includes/parser/LinkHolderArray.php

   1 <?php
   2
   3 class LinkHolderArray {
   4         var $internals = array(), $interwikis = array();
   5         var $size = 0;
   6         var $parent;
   7
   8         function __construct( $parent ) {
   9                 $this->parent = $parent;
  10         }
  11
  12         /**
  13          * Reduce memory usage to reduce the impact of circular references
  14          */
  15         function __destruct() {
  16                 foreach ( $this as $name => $value ) {
  17                         unset( $this->$name );
  18                 }
  19         }
  20
  21         /**
  22          * Merge another LinkHolderArray into this one
  23          */
  24         function merge( $other ) {
  25                 foreach ( $other->internals as $ns => $entries ) {
  26                         $this->size += count( $entries );
  27                         if ( !isset( $this->internals[$ns] ) ) {
  28                                 $this->internals[$ns] = $entries;
  29                         } else {
  30                                 $this->internals[$ns] += $entries;
  31                         }
  32                 }
  33                 $this->interwikis += $other->interwikis;
  34         }
  35
  36         /**
  37          * Returns true if the memory requirements of this object are getting large
  38          */
  39         function isBig() {
  40                 global $wgLinkHolderBatchSize;
  41                 return $this->size > $wgLinkHolderBatchSize;
  42         }
  43
  44         /**
  45          * Clear all stored link holders.
  46          * Make sure you don't have any text left using these link holders, before you call this
  47          */
  48         function clear() {
  49                 $this->internals = array();
  50                 $this->interwikis = array();
  51                 $this->size = 0;
  52         }
  53
  54         /**
  55          * Make a link placeholder. The text returned can be later resolved to a real link with
  56          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
  57          * parsing of interwiki links, and secondly to allow all existence checks and
  58          * article length checks (for stub links) to be bundled into a single query.
  59          *
  60          */
  61         function makeHolder( $nt, $text = '', $query = '', $trail = '', $prefix = ''  ) {
  62                 wfProfileIn( __METHOD__ );
  63                 if ( ! is_object($nt) ) {
  64                         # Fail gracefully
  65                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
  66                 } else {
  67                         # Separate the link trail from the rest of the link
  68                         list( $inside, $trail ) = Linker::splitTrail( $trail );
  69
  70                         $entry = array(
  71                                 'title' => $nt,
  72                                 'text' => $prefix.$text.$inside,
  73                                 'pdbk' => $nt->getPrefixedDBkey(),
  74                         );
  75                         if ( $query !== '' ) {
  76                                 $entry['query'] = $query;
  77                         }
  78
  79                         if ( $nt->isExternal() ) {
  80                                 // Use a globally unique ID to keep the objects mergable
  81                                 $key = $this->parent->nextLinkID();
  82                                 $this->interwikis[$key] = $entry;
  83                                 $retVal = "<!--IWLINK $key-->{$trail}";
  84                         } else {
  85                                 $key = $this->parent->nextLinkID();
  86                                 $ns = $nt->getNamespace();
  87                                 $this->internals[$ns][$key] = $entry;
  88                                 $retVal = "<!--LINK $ns:$key-->{$trail}";
  89                         }
  90                         $this->size++;
  91                 }
  92                 wfProfileOut( __METHOD__ );
  93                 return $retVal;
  94         }
  95
  96         /**
  97          * Get the stub threshold
  98          */
  99         function getStubThreshold() {
 100                 global $wgUser;
 101                 if ( !isset( $this->stubThreshold ) ) {
 102                         $this->stubThreshold = $wgUser->getStubThreshold();
 103                 }
 104                 return $this->stubThreshold;
 105         }
 106
 107         /**
 108          * FIXME: update documentation. makeLinkObj() is deprecated.
 109          * Replace <!--LINK--> link placeholders with actual links, in the buffer
 110          * Placeholders created in Skin::makeLinkObj()
 111          * Returns an array of link CSS classes, indexed by PDBK.
 112          */
 113         function replace( &$text ) {
 114                 wfProfileIn( __METHOD__ );
 115
 116                 $colours = $this->replaceInternal( $text );
 117                 $this->replaceInterwiki( $text );
 118
 119                 wfProfileOut( __METHOD__ );
 120                 return $colours;
 121         }
 122
 123         /**
 124          * Replace internal links
 125          */
 126         protected function replaceInternal( &$text ) {
 127                 if ( !$this->internals ) {
 128                         return;
 129                 }
 130
 131                 wfProfileIn( __METHOD__ );
 132                 global $wgContLang;
 133
 134                 $colours = array();
 135                 $sk = $this->parent->getOptions()->getSkin();
 136                 $linkCache = LinkCache::singleton();
 137                 $output = $this->parent->getOutput();
 138
 139                 wfProfileIn( __METHOD__.'-check' );
 140                 $dbr = wfGetDB( DB_SLAVE );
 141                 $page = $dbr->tableName( 'page' );
 142                 $threshold = $this->getStubThreshold();
 143
 144                 # Sort by namespace
 145                 ksort( $this->internals );
 146
 147                 $linkcolour_ids = array();
 148
 149                 # Generate query
 150                 $query = false;
 151                 $current = null;
 152                 foreach ( $this->internals as $ns => $entries ) {
 153                         foreach ( $entries as $index => $entry ) {
 154                                 $key =  "$ns:$index";
 155                                 $title = $entry['title'];
 156                                 $pdbk = $entry['pdbk'];
 157
 158                                 # Skip invalid entries.
 159                                 # Result will be ugly, but prevents crash.
