includes/api/ApiPageSet.php

   1 <?php
   2 /**
   3  *
   4  *
   5  * Created on Sep 24, 2006
   6  *
   7  * Copyright © 2006, 2013 Yuri Astrakhan "<Firstname><Lastname>@gmail.com"
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License along
  20  * with this program; if not, write to the Free Software Foundation, Inc.,
  21  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  22  * http://www.gnu.org/copyleft/gpl.html
  23  *
  24  * @file
  25  */
  26
  27 /**
  28  * This class contains a list of pages that the client has requested.
  29  * Initially, when the client passes in titles=, pageids=, or revisions=
  30  * parameter, an instance of the ApiPageSet class will normalize titles,
  31  * determine if the pages/revisions exist, and prefetch any additional page
  32  * data requested.
  33  *
  34  * When a generator is used, the result of the generator will become the input
  35  * for the second instance of this class, and all subsequent actions will use
  36  * the second instance for all their work.
  37  *
  38  * @ingroup API
  39  * @since 1.21 derives from ApiBase instead of ApiQueryBase
  40  */
  41 class ApiPageSet extends ApiBase {
  42         /**
  43          * Constructor flag: The new instance of ApiPageSet will ignore the 'generator=' parameter
  44          * @since 1.21
  45          */
  46         const DISABLE_GENERATORS = 1;
  47
  48         private $mDbSource;
  49         private $mParams;
  50         private $mResolveRedirects;
  51         private $mConvertTitles;
  52         private $mAllowGenerator;
  53
  54         private $mAllPages = array(); // [ns][dbkey] => page_id or negative when missing
  55         private $mTitles = array();
  56         private $mGoodAndMissingPages = array(); // [ns][dbkey] => page_id or negative when missing
  57         private $mGoodPages = array(); // [ns][dbkey] => page_id
  58         private $mGoodTitles = array();
  59         private $mMissingPages = array(); // [ns][dbkey] => fake page_id
  60         private $mMissingTitles = array();
  61         private $mInvalidTitles = array();
  62         private $mMissingPageIDs = array();
  63         private $mRedirectTitles = array();
  64         private $mSpecialTitles = array();
  65         private $mNormalizedTitles = array();
  66         private $mInterwikiTitles = array();
  67         /** @var Title[] */
  68         private $mPendingRedirectIDs = array();
  69         private $mConvertedTitles = array();
  70         private $mGoodRevIDs = array();
  71         private $mLiveRevIDs = array();
  72         private $mDeletedRevIDs = array();
  73         private $mMissingRevIDs = array();
  74         private $mFakePageId = -1;
  75         private $mCacheMode = 'public';
  76         private $mRequestedPageFields = array();
  77         /** @var int */
  78         private $mDefaultNamespace = NS_MAIN;
  79
  80         /**
  81          * Add all items from $values into the result
  82          * @param array $result Output
  83          * @param array $values Values to add
  84          * @param string $flag The name of the boolean flag to mark this element
  85          * @param string $name If given, name of the value
  86          */
  87         private static function addValues( array &$result, $values, $flag = null, $name = null ) {
  88                 foreach ( $values as $val ) {
  89                         if ( $val instanceof Title ) {
  90                                 $v = array();
  91                                 ApiQueryBase::addTitleInfo( $v, $val );
  92                         } elseif ( $name !== null ) {
  93                                 $v = array( $name => $val );
  94                         } else {
  95                                 $v = $val;
  96                         }
  97                         if ( $flag !== null ) {
  98                                 $v[$flag] = '';
  99                         }
 100                         $result[] = $v;
 101                 }
 102         }
 103
 104         /**
 105          * @param ApiBase $dbSource Module implementing getDB().
 106          *        Allows PageSet to reuse existing db connection from the shared state like ApiQuery.
 107          * @param int $flags Zero or more flags like DISABLE_GENERATORS
 108          * @param int $defaultNamespace The namespace to use if none is specified by a prefix.
 109          * @since 1.21 accepts $flags instead of two boolean values
 110          */
 111         public function __construct( ApiBase $dbSource, $flags = 0, $defaultNamespace = NS_MAIN ) {
 112                 parent::__construct( $dbSource->getMain(), $dbSource->getModuleName() );
 113                 $this->mDbSource = $dbSource;
 114                 $this->mAllowGenerator = ( $flags & ApiPageSet::DISABLE_GENERATORS ) == 0;
 115                 $this->mDefaultNamespace = $defaultNamespace;
 116
 117                 $this->profileIn();
 118                 $this->mParams = $this->extractRequestParams();
 119                 $this->mResolveRedirects = $this->mParams['redirects'];
 120                 $this->mConvertTitles = $this->mParams['converttitles'];
 121                 $this->profileOut();
 122         }
 123
 124         /**
 125          * In case execute() is not called, call this method to mark all relevant parameters as used
 126          * This prevents unused parameters from being reported as warnings
 127          */
 128         public function executeDryRun() {
 129                 $this->executeInternal( true );
 130         }
 131
 132         /**
 133          * Populate the PageSet from the request parameters.
 134          */
 135         public function execute() {
 136                 $this->executeInternal( false );
 137         }
 138
 139         /**
 140          * Populate the PageSet from the request parameters.
