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