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