includes/parser/ParserOutput.php

   1 <?php
   2 /**
   3  * Output of the PHP parser
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Parser
  22  */
  23
  24 /**
  25  * @todo document
  26  * @ingroup Parser
  27  */
  28 class CacheTime {
  29         var     $mVersion = Parser::VERSION,  # Compatibility check
  30                 $mCacheTime = '',             # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
  31                 $mCacheExpiry = null,         # Seconds after which the object should expire, use 0 for uncachable. Used in ParserCache.
  32                 $mContainsOldMagic;           # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
  33
  34         function getCacheTime()              { return $this->mCacheTime; }
  35
  36         function containsOldMagic()          { return $this->mContainsOldMagic; }
  37         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
  38
  39         /**
  40          * setCacheTime() sets the timestamp expressing when the page has been rendered.
  41          * This doesn not control expiry, see updateCacheExpiry() for that!
  42          * @param $t string
  43          * @return string
  44          */
  45         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
  46
  47         /**abstract
  48          * Sets the number of seconds after which this object should expire.
  49          * This value is used with the ParserCache.
  50          * If called with a value greater than the value provided at any previous call,
  51          * the new call has no effect. The value returned by getCacheExpiry is smaller
  52          * or equal to the smallest number that was provided as an argument to
  53          * updateCacheExpiry().
  54          *
  55          * @param $seconds number
  56          */
  57         function updateCacheExpiry( $seconds ) {
  58                 $seconds = (int)$seconds;
  59
  60                 if ( $this->mCacheExpiry === null || $this->mCacheExpiry > $seconds ) {
  61                         $this->mCacheExpiry = $seconds;
  62                 }
  63
  64                 // hack: set old-style marker for uncacheable entries.
  65                 if ( $this->mCacheExpiry !== null && $this->mCacheExpiry <= 0 ) {
  66                         $this->mCacheTime = -1;
  67                 }
  68         }
  69
  70         /**
  71          * Returns the number of seconds after which this object should expire.
  72          * This method is used by ParserCache to determine how long the ParserOutput can be cached.
  73          * The timestamp of expiry can be calculated by adding getCacheExpiry() to getCacheTime().
  74          * The value returned by getCacheExpiry is smaller or equal to the smallest number
  75          * that was provided to a call of updateCacheExpiry(), and smaller or equal to the
  76          * value of $wgParserCacheExpireTime.
  77          * @return int|mixed|null
  78          */
  79         function getCacheExpiry() {
  80                 global $wgParserCacheExpireTime;
  81
  82                 if ( $this->mCacheTime < 0 ) {
  83                         return 0;
  84                 } // old-style marker for "not cachable"
  85
  86                 $expire = $this->mCacheExpiry;
  87
  88                 if ( $expire === null ) {
  89                         $expire = $wgParserCacheExpireTime;
  90                 } else {
  91                         $expire = min( $expire, $wgParserCacheExpireTime );
  92                 }
  93
  94                 if( $this->containsOldMagic() ) { //compatibility hack
  95                         $expire = min( $expire, 3600 ); # 1 hour
  96                 }
  97
  98                 if ( $expire <= 0 ) {
  99                         return 0; // not cachable
 100                 } else {
 101                         return $expire;
 102                 }
 103         }
 104
 105         /**
 106          * @return bool
 107          */
 108         function isCacheable() {
 109                 return $this->getCacheExpiry() > 0;
 110         }
 111
 112         /**
 113          * Return true if this cached output object predates the global or
 114          * per-article cache invalidation timestamps, or if it comes from
 115          * an incompatible older version.
 116          *
 117          * @param $touched String: the affected article's last touched timestamp
 118          * @return Boolean
 119          */
 120         public function expired( $touched ) {
 121                 global $wgCacheEpoch;
 122                 return !$this->isCacheable() || // parser says it's uncacheable
 123                            $this->getCacheTime() < $touched ||
 124                            $this->getCacheTime() <= $wgCacheEpoch ||
 125                            $this->getCacheTime() < wfTimestamp( TS_MW, time() - $this->getCacheExpiry() ) || // expiry period has passed
 126                            !isset( $this->mVersion ) ||
 127                            version_compare( $this->mVersion, Parser::VERSION, "lt" );
 128         }
 129 }
 130
 131 class ParserOutput extends CacheTime {
 132         var $mText,                       # The output text
 133                 $mLanguageLinks,              # List of the full text of language links, in the order they appear
 134                 $mCategories,                 # Map of category names to sort keys
 135                 $mTitleText,                  # title text of the chosen language variant
 136                 $mLinks = array(),            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
 137                 $mTemplates = array(),        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
 138                 $mTemplateIds = array(),      # 2-D map of NS/DBK to rev ID for the template references. ID=zero for broken.
