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 $this->stubThreshold;
 105         }
 106
 107         /**
 108          * Replace <!--LINK--> link placeholders with actual links, in the buffer
 109          * Placeholders created in Skin::makeLinkObj()
 110          * Returns an array of link CSS classes, indexed by PDBK.
 111          */
 112         function replace( &$text ) {
 113                 wfProfileIn( __METHOD__ );
 114
 115                 $colours = $this->replaceInternal( $text );
 116                 $this->replaceInterwiki( $text );
 117
 118                 wfProfileOut( __METHOD__ );
 119                 return $colours;
 120         }
 121
 122         /**
 123          * Replace internal links
 124          */
 125         protected function replaceInternal( &$text ) {
 126                 if ( !$this->internals ) {
 127                         return;
 128                 }
 129
 130                 wfProfileIn( __METHOD__ );
 131                 global $wgContLang;
 132
 133                 $colours = array();
 134                 $sk = $this->parent->getOptions()->getSkin();
 135                 $linkCache = LinkCache::singleton();
 136                 $output = $this->parent->getOutput();
 137
 138                 wfProfileIn( __METHOD__.'-check' );
 139                 $dbr = wfGetDB( DB_SLAVE );
 140                 $page = $dbr->tableName( 'page' );
 141                 $threshold = $this->getStubThreshold();
 142
 143                 # Sort by namespace
 144                 ksort( $this->internals );
 145
 146                 # Generate query
 147                 $query = false;
 148                 $current = null;
 149                 foreach ( $this->internals as $ns => $entries ) {
 150                         foreach ( $entries as $index => $entry ) {
 151                                 $key =  "$ns:$index";
 152                                 $title = $entry['title'];
 153                                 $pdbk = $entry['pdbk'];
 154
 155                                 # Skip invalid entries.
 156                                 # Result will be ugly, but prevents crash.
 157                                 if ( is_null( $title ) ) {
 158                                         continue;
 159                                 }
 160
 161                                 # Check if it's a static known link, e.g. interwiki
 162                                 if ( $title->isAlwaysKnown() ) {
 163                                         $colours[$pdbk] = '';
 164                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
 165                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 166                                         $output->addLink( $title, $id );
 167                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
 168                                         $colours[$pdbk] = 'new';
 169                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
 170                                         $colours[$pdbk] = 'new';
 171                                 } else {
 172                                         # Not in the link cache, add it to the query
 173                                         if ( !isset( $current ) ) {
 174                                                 $current = $ns;
 175                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 176                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
 177                                         } elseif ( $current != $ns ) {
 178                                                 $current = $ns;
 179                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
 180                                         } else {
 181                                                 $query .= ', ';
 182                                         }
 183
 184                                         $query .= $dbr->addQuotes( $title->getDBkey() );
 185                                 }
 186                         }
 187                 }
 188                 if ( $query ) {
 189                         $query .= '))';
 190
 191                         $res = $dbr->query( $query, __METHOD__ );
 192
 193                         # Fetch data and form into an associative array
 194                         # non-existent = broken
 195                         $linkcolour_ids = array();
 196                         while ( $s = $dbr->fetchObject($res) ) {
 197                                 $title = Title::makeTitle( $s->page_namespace, $s->page_title );
 198                                 $pdbk = $title->getPrefixedDBkey();
 199                                 $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
 200                                 $output->addLink( $title, $s->page_id );
 201                                 # FIXME: convoluted data flow
 202                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 203                                 # Use formal parameters instead
 204                                 $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
 205                                 //add id to the extension todolist
 206                                 $linkcolour_ids[$s->page_id] = $pdbk;
 207                         }
 208                         unset( $res );
 209                         //pass an array of page_ids to an extension
 210                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 211                 }
 212                 wfProfileOut( __METHOD__.'-check' );
 213
 214                 # Do a second query for different language variants of links and categories
 215                 if($wgContLang->hasVariants()) {
 216                         $this->doVariants( $colours );
 217                 }
 218
 219                 # Construct search and replace arrays
 220                 wfProfileIn( __METHOD__.'-construct' );
 221                 $replacePairs = array();
