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 <FirstnameLastname@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  * @addtogroup API
  33  */
  34 class ApiPageSet extends ApiQueryBase {
  35
  36         private $mAllPages; // [ns][dbkey] => page_id or 0 when missing
  37         private $mTitles, $mGoodTitles, $mMissingTitles, $mMissingPageIDs, $mRedirectTitles, $mNormalizedTitles;
  38         private $mResolveRedirects, $mPendingRedirectIDs;
  39         private $mGoodRevIDs, $mMissingRevIDs;
  40
  41         private $mRequestedPageFields;
  42
  43         public function __construct($query, $resolveRedirects = false) {
  44                 parent :: __construct($query, __CLASS__);
  45
  46                 $this->mAllPages = array ();
  47                 $this->mTitles = array();
  48                 $this->mGoodTitles = array ();
  49                 $this->mMissingTitles = array ();
  50                 $this->mMissingPageIDs = array ();
  51                 $this->mRedirectTitles = array ();
  52                 $this->mNormalizedTitles = array ();
  53                 $this->mGoodRevIDs = array();
  54                 $this->mMissingRevIDs = array();
  55
  56                 $this->mRequestedPageFields = array ();
  57                 $this->mResolveRedirects = $resolveRedirects;
  58                 if($resolveRedirects)
  59                         $this->mPendingRedirectIDs = array();
  60         }
  61
  62         public function isResolvingRedirects() {
  63                 return $this->mResolveRedirects;
  64         }
  65
  66         public function requestField($fieldName) {
  67                 $this->mRequestedPageFields[$fieldName] = null;
  68         }
  69
  70         public function getCustomField($fieldName) {
  71                 return $this->mRequestedPageFields[$fieldName];
  72         }
  73
  74         /**
  75          * Get fields that modules have requested from the page table
  76          */
  77         public function getPageTableFields() {
  78                 // Ensure we get minimum required fields
  79                 $pageFlds = array (
  80                         'page_id' => null,
  81                         'page_namespace' => null,
  82                         'page_title' => null
  83                 );
  84
  85                 // only store non-default fields
  86                 $this->mRequestedPageFields = array_diff_key($this->mRequestedPageFields, $pageFlds);
  87
  88                 if ($this->mResolveRedirects)
  89                         $pageFlds['page_is_redirect'] = null;
  90
  91                 return array_keys(array_merge($pageFlds, $this->mRequestedPageFields));
  92         }
  93
  94         /**
  95          * All Title objects provided.
  96          * @return array of Title objects
  97          */
  98         public function getTitles() {
  99                 return $this->mTitles;
 100         }
 101
 102         /**
 103          * Returns the number of unique pages (not revisions) in the set.
 104          */
 105         public function getTitleCount() {
 106                 return count($this->mTitles);
 107         }
 108
 109         /**
 110          * Title objects that were found in the database.
 111          * @return array page_id (int) => Title (obj)
 112          */
 113         public function getGoodTitles() {
 114                 return $this->mGoodTitles;
 115         }
 116
 117         /**
 118          * Returns the number of found unique pages (not revisions) in the set.
 119          */
 120         public function getGoodTitleCount() {
 121                 return count($this->mGoodTitles);
 122         }
 123
 124         /**
 125          * Title objects that were NOT found in the database.
 126          * @return array of Title objects
 127          */
 128         public function getMissingTitles() {
 129                 return $this->mMissingTitles;
 130         }
 131
 132         /**
 133          * Page IDs that were not found in the database
 134          * @return array of page IDs
 135          */
 136         public function getMissingPageIDs() {
 137                 return $this->mMissingPageIDs;
 138         }
 139
 140         /**
 141          * Get a list of redirects when doing redirect resolution
 142          * @return array prefixed_title (string) => prefixed_title (string)
 143          */
 144         public function getRedirectTitles() {
 145                 return $this->mRedirectTitles;
 146         }
 147
 148         /**
 149          * Get a list of title normalizations - maps the title given
 150          * with its normalized version.
