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