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
  16         function __construct( $parent ) {
  17                 $this->parent = $parent;
  18         }
  19
  20         /**
  21          * Reduce memory usage to reduce the impact of circular references
  22          */
  23         function __destruct() {
  24                 foreach ( $this as $name => $value ) {
  25                         unset( $this->$name );
  26                 }
  27         }
  28
  29         /**
  30          * Merge another LinkHolderArray into this one
  31          */
  32         function merge( $other ) {
  33                 foreach ( $other->internals as $ns => $entries ) {
  34                         $this->size += count( $entries );
  35                         if ( !isset( $this->internals[$ns] ) ) {
  36                                 $this->internals[$ns] = $entries;
  37                         } else {
  38                                 $this->internals[$ns] += $entries;
  39                         }
  40                 }
  41                 $this->interwikis += $other->interwikis;
  42         }
  43
  44         /**
  45          * Returns true if the memory requirements of this object are getting large
  46          */
  47         function isBig() {
  48                 global $wgLinkHolderBatchSize;
  49                 return $this->size > $wgLinkHolderBatchSize;
  50         }
  51
  52         /**
  53          * Clear all stored link holders.
  54          * Make sure you don't have any text left using these link holders, before you call this
  55          */
  56         function clear() {
  57                 $this->internals = array();
  58                 $this->interwikis = array();
  59                 $this->size = 0;
  60         }
  61
  62         /**
  63          * Make a link placeholder. The text returned can be later resolved to a real link with
  64          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
  65          * parsing of interwiki links, and secondly to allow all existence checks and
  66          * article length checks (for stub links) to be bundled into a single query.
  67          *
  68          */
  69         function makeHolder( $nt, $text = '', $query = '', $trail = '', $prefix = ''  ) {
  70                 wfProfileIn( __METHOD__ );
  71                 if ( ! is_object($nt) ) {
  72                         # Fail gracefully
  73                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
  74                 } else {
  75                         # Separate the link trail from the rest of the link
  76                         list( $inside, $trail ) = Linker::splitTrail( $trail );
  77
  78                         $entry = array(
  79                                 'title' => $nt,
  80                                 'text' => $prefix.$text.$inside,
  81                                 'pdbk' => $nt->getPrefixedDBkey(),
  82                         );
  83                         if ( $query !== '' ) {
  84                                 $entry['query'] = $query;
  85                         }
  86
  87                         if ( $nt->isExternal() ) {
  88                                 // Use a globally unique ID to keep the objects mergable
  89                                 $key = $this->parent->nextLinkID();
  90                                 $this->interwikis[$key] = $entry;
  91                                 $retVal = "<!--IWLINK $key-->{$trail}";
  92                         } else {
  93                                 $key = $this->parent->nextLinkID();
  94                                 $ns = $nt->getNamespace();
  95                                 $this->internals[$ns][$key] = $entry;
  96                                 $retVal = "<!--LINK $ns:$key-->{$trail}";
  97                         }
  98                         $this->size++;
  99                 }
 100                 wfProfileOut( __METHOD__ );
 101                 return $retVal;
 102         }
 103
 104         /**
 105          * Get the stub threshold
 106          */
 107         function getStubThreshold() {
 108                 global $wgUser;
 109                 if ( !isset( $this->stubThreshold ) ) {
 110                         $this->stubThreshold = $wgUser->getStubThreshold();
 111                 }
 112                 return $this->stubThreshold;
 113         }
 114
 115         /**
 116          * FIXME: update documentation. makeLinkObj() is deprecated.
 117          * Replace <!--LINK--> link placeholders with actual links, in the buffer
 118          * Placeholders created in Skin::makeLinkObj()
 119          * Returns an array of link CSS classes, indexed by PDBK.
