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