includes/api/ApiPageSet.php

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