includes/parser/LinkHolderArray.php

   1 <?php
   2
   3 class LinkHolderArray {
   4         var $batchSize = 1000;
   5
   6         var $internals = array(), $interwikis = array();
   7         var $size = 0;
   8         var $parent;
   9
  10         function __construct( $parent ) {
  11                 $this->parent = $parent;
  12         }
  13
  14         /**
  15          * Reduce memory usage to reduce the impact of circular references
  16          */
  17         function __destruct() {
  18                 foreach ( $this as $name => $value ) {
  19                         unset( $this->$name );
  20                 }
  21         }
  22
  23         /**
  24          * Merge another LinkHolderArray into this one
  25          */
  26         function merge( $other ) {
  27                 foreach ( $other->internals as $ns => $entries ) {
  28                         $this->size += count( $entries );
  29                         if ( !isset( $this->internals[$ns] ) ) {
  30                                 $this->internals[$ns] = $entries;
  31                         } else {
  32                                 $this->internals[$ns] += $entries;
  33                         }
  34                 }
  35                 $this->interwikis += $other->interwikis;
  36         }
  37
  38         /**
  39          * Returns true if the memory requirements of this object are getting large
  40          */
  41         function isBig() {
  42                 return $this->size > $this->batchSize;
  43         }
  44
  45         /**
  46          * Clear all stored link holders.
  47          * Make sure you don't have any text left using these link holders, before you call this
  48          */
  49         function clear() {
  50                 $this->internals = array();
  51                 $this->interwikis = array();
  52                 $this->size = 0;
  53         }
  54
  55         /**
  56          * Make a link placeholder. The text returned can be later resolved to a real link with
  57          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
  58          * parsing of interwiki links, and secondly to allow all existence checks and
  59          * article length checks (for stub links) to be bundled into a single query.
  60          *
  61          */
  62         function makeHolder( $nt, $text = '', $query = '', $trail = '', $prefix = ''  ) {
  63                 wfProfileIn( __METHOD__ );
  64                 if ( ! is_object($nt) ) {
  65                         # Fail gracefully
  66                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
  67                 } else {
  68                         # Separate the link trail from the rest of the link
  69                         list( $inside, $trail ) = Linker::splitTrail( $trail );
  70
  71                         $entry = array(
  72                                 'title' => $nt,
  73                                 'text' => $prefix.$text.$inside,
  74                                 'pdbk' => $nt->getPrefixedDBkey(),
  75                         );
  76                         if ( $query !== '' ) {
  77                                 $entry['query'] = $query;
  78                         }
  79
  80                         if ( $nt->isExternal() ) {
  81                                 // Use a globally unique ID to keep the objects mergable
  82                                 $key = $this->parent->nextLinkID();
  83                                 $this->interwikis[$key] = $entry;
  84                                 $retVal = "<!--IWLINK $key-->{$trail}";
  85                         } else {
  86                                 $key = $this->parent->nextLinkID();
  87                                 $ns = $nt->getNamespace();
  88                                 $this->internals[$ns][$key] = $entry;
  89                                 $retVal = "<!--LINK $ns:$key-->{$trail}";
  90                         }
  91                         $this->size++;
  92                 }
  93                 wfProfileOut( __METHOD__ );
  94                 return $retVal;
  95         }
  96
  97         /**
  98          * Replace <!--LINK--> link placeholders with actual links, in the buffer
  99          * Placeholders created in Skin::makeLinkObj()
 100          * Returns an array of link CSS classes, indexed by PDBK.
 101          */
 102         function replace( &$text ) {
 103                 wfProfileIn( __METHOD__ );
 104
 105                 $colours = $this->replaceInternal( $text );
 106                 $this->replaceInterwiki( $text );
 107
 108                 wfProfileOut( __METHOD__ );
 109                 return $colours;
 110         }
 111
 112         /**
 113          * Replace internal links
 114          */
 115         protected function replaceInternal( &$text ) {
 116                 if ( !$this->internals ) {
 117                         return;
 118                 }
 119
 120                 wfProfileIn( __METHOD__ );
 121                 global $wgUser, $wgContLang;
 122
 123                 $pdbks = array();
 124                 $colours = array();
 125                 $linkcolour_ids = array();
 126                 $sk = $this->parent->getOptions()->getSkin();
 127                 $linkCache = LinkCache::singleton();
 128                 $output = $this->parent->getOutput();
 129
 130                 wfProfileIn( __METHOD__.'-check' );
 131                 $dbr = wfGetDB( DB_SLAVE );
 132                 $page = $dbr->tableName( 'page' );
 133                 $threshold = $wgUser->getOption('stubthreshold');
 134
 135                 # Sort by namespace
 136                 ksort( $this->internals );
 137
 138                 # Generate query
 139                 $query = false;
 140                 $current = null;
 141                 foreach ( $this->internals as $ns => $entries ) {
 142                         foreach ( $entries as $index => $entry ) {
 143                                 $key =  "$ns:$index";
 144                                 $title = $entry['title'];
 145                                 $pdbk = $entry['pdbk'];
 146
 147                                 # Skip invalid entries.