 139                 $mImages = array(),           # DB keys of the images used, in the array key only
 140                 $mFileSearchOptions = array(), # DB keys of the images used mapped to sha1 and MW timestamp
 141                 $mExternalLinks = array(),    # External link URLs, in the key only
 142                 $mInterwikiLinks = array(),   # 2-D map of prefix/DBK (in keys only) for the inline interwiki links in the document.
 143                 $mNewSection = false,         # Show a new section link?
 144                 $mHideNewSection = false,     # Hide the new section link?
 145                 $mNoGallery = false,          # No gallery on category page? (__NOGALLERY__)
 146                 $mHeadItems = array(),        # Items to put in the <head> section
 147                 $mModules = array(),          # Modules to be loaded by the resource loader
 148                 $mModuleScripts = array(),    # Modules of which only the JS will be loaded by the resource loader
 149                 $mModuleStyles = array(),     # Modules of which only the CSSS will be loaded by the resource loader
 150                 $mModuleMessages = array(),   # Modules of which only the messages will be loaded by the resource loader
 151                 $mOutputHooks = array(),      # Hook tags as per $wgParserOutputHooks
 152                 $mWarnings = array(),         # Warning text to be returned to the user. Wikitext formatted, in the key only
 153                 $mSections = array(),         # Table of contents
 154                 $mEditSectionTokens = false,  # prefix/suffix markers if edit sections were output as tokens
 155                 $mProperties = array(),       # Name/value pairs to be cached in the DB
 156                 $mTOCHTML = '',               # HTML of the TOC
 157                 $mTimestamp;                  # Timestamp of the revision
 158
 159         /**
 160          * 'index' or 'noindex'?  Any other value will result in no change.
 161          *
 162          * @var string
 163          */
 164         protected $mIndexPolicy = '';
 165
 166         /**
 167          * List of ParserOptions (stored in the keys)
 168          *
 169          * @var array
 170          */
 171         protected $mAccessedOptions = array();
 172
 173         /**
 174          * List of instances of SecondaryDataObject(), used to cause some information extracted from the page in a custom place.
 175          * @since WD.1
 176          * @var array of SecondaryDataObject
 177          */
 178         protected $mSecondaryDataUpdates = array();
 179
 180         const EDITSECTION_REGEX = '#<(?:mw:)?editsection page="(.*?)" section="(.*?)"(?:/>|>(.*?)(</(?:mw:)?editsection>))#';
 181
 182         function __construct( $text = '', $languageLinks = array(), $categoryLinks = array(),
 183                 $containsOldMagic = false, $titletext = '' )
 184         {
 185                 $this->mText = $text;
 186                 $this->mLanguageLinks = $languageLinks;
 187                 $this->mCategories = $categoryLinks;
 188                 $this->mContainsOldMagic = $containsOldMagic;
 189                 $this->mTitleText = $titletext;
 190         }
 191
 192         function getText() {
 193                 if ( $this->mEditSectionTokens ) {
 194                         return preg_replace_callback( ParserOutput::EDITSECTION_REGEX,
 195                                 array( &$this, 'replaceEditSectionLinksCallback' ), $this->mText );
 196                 }
 197                 return preg_replace( ParserOutput::EDITSECTION_REGEX, '', $this->mText );
 198         }
 199
 200         /**
 201          * callback used by getText to replace editsection tokens
 202          * @private
 203          * @return mixed
 204          */
 205         function replaceEditSectionLinksCallback( $m ) {
 206                 global $wgOut, $wgLang;
 207                 $args = array(
 208                         htmlspecialchars_decode($m[1]),
 209                         htmlspecialchars_decode($m[2]),
 210                         isset($m[4]) ? $m[3] : null,
 211                 );
 212                 $args[0] = Title::newFromText( $args[0] );
 213                 if ( !is_object($args[0]) ) {
 214                         throw new MWException("Bad parser output text.");
