includes/api/ApiPageSet.php

   1 <?php
   2
   3 /*
   4  * Created on Sep 24, 2006
   5  *
   6  * API for MediaWiki 1.8+
   7  *
   8  * Copyright (C) 2006 Yuri Astrakhan <Firstname><Lastname>@gmail.com
   9  *
  10  * This program is free software; you can redistribute it and/or modify
  11  * it under the terms of the GNU General Public License as published by
  12  * the Free Software Foundation; either version 2 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  18  * GNU General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU General Public License along
  21  * with this program; if not, write to the Free Software Foundation, Inc.,
  22  * 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  23  * http://www.gnu.org/copyleft/gpl.html
  24  */
  25
  26 if (!defined('MEDIAWIKI')) {
  27         // Eclipse helper - will be ignored in production
  28         require_once ('ApiQueryBase.php');
  29 }
  30
  31 /**
  32  * This class contains a list of pages that the client has requested.
  33  * Initially, when the client passes in titles=, pageids=, or revisions= parameter,
  34  * an instance of the ApiPageSet class will normalize titles,
  35  * determine if the pages/revisions exist, and prefetch any additional data page data requested.
  36  *
  37  * When generator is used, the result of the generator will become the input for the
  38  * second instance of this class, and all subsequent actions will go use the second instance
  39  * for all their work.
  40  *
  41  * @ingroup API
  42  */
  43 class ApiPageSet extends ApiQueryBase {
  44
  45         private $mAllPages; // [ns][dbkey] => page_id or negative when missing
  46         private $mTitles, $mGoodTitles, $mMissingTitles, $mInvalidTitles;
  47         private $mMissingPageIDs, $mRedirectTitles;
  48         private $mNormalizedTitles, $mInterwikiTitles;
  49         private $mResolveRedirects, $mPendingRedirectIDs;
  50         private $mGoodRevIDs, $mMissingRevIDs;
  51         private $mFakePageId;
  52
  53         private $mRequestedPageFields;
  54
  55         public function __construct($query, $resolveRedirects = false) {
  56                 parent :: __construct($query, 'query');
  57
  58                 $this->mAllPages = array ();
  59                 $this->mTitles = array();
  60                 $this->mGoodTitles = array ();
  61                 $this->mMissingTitles = array ();
  62                 $this->mInvalidTitles = array ();
  63                 $this->mMissingPageIDs = array ();
  64                 $this->mRedirectTitles = array ();
  65                 $this->mNormalizedTitles = array ();
  66                 $this->mInterwikiTitles = array ();
  67                 $this->mGoodRevIDs = array();
  68                 $this->mMissingRevIDs = array();
  69
  70                 $this->mRequestedPageFields = array ();
  71                 $this->mResolveRedirects = $resolveRedirects;
  72                 if($resolveRedirects)
  73                         $this->mPendingRedirectIDs = array();
  74
  75                 $this->mFakePageId = -1;
  76         }
  77
  78         public function isResolvingRedirects() {
  79                 return $this->mResolveRedirects;
  80         }
  81
  82         public function requestField($fieldName) {
  83                 $this->mRequestedPageFields[$fieldName] = null;
  84         }
  85
  86         public function getCustomField($fieldName) {
  87                 return $this->mRequestedPageFields[$fieldName];
  88         }
  89
  90         /**
  91          * Get fields that modules have requested from the page table
  92          */
  93         public function getPageTableFields() {
  94                 // Ensure we get minimum required fields
  95                 // DON'T change this order
  96                 $pageFlds = array (
  97                         'page_namespace' => null,
  98                         'page_title' => null,
  99                         'page_id' => null,
 100                 );
 101
 102                 // only store non-default fields
 103                 $this->mRequestedPageFields = array_diff_key($this->mRequestedPageFields, $pageFlds);
 104
 105                 if ($this->mResolveRedirects)
 106                         $pageFlds['page_is_redirect'] = null;
 107
 108                 $pageFlds = array_merge($pageFlds, $this->mRequestedPageFields);
 109                 return array_keys($pageFlds);
 110         }
 111
 112         /**
 113          * Returns an array [ns][dbkey] => page_id for all requested titles.
 114          * page_id is a unique negative number in case title was not found.
 115          * Invalid titles will also have negative page IDs and will be in namespace 0
 116          */
 117         public function getAllTitlesByNamespace() {
 118                 return $this->mAllPages;
 119         }
 120
 121         /**
 122          * All Title objects provided.