 148                                 # Result will be ugly, but prevents crash.
 149                                 if ( is_null( $title ) ) {
 150                                         continue;
 151                                 }
 152
 153                                 # Check if it's a static known link, e.g. interwiki
 154                                 if ( $title->isAlwaysKnown() ) {
 155                                         $colours[$pdbk] = '';
 156                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
 157                                         $colours[$pdbk] = '';
 158                                         $output->addLink( $title, $id );
 159                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
 160                                         $colours[$pdbk] = 'new';
 161                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
 162                                         $colours[$pdbk] = 'new';
 163                                 } else {
 164                                         # Not in the link cache, add it to the query
 165                                         if ( !isset( $current ) ) {
 166                                                 $current = $ns;
 167                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 168                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
 169                                         } elseif ( $current != $ns ) {
 170                                                 $current = $ns;
 171                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
 172                                         } else {
 173                                                 $query .= ', ';
 174                                         }
 175
 176                                         $query .= $dbr->addQuotes( $title->getDBkey() );
 177                                 }
 178                         }
 179                 }
 180                 if ( $query ) {
 181                         $query .= '))';
 182
 183                         $res = $dbr->query( $query, __METHOD__ );
 184
 185                         # Fetch data and form into an associative array
 186                         # non-existent = broken
 187                         while ( $s = $dbr->fetchObject($res) ) {
 188                                 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
 189                                 $pdbk = $title->getPrefixedDBkey();
 190                                 $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
 191                                 $output->addLink( $title, $s->page_id );
 192                                 $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 193                                 //add id to the extension todolist
 194                                 $linkcolour_ids[$s->page_id] = $pdbk;
 195                         }
 196                         unset( $res );
 197                         //pass an array of page_ids to an extension
 198                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 199                 }
 200                 wfProfileOut( __METHOD__.'-check' );
 201
 202                 # Do a second query for different language variants of links and categories
 203                 if($wgContLang->hasVariants()){
 204                         $linkBatch = new LinkBatch();
 205                         $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
 206                         $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 207                         $varCategories = array(); // category replacements oldDBkey => newDBkey
 208
 209                         $categories = $output->getCategoryLinks();
 210
 211                         // Add variants of links to link batch
 212                         foreach ( $this->internals as $ns => $entries ) {
 213                                 foreach ( $entries as $index => $entry ) {
 214                                         $key = "$ns:$index";
 215                                         $pdbk = $entry['pdbk'];
 216                                         $title = $entry['title'];
 217                                         $titleText = $title->getText();
 218
 219                                         // generate all variants of the link title text
 220                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
 221
 222                                         // if link was not found (in first query), add all variants to query
 223                                         if ( !isset($colours[$pdbk]) ){
 224                                                 foreach($allTextVariants as $textVariant){
 225                                                         if($textVariant != $titleText){
 226                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
 227                                                                 if(is_null($variantTitle)) continue;
 228                                                                 $linkBatch->addObj( $variantTitle );
 229                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
 230                                                         }
 231                                                 }
 232                                         }
 233                                 }
 234                         }
 235
 236                         // process categories, check if a category exists in some variant
 237                         foreach( $categories as $category ){
 238                                 $variants = $wgContLang->convertLinkToAllVariants($category);
 239                                 foreach($variants as $variant){
 240                                         if($variant != $category){
 241                                                 $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 242                                                 if(is_null($variantTitle)) continue;
 243                                                 $linkBatch->addObj( $variantTitle );
 244                                                 $categoryMap[$variant] = $category;
 245                                         }
 246                                 }
 247                         }
 248
 249
 250                         if(!$linkBatch->isEmpty()){
 251                                 // construct query
 252                                 $titleClause = $linkBatch->constructSet('page', $dbr);
 253
 254                                 $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 255
 256                                 $variantQuery .= " FROM $page WHERE $titleClause";
 257
 258                                 $varRes = $dbr->query( $variantQuery, __METHOD__ );
 259
 260                                 // for each found variants, figure out link holders and replace
 261                                 while ( $s = $dbr->fetchObject($varRes) ) {
 262
 263                                         $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 264                                         $varPdbk = $variantTitle->getPrefixedDBkey();
 265                                         $vardbk = $variantTitle->getDBkey();
 266
 267                                         $holderKeys = array();
 268                                         if(isset($variantMap[$varPdbk])){
 269                                                 $holderKeys = $variantMap[$varPdbk];
 270                                                 $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
