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 $retVal;
  94         }
  95
  96         /**
  97          * Replace <!--LINK--> link placeholders with actual links, in the buffer
  98          * Placeholders created in Skin::makeLinkObj()
  99          * Returns an array of link CSS classes, indexed by PDBK.
 100          */
 101         function replace( &$text ) {
 102                 wfProfileIn( __METHOD__ );
 103
 104                 $colours = $this->replaceInternal( $text );
 105                 $this->replaceInterwiki( $text );
 106
 107                 wfProfileOut( __METHOD__ );
 108                 return $colours;
 109         }
 110
 111         /**
 112          * Replace internal links
 113          */
 114         protected function replaceInternal( &$text ) {
 115                 if ( !$this->internals ) {
 116                         return;
 117                 }
 118
 119                 wfProfileIn( __METHOD__ );
 120                 global $wgUser, $wgContLang;
 121
 122                 $pdbks = array();
 123                 $colours = array();
 124                 $linkcolour_ids = array();
 125                 $sk = $this->parent->getOptions()->getSkin();
 126                 $linkCache = LinkCache::singleton();
 127                 $output = $this->parent->getOutput();
 128
 129                 wfProfileIn( __METHOD__.'-check' );
 130                 $dbr = wfGetDB( DB_SLAVE );
 131                 $page = $dbr->tableName( 'page' );
 132                 $threshold = $wgUser->getOption('stubthreshold');
 133
 134                 # Sort by namespace
 135                 ksort( $this->internals );
 136
 137                 # Generate query
 138                 $query = false;
 139                 $current = null;
 140                 foreach ( $this->internals as $ns => $entries ) {
 141                         foreach ( $entries as $index => $entry ) {
 142                                 $key =  "$ns:$index";
 143                                 $title = $entry['title'];
 144                                 $pdbk = $entry['pdbk'];
 145
 146                                 # Skip invalid entries.
 147                                 # Result will be ugly, but prevents crash.
 148                                 if ( is_null( $title ) ) {
 149                                         continue;
 150                                 }
 151
 152                                 # Check if it's a static known link, e.g. interwiki
 153                                 if ( $title->isAlwaysKnown() ) {
 154                                         $colours[$pdbk] = '';
 155                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
 156                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 157                                         $output->addLink( $title, $id );
 158                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
 159                                         $colours[$pdbk] = 'new';
 160                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
 161                                         $colours[$pdbk] = 'new';
 162                                 } else {
 163                                         # Not in the link cache, add it to the query
 164                                         if ( !isset( $current ) ) {
 165                                                 $current = $ns;
 166                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 167                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
 168                                         } elseif ( $current != $ns ) {
 169                                                 $current = $ns;
 170                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
 171                                         } else {
 172                                                 $query .= ', ';
 173                                         }
 174
 175                                         $query .= $dbr->addQuotes( $title->getDBkey() );
 176                                 }
 177                         }
 178                 }
 179                 if ( $query ) {
 180                         $query .= '))';
 181
 182                         $res = $dbr->query( $query, __METHOD__ );
 183
 184                         # Fetch data and form into an associative array
 185                         # non-existent = broken
 186                         while ( $s = $dbr->fetchObject($res) ) {
 187                                 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
 188                                 $pdbk = $title->getPrefixedDBkey();
 189                                 $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
 190                                 $output->addLink( $title, $s->page_id );
 191                                 # FIXME: convoluted data flow
 192                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 193                                 # Use formal parameters instead
 194                                 $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 195                                 //add id to the extension todolist
 196                                 $linkcolour_ids[$s->page_id] = $pdbk;
 197                         }
 198                         unset( $res );
 199                         //pass an array of page_ids to an extension
 200                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 201                 }
 202                 wfProfileOut( __METHOD__.'-check' );
 203
 204                 # Do a second query for different language variants of links and categories
 205                 if($wgContLang->hasVariants()){
 206                         $linkBatch = new LinkBatch();
 207                         $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
 208                         $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 209                         $varCategories = array(); // category replacements oldDBkey => newDBkey
