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