 151          * @return array raw_prefixed_title (string) => prefixed_title (string)
 152          */
 153         public function getNormalizedTitles() {
 154                 return $this->mNormalizedTitles;
 155         }
 156
 157         /**
 158          * Get the list of revision IDs (requested with revids= parameter)
 159          * @return array revID (int) => pageID (int)
 160          */
 161         public function getRevisionIDs() {
 162                 return $this->mGoodRevIDs;
 163         }
 164
 165         /**
 166          * Revision IDs that were not found in the database
 167          * @return array of revision IDs
 168          */
 169         public function getMissingRevisionIDs() {
 170                 return $this->mMissingRevIDs;
 171         }
 172
 173         /**
 174          * Returns the number of revisions (requested with revids= parameter)
 175          */
 176         public function getRevisionCount() {
 177                 return count($this->getRevisionIDs());
 178         }
 179
 180         /**
 181          * Populate from the request parameters
 182          */
 183         public function execute() {
 184                 $this->profileIn();
 185                 $titles = $pageids = $revids = null;
 186                 extract($this->extractRequestParams());
 187
 188                 // Only one of the titles/pageids/revids is allowed at the same time
 189                 $dataSource = null;
 190                 if (isset ($titles))
 191                         $dataSource = 'titles';
 192                 if (isset ($pageids)) {
 193                         if (isset ($dataSource))
 194                                 $this->dieUsage("Cannot use 'pageids' at the same time as '$dataSource'", 'multisource');
 195                         $dataSource = 'pageids';
 196                 }
 197                 if (isset ($revids)) {
 198                         if (isset ($dataSource))
 199                                 $this->dieUsage("Cannot use 'revids' at the same time as '$dataSource'", 'multisource');
 200                         $dataSource = 'revids';
 201                 }
 202
 203                 switch ($dataSource) {
 204                         case 'titles' :
 205                                 $this->initFromTitles($titles);
 206                                 break;
 207                         case 'pageids' :
 208                                 $this->initFromPageIds($pageids);
 209                                 break;
 210                         case 'revids' :
 211                                 if($this->mResolveRedirects)
 212                                         $this->dieUsage('revids may not be used with redirect resolution', 'params');
 213                                 $this->initFromRevIDs($revids);
 214                                 break;
 215                         default :
 216                                 // Do nothing - some queries do not need any of the data sources.
 217                                 break;
 218                 }
 219                 $this->profileOut();
 220         }
 221
 222         /**
 223          * Initialize PageSet from a list of Titles
 224          */
 225         public function populateFromTitles($titles) {
 226                 $this->profileIn();
 227                 $this->initFromTitles($titles);
 228                 $this->profileOut();
 229         }
 230
 231         /**
 232          * Initialize PageSet from a list of Page IDs
 233          */
 234         public function populateFromPageIDs($pageIDs) {
 235                 $this->profileIn();
 236                 $pageIDs = array_map('intval', $pageIDs); // paranoia
 237                 $this->initFromPageIds($pageIDs);
 238                 $this->profileOut();
 239         }
 240
 241         /**
 242          * Initialize PageSet from a rowset returned from the database
 243          */
 244         public function populateFromQueryResult($db, $queryResult) {
 245                 $this->profileIn();
 246                 $this->initFromQueryResult($db, $queryResult);
 247                 $this->profileOut();
 248         }
 249
 250         /**
 251          * Initialize PageSet from a list of Revision IDs
 252          */
 253         public function populateFromRevisionIDs($revIDs) {
 254                 $this->profileIn();
 255                 $revIDs = array_map('intval', $revIDs); // paranoia
 256                 $this->initFromRevIDs($revIDs);
 257                 $this->profileOut();
 258         }
 259
 260         /**
 261          * Extract all requested fields from the row received from the database
 262          */
 263         public function processDbRow($row) {
 264
 265                 // Store Title object in various data structures
 266                 $title = Title :: makeTitle($row->page_namespace, $row->page_title);
 267
 268                 // skip any pages that user has no rights to read
 269                 if ($title->userCanRead()) {
 270
 271                         $pageId = intval($row->page_id);
 272                         $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
 273                         $this->mTitles[] = $title;
 274
 275                         if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
 276                                 $this->mPendingRedirectIDs[$pageId] = $title;
 277                         } else {
 278                                 $this->mGoodTitles[$pageId] = $title;
 279                         }
 280
 281                         foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
 282                                 $fieldValues[$pageId] = $row-> $fieldName;
 283                 }
 284         }
 285
 286         public function finishPageSetGeneration() {
 287                 $this->profileIn();
 288                 $this->resolvePendingRedirects();
 289                 $this->profileOut();
 290         }
 291
 292         /**
 293          * This method populates internal variables with page information
 294          * based on the given array of title strings.
