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  * @addtogroup API
  42  */
  43 class ApiPageSet extends ApiQueryBase {
  44
  45         private $mAllPages; // [ns][dbkey] => page_id or 0 when missing
  46         private $mTitles, $mGoodTitles, $mMissingTitles, $mMissingPageIDs, $mRedirectTitles;
  47         private $mNormalizedTitles, $mInterwikiTitles;
  48         private $mResolveRedirects, $mPendingRedirectIDs;
  49         private $mGoodRevIDs, $mMissingRevIDs;
  50         private $mFakePageId;
  51
  52         private $mRequestedPageFields;
  53
  54         public function __construct($query, $resolveRedirects = false) {
  55                 parent :: __construct($query, __CLASS__);
  56
  57                 $this->mAllPages = array ();
  58                 $this->mTitles = array();
  59                 $this->mGoodTitles = array ();
  60                 $this->mMissingTitles = array ();
  61                 $this->mMissingPageIDs = array ();
  62                 $this->mRedirectTitles = array ();
  63                 $this->mNormalizedTitles = array ();
  64                 $this->mInterwikiTitles = array ();
  65                 $this->mGoodRevIDs = array();
  66                 $this->mMissingRevIDs = array();
  67
  68                 $this->mRequestedPageFields = array ();
  69                 $this->mResolveRedirects = $resolveRedirects;
  70                 if($resolveRedirects)
  71                         $this->mPendingRedirectIDs = array();
  72
  73                 $this->mFakePageId = -1;
  74         }
  75
  76         public function isResolvingRedirects() {
  77                 return $this->mResolveRedirects;
  78         }
  79
  80         public function requestField($fieldName) {
  81                 $this->mRequestedPageFields[$fieldName] = null;
  82         }
  83
  84         public function getCustomField($fieldName) {
  85                 return $this->mRequestedPageFields[$fieldName];
  86         }
  87
  88         /**
  89          * Get fields that modules have requested from the page table
  90          */
  91         public function getPageTableFields() {
  92                 // Ensure we get minimum required fields
  93                 $pageFlds = array (
  94                         'page_id' => null,
  95                         'page_namespace' => null,
  96                         'page_title' => null
  97                 );
  98
  99                 // only store non-default fields
 100                 $this->mRequestedPageFields = array_diff_key($this->mRequestedPageFields, $pageFlds);
 101
 102                 if ($this->mResolveRedirects)
 103                         $pageFlds['page_is_redirect'] = null;
 104
 105                 $pageFlds = array_merge($pageFlds, $this->mRequestedPageFields);
 106                 return array_keys($pageFlds);
 107         }
 108
 109         /**
 110          * Returns an array [ns][dbkey] => page_id for all requested titles
 111          * page_id is a unique negative number in case title was not found
 112          */
 113         public function getAllTitlesByNamespace() {
 114                 return $this->mAllPages;
 115         }
 116
 117         /**
 118          * All Title objects provided.
 119          * @return array of Title objects
 120          */
 121         public function getTitles() {
 122                 return $this->mTitles;
 123         }
 124
 125         /**
 126          * Returns the number of unique pages (not revisions) in the set.
 127          */
 128         public function getTitleCount() {
 129                 return count($this->mTitles);
 130         }
 131
 132         /**
 133          * Title objects that were found in the database.
 134          * @return array page_id (int) => Title (obj)
 135          */
 136         public function getGoodTitles() {
 137                 return $this->mGoodTitles;
 138         }
 139
 140         /**
 141          * Returns the number of found unique pages (not revisions) in the set.
 142          */
 143         public function getGoodTitleCount() {
 144                 return count($this->mGoodTitles);
 145         }
 146
 147         /**
 148          * Title objects that were NOT found in the database.
 149          * The array's index will be negative for each item
 150          * @return array of Title objects
 151          */
 152         public function getMissingTitles() {
 153                 return $this->mMissingTitles;
 154         }
 155
 156         /**
 157          * Page IDs that were not found in the database
 158          * @return array of page IDs
 159          */
 160         public function getMissingPageIDs() {
 161                 return $this->mMissingPageIDs;
 162         }
 163
 164         /**
 165          * Get a list of redirects when doing redirect resolution
 166          * @return array prefixed_title (string) => prefixed_title (string)
 167          */
 168         public function getRedirectTitles() {
 169                 return $this->mRedirectTitles;
 170         }
 171
 172         /**
 173          * Get a list of title normalizations - maps the title given
 174          * with its normalized version.