 123          * @return array of Title objects
 124          */
 125         public function getTitles() {
 126                 return $this->mTitles;
 127         }
 128
 129         /**
 130          * Returns the number of unique pages (not revisions) in the set.
 131          */
 132         public function getTitleCount() {
 133                 return count($this->mTitles);
 134         }
 135
 136         /**
 137          * Title objects that were found in the database.
 138          * @return array page_id (int) => Title (obj)
 139          */
 140         public function getGoodTitles() {
 141                 return $this->mGoodTitles;
 142         }
 143
 144         /**
 145          * Returns the number of found unique pages (not revisions) in the set.
 146          */
 147         public function getGoodTitleCount() {
 148                 return count($this->mGoodTitles);
 149         }
 150
 151         /**
 152          * Title objects that were NOT found in the database.
 153          * The array's index will be negative for each item
 154          * @return array of Title objects
 155          */
 156         public function getMissingTitles() {
 157                 return $this->mMissingTitles;
 158         }
 159
 160         /**
 161          * Titles that were deemed invalid by Title::newFromText()
 162          * The array's index will be unique and negative for each item
 163          * @return array of strings (not Title objects)
 164          */
 165         public function getInvalidTitles() {
 166                 return $this->mInvalidTitles;
 167         }
 168
 169         /**
 170          * Page IDs that were not found in the database
 171          * @return array of page IDs
 172          */
 173         public function getMissingPageIDs() {
 174                 return $this->mMissingPageIDs;
 175         }
 176
 177         /**
 178          * Get a list of redirects when doing redirect resolution
 179          * @return array prefixed_title (string) => prefixed_title (string)
 180          */
 181         public function getRedirectTitles() {
 182                 return $this->mRedirectTitles;
 183         }
 184
 185         /**
 186          * Get a list of title normalizations - maps the title given
 187          * with its normalized version.
 188          * @return array raw_prefixed_title (string) => prefixed_title (string)
 189          */
 190         public function getNormalizedTitles() {
 191                 return $this->mNormalizedTitles;
 192         }
 193
 194         /**
 195          * Get a list of interwiki titles - maps the title given
 196          * with to the interwiki prefix.
 197          * @return array raw_prefixed_title (string) => interwiki_prefix (string)
 198          */
 199         public function getInterwikiTitles() {
 200                 return $this->mInterwikiTitles;
 201         }
 202
 203         /**
 204          * Get the list of revision IDs (requested with revids= parameter)
 205          * @return array revID (int) => pageID (int)
 206          */
 207         public function getRevisionIDs() {
 208                 return $this->mGoodRevIDs;
 209         }
 210
 211         /**
 212          * Revision IDs that were not found in the database
 213          * @return array of revision IDs
 214          */
 215         public function getMissingRevisionIDs() {
 216                 return $this->mMissingRevIDs;
 217         }
 218
 219         /**
 220          * Returns the number of revisions (requested with revids= parameter)
 221          */
 222         public function getRevisionCount() {
 223                 return count($this->getRevisionIDs());
 224         }
 225
 226         /**
 227          * Populate from the request parameters
 228          */
 229         public function execute() {
 230                 $this->profileIn();
 231                 $params = $this->extractRequestParams();
 232
 233                 // Only one of the titles/pageids/revids is allowed at the same time
 234                 $dataSource = null;
 235                 if (isset ($params['titles']))
 236                         $dataSource = 'titles';
 237                 if (isset ($params['pageids'])) {
 238                         if (isset ($dataSource))
 239                                 $this->dieUsage("Cannot use 'pageids' at the same time as '$dataSource'", 'multisource');
 240                         $dataSource = 'pageids';
 241                 }
 242                 if (isset ($params['revids'])) {
 243                         if (isset ($dataSource))
 244                                 $this->dieUsage("Cannot use 'revids' at the same time as '$dataSource'", 'multisource');
 245                         $dataSource = 'revids';
 246                 }
 247
 248                 switch ($dataSource) {
 249                         case 'titles' :
 250                                 $this->initFromTitles($params['titles']);
 251                                 break;
 252                         case 'pageids' :
 253                                 $this->initFromPageIds($params['pageids']);
 254                                 break;
 255                         case 'revids' :
 256                                 if($this->mResolveRedirects)
 257                                         $this->setWarning('Redirect resolution cannot be used together with the revids= parameter. '.
 258                                         'Any redirects the revids= point to have not been resolved.');
 259                                 $this->mResolveRedirects = false;
 260                                 $this->initFromRevIDs($params['revids']);
 261                                 break;
 262                         default :
 263                                 // Do nothing - some queries do not need any of the data sources.
