includes/api/ApiPageSet.php

   1 <?php
   2
   3
   4 /*
   5  * Created on Sep 24, 2006
   6  *
   7  * API for MediaWiki 1.8+
   8  *
   9  * Copyright (C) 2006 Yuri Astrakhan <FirstnameLastname@gmail.com>
  10  *
  11  * This program is free software; you can redistribute it and/or modify
  12  * it under the terms of the GNU General Public License as published by
  13  * the Free Software Foundation; either version 2 of the License, or
  14  * (at your option) any later version.
  15  *
  16  * This program is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  19  * GNU General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU General Public License along
  22  * with this program; if not, write to the Free Software Foundation, Inc.,
  23  * 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  24  * http://www.gnu.org/copyleft/gpl.html
  25  */
  26
  27 if (!defined('MEDIAWIKI')) {
  28         // Eclipse helper - will be ignored in production
  29         require_once ('ApiQueryBase.php');
  30 }
  31
  32 class ApiPageSet extends ApiQueryBase {
  33
  34         private $mAllPages; // [ns][dbkey] => page_id or 0 when missing
  35         private $mTitles, $mGoodTitles, $mMissingTitles, $mMissingPageIDs, $mRedirectTitles, $mNormalizedTitles;
  36         private $mResolveRedirects, $mPendingRedirectIDs;
  37         private $mGoodRevIDs, $mMissingRevIDs;
  38
  39         private $mRequestedPageFields;
  40
  41         public function __construct($query, $resolveRedirects = false) {
  42                 parent :: __construct($query, __CLASS__);
  43
  44                 $this->mAllPages = array ();
  45                 $this->mTitles = array();
  46                 $this->mGoodTitles = array ();
  47                 $this->mMissingTitles = array ();
  48                 $this->mMissingPageIDs = array ();
  49                 $this->mRedirectTitles = array ();
  50                 $this->mNormalizedTitles = array ();
  51                 $this->mGoodRevIDs = array();
  52                 $this->mMissingRevIDs = array();
  53
  54                 $this->mRequestedPageFields = array ();
  55                 $this->mResolveRedirects = $resolveRedirects;
  56                 if($resolveRedirects)
  57                         $this->mPendingRedirectIDs = array();
  58         }
  59
  60         public function isResolvingRedirects() {
  61                 return $this->mResolveRedirects;
  62         }
  63
  64         public function requestField($fieldName) {
  65                 $this->mRequestedPageFields[$fieldName] = null;
  66         }
  67
  68         public function getCustomField($fieldName) {
  69                 return $this->mRequestedPageFields[$fieldName];
  70         }
  71
  72         /**
  73          * Get fields that modules have requested from the page table
  74          */
  75         public function getPageTableFields() {
  76                 // Ensure we get minimum required fields
  77                 $pageFlds = array (
  78                         'page_id' => null,
  79                         'page_namespace' => null,
  80                         'page_title' => null
  81                 );
  82
  83                 // only store non-default fields
  84                 $this->mRequestedPageFields = array_diff_key($this->mRequestedPageFields, $pageFlds);
  85
  86                 if ($this->mResolveRedirects)
  87                         $pageFlds['page_is_redirect'] = null;
  88
  89                 return array_keys(array_merge($pageFlds, $this->mRequestedPageFields));
  90         }
  91
  92         /**
  93          * All Title objects provided.
  94          * @return array of Title objects
  95          */
  96         public function getTitles() {
  97                 return $this->mTitles;
  98         }
  99
 100         /**
 101          * Returns the number of unique pages (not revisions) in the set.
 102          */
 103         public function getTitleCount() {
 104                 return count($this->mTitles);
 105         }
 106
 107         /**
 108          * Title objects that were found in the database.
 109          * @return array page_id (int) => Title (obj)
 110          */
 111         public function getGoodTitles() {
 112                 return $this->mGoodTitles;
 113         }
 114
 115         /**
 116          * Returns the number of found unique pages (not revisions) in the set.
 117          */
 118         public function getGoodTitleCount() {
 119                 return count($this->mGoodTitles);
 120         }
 121
 122         /**
 123          * Title objects that were NOT found in the database.
 124          * @return array of Title objects
 125          */
 126         public function getMissingTitles() {
 127                 return $this->mMissingTitles;
 128         }
 129
 130         /**
 131          * Page IDs that were not found in the database
 132          * @return array of page IDs
 133          */
 134         public function getMissingPageIDs() {
 135                 return $this->mMissingPageIDs;
 136         }
 137
 138         /**
 139          * Get a list of redirects when doing redirect resolution
 140          * @return array prefixed_title (string) => prefixed_title (string)
 141          */
 142         public function getRedirectTitles() {
 143                 return $this->mRedirectTitles;
 144         }
 145
 146         /**
 147          * Get a list of title normalizations - maps the title given
 148          * with its normalized version.
