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