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          * Get the stub threshold
 108          */
 109         function getStubThreshold() {
 110                 if ( !isset( $this->stubThreshold ) ) {
 111                         $this->stubThreshold = $this->parent->getUser()->getStubThreshold();
 112                 }
 113                 return $this->stubThreshold;
 114         }
 115
 116         /**
 117          * FIXME: update documentation. makeLinkObj() is deprecated.
 118          * Replace <!--LINK--> link placeholders with actual links, in the buffer
 119          * Placeholders created in Skin::makeLinkObj()
 120          * Returns an array of link CSS classes, indexed by PDBK.
 121          */
 122         function replace( &$text ) {
 123                 wfProfileIn( __METHOD__ );
 124
 125                 $colours = $this->replaceInternal( $text );
 126                 $this->replaceInterwiki( $text );
 127
 128                 wfProfileOut( __METHOD__ );
 129                 return $colours;
 130         }
 131
 132         /**
 133          * Replace internal links
 134          */
 135         protected function replaceInternal( &$text ) {
 136                 if ( !$this->internals ) {
 137                         return;
 138                 }
 139
 140                 wfProfileIn( __METHOD__ );
 141                 global $wgContLang;
 142
 143                 $colours = array();
 144                 $sk = $this->parent->getOptions()->getSkin( $this->parent->mTitle );
 145                 $linkCache = LinkCache::singleton();
 146                 $output = $this->parent->getOutput();
 147
 148                 wfProfileIn( __METHOD__.'-check' );
 149                 $dbr = wfGetDB( DB_SLAVE );
 150                 $page = $dbr->tableName( 'page' );
 151                 $threshold = $this->getStubThreshold();
 152
 153                 # Sort by namespace
 154                 ksort( $this->internals );
 155
 156                 $linkcolour_ids = array();
 157
 158                 # Generate query
 159                 $query = false;
 160                 $current = null;
 161                 foreach ( $this->internals as $ns => $entries ) {
 162                         foreach ( $entries as $entry ) {
 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 ( $ns == NS_SPECIAL ) {
 176                                         $colours[$pdbk] = 'new';
 177                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
 178                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 179                                         $output->addLink( $title, $id );
 180                                         $linkcolour_ids[$id] = $pdbk;
 181                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
 182                                         $colours[$pdbk] = 'new';
 183                                 } else {
 184                                         # Not in the link cache, add it to the query
 185                                         if ( !isset( $current ) ) {
 186                                                 $current = $ns;
 187                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len, page_latest";
 188                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
 189                                         } elseif ( $current != $ns ) {
 190                                                 $current = $ns;
 191                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
 192                                         } else {
 193                                                 $query .= ', ';
 194                                         }
 195
 196                                         $query .= $dbr->addQuotes( $title->getDBkey() );
 197                                 }
 198                         }
 199                 }
 200                 if ( $query ) {
 201                         $query .= '))';
 202
 203                         $res = $dbr->query( $query, __METHOD__ );
 204
 205                         # Fetch data and form into an associative array
 206                         # non-existent = broken
 207                         foreach ( $res as $s ) {
 208                                 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
 209                                 $pdbk = $title->getPrefixedDBkey();
 210                                 $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect, $s->page_latest );
 211                                 $output->addLink( $title, $s->page_id );
 212                                 # FIXME: convoluted data flow
 213                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 214                                 # Use formal parameters instead
 215                                 $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 216                                 //add id to the extension todolist
 217                                 $linkcolour_ids[$s->page_id] = $pdbk;
 218                         }
 219                         unset( $res );
 220                 }
 221                 if ( count($linkcolour_ids) ) {
 222                         //pass an array of page_ids to an extension
 223                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 224                 }
 225                 wfProfileOut( __METHOD__.'-check' );
 226
 227                 # Do a second query for different language variants of links and categories
 228                 if($wgContLang->hasVariants()) {
 229                         $this->doVariants( $colours );
 230                 }
 231
 232                 # Construct search and replace arrays
 233                 wfProfileIn( __METHOD__.'-construct' );
 234                 $replacePairs = array();
 235                 foreach ( $this->internals as $ns => $entries ) {
 236                         foreach ( $entries as $index => $entry ) {
 237                                 $pdbk = $entry['pdbk'];
 238                                 $title = $entry['title'];
 239                                 $query = isset( $entry['query'] ) ? $entry['query'] : '';
 240                                 $key = "$ns:$index";
 241                                 $searchkey = "<!--LINK $key-->";
 242                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
 243                                         $linkCache->addBadLinkObj( $title );
 244                                         $colours[$pdbk] = 'new';
