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