 215                 }
 216                 $args[] = $wgLang->getCode();
 217                 $skin = $wgOut->getSkin();
 218                 return call_user_func_array( array( $skin, 'doEditSectionLink' ), $args );
 219         }
 220
 221         function &getLanguageLinks()         { return $this->mLanguageLinks; }
 222         function getInterwikiLinks()         { return $this->mInterwikiLinks; }
 223         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
 224         function &getCategories()            { return $this->mCategories; }
 225         function getTitleText()              { return $this->mTitleText; }
 226         function getSections()               { return $this->mSections; }
 227         function getEditSectionTokens()      { return $this->mEditSectionTokens; }
 228         function &getLinks()                 { return $this->mLinks; }
 229         function &getTemplates()             { return $this->mTemplates; }
 230         function &getTemplateIds()           { return $this->mTemplateIds; }
 231         function &getImages()                { return $this->mImages; }
 232         function &getFileSearchOptions()     { return $this->mFileSearchOptions; }
 233         function &getExternalLinks()         { return $this->mExternalLinks; }
 234         function getNoGallery()              { return $this->mNoGallery; }
 235         function getHeadItems()              { return $this->mHeadItems; }
 236         function getModules()                { return $this->mModules; }
 237         function getModuleScripts()          { return $this->mModuleScripts; }
 238         function getModuleStyles()           { return $this->mModuleStyles; }
 239         function getModuleMessages()         { return $this->mModuleMessages; }
 240         function getOutputHooks()            { return (array)$this->mOutputHooks; }
 241         function getWarnings()               { return array_keys( $this->mWarnings ); }
 242         function getIndexPolicy()            { return $this->mIndexPolicy; }
 243         function getTOCHTML()                { return $this->mTOCHTML; }
 244         function getTimestamp()              { return $this->mTimestamp; }
 245
 246         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
 247         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
 248         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
 249
 250         function setTitleText( $t )          { return wfSetVar( $this->mTitleText, $t ); }
 251         function setSections( $toc )         { return wfSetVar( $this->mSections, $toc ); }
 252         function setEditSectionTokens( $t )  { return wfSetVar( $this->mEditSectionTokens, $t ); }
 253         function setIndexPolicy( $policy )   { return wfSetVar( $this->mIndexPolicy, $policy ); }
 254         function setTOCHTML( $tochtml )      { return wfSetVar( $this->mTOCHTML, $tochtml ); }
 255         function setTimestamp( $timestamp )  { return wfSetVar( $this->mTimestamp, $timestamp ); }
 256
 257         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
 258         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
 259         function addWarning( $s )            { $this->mWarnings[$s] = 1; }
 260
 261         function addOutputHook( $hook, $data = false ) {
 262                 $this->mOutputHooks[] = array( $hook, $data );
 263         }
 264
 265         function setNewSection( $value ) {
 266                 $this->mNewSection = (bool)$value;
 267         }
 268         function hideNewSection ( $value ) {
 269                 $this->mHideNewSection = (bool)$value;
 270         }
 271         function getHideNewSection () {
 272                 return (bool)$this->mHideNewSection;
 273         }
 274         function getNewSection() {
 275                 return (bool)$this->mNewSection;
 276         }
 277
 278         function addExternalLink( $url ) {
 279                 # We don't register links pointing to our own server, unless... :-)
 280                 global $wgServer, $wgRegisterInternalExternals;
 281                 if( $wgRegisterInternalExternals or stripos($url,$wgServer.'/')!==0)
 282                         $this->mExternalLinks[$url] = 1;
 283         }
 284
 285         /**
 286          * Record a local or interwiki inline link for saving in future link tables.
