28         }
  29
  30         /**
  31          * Don't serialize the parent object, it is big, and not needed when it is
  32          * a parameter to mergeForeign(), which is the only application of
  33          * serializing at present.
  34          *
  35          * Compact the titles, only serialize the text form.
  36          */
  37         function __sleep() {
  38                 foreach ( $this->internals as $ns => &$nsLinks ) {
  39                         foreach ( $nsLinks as $key => &$entry ) {
  40                                 unset( $entry['title'] );
  41                         }
  42                 }
  43                 unset( $nsLinks );
  44                 unset( $entry );
  45
  46                 foreach ( $this->interwikis as $key => &$entry ) {
  47                         unset( $entry['title'] );
  48                 }
  49                 unset( $entry );
  50
  51                 return array( 'internals', 'interwikis', 'size' );
  52         }
  53
  54         /**
  55          * Recreate the Title objects
  56          */
  57         function __wakeup() {
  58                 foreach ( $this->internals as $ns => &$nsLinks ) {
  59                         foreach ( $nsLinks as $key => &$entry ) {
  60                                 $entry['title'] = Title::newFromText( $entry['pdbk'] );
  61                         }
  62                 }
  63                 unset( $nsLinks );
  64                 unset( $entry );
  65
  66                 foreach ( $this->interwikis as $key => &$entry ) {
  67                         $entry['title'] = Title::newFromText( $entry['pdbk'] );
  68                 }
  69                 unset( $entry );
  70         }
  71
  72         /**
  73          * Merge another LinkHolderArray into this one
  74          * @param $other LinkHolderArray
  75          */
  76         function merge( $other ) {
  77                 foreach ( $other->internals as $ns => $entries ) {
  78                         $this->size += count( $entries );
  79                         if ( !isset( $this->internals[$ns] ) ) {
  80                                 $this->internals[$ns] = $entries;
  81                         } else {
  82                                 $this->internals[$ns] += $entries;
  83                         }
  84                 }
  85                 $this->interwikis += $other->interwikis;
  86         }
  87
  88         /**
  89          * Merge a LinkHolderArray from another parser instance into this one. The
  90          * keys will not be preserved. Any text which went with the old
  91          * LinkHolderArray and needs to work with the new one should be passed in
  92          * the $texts array. The strings in this array will have their link holders
  93          * converted for use in the destination link holder. The resulting array of
  94          * strings will be returned.
  95          *
  96          * @param $other LinkHolderArray
  97          * @param $text Array of strings
  98          * @return Array
  99          */
 100         function mergeForeign( $other, $texts ) {
 101                 $this->tempIdOffset = $idOffset = $this->parent->nextLinkID();
 102                 $maxId = 0;
 103
 104                 # Renumber internal links
 105                 foreach ( $other->internals as $ns => $nsLinks ) {
 106                         foreach ( $nsLinks as $key => $entry ) {
 107                                 $newKey = $idOffset + $key;
 108                                 $this->internals[$ns][$newKey] = $entry;
 109                                 $maxId = $newKey > $maxId ? $newKey : $maxId;
 110                         }
 111                 }
 112                 $texts = preg_replace_callback( '/(<!--LINK \d+:)(\d+)(-->)/',
 113                         array( $this, 'mergeForeignCallback' ), $texts );
 114
 115                 # Renumber interwiki links
 116                 foreach ( $other->interwikis as $key => $entry ) {
 117                         $newKey = $idOffset + $key;
 118                         $this->interwikis[$newKey] = $entry;
 119                         $maxId = $newKey > $maxId ? $newKey : $maxId;
 120                 }
 121                 $texts = preg_replace_callback( '/(<!--IWLINK )(\d+)(-->)/',
 122                         array( $this, 'mergeForeignCallback' ), $texts );
 123
 124                 # Set the parent link ID to be beyond the highest used ID
 125                 $this->parent->setLinkID( $maxId + 1 );
 126                 $this->tempIdOffset = null;
 127                 return $texts;
 128         }
 129
 130         protected function mergeForeignCallback( $m ) {
 131                 return $m[1] . ( $m[2] + $this->tempIdOffset ) . $m[3];
 132         }
 133
 134         /**
 135          * Get a subset of the current LinkHolderArray which is sufficient to
 136          * interpret the given text.