 210
 211                         $categories = $output->getCategoryLinks();
 212
 213                         // Add variants of links to link batch
 214                         foreach ( $this->internals as $ns => $entries ) {
 215                                 foreach ( $entries as $index => $entry ) {
 216                                         $key = "$ns:$index";
 217                                         $pdbk = $entry['pdbk'];
 218                                         $title = $entry['title'];
 219                                         $titleText = $title->getText();
 220
 221                                         // generate all variants of the link title text
 222                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
 223
 224                                         // if link was not found (in first query), add all variants to query
 225                                         if ( !isset($colours[$pdbk]) ){
 226                                                 foreach($allTextVariants as $textVariant){
 227                                                         if($textVariant != $titleText){
 228                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
 229                                                                 if(is_null($variantTitle)) continue;
 230                                                                 $linkBatch->addObj( $variantTitle );
 231                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
 232                                                         }
 233                                                 }
 234                                         }
 235                                 }
 236                         }
 237
 238                         // process categories, check if a category exists in some variant
 239                         foreach( $categories as $category ){
 240                                 $variants = $wgContLang->convertLinkToAllVariants($category);
 241                                 foreach($variants as $variant){
 242                                         if($variant != $category){
 243                                                 $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 244                                                 if(is_null($variantTitle)) continue;
 245                                                 $linkBatch->addObj( $variantTitle );
 246                                                 $categoryMap[$variant] = $category;
 247                                         }
 248                                 }
 249                         }
 250
 251
 252                         if(!$linkBatch->isEmpty()){
 253                                 // construct query
 254                                 $titleClause = $linkBatch->constructSet('page', $dbr);
 255
 256                                 $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 257
 258                                 $variantQuery .= " FROM $page WHERE $titleClause";
 259
 260                                 $varRes = $dbr->query( $variantQuery, __METHOD__ );
 261
 262                                 // for each found variants, figure out link holders and replace
 263                                 while ( $s = $dbr->fetchObject($varRes) ) {
 264
 265                                         $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 266                                         $varPdbk = $variantTitle->getPrefixedDBkey();
 267                                         $vardbk = $variantTitle->getDBkey();
 268
 269                                         $holderKeys = array();
 270                                         if(isset($variantMap[$varPdbk])){
 271                                                 $holderKeys = $variantMap[$varPdbk];
 272                                                 $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
 273                                                 $output->addLink( $variantTitle, $s->page_id );
 274                                         }
 275
 276                                         // loop over link holders
 277                                         foreach($holderKeys as $key){
 278                                                 list( $ns, $index ) = explode( ':', $key, 2 );
 279                                                 $entry =& $this->internals[$ns][$index];
 280                                                 $pdbk = $entry['pdbk'];
 281
 282                                                 if(!isset($colours[$pdbk])){
 283                                                         // found link in some of the variants, replace the link holder data
 284                                                         $entry['title'] = $variantTitle;
 285                                                         $entry['pdbk'] = $varPdbk;
 286
 287                                                         // set pdbk and colour
 288                                                         # FIXME: convoluted data flow
 289                                                         # The redirect status and length is passed to getLinkColour via the LinkCache
 290                                                         # Use formal parameters instead
 291                                                         $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
 292                                                         $linkcolour_ids[$s->page_id] = $pdbk;
 293                                                 }
 294                                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 295                                         }
 296
 297                                         // check if the object is a variant of a category
 298                                         if(isset($categoryMap[$vardbk])){
 299                                                 $oldkey = $categoryMap[$vardbk];