 295          *
 296          * Steps:
 297          * #1 For each title, get data from `page` table
 298          * #2 If page was not found in the DB, store it as missing
 299          *
 300          * Additionally, when resolving redirects:
 301          * #3 If no more redirects left, stop.
 302          * #4 For each redirect, get its links from `pagelinks` table.
 303          * #5 Substitute the original LinkBatch object with the new list
 304          * #6 Repeat from step #1
 305          */
 306         private function initFromTitles($titles) {
 307
 308                 // Get validated and normalized title objects
 309                 $linkBatch = $this->processTitlesArray($titles);
 310                 if($linkBatch->isEmpty())
 311                         return;
 312
 313                 $db = $this->getDB();
 314                 $set = $linkBatch->constructSet('page', $db);
 315
 316                 // Get pageIDs data from the `page` table
 317                 $this->profileDBIn();
 318                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 319                 $this->profileDBOut();
 320
 321                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 322                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);  // process Titles
 323
 324                 // Resolve any found redirects
 325                 $this->resolvePendingRedirects();
 326         }
 327
 328         private function initFromPageIds($pageids) {
 329                 if(empty($pageids))
 330                         return;
 331
 332                 $set = array (
 333                         'page_id' => $pageids
 334                 );
 335
 336                 $db = $this->getDB();
 337
 338                 // Get pageIDs data from the `page` table
 339                 $this->profileDBIn();
 340                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 341                 $this->profileDBOut();
 342
 343                 $this->initFromQueryResult($db, $res, array_flip($pageids), false);     // process PageIDs
 344
 345                 // Resolve any found redirects
 346                 $this->resolvePendingRedirects();
 347         }
 348
 349         /**
 350          * Iterate through the result of the query on 'page' table,
 351          * and for each row create and store title object and save any extra fields requested.
 352          * @param $db Database
 353          * @param $res DB Query result
 354          * @param $remaining Array of either pageID or ns/title elements (optional).
 355          *        If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
 356          * @param $processTitles bool Must be provided together with $remaining.
 357          *        If true, treat $remaining as an array of [ns][title]
 358          *        If false, treat it as an array of [pageIDs]
 359          * @return Array of redirect IDs (only when resolving redirects)
 360          */
 361         private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
 362                 if (!is_null($remaining) && is_null($processTitles))
 363                         ApiBase :: dieDebug(__METHOD__, 'Missing $processTitles parameter when $remaining is provided');
 364
 365                 while ($row = $db->fetchObject($res)) {
 366
 367                         $pageId = intval($row->page_id);
 368
 369                         // Remove found page from the list of remaining items
 370                         if (isset($remaining)) {
 371                                 if ($processTitles)
 372                                         unset ($remaining[$row->page_namespace][$row->page_title]);
 373                                 else
 374                                         unset ($remaining[$pageId]);
 375                         }
 376
 377                         // Store any extra fields requested by modules
 378                         $this->processDbRow($row);
 379                 }
 380                 $db->freeResult($res);
 381
 382                 if(isset($remaining)) {
 383                         // Any items left in the $remaining list are added as missing
 384                         if($processTitles) {
 385                                 // The remaining titles in $remaining are non-existant pages
 386                                 foreach ($remaining as $ns => $dbkeys) {
 387                                         foreach ( $dbkeys as $dbkey => $unused ) {
 388                                                 $title = Title :: makeTitle($ns, $dbkey);
 389                                                 $this->mMissingTitles[] = $title;
 390                                                 $this->mAllPages[$ns][$dbkey] = 0;
 391                                                 $this->mTitles[] = $title;
 392                                         }
 393                                 }
 394                         }
 395                         else
 396                         {
 397                                 // The remaining pageids do not exist
 398                                 if(empty($this->mMissingPageIDs))
 399                                         $this->mMissingPageIDs = array_keys($remaining);
 400                                 else
 401                                         $this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