 120          */
 121         function replace( &$text ) {
 122                 wfProfileIn( __METHOD__ );
 123
 124                 $colours = $this->replaceInternal( $text );
 125                 $this->replaceInterwiki( $text );
 126
 127                 wfProfileOut( __METHOD__ );
 128                 return $colours;
 129         }
 130
 131         /**
 132          * Replace internal links
 133          */
 134         protected function replaceInternal( &$text ) {
 135                 if ( !$this->internals ) {
 136                         return;
 137                 }
 138
 139                 wfProfileIn( __METHOD__ );
 140                 global $wgContLang;
 141
 142                 $colours = array();
 143                 $sk = $this->parent->getOptions()->getSkin( $this->parent->mTitle );
 144                 $linkCache = LinkCache::singleton();
 145                 $output = $this->parent->getOutput();
 146
 147                 wfProfileIn( __METHOD__.'-check' );
 148                 $dbr = wfGetDB( DB_SLAVE );
 149                 $page = $dbr->tableName( 'page' );
 150                 $threshold = $this->getStubThreshold();
 151
 152                 # Sort by namespace
 153                 ksort( $this->internals );
 154
 155                 $linkcolour_ids = array();
 156
 157                 # Generate query
 158                 $query = false;
 159                 $current = null;
 160                 foreach ( $this->internals as $ns => $entries ) {
 161                         foreach ( $entries as $index => $entry ) {
 162                                 $key =  "$ns:$index";
 163                                 $title = $entry['title'];
 164                                 $pdbk = $entry['pdbk'];
 165
 166                                 # Skip invalid entries.
 167                                 # Result will be ugly, but prevents crash.
 168                                 if ( is_null( $title ) ) {
 169                                         continue;
 170                                 }
 171
 172                                 # Check if it's a static known link, e.g. interwiki
 173                                 if ( $title->isAlwaysKnown() ) {
 174                                         $colours[$pdbk] = '';
 175                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
 176                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 177                                         $output->addLink( $title, $id );
 178                                         $linkcolour_ids[$id] = $pdbk;
 179                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
 180                                         $colours[$pdbk] = 'new';
 181                                 } else {
 182                                         # Not in the link cache, add it to the query
 183                                         if ( !isset( $current ) ) {
 184                                                 $current = $ns;
 185                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len, page_latest";
 186                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
 187                                         } elseif ( $current != $ns ) {
 188                                                 $current = $ns;
 189                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
 190                                         } else {
 191                                                 $query .= ', ';
 192                                         }
 193
 194                                         $query .= $dbr->addQuotes( $title->getDBkey() );
 195                                 }
 196                         }
 197                 }
 198                 if ( $query ) {
 199                         $query .= '))';
 200
 201                         $res = $dbr->query( $query, __METHOD__ );
 202
 203                         # Fetch data and form into an associative array
 204                         # non-existent = broken
 205                         while ( $s = $dbr->fetchObject($res) ) {
 206                                 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
 207                                 $pdbk = $title->getPrefixedDBkey();
 208                                 $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect, $s->page_latest );
 209                                 $output->addLink( $title, $s->page_id );
 210                                 # FIXME: convoluted data flow
 211                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 212                                 # Use formal parameters instead
 213                                 $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 214                                 //add id to the extension todolist
 215                                 $linkcolour_ids[$s->page_id] = $pdbk;
 216                         }
 217                         unset( $res );
 218                 }
 219                 if ( count($linkcolour_ids) ) {
 220                         //pass an array of page_ids to an extension
 221                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 222                 }
 223                 wfProfileOut( __METHOD__.'-check' );
 224
 225                 # Do a second query for different language variants of links and categories
 226                 if($wgContLang->hasVariants()) {
 227                         $this->doVariants( $colours );
 228                 }
 229
 230                 # Construct search and replace arrays
 231                 wfProfileIn( __METHOD__.'-construct' );
 232                 $replacePairs = array();
 233                 foreach ( $this->internals as $ns => $entries ) {
 234                         foreach ( $entries as $index => $entry ) {
 235                                 $pdbk = $entry['pdbk'];
 236                                 $title = $entry['title'];
 237                                 $query = isset( $entry['query'] ) ? $entry['query'] : '';
 238                                 $key = "$ns:$index";
 239                                 $searchkey = "<!--LINK $key-->";
 240                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
 241                                         $linkCache->addBadLinkObj( $title );
 242                                         $colours[$pdbk] = 'new';