 137          */
 138         function getSubArray( $text ) {
 139                 $sub = new LinkHolderArray( $this->parent );
 140
 141                 # Internal links
 142                 $pos = 0;
 143                 while ( $pos < strlen( $text ) ) {
 144                         if ( !preg_match( '/<!--LINK (\d+):(\d+)-->/',
 145                                 $text, $m, PREG_OFFSET_CAPTURE, $pos ) )
 146                         {
 147                                 break;
 148                         }
 149                         $ns = $m[1][0];
 150                         $key = $m[2][0];
 151                         $sub->internals[$ns][$key] = $this->internals[$ns][$key];
 152                         $pos = $m[0][1] + strlen( $m[0][0] );
 153                 }
 154
 155                 # Interwiki links
 156                 $pos = 0;
 157                 while ( $pos < strlen( $text ) ) {
 158                         if ( !preg_match( '/<!--IWLINK (\d+)-->/', $text, $m, PREG_OFFSET_CAPTURE, $pos ) ) {
 159                                 break;
 160                         }
 161                         $key = $m[1][0];
 162                         $sub->interwikis[$key] = $this->interwikis[$key];
 163                         $pos = $m[0][1] + strlen( $m[0][0] );
 164                 }
 165                 return $sub;
 166         }
 167
 168         /**
 169          * Returns true if the memory requirements of this object are getting large
 170          */
 171         function isBig() {
 172                 global $wgLinkHolderBatchSize;
 173                 return $this->size > $wgLinkHolderBatchSize;
 174         }
 175
 176         /**
 177          * Clear all stored link holders.
 178          * Make sure you don't have any text left using these link holders, before you call this
 179          */
 180         function clear() {
 181                 $this->internals = array();
 182                 $this->interwikis = array();
 183                 $this->size = 0;
 184         }
 185
 186         /**
 187          * Make a link placeholder. The text returned can be later resolved to a real link with
 188          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
 189          * parsing of interwiki links, and secondly to allow all existence checks and
 190          * article length checks (for stub links) to be bundled into a single query.
 191          *
 192          * @param $nt Title
 193          */
 194         function makeHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = ''  ) {
 195                 wfProfileIn( __METHOD__ );
 196                 if ( ! is_object($nt) ) {
 197                         # Fail gracefully
 198                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
 199                 } else {
 200                         # Separate the link trail from the rest of the link
 201                         list( $inside, $trail ) = Linker::splitTrail( $trail );
 202
 203                         $entry = array(
 204                                 'title' => $nt,
 205                                 'text' => $prefix.$text.$inside,
 206                                 'pdbk' => $nt->getPrefixedDBkey(),
 207                         );
 208                         if ( $query !== array() ) {
 209                                 $entry['query'] = $query;
 210                         }
 211
 212                         if ( $nt->isExternal() ) {
 213                                 // Use a globally unique ID to keep the objects mergable
 214                                 $key = $this->parent->nextLinkID();
 215                                 $this->interwikis[$key] = $entry;
 216                                 $retVal = "<!--IWLINK $key-->{$trail}";
 217                         } else {
 218                                 $key = $this->parent->nextLinkID();
 219                                 $ns = $nt->getNamespace();
 220                                 $this->internals[$ns][$key] = $entry;
 221                                 $retVal = "<!--LINK $ns:$key-->{$trail}";
 222                         }
 223                         $this->size++;
 224                 }
 225                 wfProfileOut( __METHOD__ );
 226                 return $retVal;
 227         }
 228
 229         /**
 230          * FIXME: update documentation. makeLinkObj() is deprecated.
 231          * Replace <!--LINK--> link placeholders with actual links, in the buffer
 232          * Placeholders created in Skin::makeLinkObj()
 233          * Returns an array of link CSS classes, indexed by PDBK.
 234          */
 235         function replace( &$text ) {
 236                 wfProfileIn( __METHOD__ );
 237
 238                 $colours = $this->replaceInternal( $text );
 239                 $this->replaceInterwiki( $text );
 240
 241                 wfProfileOut( __METHOD__ );
 242                 return $colours;
 243         }
 244
 245         /**
 246          * Replace internal links
 247          */
 248         protected function replaceInternal( &$text ) {
 249                 if ( !$this->internals ) {
 250                         return;
 251                 }
 252
 253                 wfProfileIn( __METHOD__ );
 254                 global $wgContLang;
 255
 256                 $colours = array();
 257                 $linkCache = LinkCache::singleton();
 258                 $output = $this->parent->getOutput();
 259
 260                 wfProfileIn( __METHOD__.'-check' );
 261                 $dbr = wfGetDB( DB_SLAVE );
 262                 $page = $dbr->tableName( 'page' );
 263                 $threshold = $this->parent->getOptions()->getStubThreshold();
 264
 265                 # Sort by namespace
 266                 ksort( $this->internals );
 267
 268                 $linkcolour_ids = array();
 269
 270                 # Generate query
 271                 $query = false;
 272                 $current = null;
 273                 $queries = array();
 274                 foreach ( $this->internals as $ns => $entries ) {
 275                         foreach ( $entries as $entry ) {
 276                                 $title = $entry['title'];
 277                                 $pdbk = $entry['pdbk'];
 278
 279                                 # Skip invalid entries.