 402                         }
 403                 }
 404         }
 405
 406         private function initFromRevIDs($revids) {
 407
 408                 if(empty($revids))
 409                         return;
 410
 411                 $db = $this->getDB();
 412                 $pageids = array();
 413                 $remaining = array_flip($revids);
 414
 415                 $tables = array('revision');
 416                 $fields = array('rev_id','rev_page');
 417                 $where = array('rev_deleted' => 0, 'rev_id' => $revids);
 418
 419                 // Get pageIDs data from the `page` table
 420                 $this->profileDBIn();
 421                 $res = $db->select( $tables, $fields, $where,  __METHOD__ );
 422                 while ( $row = $db->fetchObject( $res ) ) {
 423                         $revid = intval($row->rev_id);
 424                         $pageid = intval($row->rev_page);
 425                         $this->mGoodRevIDs[$revid] = $pageid;
 426                         $pageids[$pageid] = '';
 427                         unset($remaining[$revid]);
 428                 }
 429                 $db->freeResult( $res );
 430                 $this->profileDBOut();
 431
 432                 $this->mMissingRevIDs = array_keys($remaining);
 433
 434                 // Populate all the page information
 435                 if($this->mResolveRedirects)
 436                         ApiBase :: dieDebug(__METHOD__, 'revids may not be used with redirect resolution');
 437                 $this->initFromPageIds(array_keys($pageids));
 438         }
 439
 440         private function resolvePendingRedirects() {
 441
 442                 if($this->mResolveRedirects) {
 443                         $db = $this->getDB();
 444                         $pageFlds = $this->getPageTableFields();
 445
 446                         // Repeat until all redirects have been resolved
 447                         // The infinite loop is prevented by keeping all known pages in $this->mAllPages
 448                         while (!empty ($this->mPendingRedirectIDs)) {
 449
 450                                 // Resolve redirects by querying the pagelinks table, and repeat the process
 451                                 // Create a new linkBatch object for the next pass
 452                                 $linkBatch = $this->getRedirectTargets();
 453
 454                                 if ($linkBatch->isEmpty())
 455                                         break;
 456
 457                                 $set = $linkBatch->constructSet('page', $db);
 458                                 if(false === $set)
 459                                         break;
 460
 461                                 // Get pageIDs data from the `page` table
 462                                 $this->profileDBIn();
 463                                 $res = $db->select('page', $pageFlds, $set, __METHOD__);
 464                                 $this->profileDBOut();
 465
 466                                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 467                                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);
 468                         }
 469                 }
 470         }
 471
 472         private function getRedirectTargets() {
 473
 474                 $linkBatch = new LinkBatch();
 475                 $db = $this->getDB();
 476
 477                 // find redirect targets for all redirect pages
 478                 $this->profileDBIn();
 479                 $res = $db->select('pagelinks', array (
 480                         'pl_from',
 481                         'pl_namespace',
 482                         'pl_title'
 483                 ), array (
 484                         'pl_from' => array_keys($this->mPendingRedirectIDs
 485                 )), __METHOD__);
 486                 $this->profileDBOut();
 487
 488                 while ($row = $db->fetchObject($res)) {
 489
 490                         $plfrom = intval($row->pl_from);
 491
 492                         // Bug 7304 workaround
 493                         // ( http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 )
 494                         // A redirect page may have more than one link.
 495                         // This code will only use the first link returned.
 496                         if (isset ($this->mPendingRedirectIDs[$plfrom])) { // remove line when bug 7304 is fixed
 497
 498                                 $titleStrFrom = $this->mPendingRedirectIDs[$plfrom]->getPrefixedText();
 499                                 $titleStrTo = Title :: makeTitle($row->pl_namespace, $row->pl_title)->getPrefixedText();
 500                                 unset ($this->mPendingRedirectIDs[$plfrom]); // remove line when bug 7304 is fixed
 501
 502                                 // Avoid an infinite loop by checking if we have already processed this target
 503                                 if (!isset ($this->mAllPages[$row->pl_namespace][$row->pl_title])) {
 504                                         $linkBatch->add($row->pl_namespace, $row->pl_title);
 505                                 }
 506                         } else {
 507                                 // This redirect page has more than one link.