 222                 foreach ( $this->internals as $ns => $entries ) {
 223                         foreach ( $entries as $index => $entry ) {
 224                                 $pdbk = $entry['pdbk'];
 225                                 $title = $entry['title'];
 226                                 $query = isset( $entry['query'] ) ? $entry['query'] : '';
 227                                 $key = "$ns:$index";
 228                                 $searchkey = "<!--LINK $key-->";
 229                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
 230                                         $linkCache->addBadLinkObj( $title );
 231                                         $colours[$pdbk] = 'new';
 232                                         $output->addLink( $title, 0 );
 233                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
 234                                                                         $entry['text'],
 235                                                                         $query );
 236                                 } else {
 237                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
 238                                                                         $entry['text'],
 239                                                                         $query );
 240                                 }
 241                         }
 242                 }
 243                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 244                 wfProfileOut( __METHOD__.'-construct' );
 245
 246                 # Do the thing
 247                 wfProfileIn( __METHOD__.'-replace' );
 248                 $text = preg_replace_callback(
 249                         '/(<!--LINK .*?-->)/',
 250                         $replacer->cb(),
 251                         $text);
 252
 253                 wfProfileOut( __METHOD__.'-replace' );
 254                 wfProfileOut( __METHOD__ );
 255         }
 256
 257         /**
 258          * Replace interwiki links
 259          */
 260         protected function replaceInterwiki( &$text ) {
 261                 if ( empty( $this->interwikis ) ) {
 262                         return;
 263                 }
 264
 265                 wfProfileIn( __METHOD__ );
 266                 # Make interwiki link HTML
 267                 $sk = $this->parent->getOptions()->getSkin();
 268                 $replacePairs = array();
 269                 foreach( $this->interwikis as $key => $link ) {
 270                         $replacePairs[$key] = $sk->link( $link['title'], $link['text'] );
 271                 }
 272                 $replacer = new HashtableReplacer( $replacePairs, 1 );
 273
 274                 $text = preg_replace_callback(
 275                         '/<!--IWLINK (.*?)-->/',
 276                         $replacer->cb(),
 277                         $text );
 278                 wfProfileOut( __METHOD__ );
 279         }
 280
 281         /**
 282          * Modify $this->internals and $colours according to language variant linking rules
 283          */
 284         protected function doVariants( &$colours ) {
 285                 global $wgContLang;
 286                 $linkBatch = new LinkBatch();
 287                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
 288                 $output = $this->parent->getOutput();
 289                 $linkCache = LinkCache::singleton();
 290                 $sk = $this->parent->getOptions()->getSkin();
 291                 $threshold = $this->getStubThreshold();
 292
 293                 // Add variants of links to link batch
 294                 foreach ( $this->internals as $ns => $entries ) {
 295                         foreach ( $entries as $index => $entry ) {
 296                                 $key = "$ns:$index";
 297                                 $pdbk = $entry['pdbk'];
 298                                 $title = $entry['title'];
 299                                 $titleText = $title->getText();
 300
 301                                 // generate all variants of the link title text
 302                                 $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
 303
 304                                 // if link was not found (in first query), add all variants to query
 305                                 if ( !isset($colours[$pdbk]) ){
 306                                         foreach($allTextVariants as $textVariant){
 307                                                 if($textVariant != $titleText){
 308                                                         $variantTitle = Title::makeTitle( $ns, $textVariant );
 309                                                         if(is_null($variantTitle)) continue;
 310                                                         $linkBatch->addObj( $variantTitle );
 311                                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
 312                                                 }
 313                                         }
 314                                 }
 315                         }
 316                 }
 317
 318                 // process categories, check if a category exists in some variant
 319                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 320                 $varCategories = array(); // category replacements oldDBkey => newDBkey
 321                 foreach( $output->getCategoryLinks() as $category ){
 322                         $variants = $wgContLang->convertLinkToAllVariants($category);
 323                         foreach($variants as $variant){
 324                                 if($variant != $category){
 325                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 326                                         if(is_null($variantTitle)) continue;
 327                                         $linkBatch->addObj( $variantTitle );
