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