 264                                 break;
 265                 }
 266                 $this->profileOut();
 267         }
 268
 269         /**
 270          * Initialize PageSet from a list of Titles
 271          */
 272         public function populateFromTitles($titles) {
 273                 $this->profileIn();
 274                 $this->initFromTitles($titles);
 275                 $this->profileOut();
 276         }
 277
 278         /**
 279          * Initialize PageSet from a list of Page IDs
 280          */
 281         public function populateFromPageIDs($pageIDs) {
 282                 $this->profileIn();
 283                 $this->initFromPageIds($pageIDs);
 284                 $this->profileOut();
 285         }
 286
 287         /**
 288          * Initialize PageSet from a rowset returned from the database
 289          */
 290         public function populateFromQueryResult($db, $queryResult) {
 291                 $this->profileIn();
 292                 $this->initFromQueryResult($db, $queryResult);
 293                 $this->profileOut();
 294         }
 295
 296         /**
 297          * Initialize PageSet from a list of Revision IDs
 298          */
 299         public function populateFromRevisionIDs($revIDs) {
 300                 $this->profileIn();
 301                 $revIDs = array_map('intval', $revIDs); // paranoia
 302                 $this->initFromRevIDs($revIDs);
 303                 $this->profileOut();
 304         }
 305
 306         /**
 307          * Extract all requested fields from the row received from the database
 308          */
 309         public function processDbRow($row) {
 310
 311                 // Store Title object in various data structures
 312                 $title = Title :: makeTitle($row->page_namespace, $row->page_title);
 313
 314                 $pageId = intval($row->page_id);
 315                 $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
 316                 $this->mTitles[] = $title;
 317
 318                 if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
 319                         $this->mPendingRedirectIDs[$pageId] = $title;
 320                 } else {
 321                         $this->mGoodTitles[$pageId] = $title;
 322                 }
 323
 324                 foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
 325                         $fieldValues[$pageId] = $row-> $fieldName;
 326         }
 327
 328         public function finishPageSetGeneration() {
 329                 $this->profileIn();
 330                 $this->resolvePendingRedirects();
 331                 $this->profileOut();
 332         }
 333
 334         /**
 335          * This method populates internal variables with page information
 336          * based on the given array of title strings.
 337          *
 338          * Steps:
 339          * #1 For each title, get data from `page` table
 340          * #2 If page was not found in the DB, store it as missing
 341          *
 342          * Additionally, when resolving redirects:
 343          * #3 If no more redirects left, stop.
 344          * #4 For each redirect, get its links from `pagelinks` table.
 345          * #5 Substitute the original LinkBatch object with the new list
 346          * #6 Repeat from step #1
 347          */
 348         private function initFromTitles($titles) {
 349
 350                 // Get validated and normalized title objects
 351                 $linkBatch = $this->processTitlesArray($titles);
 352                 if($linkBatch->isEmpty())
 353                         return;
 354
 355                 $db = $this->getDB();
 356                 $set = $linkBatch->constructSet('page', $db);
 357
 358                 // Get pageIDs data from the `page` table
 359                 $this->profileDBIn();
 360                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 361                 $this->profileDBOut();
 362
 363                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 364                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);  // process Titles
 365
 366                 // Resolve any found redirects
 367                 $this->resolvePendingRedirects();
 368         }
 369
 370         private function initFromPageIds($pageids) {
 371                 if(!count($pageids))
 372                         return;
 373
 374                 $pageids = array_map('intval', $pageids); // paranoia
 375                 $set = array (
 376                         'page_id' => $pageids
 377                 );
 378
 379                 $db = $this->getDB();
 380
 381                 // Get pageIDs data from the `page` table
 382                 $this->profileDBIn();
 383                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 384                 $this->profileDBOut();
 385
 386                 $remaining = array_flip($pageids);
 387                 $this->initFromQueryResult($db, $res, $remaining, false);       // process PageIDs
 388
 389                 // Resolve any found redirects
 390                 $this->resolvePendingRedirects();
 391         }
 392
 393         /**
 394          * Iterate through the result of the query on 'page' table,
 395          * and for each row create and store title object and save any extra fields requested.
 396          * @param $db Database
 397          * @param $res DB Query result
 398          * @param $remaining Array of either pageID or ns/title elements (optional).
 399          *        If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
 400          * @param $processTitles bool Must be provided together with $remaining.