 280                                 # Result will be ugly, but prevents crash.
 281                                 if ( is_null( $title ) ) {
 282                                         continue;
 283                                 }
 284
 285                                 # Check if it's a static known link, e.g. interwiki
 286                                 if ( $title->isAlwaysKnown() ) {
 287                                         $colours[$pdbk] = '';
 288                                 } elseif ( $ns == NS_SPECIAL ) {
 289                                         $colours[$pdbk] = 'new';
 290                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
 291                                         $colours[$pdbk] = Linker::getLinkColour( $title, $threshold );
 292                                         $output->addLink( $title, $id );
 293                                         $linkcolour_ids[$id] = $pdbk;
 294                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
 295                                         $colours[$pdbk] = 'new';
 296                                 } else {
 297                                         # Not in the link cache, add it to the query
 298                                         $queries[$ns][] = $title->getDBkey();
 299                                 }
 300                         }
 301                 }
 302                 if ( $queries ) {
 303                         $where = array();
 304                         foreach( $queries as $ns => $pages ){
 305                                 $where[] = $dbr->makeList(
 306                                         array(
 307                                                 'page_namespace' => $ns,
 308                                                 'page_title' => $pages,
 309                                         ),
 310                                         LIST_AND
 311                                 );
 312                         }
 313
 314                         $res = $dbr->select(
 315                                 'page',
 316                                 array( 'page_id', 'page_namespace', 'page_title', 'page_is_redirect', 'page_len', 'page_latest' ),
 317                                 $dbr->makeList( $where, LIST_OR ),
 318                                 __METHOD__
 319                         );
 320
 321                         # Fetch data and form into an associative array
 322                         # non-existent = broken
 323                         foreach ( $res as $s ) {
 324                                 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
 325                                 $pdbk = $title->getPrefixedDBkey();
 326                                 $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect, $s->page_latest );
 327                                 $output->addLink( $title, $s->page_id );
 328                                 # FIXME: convoluted data flow
 329                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 330                                 # Use formal parameters instead
 331                                 $colours[$pdbk] = Linker::getLinkColour( $title, $threshold );
 332                                 //add id to the extension todolist
 333                                 $linkcolour_ids[$s->page_id] = $pdbk;
 334                         }
 335                         unset( $res );
 336                 }
 337                 if ( count($linkcolour_ids) ) {
 338                         //pass an array of page_ids to an extension
 339                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 340                 }
 341                 wfProfileOut( __METHOD__.'-check' );
 342
 343                 # Do a second query for different language variants of links and categories
 344                 if($wgContLang->hasVariants()) {
 345                         $this->doVariants( $colours );
 346                 }
 347
 348                 # Construct search and replace arrays
 349                 wfProfileIn( __METHOD__.'-construct' );
 350                 $replacePairs = array();
 351                 foreach ( $this->internals as $ns => $entries ) {
 352                         foreach ( $entries as $index => $entry ) {
 353                                 $pdbk = $entry['pdbk'];
 354                                 $title = $entry['title'];
 355                                 $query = isset( $entry['query'] ) ? $entry['query'] : array();
 356                                 $key = "$ns:$index";
 357                                 $searchkey = "<!--LINK $key-->";
 358                                 $displayText = $entry['text'];
 359                                 if ( $displayText === '' ) {
 360                                         $displayText = null;
 361                                 }
 362                                 if ( !isset( $colours[$pdbk] ) ) {
 363                                         $colours[$pdbk] = 'new';
 364                                 }
 365                                 $attribs = array();
 366                                 if ( $colours[$pdbk] == 'new' ) {
 367                                         $linkCache->addBadLinkObj( $title );
 368                                         $output->addLink( $title, 0 );
 369                                         $type = array( 'broken' );
 370                                 } else {
 371                                         if ( $colours[$pdbk] != '' ) {
 372                                                 $attribs['class'] = $colours[$pdbk];
 373                                         }
 374                                         $type = array( 'known', 'noclasses' );
 375                                 }
 376                                 $replacePairs[$searchkey] = Linker::link( $title, $displayText,
 377                                                 $attribs, $query, $type );