 508                                 // This is very slow, but safer until bug 7304 is resolved
 509                                 $title = Title :: newFromID($plfrom);
 510                                 $titleStrFrom = $title->getPrefixedText();
 511
 512                                 $article = new Article($title);
 513                                 $text = $article->getContent();
 514                                 $titleTo = Title :: newFromRedirect($text);
 515                                 $titleStrTo = $titleTo->getPrefixedText();
 516
 517                                 if (is_null($titleStrTo))
 518                                         ApiBase :: dieDebug(__METHOD__, 'Bug7304 workaround: redir target from {$title->getPrefixedText()} not found');
 519
 520                                 // Avoid an infinite loop by checking if we have already processed this target
 521                                 if (!isset ($this->mAllPages[$titleTo->getNamespace()][$titleTo->getDBkey()])) {
 522                                         $linkBatch->addObj($titleTo);
 523                                 }
 524                         }
 525
 526                         $this->mRedirectTitles[$titleStrFrom] = $titleStrTo;
 527                 }
 528                 $db->freeResult($res);
 529
 530                 // All IDs must exist in the page table
 531                 if (!empty($this->mPendingRedirectIDs[$plfrom]))
 532                         ApiBase :: dieDebug(__METHOD__, 'Invalid redirect IDs were found');
 533
 534                 return $linkBatch;
 535         }
 536
 537         /**
 538          * Given an array of title strings, convert them into Title objects.
 539          * This method validates access rights for the title,
 540          * and appends normalization values to the output.
 541          *
 542          * @return LinkBatch of title objects.
 543          */
 544         private function processTitlesArray($titles) {
 545
 546                 $linkBatch = new LinkBatch();
 547
 548                 foreach ($titles as $title) {
 549
 550                         $titleObj = is_string($title) ? Title :: newFromText($title) : $title;
 551
 552                         // Validation
 553                         if (!$titleObj)
 554                                 $this->dieUsage("bad title $titleString", 'invalidtitle');
 555                         if ($titleObj->getNamespace() < 0)
 556                                 $this->dieUsage("No support for special page $titleString has been implemented", 'unsupportednamespace');
 557                         if (!$titleObj->userCanRead())
 558                                 $this->dieUsage("No read permission for $titleString", 'titleaccessdenied');
 559
 560                         $linkBatch->addObj($titleObj);
 561
 562                         // Make sure we remember the original title that was given to us
 563                         // This way the caller can correlate new titles with the originally requested,
 564                         // i.e. namespace is localized or capitalization is different
 565                         if (is_string($title) && $title !== $titleObj->getPrefixedText()) {
 566                                 $this->mNormalizedTitles[$title] = $titleObj->getPrefixedText();
 567                         }
 568                 }
 569
 570                 return $linkBatch;
 571         }
 572
 573         public static function debugPrint($name = 'unknown') {
 574                 ApiBase::debugPrint($this->mAllPages, $name);
 575         }
 576
 577         protected function getAllowedParams() {
 578                 return array (
 579                         'titles' => array (
 580                                 ApiBase :: PARAM_ISMULTI => true
 581                         ),
 582                         'pageids' => array (
 583                                 ApiBase :: PARAM_TYPE => 'integer',
 584                                 ApiBase :: PARAM_ISMULTI => true
 585                         ),
 586                         'revids' => array (
 587                                 ApiBase :: PARAM_TYPE => 'integer',
 588                                 ApiBase :: PARAM_ISMULTI => true
 589                         )
 590                 );
 591         }
 592
 593         protected function getParamDescription() {
 594                 return array (
 595                         'titles' => 'A list of titles to work on',
 596                         'pageids' => 'A list of page IDs to work on',
 597                         'revids' => 'A list of revision IDs to work on'
 598                 );
 599         }
 600
 601         public function getVersion() {
 602                 return __CLASS__ . ': $Id$';
 603         }
 604 }
 605 ?>