 401          *        If true, treat $remaining as an array of [ns][title]
 402          *        If false, treat it as an array of [pageIDs]
 403          * @return Array of redirect IDs (only when resolving redirects)
 404          */
 405         private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
 406                 if (!is_null($remaining) && is_null($processTitles))
 407                         ApiBase :: dieDebug(__METHOD__, 'Missing $processTitles parameter when $remaining is provided');
 408
 409                 while ($row = $db->fetchObject($res)) {
 410
 411                         $pageId = intval($row->page_id);
 412
 413                         // Remove found page from the list of remaining items
 414                         if (isset($remaining)) {
 415                                 if ($processTitles)
 416                                         unset ($remaining[$row->page_namespace][$row->page_title]);
 417                                 else
 418                                         unset ($remaining[$pageId]);
 419                         }
 420
 421                         // Store any extra fields requested by modules
 422                         $this->processDbRow($row);
 423                 }
 424                 $db->freeResult($res);
 425
 426                 if(isset($remaining)) {
 427                         // Any items left in the $remaining list are added as missing
 428                         if($processTitles) {
 429                                 // The remaining titles in $remaining are non-existent pages
 430                                 foreach ($remaining as $ns => $dbkeys) {
 431                                         foreach ( $dbkeys as $dbkey => $unused ) {
 432                                                 $title = Title :: makeTitle($ns, $dbkey);
 433                                                 $this->mAllPages[$ns][$dbkey] = $this->mFakePageId;
 434                                                 $this->mMissingTitles[$this->mFakePageId] = $title;
 435                                                 $this->mFakePageId--;
 436                                                 $this->mTitles[] = $title;
 437                                         }
 438                                 }
 439                         }
 440                         else
 441                         {
 442                                 // The remaining pageids do not exist
 443                                 if(!$this->mMissingPageIDs)
 444                                         $this->mMissingPageIDs = array_keys($remaining);
 445                                 else
 446                                         $this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
 447                         }
 448                 }
 449         }
 450
 451         private function initFromRevIDs($revids) {
 452
 453                 if(!count($revids))
 454                         return;
 455
 456                 $db = $this->getDB();
 457                 $pageids = array();
 458                 $remaining = array_flip($revids);
 459
 460                 $tables = array('revision','page');
 461                 $fields = array('rev_id','rev_page');
 462                 $where = array('rev_deleted' => 0, 'rev_id' => $revids,'rev_page = page_id');
 463
 464                 // Get pageIDs data from the `page` table
 465                 $this->profileDBIn();
 466                 $res = $db->select( $tables, $fields, $where,  __METHOD__ );
 467                 while ( $row = $db->fetchObject( $res ) ) {
 468                         $revid = intval($row->rev_id);
 469                         $pageid = intval($row->rev_page);
 470                         $this->mGoodRevIDs[$revid] = $pageid;
 471                         $pageids[$pageid] = '';
 472                         unset($remaining[$revid]);
 473                 }
 474                 $db->freeResult( $res );
 475                 $this->profileDBOut();
 476
 477                 $this->mMissingRevIDs = array_keys($remaining);
 478
 479                 // Populate all the page information
 480                 $this->initFromPageIds(array_keys($pageids));
 481         }
 482
 483         private function resolvePendingRedirects() {
 484
 485                 if($this->mResolveRedirects) {
 486                         $db = $this->getDB();
 487                         $pageFlds = $this->getPageTableFields();
 488
 489                         // Repeat until all redirects have been resolved
 490                         // The infinite loop is prevented by keeping all known pages in $this->mAllPages
 491                         while ($this->mPendingRedirectIDs) {
 492
 493                                 // Resolve redirects by querying the pagelinks table, and repeat the process
 494                                 // Create a new linkBatch object for the next pass
 495                                 $linkBatch = $this->getRedirectTargets();
 496
 497                                 if ($linkBatch->isEmpty())
 498                                         break;
 499
 500                                 $set = $linkBatch->constructSet('page', $db);
 501                                 if(false === $set)
 502                                         break;
 503
 504                                 // Get pageIDs data from the `page` table
 505                                 $this->profileDBIn();
 506                                 $res = $db->select('page', $pageFlds, $set, __METHOD__);
 507                                 $this->profileDBOut();
 508
 509                                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 510                                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);
 511                         }
 512                 }
 513         }
 514
 515         private function getRedirectTargets() {