 271                                                 $output->addLink( $variantTitle, $s->page_id );
 272                                         }
 273
 274                                         // loop over link holders
 275                                         foreach($holderKeys as $key){
 276                                                 list( $ns, $index ) = explode( ':', $key, 2 );
 277                                                 $entry =& $this->internals[$ns][$index];
 278                                                 $pdbk = $entry['pdbk'];
 279
 280                                                 if(!isset($colours[$pdbk])){
 281                                                         // found link in some of the variants, replace the link holder data
 282                                                         $entry['title'] = $variantTitle;
 283                                                         $entry['pdbk'] = $varPdbk;
 284
 285                                                         // set pdbk and colour
 286                                                         $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
 287                                                         $linkcolour_ids[$s->page_id] = $pdbk;
 288                                                 }
 289                                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 290                                         }
 291
 292                                         // check if the object is a variant of a category
 293                                         if(isset($categoryMap[$vardbk])){
 294                                                 $oldkey = $categoryMap[$vardbk];
 295                                                 if($oldkey != $vardbk)
 296                                                         $varCategories[$oldkey]=$vardbk;
 297                                         }
 298                                 }
 299
 300                                 // rebuild the categories in original order (if there are replacements)
 301                                 if(count($varCategories)>0){
 302                                         $newCats = array();
 303                                         $originalCats = $output->getCategories();
 304                                         foreach($originalCats as $cat => $sortkey){
 305                                                 // make the replacement
 306                                                 if( array_key_exists($cat,$varCategories) )
 307                                                         $newCats[$varCategories[$cat]] = $sortkey;
 308                                                 else $newCats[$cat] = $sortkey;
 309                                         }
 310                                         $this->parent->mOutput->setCategoryLinks($newCats);
 311                                 }
 312                         }
 313                 }
 314
 315                 # Construct search and replace arrays
 316                 wfProfileIn( __METHOD__.'-construct' );
 317                 $replacePairs = array();
 318                 foreach ( $this->internals as $ns => $entries ) {
 319                         foreach ( $entries as $index => $entry ) {
 320                                 $pdbk = $entry['pdbk'];
 321                                 $title = $entry['title'];
 322                                 $query = isset( $entry['query'] ) ? $entry['query'] : '';
 323                                 $key = "$ns:$index";
 324                                 $searchkey = "<!--LINK $key-->";
 325                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
 326                                         $linkCache->addBadLinkObj( $title );
 327                                         $colours[$pdbk] = 'new';
 328                                         $output->addLink( $title, 0 );
 329                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
 330                                                                         $entry['text'],
 331                                                                         $query );
 332                                 } else {
 333                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
 334                                                                         $entry['text'],
 335                                                                         $query );
 336                                 }
 337                         }
 338                 }
 339                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 340                 wfProfileOut( __METHOD__.'-construct' );
 341
 342                 # Do the thing
 343                 wfProfileIn( __METHOD__.'-replace' );
 344                 $text = preg_replace_callback(
 345                         '/(<!--LINK .*?-->)/',
 346                         $replacer->cb(),
 347                         $text);
 348
 349                 wfProfileOut( __METHOD__.'-replace' );
 350                 wfProfileOut( __METHOD__ );
 351         }
 352
 353         /**
 354          * Replace interwiki links
 355          */
 356         protected function replaceInterwiki( &$text ) {
 357                 if ( empty( $this->interwikis ) ) {
 358                         return;
 359                 }
 360
 361                 wfProfileIn( __METHOD__ );
 362                 # Make interwiki link HTML
 363                 $sk = $this->parent->getOptions()->getSkin();
 364                 $replacePairs = array();
 365                 foreach( $this->interwikis as $key => $link ) {
 366                         $replacePairs[$key] = $sk->link( $link['title'], $link['text'] );
 367                 }
 368                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 369
 370                 $text = preg_replace_callback(
 371                         '/<!--IWLINK (.*?)-->/',
 372                         $replacer->cb(),
 373                         $text );
 374                 wfProfileOut( __METHOD__ );
 375         }
 376
 377         /**
 378          * Replace <!--LINK--> link placeholders with plain text of links
 379          * (not HTML-formatted).
 380          * @param string $text
 381          * @return string
 382          */
 383         function replaceText( $text ) {
 384                 wfProfileIn( __METHOD__ );
 385
 386                 $text = preg_replace_callback(
 387                         '/<!--(LINK|IWLINK) (.*?)-->/',
 388                         array( &$this, 'replaceTextCallback' ),
 389                         $text );
 390
 391                 wfProfileOut( __METHOD__ );
 392                 return $text;
 393         }
 394
 395         /**
 396          * @param array $matches
 397          * @return string
 398          * @private
 399          */
 400         function replaceTextCallback( $matches ) {
 401                 $type = $matches[1];
 402                 $key  = $matches[2];
 403                 if( $type == 'LINK' ) {
 404                         list( $ns, $index ) = explode( ':', $key, 2 );
 405                         if( isset( $this->internals[$ns][$index]['text'] ) ) {
 406                                 return $this->internals[$ns][$index]['text'];
 407                         }
 408                 } elseif( $type == 'IWLINK' ) {
 409                         if( isset( $this->interwikis[$key]['text'] ) ) {
 410                                 return $this->interwikis[$key]['text'];
 411                         }
 412                 }
 413                 return $matches[0];
 414         }
 415 }