 141          * @param bool $isDryRun If true, instantiates generator, but only to mark
 142          *    relevant parameters as used
 143          */
 144         private function executeInternal( $isDryRun ) {
 145                 $this->profileIn();
 146
 147                 $generatorName = $this->mAllowGenerator ? $this->mParams['generator'] : null;
 148                 if ( isset( $generatorName ) ) {
 149                         $dbSource = $this->mDbSource;
 150                         $isQuery = $dbSource instanceof ApiQuery;
 151                         if ( !$isQuery ) {
 152                                 // If the parent container of this pageset is not ApiQuery, we must create it to run generator
 153                                 $dbSource = $this->getMain()->getModuleManager()->getModule( 'query' );
 154                                 // Enable profiling for query module because it will be used for db sql profiling
 155                                 $dbSource->profileIn();
 156                         }
 157                         $generator = $dbSource->getModuleManager()->getModule( $generatorName, null, true );
 158                         if ( $generator === null ) {
 159                                 $this->dieUsage( 'Unknown generator=' . $generatorName, 'badgenerator' );
 160                         }
 161                         if ( !$generator instanceof ApiQueryGeneratorBase ) {
 162                                 $this->dieUsage( "Module $generatorName cannot be used as a generator", 'badgenerator' );
 163                         }
 164                         // Create a temporary pageset to store generator's output,
 165                         // add any additional fields generator may need, and execute pageset to populate titles/pageids
 166                         $tmpPageSet = new ApiPageSet( $dbSource, ApiPageSet::DISABLE_GENERATORS );
 167                         $generator->setGeneratorMode( $tmpPageSet );
 168                         $this->mCacheMode = $generator->getCacheMode( $generator->extractRequestParams() );
 169
 170                         if ( !$isDryRun ) {
 171                                 $generator->requestExtraData( $tmpPageSet );
 172                         }
 173                         $tmpPageSet->executeInternal( $isDryRun );
 174
 175                         // populate this pageset with the generator output
 176                         $this->profileOut();
 177                         $generator->profileIn();
 178
 179                         if ( !$isDryRun ) {
 180                                 $generator->executeGenerator( $this );
 181                                 wfRunHooks( 'APIQueryGeneratorAfterExecute', array( &$generator, &$this ) );
 182                         } else {
 183                                 // Prevent warnings from being reported on these parameters
 184                                 $main = $this->getMain();
 185                                 foreach ( $generator->extractRequestParams() as $paramName => $param ) {
 186                                         $main->getVal( $generator->encodeParamName( $paramName ) );
 187                                 }
 188                         }
 189                         $generator->profileOut();
 190                         $this->profileIn();
 191
 192                         if ( !$isDryRun ) {
 193                                 $this->resolvePendingRedirects();
 194                         }
 195
 196                         if ( !$isQuery ) {
 197                                 // If this pageset is not part of the query, we called profileIn() above
 198                                 $dbSource->profileOut();
 199                         }
 200                 } else {
 201                         // Only one of the titles/pageids/revids is allowed at the same time
 202                         $dataSource = null;
 203                         if ( isset( $this->mParams['titles'] ) ) {
 204                                 $dataSource = 'titles';
 205                         }
 206                         if ( isset( $this->mParams['pageids'] ) ) {
 207                                 if ( isset( $dataSource ) ) {
 208                                         $this->dieUsage( "Cannot use 'pageids' at the same time as '$dataSource'", 'multisource' );
 209                                 }
 210                                 $dataSource = 'pageids';
 211                         }
 212                         if ( isset( $this->mParams['revids'] ) ) {
 213                                 if ( isset( $dataSource ) ) {
 214                                         $this->dieUsage( "Cannot use 'revids' at the same time as '$dataSource'", 'multisource' );
 215                                 }
 216                                 $dataSource = 'revids';
 217                         }
 218
 219                         if ( !$isDryRun ) {
 220                                 // Populate page information with the original user input
 221                                 switch ( $dataSource ) {
 222                                         case 'titles':
 223                                                 $this->initFromTitles( $this->mParams['titles'] );
 224                                                 break;
 225                                         case 'pageids':
 226                                                 $this->initFromPageIds( $this->mParams['pageids'] );
 227                                                 break;
 228                                         case 'revids':
 229                                                 if ( $this->mResolveRedirects ) {
 230                                                         $this->setWarning( 'Redirect resolution cannot be used ' .
 231                                                                 'together with the revids= parameter. Any redirects ' .
 232                                                                 'the revids= point to have not been resolved.' );
 233                                                 }
 234                                                 $this->mResolveRedirects = false;
 235                                                 $this->initFromRevIDs( $this->mParams['revids'] );
 236                                                 break;
 237                                         default:
 238                                                 // Do nothing - some queries do not need any of the data sources.
 239                                                 break;
 240                                 }
 241                         }
 242                 }
 243                 $this->profileOut();
 244         }
 245
 246         /**
 247          * Check whether this PageSet is resolving redirects
 248          * @return bool
 249          */
 250         public function isResolvingRedirects() {
 251                 return $this->mResolveRedirects;
 252         }
 253
 254         /**
 255          * Return the parameter name that is the source of data for this PageSet
 256          *
 257          * If multiple source parameters are specified (e.g. titles and pageids),
 258          * one will be named arbitrarily.
 259          *
 260          * @return string|null
 261          */
 262         public function getDataSource() {
 263                 if ( $this->mAllowGenerator && isset( $this->mParams['generator'] ) ) {
 264                         return 'generator';
 265                 }
 266                 if ( isset( $this->mParams['titles'] ) ) {
 267                         return 'titles';
 268                 }
 269                 if ( isset( $this->mParams['pageids'] ) ) {
 270                         return 'pageids';
 271                 }
 272                 if ( isset( $this->mParams['revids'] ) ) {
 273                         return 'revids';
 274                 }
 275
 276                 return null;
 277         }
 278
 279         /**
 280          * Request an additional field from the page table.
 281          * Must be called before execute()
 282          * @param string $fieldName Field name
 283          */
 284         public function requestField( $fieldName ) {
 285                 $this->mRequestedPageFields[$fieldName] = null;
 286         }
 287
 288         /**
 289          * Get the value of a custom field previously requested through
 290          * requestField()
 291          * @param string $fieldName Field name
 292          * @return mixed Field value
 293          */
 294         public function getCustomField( $fieldName ) {
 295                 return $this->mRequestedPageFields[$fieldName];
 296         }
 297
 298         /**
 299          * Get the fields that have to be queried from the page table:
 300          * the ones requested through requestField() and a few basic ones
 301          * we always need
 302          * @return array Array of field names
 303          */
 304         public function getPageTableFields() {
 305                 // Ensure we get minimum required fields
 306                 // DON'T change this order
 307                 $pageFlds = array(
 308                         'page_namespace' => null,
 309                         'page_title' => null,
 310                         'page_id' => null,
 311                 );
 312
 313                 if ( $this->mResolveRedirects ) {
 314                         $pageFlds['page_is_redirect'] = null;
 315                 }
 316
 317                 // only store non-default fields
 318                 $this->mRequestedPageFields = array_diff_key( $this->mRequestedPageFields, $pageFlds );
 319
 320                 $pageFlds = array_merge( $pageFlds, $this->mRequestedPageFields );
 321
 322                 return array_keys( $pageFlds );
 323         }
 324
 325         /**
 326          * Returns an array [ns][dbkey] => page_id for all requested titles.