 378                         }
 379                 }
 380                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 381                 wfProfileOut( __METHOD__.'-construct' );
 382
 383                 # Do the thing
 384                 wfProfileIn( __METHOD__.'-replace' );
 385                 $text = preg_replace_callback(
 386                         '/(<!--LINK .*?-->)/',
 387                         $replacer->cb(),
 388                         $text);
 389
 390                 wfProfileOut( __METHOD__.'-replace' );
 391                 wfProfileOut( __METHOD__ );
 392         }
 393
 394         /**
 395          * Replace interwiki links
 396          */
 397         protected function replaceInterwiki( &$text ) {
 398                 if ( empty( $this->interwikis ) ) {
 399                         return;
 400                 }
 401
 402                 wfProfileIn( __METHOD__ );
 403                 # Make interwiki link HTML
 404                 $output = $this->parent->getOutput();
 405                 $replacePairs = array();
 406                 foreach( $this->interwikis as $key => $link ) {
 407                         $replacePairs[$key] = Linker::link( $link['title'], $link['text'] );
 408                         $output->addInterwikiLink( $link['title'] );
 409                 }
 410                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 411
 412                 $text = preg_replace_callback(
 413                         '/<!--IWLINK (.*?)-->/',
 414                         $replacer->cb(),
 415                         $text );
 416                 wfProfileOut( __METHOD__ );
 417         }
 418
 419         /**
 420          * Modify $this->internals and $colours according to language variant linking rules
 421          */
 422         protected function doVariants( &$colours ) {
 423                 global $wgContLang;
 424                 $linkBatch = new LinkBatch();
 425                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
 426                 $output = $this->parent->getOutput();
 427                 $linkCache = LinkCache::singleton();
 428                 $threshold = $this->parent->getOptions()->getStubThreshold();
 429                 $titlesToBeConverted = '';
 430                 $titlesAttrs = array();
 431
 432                 // Concatenate titles to a single string, thus we only need auto convert the
 433                 // single string to all variants. This would improve parser's performance
 434                 // significantly.
 435                 foreach ( $this->internals as $ns => $entries ) {
 436                         foreach ( $entries as $index => $entry ) {
 437                                 $pdbk = $entry['pdbk'];
 438                                 // we only deal with new links (in its first query)
 439                                 if ( !isset( $colours[$pdbk] ) ) {
 440                                         $title = $entry['title'];
 441                                         $titleText = $title->getText();
 442                                         $titlesAttrs[] = array(
 443                                                 'ns' => $ns,
 444                                                 'key' => "$ns:$index",
 445                                                 'titleText' => $titleText,
 446                                         );
 447                                         // separate titles with \0 because it would never appears
 448                                         // in a valid title
 449                                         $titlesToBeConverted .= $titleText . "\0";
 450                                 }
 451                         }
 452                 }
 453
 454                 // Now do the conversion and explode string to text of titles
 455                 $titlesAllVariants = $wgContLang->autoConvertToAllVariants( $titlesToBeConverted );
 456                 $allVariantsName = array_keys( $titlesAllVariants );
 457                 foreach ( $titlesAllVariants as &$titlesVariant ) {
 458                         $titlesVariant = explode( "\0", $titlesVariant );
 459                 }
 460                 $l = count( $titlesAttrs );
 461                 // Then add variants of links to link batch
 462                 for ( $i = 0; $i < $l; $i ++ ) {
 463                         foreach ( $allVariantsName as $variantName ) {
 464                                 $textVariant = $titlesAllVariants[$variantName][$i];
 465                                 if ( $textVariant != $titlesAttrs[$i]['titleText'] ) {
 466                                         $variantTitle = Title::makeTitle( $titlesAttrs[$i]['ns'], $textVariant );
 467                                         if( is_null( $variantTitle ) ) {
 468                                                 continue;
 469                                         }
 470                                         $linkBatch->addObj( $variantTitle );
 471                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $titlesAttrs[$i]['key'];
 472                                 }
 473                         }
 474                 }
 475
 476                 // process categories, check if a category exists in some variant
 477                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 478                 $varCategories = array(); // category replacements oldDBkey => newDBkey
 479                 foreach( $output->getCategoryLinks() as $category ){
 480                         $variants = $wgContLang->autoConvertToAllVariants( $category );
 481                         foreach($variants as $variant){
 482                                 if($variant != $category){
 483                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 484                                         if(is_null($variantTitle)) continue;
 485                                         $linkBatch->addObj( $variantTitle );