 245                                         $output->addLink( $title, 0 );
 246                                         // FIXME: replace deprecated makeBrokenLinkObj() by link()
 247                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
 248                                                                         $entry['text'],
 249                                                                         $query );
 250                                 } else {
 251                                         // FIXME: replace deprecated makeColouredLinkObj() by link()
 252                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
 253                                                                         $entry['text'],
 254                                                                         $query );
 255                                 }
 256                         }
 257                 }
 258                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 259                 wfProfileOut( __METHOD__.'-construct' );
 260
 261                 # Do the thing
 262                 wfProfileIn( __METHOD__.'-replace' );
 263                 $text = preg_replace_callback(
 264                         '/(<!--LINK .*?-->)/',
 265                         $replacer->cb(),
 266                         $text);
 267
 268                 wfProfileOut( __METHOD__.'-replace' );
 269                 wfProfileOut( __METHOD__ );
 270         }
 271
 272         /**
 273          * Replace interwiki links
 274          */
 275         protected function replaceInterwiki( &$text ) {
 276                 if ( empty( $this->interwikis ) ) {
 277                         return;
 278                 }
 279
 280                 wfProfileIn( __METHOD__ );
 281                 # Make interwiki link HTML
 282                 $sk = $this->parent->getOptions()->getSkin( $this->parent->mTitle );
 283                 $output = $this->parent->getOutput();
 284                 $replacePairs = array();
 285                 foreach( $this->interwikis as $key => $link ) {
 286                         $replacePairs[$key] = $sk->link( $link['title'], $link['text'] );
 287                         $output->addInterwikiLink( $link['title'] );
 288                 }
 289                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 290
 291                 $text = preg_replace_callback(
 292                         '/<!--IWLINK (.*?)-->/',
 293                         $replacer->cb(),
 294                         $text );
 295                 wfProfileOut( __METHOD__ );
 296         }
 297
 298         /**
 299          * Modify $this->internals and $colours according to language variant linking rules
 300          */
 301         protected function doVariants( &$colours ) {
 302                 global $wgContLang;
 303                 $linkBatch = new LinkBatch();
 304                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
 305                 $output = $this->parent->getOutput();
 306                 $linkCache = LinkCache::singleton();
 307                 $sk = $this->parent->getOptions()->getSkin( $this->parent->mTitle );
 308                 $threshold = $this->getStubThreshold();
 309                 $titlesToBeConverted = '';
 310                 $titlesAttrs = array();
 311
 312                 // Concatenate titles to a single string, thus we only need auto convert the
 313                 // single string to all variants. This would improve parser's performance
 314                 // significantly.
 315                 foreach ( $this->internals as $ns => $entries ) {
 316                         foreach ( $entries as $index => $entry ) {
 317                                 $pdbk = $entry['pdbk'];
 318                                 // we only deal with new links (in its first query)
 319                                 if ( !isset( $colours[$pdbk] ) ) {
 320                                         $title = $entry['title'];
 321                                         $titleText = $title->getText();
 322                                         $titlesAttrs[] = array(
 323                                                 'ns' => $ns,
 324                                                 'key' => "$ns:$index",
 325                                                 'titleText' => $titleText,
 326                                         );
 327                                         // separate titles with \0 because it would never appears
 328                                         // in a valid title
 329                                         $titlesToBeConverted .= $titleText . "\0";
 330                                 }
 331                         }
 332                 }
 333
 334                 // Now do the conversion and explode string to text of titles
 335                 $titlesAllVariants = $wgContLang->autoConvertToAllVariants( $titlesToBeConverted );
 336                 $allVariantsName = array_keys( $titlesAllVariants );
 337                 foreach ( $titlesAllVariants as &$titlesVariant ) {
 338                         $titlesVariant = explode( "\0", $titlesVariant );
 339                 }
 340                 $l = count( $titlesAttrs );
 341                 // Then add variants of links to link batch
 342                 for ( $i = 0; $i < $l; $i ++ ) {
 343                         foreach ( $allVariantsName as $variantName ) {
 344                                 $textVariant = $titlesAllVariants[$variantName][$i];
 345                                 extract( $titlesAttrs[$i] );
 346                                 if($textVariant != $titleText){
 347                                         $variantTitle = Title::makeTitle( $ns, $textVariant );
 348                                         if( is_null( $variantTitle ) ) {
 349                                                 continue;
 350                                         }
 351                                         $linkBatch->addObj( $variantTitle );
 352                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
 353                                 }
 354                         }
 355                 }
 356
 357                 // process categories, check if a category exists in some variant
 358                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 359                 $varCategories = array(); // category replacements oldDBkey => newDBkey