 327          * page_id is a unique negative number in case title was not found.
 328          * Invalid titles will also have negative page IDs and will be in namespace 0
 329          * @return array
 330          */
 331         public function getAllTitlesByNamespace() {
 332                 return $this->mAllPages;
 333         }
 334
 335         /**
 336          * All Title objects provided.
 337          * @return Title[]
 338          */
 339         public function getTitles() {
 340                 return $this->mTitles;
 341         }
 342
 343         /**
 344          * Returns the number of unique pages (not revisions) in the set.
 345          * @return int
 346          */
 347         public function getTitleCount() {
 348                 return count( $this->mTitles );
 349         }
 350
 351         /**
 352          * Returns an array [ns][dbkey] => page_id for all good titles.
 353          * @return array
 354          */
 355         public function getGoodTitlesByNamespace() {
 356                 return $this->mGoodPages;
 357         }
 358
 359         /**
 360          * Title objects that were found in the database.
 361          * @return Title[] Array page_id (int) => Title (obj)
 362          */
 363         public function getGoodTitles() {
 364                 return $this->mGoodTitles;
 365         }
 366
 367         /**
 368          * Returns the number of found unique pages (not revisions) in the set.
 369          * @return int
 370          */
 371         public function getGoodTitleCount() {
 372                 return count( $this->mGoodTitles );
 373         }
 374
 375         /**
 376          * Returns an array [ns][dbkey] => fake_page_id for all missing titles.
 377          * fake_page_id is a unique negative number.
 378          * @return array
 379          */
 380         public function getMissingTitlesByNamespace() {
 381                 return $this->mMissingPages;
 382         }
 383
 384         /**
 385          * Title objects that were NOT found in the database.
 386          * The array's index will be negative for each item
 387          * @return Title[]
 388          */
 389         public function getMissingTitles() {
 390                 return $this->mMissingTitles;
 391         }
 392
 393         /**
 394          * Returns an array [ns][dbkey] => page_id for all good and missing titles.
 395          * @return array
 396          */
 397         public function getGoodAndMissingTitlesByNamespace() {
 398                 return $this->mGoodAndMissingPages;
 399         }
 400
 401         /**
 402          * Title objects for good and missing titles.
 403          * @return array
 404          */
 405         public function getGoodAndMissingTitles() {
 406                 return $this->mGoodTitles + $this->mMissingTitles;
 407         }
 408
 409         /**
 410          * Titles that were deemed invalid by Title::newFromText()
 411          * The array's index will be unique and negative for each item
 412          * @return string[] Array of strings (not Title objects)
 413          */
 414         public function getInvalidTitles() {
 415                 return $this->mInvalidTitles;
 416         }
 417
 418         /**
 419          * Page IDs that were not found in the database
 420          * @return array Array of page IDs
 421          */
 422         public function getMissingPageIDs() {
 423                 return $this->mMissingPageIDs;
 424         }
 425
 426         /**
 427          * Get a list of redirect resolutions - maps a title to its redirect
 428          * target, as an array of output-ready arrays
 429          * @return Title[]
 430          */
 431         public function getRedirectTitles() {
 432                 return $this->mRedirectTitles;
 433         }
 434
 435         /**
 436          * Get a list of redirect resolutions - maps a title to its redirect
 437          * target.
 438          * @param ApiResult $result
 439          * @return array Array of prefixed_title (string) => Title object
 440          * @since 1.21
 441          */
 442         public function getRedirectTitlesAsResult( $result = null ) {
 443                 $values = array();
 444                 foreach ( $this->getRedirectTitles() as $titleStrFrom => $titleTo ) {
 445                         $r = array(
 446                                 'from' => strval( $titleStrFrom ),
 447                                 'to' => $titleTo->getPrefixedText(),
 448                         );
 449                         if ( $titleTo->hasFragment() ) {
 450                                 $r['tofragment'] = $titleTo->getFragment();
 451                         }
 452                         $values[] = $r;
 453                 }
 454                 if ( !empty( $values ) && $result ) {
 455                         $result->setIndexedTagName( $values, 'r' );
 456                 }
 457
 458                 return $values;
 459         }
 460
 461         /**
 462          * Get a list of title normalizations - maps a title to its normalized
 463          * version.
 464          * @return array Array of raw_prefixed_title (string) => prefixed_title (string)
 465          */
 466         public function getNormalizedTitles() {
 467                 return $this->mNormalizedTitles;
 468         }
 469
 470         /**
 471          * Get a list of title normalizations - maps a title to its normalized
 472          * version in the form of result array.
 473          * @param ApiResult $result
 474          * @return array Array of raw_prefixed_title (string) => prefixed_title (string)
 475          * @since 1.21
 476          */
 477         public function getNormalizedTitlesAsResult( $result = null ) {
 478                 $values = array();
 479                 foreach ( $this->getNormalizedTitles() as $rawTitleStr => $titleStr ) {
 480                         $values[] = array(
 481                                 'from' => $rawTitleStr,
 482                                 'to' => $titleStr
 483                         );
 484                 }
 485                 if ( !empty( $values ) && $result ) {
 486                         $result->setIndexedTagName( $values, 'n' );
 487                 }
 488
 489                 return $values;
 490         }
 491
 492         /**
 493          * Get a list of title conversions - maps a title to its converted
 494          * version.
 495          * @return array Array of raw_prefixed_title (string) => prefixed_title (string)
 496          */
 497         public function getConvertedTitles() {
 498                 return $this->mConvertedTitles;
 499         }
 500
 501         /**
 502          * Get a list of title conversions - maps a title to its converted
 503          * version as a result array.