 287          *
 288          * @param $title Title object
 289          * @param $id Mixed: optional known page_id so we can skip the lookup
 290          */
 291         function addLink( $title, $id = null ) {
 292                 if ( $title->isExternal() ) {
 293                         // Don't record interwikis in pagelinks
 294                         $this->addInterwikiLink( $title );
 295                         return;
 296                 }
 297                 $ns = $title->getNamespace();
 298                 $dbk = $title->getDBkey();
 299                 if ( $ns == NS_MEDIA ) {
 300                         // Normalize this pseudo-alias if it makes it down here...
 301                         $ns = NS_FILE;
 302                 } elseif( $ns == NS_SPECIAL ) {
 303                         // We don't record Special: links currently
 304                         // It might actually be wise to, but we'd need to do some normalization.
 305                         return;
 306                 } elseif( $dbk === '' ) {
 307                         // Don't record self links -  [[#Foo]]
 308                         return;
 309                 }
 310                 if ( !isset( $this->mLinks[$ns] ) ) {
 311                         $this->mLinks[$ns] = array();
 312                 }
 313                 if ( is_null( $id ) ) {
 314                         $id = $title->getArticleID();
 315                 }
 316                 $this->mLinks[$ns][$dbk] = $id;
 317         }
 318
 319         /**
 320          * Register a file dependency for this output
 321          * @param $name string Title dbKey
 322          * @param $timestamp string MW timestamp of file creation (or false if non-existing)
 323          * @param $sha1 string base 36 SHA-1 of file (or false if non-existing)
 324          * @return void
 325          */
 326         function addImage( $name, $timestamp = null, $sha1 = null ) {
 327                 $this->mImages[$name] = 1;
 328                 if ( $timestamp !== null && $sha1 !== null ) {
 329                         $this->mFileSearchOptions[$name] = array( 'time' => $timestamp, 'sha1' => $sha1 );
 330                 }
 331         }
 332
 333         /**
 334          * Register a template dependency for this output
 335          * @param $title Title
 336          * @param $page_id
 337          * @param $rev_id
 338          * @return void
 339          */
 340         function addTemplate( $title, $page_id, $rev_id ) {
 341                 $ns = $title->getNamespace();
 342                 $dbk = $title->getDBkey();
 343                 if ( !isset( $this->mTemplates[$ns] ) ) {
 344                         $this->mTemplates[$ns] = array();
 345                 }
 346                 $this->mTemplates[$ns][$dbk] = $page_id;
 347                 if ( !isset( $this->mTemplateIds[$ns] ) ) {
 348                         $this->mTemplateIds[$ns] = array();
 349                 }
 350                 $this->mTemplateIds[$ns][$dbk] = $rev_id; // For versioning
 351         }
 352
 353         /**
 354          * @param $title Title object, must be an interwiki link
 355          * @throws MWException if given invalid input
 356          */
 357         function addInterwikiLink( $title ) {
 358                 $prefix = $title->getInterwiki();
 359                 if( $prefix == '' ) {
 360                         throw new MWException( 'Non-interwiki link passed, internal parser error.' );
 361                 }
 362                 if (!isset($this->mInterwikiLinks[$prefix])) {
 363                         $this->mInterwikiLinks[$prefix] = array();
 364                 }
 365                 $this->mInterwikiLinks[$prefix][$title->getDBkey()] = 1;
 366         }
 367
 368         /**
 369          * Add some text to the <head>.
 370          * If $tag is set, the section with that tag will only be included once
 371          * in a given page.
 372          */
 373         function addHeadItem( $section, $tag = false ) {
 374                 if ( $tag !== false ) {
 375                         $this->mHeadItems[$tag] = $section;
 376                 } else {
 377                         $this->mHeadItems[] = $section;
 378                 }
 379         }
 380
 381         public function addModules( $modules ) {
 382                 $this->mModules = array_merge( $this->mModules, (array) $modules );
 383         }
 384
 385         public function addModuleScripts( $modules ) {
 386                 $this->mModuleScripts = array_merge( $this->mModuleScripts, (array)$modules );
 387         }
 388
 389         public function addModuleStyles( $modules ) {
 390                 $this->mModuleStyles = array_merge( $this->mModuleStyles, (array)$modules );
 391         }
 392
 393         public function addModuleMessages( $modules ) {
 394                 $this->mModuleMessages = array_merge( $this->mModuleMessages, (array)$modules );
 395         }
 396
 397         /**
 398          * Copy items from the OutputPage object into this one
 399          *
 400          * @param $out OutputPage object
 401          */
 402         public function addOutputPageMetadata( OutputPage $out ) {
 403                 $this->addModules( $out->getModules() );
 404                 $this->addModuleScripts( $out->getModuleScripts() );
 405                 $this->addModuleStyles( $out->getModuleStyles() );
 406                 $this->addModuleMessages( $out->getModuleMessages() );
 407
 408                 $this->mHeadItems = array_merge( $this->mHeadItems, $out->getHeadItemsArray() );
 409         }
 410
 411         /**
 412          * Override the title to be used for display
 413          * -- this is assumed to have been validated
 414          * (check equal normalisation, etc.)