 175          * @return array raw_prefixed_title (string) => prefixed_title (string)
 176          */
 177         public function getNormalizedTitles() {
 178                 return $this->mNormalizedTitles;
 179         }
 180
 181         /**
 182          * Get a list of interwiki titles - maps the title given
 183          * with to the interwiki prefix.
 184          * @return array raw_prefixed_title (string) => interwiki_prefix (string)
 185          */
 186         public function getInterwikiTitles() {
 187                 return $this->mInterwikiTitles;
 188         }
 189
 190         /**
 191          * Get the list of revision IDs (requested with revids= parameter)
 192          * @return array revID (int) => pageID (int)
 193          */
 194         public function getRevisionIDs() {
 195                 return $this->mGoodRevIDs;
 196         }
 197
 198         /**
 199          * Revision IDs that were not found in the database
 200          * @return array of revision IDs
 201          */
 202         public function getMissingRevisionIDs() {
 203                 return $this->mMissingRevIDs;
 204         }
 205
 206         /**
 207          * Returns the number of revisions (requested with revids= parameter)
 208          */
 209         public function getRevisionCount() {
 210                 return count($this->getRevisionIDs());
 211         }
 212
 213         /**
 214          * Populate from the request parameters
 215          */
 216         public function execute() {
 217                 $this->profileIn();
 218                 $titles = $pageids = $revids = null;
 219                 extract($this->extractRequestParams());
 220
 221                 // Only one of the titles/pageids/revids is allowed at the same time
 222                 $dataSource = null;
 223                 if (isset ($titles))
 224                         $dataSource = 'titles';
 225                 if (isset ($pageids)) {
 226                         if (isset ($dataSource))
 227                                 $this->dieUsage("Cannot use 'pageids' at the same time as '$dataSource'", 'multisource');
 228                         $dataSource = 'pageids';
 229                 }
 230                 if (isset ($revids)) {
 231                         if (isset ($dataSource))
 232                                 $this->dieUsage("Cannot use 'revids' at the same time as '$dataSource'", 'multisource');
 233                         $dataSource = 'revids';
 234                 }
 235
 236                 switch ($dataSource) {
 237                         case 'titles' :
 238                                 $this->initFromTitles($titles);
 239                                 break;
 240                         case 'pageids' :
 241                                 $this->initFromPageIds($pageids);
 242                                 break;
 243                         case 'revids' :
 244                                 if($this->mResolveRedirects)
 245                                         $this->dieUsage('revids may not be used with redirect resolution', 'params');
 246                                 $this->initFromRevIDs($revids);
 247                                 break;
 248                         default :
 249                                 // Do nothing - some queries do not need any of the data sources.
 250                                 break;
 251                 }
 252                 $this->profileOut();
 253         }
 254
 255         /**
 256          * Initialize PageSet from a list of Titles
 257          */
 258         public function populateFromTitles($titles) {
 259                 $this->profileIn();
 260                 $this->initFromTitles($titles);
 261                 $this->profileOut();
 262         }
 263
 264         /**
 265          * Initialize PageSet from a list of Page IDs
 266          */
 267         public function populateFromPageIDs($pageIDs) {
 268                 $this->profileIn();
 269                 $this->initFromPageIds($pageIDs);
 270                 $this->profileOut();
 271         }
 272
 273         /**
 274          * Initialize PageSet from a rowset returned from the database
 275          */
 276         public function populateFromQueryResult($db, $queryResult) {
 277                 $this->profileIn();
 278                 $this->initFromQueryResult($db, $queryResult);
 279                 $this->profileOut();
 280         }
 281
 282         /**
 283          * Initialize PageSet from a list of Revision IDs
 284          */
 285         public function populateFromRevisionIDs($revIDs) {
 286                 $this->profileIn();
 287                 $revIDs = array_map('intval', $revIDs); // paranoia
 288                 $this->initFromRevIDs($revIDs);
 289                 $this->profileOut();
 290         }
 291
 292         /**
 293          * Extract all requested fields from the row received from the database
 294          */
 295         public function processDbRow($row) {
 296
 297                 // Store Title object in various data structures
 298                 $title = Title :: makeTitle($row->page_namespace, $row->page_title);
 299
 300                 $pageId = intval($row->page_id);
 301                 $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
 302                 $this->mTitles[] = $title;
 303
 304                 if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
 305                         $this->mPendingRedirectIDs[$pageId] = $title;
 306                 } else {
 307                         $this->mGoodTitles[$pageId] = $title;
 308                 }
 309
 310                 foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
 311                         $fieldValues[$pageId] = $row-> $fieldName;
 312         }
 313
 314         public function finishPageSetGeneration() {
 315                 $this->profileIn();
 316                 $this->resolvePendingRedirects();
 317                 $this->profileOut();
 318         }
 319
 320         /**
 321          * This method populates internal variables with page information
 322          * based on the given array of title strings.