 504          * @param ApiResult $result
 505          * @return array Array of (from, to) strings
 506          * @since 1.21
 507          */
 508         public function getConvertedTitlesAsResult( $result = null ) {
 509                 $values = array();
 510                 foreach ( $this->getConvertedTitles() as $rawTitleStr => $titleStr ) {
 511                         $values[] = array(
 512                                 'from' => $rawTitleStr,
 513                                 'to' => $titleStr
 514                         );
 515                 }
 516                 if ( !empty( $values ) && $result ) {
 517                         $result->setIndexedTagName( $values, 'c' );
 518                 }
 519
 520                 return $values;
 521         }
 522
 523         /**
 524          * Get a list of interwiki titles - maps a title to its interwiki
 525          * prefix.
 526          * @return array Array of raw_prefixed_title (string) => interwiki_prefix (string)
 527          */
 528         public function getInterwikiTitles() {
 529                 return $this->mInterwikiTitles;
 530         }
 531
 532         /**
 533          * Get a list of interwiki titles - maps a title to its interwiki
 534          * prefix as result.
 535          * @param ApiResult $result
 536          * @param bool $iwUrl
 537          * @return array Array of raw_prefixed_title (string) => interwiki_prefix (string)
 538          * @since 1.21
 539          */
 540         public function getInterwikiTitlesAsResult( $result = null, $iwUrl = false ) {
 541                 $values = array();
 542                 foreach ( $this->getInterwikiTitles() as $rawTitleStr => $interwikiStr ) {
 543                         $item = array(
 544                                 'title' => $rawTitleStr,
 545                                 'iw' => $interwikiStr,
 546                         );
 547                         if ( $iwUrl ) {
 548                                 $title = Title::newFromText( $rawTitleStr );
 549                                 $item['url'] = $title->getFullURL( '', false, PROTO_CURRENT );
 550                         }
 551                         $values[] = $item;
 552                 }
 553                 if ( !empty( $values ) && $result ) {
 554                         $result->setIndexedTagName( $values, 'i' );
 555                 }
 556
 557                 return $values;
 558         }
 559
 560         /**
 561          * Get an array of invalid/special/missing titles.
 562          *
 563          * @param array $invalidChecks List of types of invalid titles to include.
 564          *   Recognized values are:
 565          *   - invalidTitles: Titles from $this->getInvalidTitles()
 566          *   - special: Titles from $this->getSpecialTitles()
 567          *   - missingIds: ids from $this->getMissingPageIDs()
 568          *   - missingRevIds: ids from $this->getMissingRevisionIDs()
 569          *   - missingTitles: Titles from $this->getMissingTitles()
 570          *   - interwikiTitles: Titles from $this->getInterwikiTitlesAsResult()
 571          * @return array Array suitable for inclusion in the response
 572          * @since 1.23
 573          */
 574         public function getInvalidTitlesAndRevisions( $invalidChecks = array( 'invalidTitles',
 575                 'special', 'missingIds', 'missingRevIds', 'missingTitles', 'interwikiTitles' )
 576         ) {
 577                 $result = array();
 578                 if ( in_array( "invalidTitles", $invalidChecks ) ) {
 579                         self::addValues( $result, $this->getInvalidTitles(), 'invalid', 'title' );
 580                 }
 581                 if ( in_array( "special", $invalidChecks ) ) {
 582                         self::addValues( $result, $this->getSpecialTitles(), 'special', 'title' );
 583                 }
 584                 if ( in_array( "missingIds", $invalidChecks ) ) {
 585                         self::addValues( $result, $this->getMissingPageIDs(), 'missing', 'pageid' );
 586                 }
 587                 if ( in_array( "missingRevIds", $invalidChecks ) ) {
 588                         self::addValues( $result, $this->getMissingRevisionIDs(), 'missing', 'revid' );
 589                 }
 590                 if ( in_array( "missingTitles", $invalidChecks ) ) {
 591                         self::addValues( $result, $this->getMissingTitles(), 'missing' );
 592                 }
 593                 if ( in_array( "interwikiTitles", $invalidChecks ) ) {
 594                         self::addValues( $result, $this->getInterwikiTitlesAsResult() );
 595                 }
 596
 597                 return $result;
 598         }
 599
 600         /**
 601          * Get the list of valid revision IDs (requested with the revids= parameter)
 602          * @return array Array of revID (int) => pageID (int)
 603          */
 604         public function getRevisionIDs() {
 605                 return $this->mGoodRevIDs;
 606         }
 607
 608         /**
 609          * Get the list of non-deleted revision IDs (requested with the revids= parameter)
 610          * @return array Array of revID (int) => pageID (int)
 611          */
 612         public function getLiveRevisionIDs() {
 613                 return $this->mLiveRevIDs;
 614         }
 615
 616         /**
 617          * Get the list of revision IDs that were associated with deleted titles.
 618          * @return array Array of revID (int) => pageID (int)
 619          */
 620         public function getDeletedRevisionIDs() {
 621                 return $this->mDeletedRevIDs;
 622         }
 623
 624         /**
 625          * Revision IDs that were not found in the database
 626          * @return array Array of revision IDs
 627          */
 628         public function getMissingRevisionIDs() {
 629                 return $this->mMissingRevIDs;
 630         }
 631
 632         /**
 633          * Revision IDs that were not found in the database as result array.
 634          * @param ApiResult $result
 635          * @return array Array of revision IDs
 636          * @since 1.21
 637          */
 638         public function getMissingRevisionIDsAsResult( $result = null ) {
 639                 $values = array();
 640                 foreach ( $this->getMissingRevisionIDs() as $revid ) {
 641                         $values[$revid] = array(
 642                                 'revid' => $revid
 643                         );
 644                 }
 645                 if ( !empty( $values ) && $result ) {
 646                         $result->setIndexedTagName( $values, 'rev' );
 647                 }
 648
 649                 return $values;
 650         }
 651
 652         /**
 653          * Get the list of titles with negative namespace
 654          * @return Title[]
 655          */
 656         public function getSpecialTitles() {
 657                 return $this->mSpecialTitles;
 658         }
 659
 660         /**
 661          * Returns the number of revisions (requested with revids= parameter).
 662          * @return int Number of revisions.