 243                                         $output->addLink( $title, 0 );
 244                                         // FIXME: replace deprecated makeBrokenLinkObj() by link()
 245                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
 246                                                                         $entry['text'],
 247                                                                         $query );
 248                                 } else {
 249                                         // FIXME: replace deprecated makeColouredLinkObj() by link()
 250                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
 251                                                                         $entry['text'],
 252                                                                         $query );
 253                                 }
 254                         }
 255                 }
 256                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 257                 wfProfileOut( __METHOD__.'-construct' );
 258
 259                 # Do the thing
 260                 wfProfileIn( __METHOD__.'-replace' );
 261                 $text = preg_replace_callback(
 262                         '/(<!--LINK .*?-->)/',
 263                         $replacer->cb(),
 264                         $text);
 265
 266                 wfProfileOut( __METHOD__.'-replace' );
 267                 wfProfileOut( __METHOD__ );
 268         }
 269
 270         /**
 271          * Replace interwiki links
 272          */
 273         protected function replaceInterwiki( &$text ) {
 274                 if ( empty( $this->interwikis ) ) {
 275                         return;
 276                 }
 277
 278                 wfProfileIn( __METHOD__ );
 279                 # Make interwiki link HTML
 280                 $sk = $this->parent->getOptions()->getSkin( $this->parent->mTitle );
 281                 $output = $this->parent->getOutput();
 282                 $replacePairs = array();
 283                 foreach( $this->interwikis as $key => $link ) {
 284                         $replacePairs[$key] = $sk->link( $link['title'], $link['text'] );
 285                         $output->addInterwikiLink( $link['title'] );
 286                 }
 287                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 288
 289                 $text = preg_replace_callback(
 290                         '/<!--IWLINK (.*?)-->/',
 291                         $replacer->cb(),
 292                         $text );
 293                 wfProfileOut( __METHOD__ );
 294         }
 295
 296         /**
 297          * Modify $this->internals and $colours according to language variant linking rules
 298          */
 299         protected function doVariants( &$colours ) {
 300                 global $wgContLang;
 301                 $linkBatch = new LinkBatch();
 302                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
 303                 $output = $this->parent->getOutput();
 304                 $linkCache = LinkCache::singleton();
 305                 $sk = $this->parent->getOptions()->getSkin( $this->parent->mTitle );
 306                 $threshold = $this->getStubThreshold();
 307                 $titlesToBeConverted = '';
 308                 $titlesAttrs = array();
 309
 310                 // Concatenate titles to a single string, thus we only need auto convert the
 311                 // single string to all variants. This would improve parser's performance
 312                 // significantly.
 313                 foreach ( $this->internals as $ns => $entries ) {
 314                         foreach ( $entries as $index => $entry ) {
 315                                 $pdbk = $entry['pdbk'];
 316                                 // we only deal with new links (in its first query)
 317                                 if ( !isset( $colours[$pdbk] ) ) {
 318                                         $title = $entry['title'];
 319                                         $titleText = $title->getText();
 320                                         $titlesAttrs[] = array(
 321                                                 'ns' => $ns,
 322                                                 'key' => "$ns:$index",
 323                                                 'titleText' => $titleText,
 324                                         );
 325                                         // separate titles with \0 because it would never appears
 326                                         // in a valid title
 327                                         $titlesToBeConverted .= $titleText . "\0";
 328                                 }
 329                         }
 330                 }
 331
 332                 // Now do the conversion and explode string to text of titles
 333                 $titlesAllVariants = $wgContLang->convertLinkToAllVariants( $titlesToBeConverted );
 334                 $allVariantsName = array_keys( $titlesAllVariants );
 335                 foreach ( $titlesAllVariants as &$titlesVariant ) {
 336                         $titlesVariant = explode( "\0", $titlesVariant );
 337                 }
 338
 339                 // Then add variants of links to link batch
 340                 for ( $i = 0; $l = count( $titlesAttrs ), $i < $l; $i ++ ) {
 341                         foreach ( $allVariantsName as $variantName ) {
 342                                 $textVariant = $titlesAllVariants[$variantName][$i];
 343                                 extract( $titlesAttrs[$i] );
 344                                 if($textVariant != $titleText){
 345                                         $variantTitle = Title::makeTitle( $ns, $textVariant );
 346                                         if( is_null( $variantTitle ) ) { continue; }
 347                                         $linkBatch->addObj( $variantTitle );
 348                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
 349                                 }
 350                         }
 351                 }
 352
 353                 // process categories, check if a category exists in some variant
 354                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 355                 $varCategories = array(); // category replacements oldDBkey => newDBkey
 356                 foreach( $output->getCategoryLinks() as $category ){
