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