 360                 foreach( $output->getCategoryLinks() as $category ){
 361                         $variants = $wgContLang->autoConvertToAllVariants( $category );
 362                         foreach($variants as $variant){
 363                                 if($variant != $category){
 364                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 365                                         if(is_null($variantTitle)) continue;
 366                                         $linkBatch->addObj( $variantTitle );
 367                                         $categoryMap[$variant] = $category;
 368                                 }
 369                         }
 370                 }
 371
 372
 373                 if(!$linkBatch->isEmpty()){
 374                         // construct query
 375                         $dbr = wfGetDB( DB_SLAVE );
 376                         $varRes = $dbr->select( 'page',
 377                                 array( 'page_id', 'page_namespace', 'page_title', 'page_is_redirect', 'page_len' ),
 378                                 $linkBatch->constructSet( 'page', $dbr ),
 379                                 __METHOD__
 380                         );
 381
 382                         $linkcolour_ids = array();
 383
 384                         // for each found variants, figure out link holders and replace
 385                         foreach ( $varRes as $s ) {
 386
 387                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 388                                 $varPdbk = $variantTitle->getPrefixedDBkey();
 389                                 $vardbk = $variantTitle->getDBkey();
 390
 391                                 $holderKeys = array();
 392                                 if( isset( $variantMap[$varPdbk] ) ) {
 393                                         $holderKeys = $variantMap[$varPdbk];
 394                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
 395                                         $output->addLink( $variantTitle, $s->page_id );
 396                                 }
 397
 398                                 // loop over link holders
 399                                 foreach( $holderKeys as $key ) {
 400                                         list( $ns, $index ) = explode( ':', $key, 2 );
 401                                         $entry =& $this->internals[$ns][$index];
 402                                         $pdbk = $entry['pdbk'];
 403
 404                                         if(!isset($colours[$pdbk])){
 405                                                 // found link in some of the variants, replace the link holder data
 406                                                 $entry['title'] = $variantTitle;
 407                                                 $entry['pdbk'] = $varPdbk;
 408
 409                                                 // set pdbk and colour
 410                                                 # FIXME: convoluted data flow
 411                                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 412                                                 # Use formal parameters instead
 413                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
 414                                                 $linkcolour_ids[$s->page_id] = $pdbk;
 415                                         }
 416                                 }
 417
 418                                 // check if the object is a variant of a category
 419                                 if(isset($categoryMap[$vardbk])){
 420                                         $oldkey = $categoryMap[$vardbk];
 421                                         if($oldkey != $vardbk)
 422                                                 $varCategories[$oldkey]=$vardbk;
 423                                 }
 424                         }
 425                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 426
 427                         // rebuild the categories in original order (if there are replacements)
 428                         if(count($varCategories)>0){
 429                                 $newCats = array();
 430                                 $originalCats = $output->getCategories();
 431                                 foreach($originalCats as $cat => $sortkey){
 432                                         // make the replacement
 433                                         if( array_key_exists($cat,$varCategories) )
 434                                                 $newCats[$varCategories[$cat]] = $sortkey;
 435                                         else $newCats[$cat] = $sortkey;
 436                                 }
 437                                 $output->setCategoryLinks($newCats);
 438                         }
 439                 }
 440         }
 441
 442         /**
 443          * Replace <!--LINK--> link placeholders with plain text of links
 444          * (not HTML-formatted).
 445          *
 446          * @param $text String
 447          * @return String
 448          */
 449         function replaceText( $text ) {
 450                 wfProfileIn( __METHOD__ );
 451
 452                 $text = preg_replace_callback(
 453                         '/<!--(LINK|IWLINK) (.*?)-->/',
 454                         array( &$this, 'replaceTextCallback' ),
 455                         $text );
 456
 457                 wfProfileOut( __METHOD__ );
 458                 return $text;
 459         }
 460
 461         /**
 462          * Callback for replaceText()
 463          *
 464          * @param $matches Array
 465          * @return string
 466          * @private
 467          */
 468         function replaceTextCallback( $matches ) {
 469                 $type = $matches[1];
 470                 $key  = $matches[2];
 471                 if( $type == 'LINK' ) {
 472                         list( $ns, $index ) = explode( ':', $key, 2 );
 473                         if( isset( $this->internals[$ns][$index]['text'] ) ) {
 474                                 return $this->internals[$ns][$index]['text'];
 475                         }
 476                 } elseif( $type == 'IWLINK' ) {
 477                         if( isset( $this->interwikis[$key]['text'] ) ) {
 478                                 return $this->interwikis[$key]['text'];
 479                         }
 480                 }
 481                 return $matches[0];
 482         }
 483 }