 323          *
 324          * Steps:
 325          * #1 For each title, get data from `page` table
 326          * #2 If page was not found in the DB, store it as missing
 327          *
 328          * Additionally, when resolving redirects:
 329          * #3 If no more redirects left, stop.
 330          * #4 For each redirect, get its links from `pagelinks` table.
 331          * #5 Substitute the original LinkBatch object with the new list
 332          * #6 Repeat from step #1
 333          */
 334         private function initFromTitles($titles) {
 335
 336                 // Get validated and normalized title objects
 337                 $linkBatch = $this->processTitlesArray($titles);
 338                 if($linkBatch->isEmpty())
 339                         return;
 340
 341                 $db = $this->getDB();
 342                 $set = $linkBatch->constructSet('page', $db);
 343
 344                 // Get pageIDs data from the `page` table
 345                 $this->profileDBIn();
 346                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 347                 $this->profileDBOut();
 348
 349                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 350                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);  // process Titles
 351
 352                 // Resolve any found redirects
 353                 $this->resolvePendingRedirects();
 354         }
 355
 356         private function initFromPageIds($pageids) {
 357                 if(empty($pageids))
 358                         return;
 359
 360                 $pageids = array_map('intval', $pageids); // paranoia
 361                 $set = array (
 362                         'page_id' => $pageids
 363                 );
 364
 365                 $db = $this->getDB();
 366
 367                 // Get pageIDs data from the `page` table
 368                 $this->profileDBIn();
 369                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 370                 $this->profileDBOut();
 371
 372                 $this->initFromQueryResult($db, $res, array_flip($pageids), false);     // process PageIDs
 373
 374                 // Resolve any found redirects
 375                 $this->resolvePendingRedirects();
 376         }
 377
 378         /**
 379          * Iterate through the result of the query on 'page' table,
 380          * and for each row create and store title object and save any extra fields requested.
 381          * @param $db Database
 382          * @param $res DB Query result
 383          * @param $remaining Array of either pageID or ns/title elements (optional).
 384          *        If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
 385          * @param $processTitles bool Must be provided together with $remaining.
 386          *        If true, treat $remaining as an array of [ns][title]
 387          *        If false, treat it as an array of [pageIDs]
 388          * @return Array of redirect IDs (only when resolving redirects)
 389          */
 390         private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
 391                 if (!is_null($remaining) && is_null($processTitles))
 392                         ApiBase :: dieDebug(__METHOD__, 'Missing $processTitles parameter when $remaining is provided');
 393
 394                 while ($row = $db->fetchObject($res)) {
 395
 396                         $pageId = intval($row->page_id);
 397
 398                         // Remove found page from the list of remaining items
 399                         if (isset($remaining)) {
 400                                 if ($processTitles)
 401                                         unset ($remaining[$row->page_namespace][$row->page_title]);
 402                                 else
 403                                         unset ($remaining[$pageId]);
 404                         }
 405
 406                         // Store any extra fields requested by modules
 407                         $this->processDbRow($row);
 408                 }
 409                 $db->freeResult($res);
 410
 411                 if(isset($remaining)) {
 412                         // Any items left in the $remaining list are added as missing
 413                         if($processTitles) {
 414                                 // The remaining titles in $remaining are non-existant pages
 415                                 foreach ($remaining as $ns => $dbkeys) {
 416                                         foreach ( $dbkeys as $dbkey => $unused ) {
 417                                                 $title = Title :: makeTitle($ns, $dbkey);
 418                                                 $this->mAllPages[$ns][$dbkey] = $this->mFakePageId;
 419                                                 $this->mMissingTitles[$this->mFakePageId] = $title;
 420                                                 $this->mFakePageId--;
 421                                                 $this->mTitles[] = $title;
 422                                         }
 423                                 }
 424                         }
 425                         else
 426                         {
 427                                 // The remaining pageids do not exist
 428                                 if(empty($this->mMissingPageIDs))
 429                                         $this->mMissingPageIDs = array_keys($remaining);
