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->getOption('stubthreshold');
 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                 # Generate query
 148                 $query = false;
 149                 $current = null;
 150                 foreach ( $this->internals as $ns => $entries ) {
 151                         foreach ( $entries as $index => $entry ) {
 152                                 $key =  "$ns:$index";
 153                                 $title = $entry['title'];
 154                                 $pdbk = $entry['pdbk'];
 155
 156                                 # Skip invalid entries.
 157                                 # Result will be ugly, but prevents crash.
 158                                 if ( is_null( $title ) ) {
 159                                         continue;
 160                                 }
 161
 162                                 # Check if it's a static known link, e.g. interwiki
 163                                 if ( $title->isAlwaysKnown() ) {
 164                                         $colours[$pdbk] = '';
 165                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
 166                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 167                                         $output->addLink( $title, $id );
 168                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
 169                                         $colours[$pdbk] = 'new';
 170                                 } else {
 171                                         # Not in the link cache, add it to the query
 172                                         if ( !isset( $current ) ) {
 173                                                 $current = $ns;
 174                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 175                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
 176                                         } elseif ( $current != $ns ) {
 177                                                 $current = $ns;
 178                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
 179                                         } else {
 180                                                 $query .= ', ';
 181                                         }
 182
 183                                         $query .= $dbr->addQuotes( $title->getDBkey() );
 184                                 }
 185                         }
 186                 }
 187                 if ( $query ) {
 188                         $query .= '))';
 189
 190                         $res = $dbr->query( $query, __METHOD__ );
 191
 192                         # Fetch data and form into an associative array
 193                         # non-existent = broken
 194                         $linkcolour_ids = array();
 195                         while ( $s = $dbr->fetchObject($res) ) {
 196                                 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
 197                                 $pdbk = $title->getPrefixedDBkey();
 198                                 $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
 199                                 $output->addLink( $title, $s->page_id );
 200                                 # FIXME: convoluted data flow
 201                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 202                                 # Use formal parameters instead
 203                                 $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 204                                 //add id to the extension todolist
 205                                 $linkcolour_ids[$s->page_id] = $pdbk;
 206                         }
 207                         unset( $res );
 208                         //pass an array of page_ids to an extension
 209                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 210                 }
 211                 wfProfileOut( __METHOD__.'-check' );
 212
 213                 # Do a second query for different language variants of links and categories
 214                 if($wgContLang->hasVariants()) {
 215                         $this->doVariants( $colours );
 216                 }
 217
 218                 # Construct search and replace arrays
 219                 wfProfileIn( __METHOD__.'-construct' );
 220                 $replacePairs = array();
 221                 foreach ( $this->internals as $ns => $entries ) {
 222                         foreach ( $entries as $index => $entry ) {
 223                                 $pdbk = $entry['pdbk'];
 224                                 $title = $entry['title'];
 225                                 $query = isset( $entry['query'] ) ? $entry['query'] : '';
 226                                 $key = "$ns:$index";
 227                                 $searchkey = "<!--LINK $key-->";
 228                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
 229                                         $linkCache->addBadLinkObj( $title );
 230                                         $colours[$pdbk] = 'new';
 231                                         $output->addLink( $title, 0 );
 232                                         // FIXME: replace deprecated makeBrokenLinkObj() by link()
 233                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
 234                                                                         $entry['text'],
 235                                                                         $query );
 236                                 } else {
 237                                         // FIXME: replace deprecated makeColouredLinkObj() by link()
 238                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
 239                                                                         $entry['text'],
 240                                                                         $query );
 241                                 }
 242                         }
 243                 }
 244                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 245                 wfProfileOut( __METHOD__.'-construct' );
 246
 247                 # Do the thing
 248                 wfProfileIn( __METHOD__.'-replace' );
 249                 $text = preg_replace_callback(
 250                         '/(<!--LINK .*?-->)/',
 251                         $replacer->cb(),
 252                         $text);
 253
 254                 wfProfileOut( __METHOD__.'-replace' );
 255                 wfProfileOut( __METHOD__ );
 256         }
 257
 258         /**
 259          * Replace interwiki links
 260          */
 261         protected function replaceInterwiki( &$text ) {
 262                 if ( empty( $this->interwikis ) ) {
 263                         return;
 264                 }
 265
 266                 wfProfileIn( __METHOD__ );
 267                 # Make interwiki link HTML
 268                 $sk = $this->parent->getOptions()->getSkin();
 269                 $replacePairs = array();
 270                 foreach( $this->interwikis as $key => $link ) {
 271                         $replacePairs[$key] = $sk->link( $link['title'], $link['text'] );
 272                 }
 273                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 274
 275                 $text = preg_replace_callback(
 276                         '/<!--IWLINK (.*?)-->/',
 277                         $replacer->cb(),
 278                         $text );
 279                 wfProfileOut( __METHOD__ );
 280         }
 281
 282         /**
 283          * Modify $this->internals and $colours according to language variant linking rules
 284          */
 285         protected function doVariants( &$colours ) {
 286                 global $wgContLang;
 287                 $linkBatch = new LinkBatch();
 288                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
 289                 $output = $this->parent->getOutput();
 290                 $linkCache = LinkCache::singleton();
 291                 $sk = $this->parent->getOptions()->getSkin();
 292                 $threshold = $this->getStubThreshold();
 293
 294                 // Add variants of links to link batch