 160                                 if ( is_null( $title ) ) {
 161                                         continue;
 162                                 }
 163
 164                                 # Check if it's a static known link, e.g. interwiki
 165                                 if ( $title->isAlwaysKnown() ) {
 166                                         $colours[$pdbk] = '';
 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                         while ( $s = $dbr->fetchObject($res) ) {
 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();
 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();
 298                 $threshold = $this->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->convertLinkToAllVariants( $titlesToBeConverted );
 326                 $allVariantsName = array_keys( $titlesAllVariants );
 327                 foreach ( $titlesAllVariants as &$titlesVariant ) {
 328                         $titlesVariant = explode( "\0", $titlesVariant );
 329                 }
 330
 331                 // Then add variants of links to link batch
 332                 for ( $i = 0; $l = count( $titlesAttrs ), $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 ) ) { continue; }
 339                                         $linkBatch->addObj( $variantTitle );
 340                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
 341                                 }
 342                         }
 343                 }
 344
 345                 // process categories, check if a category exists in some variant
 346                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 347                 $varCategories = array(); // category replacements oldDBkey => newDBkey
 348                 foreach( $output->getCategoryLinks() as $category ){
 349                         $variants = $wgContLang->convertLinkToAllVariants( $category );
 350                         foreach($variants as $variant){
 351                                 if($variant != $category){
 352                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 353                                         if(is_null($variantTitle)) continue;
 354                                         $linkBatch->addObj( $variantTitle );
 355                                         $categoryMap[$variant] = $category;
 356                                 }
 357                         }
 358                 }
 359
 360
 361                 if(!$linkBatch->isEmpty()){
 362                         // construct query
 363                         $dbr = wfGetDB( DB_SLAVE );
 364                         $page = $dbr->tableName( 'page' );
 365                         $titleClause = $linkBatch->constructSet('page', $dbr);
 366                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 367                         $variantQuery .= " FROM $page WHERE $titleClause";
 368                         $varRes = $dbr->query( $variantQuery, __METHOD__ );
 369                         $linkcolour_ids = array();
 370
 371                         // for each found variants, figure out link holders and replace
 372                         while ( $s = $dbr->fetchObject($varRes) ) {
 373
 374                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 375                                 $varPdbk = $variantTitle->getPrefixedDBkey();
 376                                 $vardbk = $variantTitle->getDBkey();
 377
 378                                 $holderKeys = array();
 379                                 if(isset($variantMap[$varPdbk])){
 380                                         $holderKeys = $variantMap[$varPdbk];
 381                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
 382                                         $output->addLink( $variantTitle, $s->page_id );
 383                                 }
 384
 385                                 // loop over link holders
 386                                 foreach($holderKeys as $key){
 387                                         list( $ns, $index ) = explode( ':', $key, 2 );
 388                                         $entry =& $this->internals[$ns][$index];
 389                                         $pdbk = $entry['pdbk'];
 390
 391                                         if(!isset($colours[$pdbk])){
 392                                                 // found link in some of the variants, replace the link holder data
 393                                                 $entry['title'] = $variantTitle;
 394                                                 $entry['pdbk'] = $varPdbk;
 395
 396                                                 // set pdbk and colour
 397                                                 # FIXME: convoluted data flow
 398                                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 399                                                 # Use formal parameters instead
 400                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
 401                                                 $linkcolour_ids[$s->page_id] = $pdbk;
 402                                         }
 403                                 }
 404
 405                                 // check if the object is a variant of a category
 406                                 if(isset($categoryMap[$vardbk])){
 407                                         $oldkey = $categoryMap[$vardbk];
 408                                         if($oldkey != $vardbk)
 409                                                 $varCategories[$oldkey]=$vardbk;
 410                                 }
 411                         }
 412                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 413
 414                         // rebuild the categories in original order (if there are replacements)
 415                         if(count($varCategories)>0){
 416                                 $newCats = array();
 417                                 $originalCats = $output->getCategories();
 418                                 foreach($originalCats as $cat => $sortkey){
 419                                         // make the replacement
 420                                         if( array_key_exists($cat,$varCategories) )
 421                                                 $newCats[$varCategories[$cat]] = $sortkey;
 422                                         else $newCats[$cat] = $sortkey;
 423                                 }
 424                                 $output->setCategoryLinks($newCats);
 425                         }
 426                 }
 427         }
 428
 429         /**
 430          * Replace <!--LINK--> link placeholders with plain text of links
 431          * (not HTML-formatted).
 432          *
 433          * @param $text String
 434          * @return String
 435          */
 436         function replaceText( $text ) {
 437                 wfProfileIn( __METHOD__ );
 438
 439                 $text = preg_replace_callback(
 440                         '/<!--(LINK|IWLINK) (.*?)-->/',
 441                         array( &$this, 'replaceTextCallback' ),
 442                         $text );
 443
 444                 wfProfileOut( __METHOD__ );
 445                 return $text;
 446         }
 447
 448         /**
 449          * Callback for replaceText()
 450          *
 451          * @param $matches Array
 452          * @return string
 453          * @private
 454          */
 455         function replaceTextCallback( $matches ) {
 456                 $type = $matches[1];
 457                 $key  = $matches[2];
 458                 if( $type == 'LINK' ) {
 459                         list( $ns, $index ) = explode( ':', $key, 2 );
 460                         if( isset( $this->internals[$ns][$index]['text'] ) ) {
 461                                 return $this->internals[$ns][$index]['text'];
 462                         }
 463                 } elseif( $type == 'IWLINK' ) {
 464                         if( isset( $this->interwikis[$key]['text'] ) ) {
 465                                 return $this->interwikis[$key]['text'];
 466                         }
 467                 }
 468                 return $matches[0];
 469         }
 470 }