 430                                 else
 431                                         $this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
 432                         }
 433                 }
 434         }
 435
 436         private function initFromRevIDs($revids) {
 437
 438                 if(empty($revids))
 439                         return;
 440
 441                 $db = $this->getDB();
 442                 $pageids = array();
 443                 $remaining = array_flip($revids);
 444
 445                 $tables = array('revision');
 446                 $fields = array('rev_id','rev_page');
 447                 $where = array('rev_deleted' => 0, 'rev_id' => $revids);
 448
 449                 // Get pageIDs data from the `page` table
 450                 $this->profileDBIn();
 451                 $res = $db->select( $tables, $fields, $where,  __METHOD__ );
 452                 while ( $row = $db->fetchObject( $res ) ) {
 453                         $revid = intval($row->rev_id);
 454                         $pageid = intval($row->rev_page);
 455                         $this->mGoodRevIDs[$revid] = $pageid;
 456                         $pageids[$pageid] = '';
 457                         unset($remaining[$revid]);
 458                 }
 459                 $db->freeResult( $res );
 460                 $this->profileDBOut();
 461
 462                 $this->mMissingRevIDs = array_keys($remaining);
 463
 464                 // Populate all the page information
 465                 if($this->mResolveRedirects)
 466                         ApiBase :: dieDebug(__METHOD__, 'revids may not be used with redirect resolution');
 467                 $this->initFromPageIds(array_keys($pageids));
 468         }
 469
 470         private function resolvePendingRedirects() {
 471
 472                 if($this->mResolveRedirects) {
 473                         $db = $this->getDB();
 474                         $pageFlds = $this->getPageTableFields();
 475
 476                         // Repeat until all redirects have been resolved
 477                         // The infinite loop is prevented by keeping all known pages in $this->mAllPages
 478                         while (!empty ($this->mPendingRedirectIDs)) {
 479
 480                                 // Resolve redirects by querying the pagelinks table, and repeat the process
 481                                 // Create a new linkBatch object for the next pass
 482                                 $linkBatch = $this->getRedirectTargets();
 483
 484                                 if ($linkBatch->isEmpty())
 485                                         break;
 486
 487                                 $set = $linkBatch->constructSet('page', $db);
 488                                 if(false === $set)
 489                                         break;
 490
 491                                 // Get pageIDs data from the `page` table
 492                                 $this->profileDBIn();
 493                                 $res = $db->select('page', $pageFlds, $set, __METHOD__);
 494                                 $this->profileDBOut();
 495
 496                                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 497                                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);
 498                         }
 499                 }
 500         }
 501
 502         private function getRedirectTargets() {
 503
 504                 $linkBatch = new LinkBatch();
 505                 $db = $this->getDB();
 506
 507                 // find redirect targets for all redirect pages
 508                 $this->profileDBIn();
 509                 $res = $db->select('pagelinks', array (
 510                         'pl_from',
 511                         'pl_namespace',
 512                         'pl_title'
 513                 ), array (
 514                         'pl_from' => array_keys($this->mPendingRedirectIDs
 515                 )), __METHOD__);
 516                 $this->profileDBOut();
 517
 518                 while ($row = $db->fetchObject($res)) {
 519
 520                         $plfrom = intval($row->pl_from);
 521
 522                         // Bug 7304 workaround
 523                         // ( http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 )
 524                         // A redirect page may have more than one link.
 525                         // This code will only use the first link returned.
 526                         if (isset ($this->mPendingRedirectIDs[$plfrom])) { // remove line when bug 7304 is fixed
 527
 528                                 $titleStrFrom = $this->mPendingRedirectIDs[$plfrom]->getPrefixedText();
 529                                 $titleStrTo = Title :: makeTitle($row->pl_namespace, $row->pl_title)->getPrefixedText();
 530                                 unset ($this->mPendingRedirectIDs[$plfrom]); // remove line when bug 7304 is fixed
 531
 532                                 // Avoid an infinite loop by checking if we have already processed this target
 533                                 if (!isset ($this->mAllPages[$row->pl_namespace][$row->pl_title])) {
 534                                         $linkBatch->add($row->pl_namespace, $row->pl_title);
 535                                 }
 536                         } else {
 537                                 // This redirect page has more than one link.