 663          */
 664         public function getRevisionCount() {
 665                 return count( $this->getRevisionIDs() );
 666         }
 667
 668         /**
 669          * Populate this PageSet from a list of Titles
 670          * @param array $titles Array of Title objects
 671          */
 672         public function populateFromTitles( $titles ) {
 673                 $this->profileIn();
 674                 $this->initFromTitles( $titles );
 675                 $this->profileOut();
 676         }
 677
 678         /**
 679          * Populate this PageSet from a list of page IDs
 680          * @param array $pageIDs Array of page IDs
 681          */
 682         public function populateFromPageIDs( $pageIDs ) {
 683                 $this->profileIn();
 684                 $this->initFromPageIds( $pageIDs );
 685                 $this->profileOut();
 686         }
 687
 688         /**
 689          * Populate this PageSet from a rowset returned from the database
 690          *
 691          * Note that the query result must include the columns returned by
 692          * $this->getPageTableFields().
 693          *
 694          * @param DatabaseBase $db
 695          * @param ResultWrapper $queryResult Query result object
 696          */
 697         public function populateFromQueryResult( $db, $queryResult ) {
 698                 $this->profileIn();
 699                 $this->initFromQueryResult( $queryResult );
 700                 $this->profileOut();
 701         }
 702
 703         /**
 704          * Populate this PageSet from a list of revision IDs
 705          * @param array $revIDs Array of revision IDs
 706          */
 707         public function populateFromRevisionIDs( $revIDs ) {
 708                 $this->profileIn();
 709                 $this->initFromRevIDs( $revIDs );
 710                 $this->profileOut();
 711         }
 712
 713         /**
 714          * Extract all requested fields from the row received from the database
 715          * @param stdClass $row Result row
 716          */
 717         public function processDbRow( $row ) {
 718                 // Store Title object in various data structures
 719                 $title = Title::newFromRow( $row );
 720
 721                 $pageId = intval( $row->page_id );
 722                 $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
 723                 $this->mTitles[] = $title;
 724
 725                 if ( $this->mResolveRedirects && $row->page_is_redirect == '1' ) {
 726                         $this->mPendingRedirectIDs[$pageId] = $title;
 727                 } else {
 728                         $this->mGoodPages[$row->page_namespace][$row->page_title] = $pageId;
 729                         $this->mGoodAndMissingPages[$row->page_namespace][$row->page_title] = $pageId;
 730                         $this->mGoodTitles[$pageId] = $title;
 731                 }
 732
 733                 foreach ( $this->mRequestedPageFields as $fieldName => &$fieldValues ) {
 734                         $fieldValues[$pageId] = $row->$fieldName;
 735                 }
 736         }
 737
 738         /**
 739          * Do not use, does nothing, will be removed
 740          * @deprecated since 1.21
 741          */
 742         public function finishPageSetGeneration() {
 743                 wfDeprecated( __METHOD__, '1.21' );
 744         }
 745
 746         /**
 747          * This method populates internal variables with page information
 748          * based on the given array of title strings.
 749          *
 750          * Steps:
 751          * #1 For each title, get data from `page` table
 752          * #2 If page was not found in the DB, store it as missing
 753          *
 754          * Additionally, when resolving redirects:
 755          * #3 If no more redirects left, stop.
 756          * #4 For each redirect, get its target from the `redirect` table.
 757          * #5 Substitute the original LinkBatch object with the new list
 758          * #6 Repeat from step #1
 759          *
 760          * @param array $titles Array of Title objects or strings
 761          */
 762         private function initFromTitles( $titles ) {
 763                 // Get validated and normalized title objects
 764                 $linkBatch = $this->processTitlesArray( $titles );
 765                 if ( $linkBatch->isEmpty() ) {
 766                         return;
 767                 }
 768
 769                 $db = $this->getDB();
 770                 $set = $linkBatch->constructSet( 'page', $db );
 771
 772                 // Get pageIDs data from the `page` table
 773                 $this->profileDBIn();
 774                 $res = $db->select( 'page', $this->getPageTableFields(), $set,
 775                         __METHOD__ );
 776                 $this->profileDBOut();
 777
 778                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 779                 $this->initFromQueryResult( $res, $linkBatch->data, true ); // process Titles
 780
 781                 // Resolve any found redirects
 782                 $this->resolvePendingRedirects();
 783         }
 784
 785         /**
 786          * Does the same as initFromTitles(), but is based on page IDs instead
 787          * @param array $pageids Array of page IDs
 788          */
 789         private function initFromPageIds( $pageids ) {
 790                 if ( !$pageids ) {
 791                         return;
 792                 }
 793
 794                 $pageids = array_map( 'intval', $pageids ); // paranoia
 795                 $remaining = array_flip( $pageids );
 796
 797                 $pageids = self::getPositiveIntegers( $pageids );
 798
 799                 $res = null;
 800                 if ( !empty( $pageids ) ) {
 801                         $set = array(
 802                                 'page_id' => $pageids
 803                         );
 804                         $db = $this->getDB();
 805
 806                         // Get pageIDs data from the `page` table
 807                         $this->profileDBIn();
 808                         $res = $db->select( 'page', $this->getPageTableFields(), $set,
 809                                 __METHOD__ );
 810                         $this->profileDBOut();
 811                 }
 812
 813                 $this->initFromQueryResult( $res, $remaining, false ); // process PageIDs
 814
 815                 // Resolve any found redirects
 816                 $this->resolvePendingRedirects();
 817         }
 818
 819         /**
 820          * Iterate through the result of the query on 'page' table,
 821          * and for each row create and store title object and save any extra fields requested.
 822          * @param ResultWrapper $res DB Query result
 823          * @param array $remaining Array of either pageID or ns/title elements (optional).
 824          *        If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
 825          * @param bool $processTitles Must be provided together with $remaining.