 328                                         $categoryMap[$variant] = $category;
 329                                 }
 330                         }
 331                 }
 332
 333
 334                 if(!$linkBatch->isEmpty()){
 335                         // construct query
 336                         $dbr = wfGetDB( DB_SLAVE );
 337                         $page = $dbr->tableName( 'page' );
 338                         $titleClause = $linkBatch->constructSet('page', $dbr);
 339                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
 340                         $variantQuery .= " FROM $page WHERE $titleClause";
 341                         $varRes = $dbr->query( $variantQuery, __METHOD__ );
 342                         $linkcolour_ids = array();
 343
 344                         // for each found variants, figure out link holders and replace
 345                         while ( $s = $dbr->fetchObject($varRes) ) {
 346
 347                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 348                                 $varPdbk = $variantTitle->getPrefixedDBkey();
 349                                 $vardbk = $variantTitle->getDBkey();
 350
 351                                 $holderKeys = array();
 352                                 if(isset($variantMap[$varPdbk])){
 353                                         $holderKeys = $variantMap[$varPdbk];
 354                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
 355                                         $output->addLink( $variantTitle, $s->page_id );
 356                                 }
 357
 358                                 // loop over link holders
 359                                 foreach($holderKeys as $key){
 360                                         list( $ns, $index ) = explode( ':', $key, 2 );
 361                                         $entry =& $this->internals[$ns][$index];
 362                                         $pdbk = $entry['pdbk'];
 363
 364                                         if(!isset($colours[$pdbk])){
 365                                                 // found link in some of the variants, replace the link holder data
 366                                                 $entry['title'] = $variantTitle;
 367                                                 $entry['pdbk'] = $varPdbk;
 368
 369                                                 // set pdbk and colour
 370                                                 # FIXME: convoluted data flow
 371                                                 # The redirect status and length is passed to getLinkColour via the LinkCache
 372                                                 # Use formal parameters instead
 373                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
 374                                                 $linkcolour_ids[$s->page_id] = $pdbk;
 375                                         }
 376                                 }
 377
 378                                 // check if the object is a variant of a category
 379                                 if(isset($categoryMap[$vardbk])){
 380                                         $oldkey = $categoryMap[$vardbk];
 381                                         if($oldkey != $vardbk)
 382                                                 $varCategories[$oldkey]=$vardbk;
 383                                 }
 384                         }
 385                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
 386
 387                         // rebuild the categories in original order (if there are replacements)
 388                         if(count($varCategories)>0){
 389                                 $newCats = array();
 390                                 $originalCats = $output->getCategories();
 391                                 foreach($originalCats as $cat => $sortkey){
 392                                         // make the replacement
 393                                         if( array_key_exists($cat,$varCategories) )
 394                                                 $newCats[$varCategories[$cat]] = $sortkey;
 395                                         else $newCats[$cat] = $sortkey;
 396                                 }
 397                                 $output->setCategoryLinks($newCats);
 398                         }
 399                 }
 400         }
 401
 402         /**
 403          * Replace <!--LINK--> link placeholders with plain text of links
 404          * (not HTML-formatted).
 405          * @param string $text
 406          * @return string
 407          */
 408         function replaceText( $text ) {
 409                 wfProfileIn( __METHOD__ );
 410
 411                 $text = preg_replace_callback(
 412                         '/<!--(LINK|IWLINK) (.*?)-->/',
 413                         array( &$this, 'replaceTextCallback' ),
 414                         $text );
 415
 416                 wfProfileOut( __METHOD__ );
 417                 return $text;
 418         }
 419
 420         /**
 421          * @param array $matches
 422          * @return string
 423          * @private
 424          */
 425         function replaceTextCallback( $matches ) {
 426                 $type = $matches[1];
 427                 $key  = $matches[2];
 428                 if( $type == 'LINK' ) {
 429                         list( $ns, $index ) = explode( ':', $key, 2 );
 430                         if( isset( $this->internals[$ns][$index]['text'] ) ) {
 431                                 return $this->internals[$ns][$index]['text'];
 432                         }
 433                 } elseif( $type == 'IWLINK' ) {
 434                         if( isset( $this->interwikis[$key]['text'] ) ) {
 435                                 return $this->interwikis[$key]['text'];
 436                         }
 437                 }
 438                 return $matches[0];
 439         }
 440 }