 538                                 // This is very slow, but safer until bug 7304 is resolved
 539                                 $title = Title :: newFromID($plfrom);
 540                                 $titleStrFrom = $title->getPrefixedText();
 541
 542                                 $article = new Article($title);
 543                                 $text = $article->getContent();
 544                                 $titleTo = Title :: newFromRedirect($text);
 545                                 $titleStrTo = $titleTo->getPrefixedText();
 546
 547                                 if (is_null($titleStrTo))
 548                                         ApiBase :: dieDebug(__METHOD__, 'Bug7304 workaround: redir target from {$title->getPrefixedText()} not found');
 549
 550                                 // Avoid an infinite loop by checking if we have already processed this target
 551                                 if (!isset ($this->mAllPages[$titleTo->getNamespace()][$titleTo->getDBkey()])) {
 552                                         $linkBatch->addObj($titleTo);
 553                                 }
 554                         }
 555
 556                         $this->mRedirectTitles[$titleStrFrom] = $titleStrTo;
 557                 }
 558                 $db->freeResult($res);
 559
 560                 // All IDs must exist in the page table
 561                 if (!empty($this->mPendingRedirectIDs[$plfrom]))
 562                         ApiBase :: dieDebug(__METHOD__, 'Invalid redirect IDs were found');
 563
 564                 return $linkBatch;
 565         }
 566
 567         /**
 568          * Given an array of title strings, convert them into Title objects.
 569          * Alternativelly, an array of Title objects may be given.
 570          * This method validates access rights for the title,
 571          * and appends normalization values to the output.
 572          *
 573          * @return LinkBatch of title objects.
 574          */
 575         private function processTitlesArray($titles) {
 576
 577                 $linkBatch = new LinkBatch();
 578
 579                 foreach ($titles as $title) {
 580
 581                         $titleObj = is_string($title) ? Title :: newFromText($title) : $title;
 582                         if (!$titleObj)
 583                                 $this->dieUsage("bad title", 'invalidtitle');
 584
 585                         $iw = $titleObj->getInterwiki();
 586                         if (!empty($iw)) {
 587                                 // This title is an interwiki link.
 588                                 $this->mInterwikiTitles[$titleObj->getPrefixedText()] = $iw;
 589                         } else {
 590
 591                                 // Validation
 592                                 if ($titleObj->getNamespace() < 0)
 593                                         $this->dieUsage("No support for special pages has been implemented", 'unsupportednamespace');
 594
 595                                 $linkBatch->addObj($titleObj);
 596                         }
 597
 598                         // Make sure we remember the original title that was given to us
 599                         // This way the caller can correlate new titles with the originally requested,
 600                         // i.e. namespace is localized or capitalization is different
 601                         if (is_string($title) && $title !== $titleObj->getPrefixedText()) {
 602                                 $this->mNormalizedTitles[$title] = $titleObj->getPrefixedText();
 603                         }
 604                 }
 605
 606                 return $linkBatch;
 607         }
 608
 609         protected function getAllowedParams() {
 610                 return array (
 611                         'titles' => array (
 612                                 ApiBase :: PARAM_ISMULTI => true
 613                         ),
 614                         'pageids' => array (
 615                                 ApiBase :: PARAM_TYPE => 'integer',
 616                                 ApiBase :: PARAM_ISMULTI => true
 617                         ),
 618                         'revids' => array (
 619                                 ApiBase :: PARAM_TYPE => 'integer',
 620                                 ApiBase :: PARAM_ISMULTI => true
 621                         )
 622                 );
 623         }
 624
 625         protected function getParamDescription() {
 626                 return array (
 627                         'titles' => 'A list of titles to work on',
 628                         'pageids' => 'A list of page IDs to work on',
 629                         'revids' => 'A list of revision IDs to work on'
 630                 );
 631         }
 632
 633         public function getVersion() {
 634                 return __CLASS__ . ': $Id$';
 635         }
 636 }
 637