 826          *        If true, treat $remaining as an array of [ns][title]
 827          *        If false, treat it as an array of [pageIDs]
 828          */
 829         private function initFromQueryResult( $res, &$remaining = null, $processTitles = null ) {
 830                 if ( !is_null( $remaining ) && is_null( $processTitles ) ) {
 831                         ApiBase::dieDebug( __METHOD__, 'Missing $processTitles parameter when $remaining is provided' );
 832                 }
 833
 834                 $usernames = array();
 835                 if ( $res ) {
 836                         foreach ( $res as $row ) {
 837                                 $pageId = intval( $row->page_id );
 838
 839                                 // Remove found page from the list of remaining items
 840                                 if ( isset( $remaining ) ) {
 841                                         if ( $processTitles ) {
 842                                                 unset( $remaining[$row->page_namespace][$row->page_title] );
 843                                         } else {
 844                                                 unset( $remaining[$pageId] );
 845                                         }
 846                                 }
 847
 848                                 // Store any extra fields requested by modules
 849                                 $this->processDbRow( $row );
 850
 851                                 // Need gender information
 852                                 if ( MWNamespace::hasGenderDistinction( $row->page_namespace ) ) {
 853                                         $usernames[] = $row->page_title;
 854                                 }
 855                         }
 856                 }
 857
 858                 if ( isset( $remaining ) ) {
 859                         // Any items left in the $remaining list are added as missing
 860                         if ( $processTitles ) {
 861                                 // The remaining titles in $remaining are non-existent pages
 862                                 foreach ( $remaining as $ns => $dbkeys ) {
 863                                         foreach ( array_keys( $dbkeys ) as $dbkey ) {
 864                                                 $title = Title::makeTitle( $ns, $dbkey );
 865                                                 $this->mAllPages[$ns][$dbkey] = $this->mFakePageId;
 866                                                 $this->mMissingPages[$ns][$dbkey] = $this->mFakePageId;
 867                                                 $this->mGoodAndMissingPages[$ns][$dbkey] = $this->mFakePageId;
 868                                                 $this->mMissingTitles[$this->mFakePageId] = $title;
 869                                                 $this->mFakePageId--;
 870                                                 $this->mTitles[] = $title;
 871
 872                                                 // need gender information
 873                                                 if ( MWNamespace::hasGenderDistinction( $ns ) ) {
 874                                                         $usernames[] = $dbkey;
 875                                                 }
 876                                         }
 877                                 }
 878                         } else {
 879                                 // The remaining pageids do not exist
 880                                 if ( !$this->mMissingPageIDs ) {
 881                                         $this->mMissingPageIDs = array_keys( $remaining );
 882                                 } else {
 883                                         $this->mMissingPageIDs = array_merge( $this->mMissingPageIDs, array_keys( $remaining ) );
 884                                 }
 885                         }
 886                 }
 887
 888                 // Get gender information
 889                 $genderCache = GenderCache::singleton();
 890                 $genderCache->doQuery( $usernames, __METHOD__ );
 891         }
 892
 893         /**
 894          * Does the same as initFromTitles(), but is based on revision IDs
 895          * instead
 896          * @param array $revids Array of revision IDs
 897          */
 898         private function initFromRevIDs( $revids ) {
 899                 if ( !$revids ) {
 900                         return;
 901                 }
 902
 903                 $revids = array_map( 'intval', $revids ); // paranoia
 904                 $db = $this->getDB();
 905                 $pageids = array();
 906                 $remaining = array_flip( $revids );
 907
 908                 $revids = self::getPositiveIntegers( $revids );
 909
 910                 if ( !empty( $revids ) ) {
 911                         $tables = array( 'revision', 'page' );
 912                         $fields = array( 'rev_id', 'rev_page' );
 913                         $where = array( 'rev_id' => $revids, 'rev_page = page_id' );
 914
 915                         // Get pageIDs data from the `page` table
 916                         $this->profileDBIn();
 917                         $res = $db->select( $tables, $fields, $where, __METHOD__ );
 918                         foreach ( $res as $row ) {
 919                                 $revid = intval( $row->rev_id );
 920                                 $pageid = intval( $row->rev_page );
 921                                 $this->mGoodRevIDs[$revid] = $pageid;
 922                                 $this->mLiveRevIDs[$revid] = $pageid;
 923                                 $pageids[$pageid] = '';
 924                                 unset( $remaining[$revid] );
 925                         }
 926                         $this->profileDBOut();
 927                 }
 928
 929                 $this->mMissingRevIDs = array_keys( $remaining );
 930
 931                 // Populate all the page information
 932                 $this->initFromPageIds( array_keys( $pageids ) );
 933
 934                 // If the user can see deleted revisions, pull out the corresponding
 935                 // titles from the archive table and include them too. We ignore
 936                 // ar_page_id because deleted revisions are tied by title, not page_id.
 937                 if ( !empty( $this->mMissingRevIDs ) && $this->getUser()->isAllowed( 'deletedhistory' ) ) {
 938                         $remaining = array_flip( $this->mMissingRevIDs );
 939                         $tables = array( 'archive' );
 940                         $fields = array( 'ar_rev_id', 'ar_namespace', 'ar_title' );
 941                         $where = array( 'ar_rev_id' => $this->mMissingRevIDs );
 942
 943                         $this->profileDBIn();
 944                         $res = $db->select( $tables, $fields, $where, __METHOD__ );
 945                         $titles = array();
 946                         foreach ( $res as $row ) {
 947                                 $revid = intval( $row->ar_rev_id );
 948                                 $titles[$revid] = Title::makeTitle( $row->ar_namespace, $row->ar_title );
 949                                 unset( $remaining[$revid] );
 950                         }
 951                         $this->profileDBOut();
 952
 953                         $this->initFromTitles( $titles );
 954
 955                         foreach ( $titles as $revid => $title ) {
 956                                 $ns = $title->getNamespace();
 957                                 $dbkey = $title->getDBkey();
 958
 959                                 // Handle converted titles
 960                                 if ( !isset( $this->mAllPages[$ns][$dbkey] ) &&
 961                                         isset( $this->mConvertedTitles[$title->getPrefixedText()] )
 962                                 ) {
 963                                         $title = Title::newFromText( $this->mConvertedTitles[$title->getPrefixedText()] );
 964                                         $ns = $title->getNamespace();
 965                                         $dbkey = $title->getDBkey();
 966                                 }
 967
 968                                 if ( isset( $this->mAllPages[$ns][$dbkey] ) ) {
 969                                         $this->mGoodRevIDs[$revid] = $this->mAllPages[$ns][$dbkey];
 970                                         $this->mDeletedRevIDs[$revid] = $this->mAllPages[$ns][$dbkey];
 971                                 } else {
 972                                         $remaining[$revid] = true;
 973                                 }
 974                         }
 975
 976                         $this->mMissingRevIDs = array_keys( $remaining );
 977                 }
 978         }
 979
 980         /**
 981          * Resolve any redirects in the result if redirect resolution was
 982          * requested. This function is called repeatedly until all redirects
 983          * have been resolved.