 295                 foreach ( $this->internals as $ns => $entries ) {
 296                         foreach ( $entries as $index => $entry ) {
 297                                 $key = "$ns:$index";
 298                                 $pdbk = $entry['pdbk'];
 299                                 $title = $entry['title'];
 300                                 $titleText = $title->getText();
 301
 302                                 // generate all variants of the link title text
 303                                 $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
 304
 305                                 // if link was not found (in first query), add all variants to query
 306                                 if ( !isset($colours[$pdbk]) ){
 307                                         foreach($allTextVariants as $textVariant){
 308                                                 if($textVariant != $titleText){
 309                                                         $variantTitle = Title::makeTitle( $ns, $textVariant );
 310                                                         if(is_null($variantTitle)) continue;
 311                                                         $linkBatch->addObj( $variantTitle );
 312                                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
 313                                                 }
 314                                         }
 315                                 }
 316                         }
 317                 }
 318
 319                 // process categories, check if a category exists in some variant
 320                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 321                 $varCategories = array(); // category replacements oldDBkey => newDBkey
 322                 foreach( $output->getCategoryLinks() as $category ){
 323                         $variants = $wgContLang->convertLinkToAllVariants($category);
 324                         foreach($variants as $variant){
 325                                 if($variant != $category){
 326                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 327                                         if(is_null($variantTitle)) continue;
 328                                         $linkBatch->addObj( $variantTitle );
 329                                         $categoryMap[$variant] = $category;
 330                                 }
 331                         }
 332                 }
 333
 334
 335                 if(!$linkBatch->isEmpty()){
 336                         // construct query
 337                         $dbr = wfGetDB( DB_SLAVE );
 338                         $page = $dbr->tableName( 'page' );
 339                         $titleClause = $linkBatch->constructSet('page', $dbr);
 340                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 341                         $variantQuery .= " FROM $page WHERE $titleClause";
 342                         $varRes = $dbr->query( $variantQuery, __METHOD__ );
 343                         $linkcolour_ids = array();
 344
 345                         // for each found variants, figure out link holders and replace
 346                         while ( $s = $dbr->fetchObject($varRes) ) {
 347
 348                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 349                                 $varPdbk = $variantTitle->getPrefixedDBkey();
 350                                 $vardbk = $variantTitle->getDBkey();
 351
 352                                 $holderKeys = array();
 353                                 if(isset($variantMap[$varPdbk])){
 354                                         $holderKeys = $variantMap[$varPdbk];
 355                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
 356                                         $output->addLink( $variantTitle, $s->page_id );
 357                                 }
 358
 359                                 // loop over link holders
 360                                 foreach($holderKeys as $key){
 361                                         list( $ns, $index ) = explode( ':', $key, 2 );
 362                                         $entry =& $this->internals[$ns][$index];
 363                                         $pdbk = $entry['pdbk'];
 364
 365                                         if(!isset($colours[$pdbk])){
 366                                                 // found link in some of the variants, replace the link holder data
 367                                                 $entry['title'] = $variantTitle;
 368                                                 $entry['pdbk'] = $varPdbk;
 369
 370                                                 // set pdbk and colour
 371                                                 # FIXME: convoluted data flow
 372                                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 373                                                 # Use formal parameters instead
 374                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
 375                                                 $linkcolour_ids[$s->page_id] = $pdbk;
 376                                         }
 377                                 }
 378
 379                                 // check if the object is a variant of a category
 380                                 if(isset($categoryMap[$vardbk])){
 381                                         $oldkey = $categoryMap[$vardbk];
 382                                         if($oldkey != $vardbk)
 383                                                 $varCategories[$oldkey]=$vardbk;
 384                                 }
 385                         }
 386                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 387
 388                         // rebuild the categories in original order (if there are replacements)
 389                         if(count($varCategories)>0){
 390                                 $newCats = array();
 391                                 $originalCats = $output->getCategories();
 392                                 foreach($originalCats as $cat => $sortkey){
 393                                         // make the replacement
 394                                         if( array_key_exists($cat,$varCategories) )
 395                                                 $newCats[$varCategories[$cat]] = $sortkey;
 396                                         else $newCats[$cat] = $sortkey;
 397                                 }
 398                                 $output->setCategoryLinks($newCats);
 399                         }
 400                 }
 401         }
 402
 403         /**
 404          * Replace <!--LINK--> link placeholders with plain text of links
 405          * (not HTML-formatted).
 406          * @param string $text
 407          * @return string
 408          */
 409         function replaceText( $text ) {
 410                 wfProfileIn( __METHOD__ );
 411
 412                 $text = preg_replace_callback(
 413                         '/<!--(LINK|IWLINK) (.*?)-->/',
 414                         array( &$this, 'replaceTextCallback' ),
 415                         $text );
 416
 417                 wfProfileOut( __METHOD__ );
 418                 return $text;
 419         }
 420
 421         /**
 422          * @param array $matches
 423          * @return string
 424          * @private
 425          */
 426         function replaceTextCallback( $matches ) {
 427                 $type = $matches[1];
 428                 $key  = $matches[2];
 429                 if( $type == 'LINK' ) {
 430                         list( $ns, $index ) = explode( ':', $key, 2 );
 431                         if( isset( $this->internals[$ns][$index]['text'] ) ) {
 432                                 return $this->internals[$ns][$index]['text'];
 433                         }
 434                 } elseif( $type == 'IWLINK' ) {
 435                         if( isset( $this->interwikis[$key]['text'] ) ) {
 436                                 return $this->interwikis[$key]['text'];
 437                         }
 438                 }
 439                 return $matches[0];
 440         }
 441 }