 300                                                 if($oldkey != $vardbk)
 301                                                         $varCategories[$oldkey]=$vardbk;
 302                                         }
 303                                 }
 304
 305                                 // rebuild the categories in original order (if there are replacements)
 306                                 if(count($varCategories)>0){
 307                                         $newCats = array();
 308                                         $originalCats = $output->getCategories();
 309                                         foreach($originalCats as $cat => $sortkey){
 310                                                 // make the replacement
 311                                                 if( array_key_exists($cat,$varCategories) )
 312                                                         $newCats[$varCategories[$cat]] = $sortkey;
 313                                                 else $newCats[$cat] = $sortkey;
 314                                         }
 315                                         $this->parent->mOutput->setCategoryLinks($newCats);
 316                                 }
 317                         }
 318                 }
 319
 320                 # Construct search and replace arrays
 321                 wfProfileIn( __METHOD__.'-construct' );
 322                 $replacePairs = array();
 323                 foreach ( $this->internals as $ns => $entries ) {
 324                         foreach ( $entries as $index => $entry ) {
 325                                 $pdbk = $entry['pdbk'];
 326                                 $title = $entry['title'];
 327                                 $query = isset( $entry['query'] ) ? $entry['query'] : '';
 328                                 $key = "$ns:$index";
 329                                 $searchkey = "<!--LINK $key-->";
 330                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
 331                                         $linkCache->addBadLinkObj( $title );
 332                                         $colours[$pdbk] = 'new';
 333                                         $output->addLink( $title, 0 );
 334                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
 335                                                                         $entry['text'],
 336                                                                         $query );
 337                                 } else {
 338                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
 339                                                                         $entry['text'],
 340                                                                         $query );
 341                                 }
 342                         }
 343                 }
 344                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 345                 wfProfileOut( __METHOD__.'-construct' );
 346
 347                 # Do the thing
 348                 wfProfileIn( __METHOD__.'-replace' );
 349                 $text = preg_replace_callback(
 350                         '/(<!--LINK .*?-->)/',
 351                         $replacer->cb(),
 352                         $text);
 353
 354                 wfProfileOut( __METHOD__.'-replace' );
 355                 wfProfileOut( __METHOD__ );
 356         }
 357
 358         /**
 359          * Replace interwiki links
 360          */
 361         protected function replaceInterwiki( &$text ) {
 362                 if ( empty( $this->interwikis ) ) {
 363                         return;
 364                 }
 365
 366                 wfProfileIn( __METHOD__ );
 367                 # Make interwiki link HTML
 368                 $sk = $this->parent->getOptions()->getSkin();
 369                 $replacePairs = array();
 370                 foreach( $this->interwikis as $key => $link ) {
 371                         $replacePairs[$key] = $sk->link( $link['title'], $link['text'] );
 372                 }
 373                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 374
 375                 $text = preg_replace_callback(
 376                         '/<!--IWLINK (.*?)-->/',
 377                         $replacer->cb(),
 378                         $text );
 379                 wfProfileOut( __METHOD__ );
 380         }
 381
 382         /**
 383          * Replace <!--LINK--> link placeholders with plain text of links
 384          * (not HTML-formatted).
 385          * @param string $text
 386          * @return string
 387          */
 388         function replaceText( $text ) {
 389                 wfProfileIn( __METHOD__ );
 390
 391                 $text = preg_replace_callback(
 392                         '/<!--(LINK|IWLINK) (.*?)-->/',
 393                         array( &$this, 'replaceTextCallback' ),
 394                         $text );
 395
 396                 wfProfileOut( __METHOD__ );
 397                 return $text;
 398         }
 399
 400         /**
 401          * @param array $matches
 402          * @return string
 403          * @private
 404          */
 405         function replaceTextCallback( $matches ) {
 406                 $type = $matches[1];
 407                 $key  = $matches[2];
 408                 if( $type == 'LINK' ) {
 409                         list( $ns, $index ) = explode( ':', $key, 2 );
 410                         if( isset( $this->internals[$ns][$index]['text'] ) ) {
 411                                 return $this->internals[$ns][$index]['text'];
 412                         }
 413                 } elseif( $type == 'IWLINK' ) {
 414                         if( isset( $this->interwikis[$key]['text'] ) ) {
 415                                 return $this->interwikis[$key]['text'];
 416                         }
 417                 }
 418                 return $matches[0];
 419         }
 420 }