 984          */
 985         private function resolvePendingRedirects() {
 986                 if ( $this->mResolveRedirects ) {
 987                         $db = $this->getDB();
 988                         $pageFlds = $this->getPageTableFields();
 989
 990                         // Repeat until all redirects have been resolved
 991                         // The infinite loop is prevented by keeping all known pages in $this->mAllPages
 992                         while ( $this->mPendingRedirectIDs ) {
 993                                 // Resolve redirects by querying the pagelinks table, and repeat the process
 994                                 // Create a new linkBatch object for the next pass
 995                                 $linkBatch = $this->getRedirectTargets();
 996
 997                                 if ( $linkBatch->isEmpty() ) {
 998                                         break;
 999                                 }
1000
1001                                 $set = $linkBatch->constructSet( 'page', $db );
1002                                 if ( $set === false ) {
1003                                         break;
1004                                 }
1005
1006                                 // Get pageIDs data from the `page` table
1007                                 $this->profileDBIn();
1008                                 $res = $db->select( 'page', $pageFlds, $set, __METHOD__ );
1009                                 $this->profileDBOut();
1010
1011                                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
1012                                 $this->initFromQueryResult( $res, $linkBatch->data, true );
1013                         }
1014                 }
1015         }
1016
1017         /**
1018          * Get the targets of the pending redirects from the database
1019          *
1020          * Also creates entries in the redirect table for redirects that don't
1021          * have one.
1022          * @return LinkBatch
1023          */
1024         private function getRedirectTargets() {
1025                 $lb = new LinkBatch();
1026                 $db = $this->getDB();
1027
1028                 $this->profileDBIn();
1029                 $res = $db->select(
1030                         'redirect',
1031                         array(
1032                                 'rd_from',
1033                                 'rd_namespace',
1034                                 'rd_fragment',
1035                                 'rd_interwiki',
1036                                 'rd_title'
1037                         ), array( 'rd_from' => array_keys( $this->mPendingRedirectIDs ) ),
1038                         __METHOD__
1039                 );
1040                 $this->profileDBOut();
1041                 foreach ( $res as $row ) {
1042                         $rdfrom = intval( $row->rd_from );
1043                         $from = $this->mPendingRedirectIDs[$rdfrom]->getPrefixedText();
1044                         $to = Title::makeTitle(
1045                                 $row->rd_namespace,
1046                                 $row->rd_title,
1047                                 $row->rd_fragment,
1048                                 $row->rd_interwiki
1049                         );
1050                         unset( $this->mPendingRedirectIDs[$rdfrom] );
1051                         if ( !$to->isExternal() && !isset( $this->mAllPages[$row->rd_namespace][$row->rd_title] ) ) {
1052                                 $lb->add( $row->rd_namespace, $row->rd_title );
1053                         }
1054                         $this->mRedirectTitles[$from] = $to;
1055                 }
1056
1057                 if ( $this->mPendingRedirectIDs ) {
1058                         // We found pages that aren't in the redirect table
1059                         // Add them
1060                         foreach ( $this->mPendingRedirectIDs as $id => $title ) {
1061                                 $page = WikiPage::factory( $title );
1062                                 $rt = $page->insertRedirect();
1063                                 if ( !$rt ) {
1064                                         // What the hell. Let's just ignore this
1065                                         continue;
1066                                 }
1067                                 $lb->addObj( $rt );
1068                                 $this->mRedirectTitles[$title->getPrefixedText()] = $rt;
1069                                 unset( $this->mPendingRedirectIDs[$id] );
1070                         }
1071                 }
1072
1073                 return $lb;
1074         }
1075
1076         /**
1077          * Get the cache mode for the data generated by this module.
1078          * All PageSet users should take into account whether this returns a more-restrictive
1079          * cache mode than the using module itself. For possible return values and other
1080          * details about cache modes, see ApiMain::setCacheMode()
1081          *
1082          * Public caching will only be allowed if *all* the modules that supply
1083          * data for a given request return a cache mode of public.
1084          *
1085          * @param array|null $params
1086          * @return string
1087          * @since 1.21
1088          */
1089         public function getCacheMode( $params = null ) {
1090                 return $this->mCacheMode;
1091         }
1092
1093         /**
1094          * Given an array of title strings, convert them into Title objects.
1095          * Alternatively, an array of Title objects may be given.
1096          * This method validates access rights for the title,
1097          * and appends normalization values to the output.
1098          *
1099          * @param array $titles Array of Title objects or strings
1100          * @return LinkBatch
1101          */
1102         private function processTitlesArray( $titles ) {
1103                 $usernames = array();
1104                 $linkBatch = new LinkBatch();
1105
1106                 foreach ( $titles as $title ) {
1107                         if ( is_string( $title ) ) {
1108                                 $titleObj = Title::newFromText( $title, $this->mDefaultNamespace );
1109                         } else {
1110                                 $titleObj = $title;
1111                         }
1112                         if ( !$titleObj ) {
1113                                 // Handle invalid titles gracefully
1114                                 $this->mAllPages[0][$title] = $this->mFakePageId;
1115                                 $this->mInvalidTitles[$this->mFakePageId] = $title;
1116                                 $this->mFakePageId--;
1117                                 continue; // There's nothing else we can do
1118                         }
1119                         $unconvertedTitle = $titleObj->getPrefixedText();
1120                         $titleWasConverted = false;
1121                         if ( $titleObj->isExternal() ) {
1122                                 // This title is an interwiki link.