 486                                         $categoryMap[$variant] = $category;
 487                                 }
 488                         }
 489                 }
 490
 491
 492                 if(!$linkBatch->isEmpty()){
 493                         // construct query
 494                         $dbr = wfGetDB( DB_SLAVE );
 495                         $varRes = $dbr->select( 'page',
 496                                 array( 'page_id', 'page_namespace', 'page_title', 'page_is_redirect', 'page_len' ),
 497                                 $linkBatch->constructSet( 'page', $dbr ),
 498                                 __METHOD__
 499                         );
 500
 501                         $linkcolour_ids = array();
 502
 503                         // for each found variants, figure out link holders and replace
 504                         foreach ( $varRes as $s ) {
 505
 506                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 507                                 $varPdbk = $variantTitle->getPrefixedDBkey();
 508                                 $vardbk = $variantTitle->getDBkey();
 509
 510                                 $holderKeys = array();
 511                                 if( isset( $variantMap[$varPdbk] ) ) {
 512                                         $holderKeys = $variantMap[$varPdbk];
 513                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
 514                                         $output->addLink( $variantTitle, $s->page_id );
 515                                 }
 516
 517                                 // loop over link holders
 518                                 foreach( $holderKeys as $key ) {
 519                                         list( $ns, $index ) = explode( ':', $key, 2 );
 520                                         $entry =& $this->internals[$ns][$index];
 521                                         $pdbk = $entry['pdbk'];
 522
 523                                         if(!isset($colours[$pdbk])){
 524                                                 // found link in some of the variants, replace the link holder data
 525                                                 $entry['title'] = $variantTitle;
 526                                                 $entry['pdbk'] = $varPdbk;
 527
 528                                                 // set pdbk and colour
 529                                                 # FIXME: convoluted data flow
 530                                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 531                                                 # Use formal parameters instead
 532                                                 $colours[$varPdbk] = Linker::getLinkColour( $variantTitle, $threshold );
 533                                                 $linkcolour_ids[$s->page_id] = $pdbk;
 534                                         }
 535                                 }
 536
 537                                 // check if the object is a variant of a category
 538                                 if(isset($categoryMap[$vardbk])){
 539                                         $oldkey = $categoryMap[$vardbk];
 540                                         if($oldkey != $vardbk)
 541                                                 $varCategories[$oldkey]=$vardbk;
 542                                 }
 543                         }
 544                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 545
 546                         // rebuild the categories in original order (if there are replacements)
 547                         if(count($varCategories)>0){
 548                                 $newCats = array();
 549                                 $originalCats = $output->getCategories();
 550                                 foreach($originalCats as $cat => $sortkey){
 551                                         // make the replacement
 552                                         if( array_key_exists($cat,$varCategories) )
 553                                                 $newCats[$varCategories[$cat]] = $sortkey;
 554                                         else $newCats[$cat] = $sortkey;
 555                                 }
 556                                 $output->setCategoryLinks($newCats);
 557                         }
 558                 }
 559         }
 560
 561         /**
 562          * Replace <!--LINK--> link placeholders with plain text of links
 563          * (not HTML-formatted).
 564          *
 565          * @param $text String
 566          * @return String
 567          */
 568         function replaceText( $text ) {
 569                 wfProfileIn( __METHOD__ );
 570
 571                 $text = preg_replace_callback(
 572                         '/<!--(LINK|IWLINK) (.*?)-->/',
 573                         array( &$this, 'replaceTextCallback' ),
 574                         $text );
 575
 576                 wfProfileOut( __METHOD__ );
 577                 return $text;
 578         }
 579
 580         /**
 581          * Callback for replaceText()
 582          *
 583          * @param $matches Array
 584          * @return string
 585          * @private
 586          */
 587         function replaceTextCallback( $matches ) {
 588                 $type = $matches[1];
 589                 $key  = $matches[2];
 590                 if( $type == 'LINK' ) {
 591                         list( $ns, $index ) = explode( ':', $key, 2 );
 592                         if( isset( $this->internals[$ns][$index]['text'] ) ) {
 593                                 return $this->internals[$ns][$index]['text'];
 594                         }
 595                 } elseif( $type == 'IWLINK' ) {
 596                         if( isset( $this->interwikis[$key]['text'] ) ) {
 597                                 return $this->interwikis[$key]['text'];
 598                         }
 599                 }
 600                 return $matches[0];
 601         }
 602 }