 357                         $variants = $wgContLang->convertLinkToAllVariants( $category );
 358                         foreach($variants as $variant){
 359                                 if($variant != $category){
 360                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 361                                         if(is_null($variantTitle)) continue;
 362                                         $linkBatch->addObj( $variantTitle );
 363                                         $categoryMap[$variant] = $category;
 364                                 }
 365                         }
 366                 }
 367
 368
 369                 if(!$linkBatch->isEmpty()){
 370                         // construct query
 371                         $dbr = wfGetDB( DB_SLAVE );
 372                         $page = $dbr->tableName( 'page' );
 373                         $titleClause = $linkBatch->constructSet('page', $dbr);
 374                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 375                         $variantQuery .= " FROM $page WHERE $titleClause";
 376                         $varRes = $dbr->query( $variantQuery, __METHOD__ );
 377                         $linkcolour_ids = array();
 378
 379                         // for each found variants, figure out link holders and replace
 380                         while ( $s = $dbr->fetchObject($varRes) ) {
 381
 382                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 383                                 $varPdbk = $variantTitle->getPrefixedDBkey();
 384                                 $vardbk = $variantTitle->getDBkey();
 385
 386                                 $holderKeys = array();
 387                                 if(isset($variantMap[$varPdbk])){
 388                                         $holderKeys = $variantMap[$varPdbk];
 389                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
 390                                         $output->addLink( $variantTitle, $s->page_id );
 391                                 }
 392
 393                                 // loop over link holders
 394                                 foreach($holderKeys as $key){
 395                                         list( $ns, $index ) = explode( ':', $key, 2 );
 396                                         $entry =& $this->internals[$ns][$index];
 397                                         $pdbk = $entry['pdbk'];
 398
 399                                         if(!isset($colours[$pdbk])){
 400                                                 // found link in some of the variants, replace the link holder data
 401                                                 $entry['title'] = $variantTitle;
 402                                                 $entry['pdbk'] = $varPdbk;
 403
 404                                                 // set pdbk and colour
 405                                                 # FIXME: convoluted data flow
 406                                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 407                                                 # Use formal parameters instead
 408                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
 409                                                 $linkcolour_ids[$s->page_id] = $pdbk;
 410                                         }
 411                                 }
 412
 413                                 // check if the object is a variant of a category
 414                                 if(isset($categoryMap[$vardbk])){
 415                                         $oldkey = $categoryMap[$vardbk];
 416                                         if($oldkey != $vardbk)
 417                                                 $varCategories[$oldkey]=$vardbk;
 418                                 }
 419                         }
 420                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 421
 422                         // rebuild the categories in original order (if there are replacements)
 423                         if(count($varCategories)>0){
 424                                 $newCats = array();
 425                                 $originalCats = $output->getCategories();
 426                                 foreach($originalCats as $cat => $sortkey){
 427                                         // make the replacement
 428                                         if( array_key_exists($cat,$varCategories) )
 429                                                 $newCats[$varCategories[$cat]] = $sortkey;
 430                                         else $newCats[$cat] = $sortkey;
 431                                 }
 432                                 $output->setCategoryLinks($newCats);
 433                         }
 434                 }
 435         }
 436
 437         /**
 438          * Replace <!--LINK--> link placeholders with plain text of links
 439          * (not HTML-formatted).
 440          *
 441          * @param $text String
 442          * @return String
 443          */
 444         function replaceText( $text ) {
 445                 wfProfileIn( __METHOD__ );
 446
 447                 $text = preg_replace_callback(
 448                         '/<!--(LINK|IWLINK) (.*?)-->/',
 449                         array( &$this, 'replaceTextCallback' ),
 450                         $text );
 451
 452                 wfProfileOut( __METHOD__ );
 453                 return $text;
 454         }
 455
 456         /**
 457          * Callback for replaceText()
 458          *
 459          * @param $matches Array
 460          * @return string
 461          * @private
 462          */
 463         function replaceTextCallback( $matches ) {
 464                 $type = $matches[1];
 465                 $key  = $matches[2];
 466                 if( $type == 'LINK' ) {
 467                         list( $ns, $index ) = explode( ':', $key, 2 );
 468                         if( isset( $this->internals[$ns][$index]['text'] ) ) {
 469                                 return $this->internals[$ns][$index]['text'];
 470                         }
 471                 } elseif( $type == 'IWLINK' ) {
 472                         if( isset( $this->interwikis[$key]['text'] ) ) {
 473                                 return $this->interwikis[$key]['text'];
 474                         }
 475                 }
 476                 return $matches[0];
 477         }
 478 }