1123                                 $this->mInterwikiTitles[$unconvertedTitle] = $titleObj->getInterwiki();
1124                         } else {
1125                                 // Variants checking
1126                                 global $wgContLang;
1127                                 if ( $this->mConvertTitles &&
1128                                         count( $wgContLang->getVariants() ) > 1 &&
1129                                         !$titleObj->exists()
1130                                 ) {
1131                                         // Language::findVariantLink will modify titleText and titleObj into
1132                                         // the canonical variant if possible
1133                                         $titleText = is_string( $title ) ? $title : $titleObj->getPrefixedText();
1134                                         $wgContLang->findVariantLink( $titleText, $titleObj );
1135                                         $titleWasConverted = $unconvertedTitle !== $titleObj->getPrefixedText();
1136                                 }
1137
1138                                 if ( $titleObj->getNamespace() < 0 ) {
1139                                         // Handle Special and Media pages
1140                                         $titleObj = $titleObj->fixSpecialName();
1141                                         $this->mSpecialTitles[$this->mFakePageId] = $titleObj;
1142                                         $this->mFakePageId--;
1143                                 } else {
1144                                         // Regular page
1145                                         $linkBatch->addObj( $titleObj );
1146                                 }
1147                         }
1148
1149                         // Make sure we remember the original title that was
1150                         // given to us. This way the caller can correlate new
1151                         // titles with the originally requested when e.g. the
1152                         // namespace is localized or the capitalization is
1153                         // different
1154                         if ( $titleWasConverted ) {
1155                                 $this->mConvertedTitles[$unconvertedTitle] = $titleObj->getPrefixedText();
1156                                 // In this case the page can't be Special.
1157                                 if ( is_string( $title ) && $title !== $unconvertedTitle ) {
1158                                         $this->mNormalizedTitles[$title] = $unconvertedTitle;
1159                                 }
1160                         } elseif ( is_string( $title ) && $title !== $titleObj->getPrefixedText() ) {
1161                                 $this->mNormalizedTitles[$title] = $titleObj->getPrefixedText();
1162                         }
1163
1164                         // Need gender information
1165                         if ( MWNamespace::hasGenderDistinction( $titleObj->getNamespace() ) ) {
1166                                 $usernames[] = $titleObj->getText();
1167                         }
1168                 }
1169                 // Get gender information
1170                 $genderCache = GenderCache::singleton();
1171                 $genderCache->doQuery( $usernames, __METHOD__ );
1172
1173                 return $linkBatch;
1174         }
1175
1176         /**
1177          * Get the database connection (read-only)
1178          * @return DatabaseBase
1179          */
1180         protected function getDB() {
1181                 return $this->mDbSource->getDB();
1182         }
1183
1184         /**
1185          * Returns the input array of integers with all values < 0 removed
1186          *
1187          * @param array $array
1188          * @return array
1189          */
1190         private static function getPositiveIntegers( $array ) {
1191                 // bug 25734 API: possible issue with revids validation
1192                 // It seems with a load of revision rows, MySQL gets upset
1193                 // Remove any < 0 integers, as they can't be valid
1194                 foreach ( $array as $i => $int ) {
1195                         if ( $int < 0 ) {
1196                                 unset( $array[$i] );
1197                         }
1198                 }
1199
1200                 return $array;
1201         }
1202
1203         public function getAllowedParams( $flags = 0 ) {
1204                 $result = array(
1205                         'titles' => array(
1206                                 ApiBase::PARAM_ISMULTI => true,
1207                                 ApiBase::PARAM_HELP_MSG => 'api-pageset-param-titles',
1208                         ),
1209                         'pageids' => array(
1210                                 ApiBase::PARAM_TYPE => 'integer',
1211                                 ApiBase::PARAM_ISMULTI => true,
1212                                 ApiBase::PARAM_HELP_MSG => 'api-pageset-param-pageids',
1213                         ),
1214                         'revids' => array(
1215                                 ApiBase::PARAM_TYPE => 'integer',
1216                                 ApiBase::PARAM_ISMULTI => true,
1217                                 ApiBase::PARAM_HELP_MSG => 'api-pageset-param-revids',
1218                         ),
1219                         'generator' => array(
1220                                 ApiBase::PARAM_TYPE => null,
1221                                 ApiBase::PARAM_VALUE_LINKS => array(),
1222                                 ApiBase::PARAM_HELP_MSG => 'api-pageset-param-generator',
1223                         ),
1224                         'redirects' => array(
1225                                 ApiBase::PARAM_DFLT => false,
1226                                 ApiBase::PARAM_HELP_MSG => $this->mAllowGenerator
1227                                         ? 'api-pageset-param-redirects-generator'
1228                                         : 'api-pageset-param-redirects-nogenerator',
1229                         ),
1230                         'converttitles' => array(
1231                                 ApiBase::PARAM_DFLT => false,
1232                                 ApiBase::PARAM_HELP_MSG => array(
1233                                         'api-pageset-param-converttitles',
1234                                         $this->getLanguage()->commaList( LanguageConverter::$languagesWithVariants ),
1235                                 ),
1236                         ),
1237                 );
1238
1239                 if ( !$this->mAllowGenerator ) {
1240                         unset( $result['generator'] );
1241                 } elseif ( $flags & ApiBase::GET_VALUES_FOR_HELP ) {
1242                         $result['generator'][ApiBase::PARAM_TYPE] = $this->getGenerators();
1243                         foreach ( $result['generator'][ApiBase::PARAM_TYPE] as $g ) {
1244                                 $result['generator'][ApiBase::PARAM_TYPE][] = $g;
1245                                 $result['generator'][ApiBase::PARAM_VALUE_LINKS][$g] = "Special:ApiHelp/query+$g";
1246                         }
1247                 }
1248
1249                 return $result;
1250         }
1251
1252         private static $generators = null;
1253
1254         /**
1255          * Get an array of all available generators
1256          * @return array
1257          */
1258         private function getGenerators() {
1259                 if ( self::$generators === null ) {
1260                         $query = $this->mDbSource;
1261                         if ( !( $query instanceof ApiQuery ) ) {
1262                                 // If the parent container of this pageset is not ApiQuery,
1263                                 // we must create it to get module manager
1264                                 $query = $this->getMain()->getModuleManager()->getModule( 'query' );
1265                         }
1266                         $gens = array();
1267                         $mgr = $query->getModuleManager();
1268                         foreach ( $mgr->getNamesWithClasses() as $name => $class ) {
1269                                 if ( is_subclass_of( $class, 'ApiQueryGeneratorBase' ) ) {
1270                                         $gens[] = $name;
1271                                 }
1272                         }
1273                         sort( $gens );
1274                         self::$generators = $gens;
1275                 }
1276
1277                 return self::$generators;
1278         }
1279 }