 415          *
 416          * @param $text String: desired title text
 417          */
 418         public function setDisplayTitle( $text ) {
 419                 $this->setTitleText( $text );
 420                 $this->setProperty( 'displaytitle', $text );
 421         }
 422
 423         /**
 424          * Get the title to be used for display
 425          *
 426          * @return String
 427          */
 428         public function getDisplayTitle() {
 429                 $t = $this->getTitleText();
 430                 if( $t === '' ) {
 431                         return false;
 432                 }
 433                 return $t;
 434         }
 435
 436         /**
 437          * Fairly generic flag setter thingy.
 438          */
 439         public function setFlag( $flag ) {
 440                 $this->mFlags[$flag] = true;
 441         }
 442
 443         public function getFlag( $flag ) {
 444                 return isset( $this->mFlags[$flag] );
 445         }
 446
 447         /**
 448          * Set a property to be cached in the DB
 449          */
 450         public function setProperty( $name, $value ) {
 451                 $this->mProperties[$name] = $value;
 452         }
 453
 454         public function getProperty( $name ){
 455                 return isset( $this->mProperties[$name] ) ? $this->mProperties[$name] : false;
 456         }
 457
 458         public function getProperties() {
 459                 if ( !isset( $this->mProperties ) ) {
 460                         $this->mProperties = array();
 461                 }
 462                 return $this->mProperties;
 463         }
 464
 465
 466         /**
 467          * Returns the options from its ParserOptions which have been taken
 468          * into account to produce this output or false if not available.
 469          * @return mixed Array
 470          */
 471          public function getUsedOptions() {
 472                 if ( !isset( $this->mAccessedOptions ) ) {
 473                         return array();
 474                 }
 475                 return array_keys( $this->mAccessedOptions );
 476          }
 477
 478          /**
 479           * Callback passed by the Parser to the ParserOptions to keep track of which options are used.
 480           * @access private
 481           */
 482          function recordOption( $option ) {
 483                  $this->mAccessedOptions[$option] = true;
 484          }
 485
 486     /**
 487      * Adds an update job to the output. Any update jobs added to the output will eventually bexecuted in order to
 488      * store any secondary information extracted from the page's content.
 489      *
 490          * @since WD.1
 491          *
 492      * @param SecondaryDataUpdate $update
 493      */
 494     public function addSecondaryDataUpdate( SecondaryDataUpdate $update ) {
 495         $this->mSecondaryDataUpdates[] = $update;
 496     }
 497
 498     /**
 499      * Returns any SecondaryDataUpdate jobs to be executed in order to store secondary information
 500      * extracted from the page's content, includingt a LinksUpdate object for all links stopred in
 501      * this ParserOutput object.
 502      *
 503          * @since WD.1
 504          *
 505      * @param $title Title of the page we're updating. If not given, a title object will be created based on $this->getTitleText()
 506      * @param $recursive Boolean: queue jobs for recursive updates?
 507      *
 508      * @return array an array of instances of SecondaryDataUpdate
 509      */
 510     public function getSecondaryDataUpdates( Title $title = null, $recursive = true ) {
 511         if ( is_null( $title ) ) {
 512             $title = Title::newFromText( $this->getTitleText() );
 513         }
 514
 515         return array_merge(
 516                         $this->mSecondaryDataUpdates,
 517                         array( new LinksUpdate( $title, $this, $recursive ) )
 518                 );
 519     }
 520 }