 516                 $lb = new LinkBatch();
 517                 $db = $this->getDB();
 518
 519                 $this->profileDBIn();
 520                 $res = $db->select('redirect', array(
 521                                 'rd_from',
 522                                 'rd_namespace',
 523                                 'rd_title'
 524                         ), array('rd_from' => array_keys($this->mPendingRedirectIDs)),
 525                         __METHOD__
 526                 );
 527                 $this->profileDBOut();
 528
 529                 while($row = $db->fetchObject($res))
 530                 {
 531                         $rdfrom = intval($row->rd_from);
 532                         $from = $this->mPendingRedirectIDs[$rdfrom]->getPrefixedText();
 533                         $to = Title::makeTitle($row->rd_namespace, $row->rd_title)->getPrefixedText();
 534                         unset($this->mPendingRedirectIDs[$rdfrom]);
 535                         if(!isset($this->mAllPages[$row->rd_namespace][$row->rd_title]))
 536                                 $lb->add($row->rd_namespace, $row->rd_title);
 537                         $this->mRedirectTitles[$from] = $to;
 538                 }
 539                 $db->freeResult($res);
 540                 if($this->mPendingRedirectIDs)
 541                 {
 542                         # We found pages that aren't in the redirect table
 543                         # Add them
 544                         foreach($this->mPendingRedirectIDs as $id => $title)
 545                         {
 546                                 $article = new Article($title);
 547                                 $rt = $article->insertRedirect();
 548                                 if(!$rt)
 549                                         # What the hell. Let's just ignore this
 550                                         continue;
 551                                 $lb->addObj($rt);
 552                                 $this->mRedirectTitles[$title->getPrefixedText()] = $rt->getPrefixedText();
 553                                 unset($this->mPendingRedirectIDs[$id]);
 554                         }
 555                 }
 556                 return $lb;
 557         }
 558
 559         /**
 560          * Given an array of title strings, convert them into Title objects.
 561          * Alternativelly, an array of Title objects may be given.
 562          * This method validates access rights for the title,
 563          * and appends normalization values to the output.
 564          *
 565          * @return LinkBatch of title objects.
 566          */
 567         private function processTitlesArray($titles) {
 568
 569                 $linkBatch = new LinkBatch();
 570
 571                 foreach ($titles as $title) {
 572
 573                         $titleObj = is_string($title) ? Title :: newFromText($title) : $title;
 574                         if (!$titleObj)
 575                         {
 576                                 # Handle invalid titles gracefully
 577                                 $this->mAllpages[0][$title] = $this->mFakePageId;
 578                                 $this->mInvalidTitles[$this->mFakePageId] = $title;
 579                                 $this->mFakePageId--;
 580                                 continue; // There's nothing else we can do
 581                         }
 582                         $iw = $titleObj->getInterwiki();
 583                         if (strval($iw) !== '') {
 584                                 // This title is an interwiki link.
 585                                 $this->mInterwikiTitles[$titleObj->getPrefixedText()] = $iw;
 586                         } else {
 587
 588                                 // Validation
 589                                 if ($titleObj->getNamespace() < 0)
 590                                         $this->setWarning("No support for special pages has been implemented");
 591                                 else
 592                                         $linkBatch->addObj($titleObj);
 593                         }
 594
 595                         // Make sure we remember the original title that was given to us
 596                         // This way the caller can correlate new titles with the originally requested,
 597                         // i.e. namespace is localized or capitalization is different
 598                         if (is_string($title) && $title !== $titleObj->getPrefixedText()) {
 599                                 $this->mNormalizedTitles[$title] = $titleObj->getPrefixedText();
 600                         }
 601                 }
 602
 603                 return $linkBatch;
 604         }
 605
 606         protected function getAllowedParams() {
 607                 return array (
 608                         'titles' => array (
 609                                 ApiBase :: PARAM_ISMULTI => true
 610                         ),
 611                         'pageids' => array (
 612                                 ApiBase :: PARAM_TYPE => 'integer',
 613                                 ApiBase :: PARAM_ISMULTI => true
 614                         ),
 615                         'revids' => array (
 616                                 ApiBase :: PARAM_TYPE => 'integer',
 617                                 ApiBase :: PARAM_ISMULTI => true
 618                         )
 619                 );
 620         }
 621
 622         protected function getParamDescription() {
 623                 return array (
 624                         'titles' => 'A list of titles to work on',
 625                         'pageids' => 'A list of page IDs to work on',
 626                         'revids' => 'A list of revision IDs to work on'
 627                 );
 628         }
 629
 630         public function getVersion() {
 631                 return __CLASS__ . ': $Id$';
 632         }
 633 }