 149          * @return array raw_prefixed_title (string) => prefixed_title (string)
 150          */
 151         public function getNormalizedTitles() {
 152                 return $this->mNormalizedTitles;
 153         }
 154
 155         /**
 156          * Get the list of revision IDs (requested with revids= parameter)
 157          * @return array revID (int) => pageID (int)
 158          */
 159         public function getRevisionIDs() {
 160                 return $this->mGoodRevIDs;
 161         }
 162
 163         /**
 164          * Revision IDs that were not found in the database
 165          * @return array of revision IDs
 166          */
 167         public function getMissingRevisionIDs() {
 168                 return $this->mMissingRevIDs;
 169         }
 170
 171         /**
 172          * Returns the number of revisions (requested with revids= parameter)
 173          */
 174         public function getRevisionCount() {
 175                 return count($this->getRevisionIDs());
 176         }
 177
 178         /**
 179          * Populate from the request parameters
 180          */
 181         public function execute() {
 182                 $this->profileIn();
 183                 $titles = $pageids = $revids = null;
 184                 extract($this->extractRequestParams());
 185
 186                 // Only one of the titles/pageids/revids is allowed at the same time
 187                 $dataSource = null;
 188                 if (isset ($titles))
 189                         $dataSource = 'titles';
 190                 if (isset ($pageids)) {
 191                         if (isset ($dataSource))
 192                                 $this->dieUsage("Cannot use 'pageids' at the same time as '$dataSource'", 'multisource');
 193                         $dataSource = 'pageids';
 194                 }
 195                 if (isset ($revids)) {
 196                         if (isset ($dataSource))
 197                                 $this->dieUsage("Cannot use 'revids' at the same time as '$dataSource'", 'multisource');
 198                         $dataSource = 'revids';
 199                 }
 200
 201                 switch ($dataSource) {
 202                         case 'titles' :
 203                                 $this->initFromTitles($titles);
 204                                 break;
 205                         case 'pageids' :
 206                                 $this->initFromPageIds($pageids);
 207                                 break;
 208                         case 'revids' :
 209                                 if($this->mResolveRedirects)
 210                                         $this->dieUsage('revids may not be used with redirect resolution', 'params');
 211                                 $this->initFromRevIDs($revids);
 212                                 break;
 213                         default :
 214                                 // Do nothing - some queries do not need any of the data sources.
 215                                 break;
 216                 }
 217                 $this->profileOut();
 218         }
 219
 220         /**
 221          * Initialize PageSet from a list of Titles
 222          */
 223         public function populateFromTitles($titles) {
 224                 $this->profileIn();
 225                 $this->initFromTitles($titles);
 226                 $this->profileOut();
 227         }
 228
 229         /**
 230          * Initialize PageSet from a list of Page IDs
 231          */
 232         public function populateFromPageIDs($pageIDs) {
 233                 $this->profileIn();
 234                 $pageIDs = array_map('intval', $pageIDs); // paranoia
 235                 $this->initFromPageIds($pageIDs);
 236                 $this->profileOut();
 237         }
 238
 239         /**
 240          * Initialize PageSet from a rowset returned from the database
 241          */
 242         public function populateFromQueryResult($db, $queryResult) {
 243                 $this->profileIn();
 244                 $this->initFromQueryResult($db, $queryResult);
 245                 $this->profileOut();
 246         }
 247
 248         /**
 249          * Initialize PageSet from a list of Revision IDs
 250          */
 251         public function populateFromRevisionIDs($revIDs) {
 252                 $this->profileIn();
 253                 $revIDs = array_map('intval', $revIDs); // paranoia
 254                 $this->initFromRevIDs($revIDs);
 255                 $this->profileOut();
 256         }
 257
 258         /**
 259          * Extract all requested fields from the row received from the database
 260          */
 261         public function processDbRow($row) {
 262
 263                 // Store Title object in various data structures
 264                 $title = Title :: makeTitle($row->page_namespace, $row->page_title);
 265
 266                 // skip any pages that user has no rights to read
 267                 if ($title->userCanRead()) {
 268
 269                         $pageId = intval($row->page_id);
 270                         $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
 271                         $this->mTitles[] = $title;
 272
 273                         if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
 274                                 $this->mPendingRedirectIDs[$pageId] = $title;
 275                         } else {
 276                                 $this->mGoodTitles[$pageId] = $title;
 277                         }
 278
 279                         foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
 280                                 $fieldValues[$pageId] = $row-> $fieldName;
 281                 }
 282         }
 283
 284         public function finishPageSetGeneration() {
 285                 $this->profileIn();
 286                 $this->resolvePendingRedirects();
 287                 $this->profileOut();
 288         }
 289
 290         /**
 291          * This method populates internal variables with page information
 292          * based on the given array of title strings.
 293          *
 294          * Steps:
 295          * #1 For each title, get data from `page` table
 296          * #2 If page was not found in the DB, store it as missing
 297          *
 298          * Additionally, when resolving redirects:
 299          * #3 If no more redirects left, stop.
 300          * #4 For each redirect, get its links from `pagelinks` table.
 301          * #5 Substitute the original LinkBatch object with the new list
 302          * #6 Repeat from step #1
 303          */
 304         private function initFromTitles($titles) {
 305
 306                 // Get validated and normalized title objects
 307                 $linkBatch = $this->processTitlesStrArray($titles);
 308                 if($linkBatch->isEmpty())
 309                         return;
 310
 311                 $db = $this->getDB();
 312                 $set = $linkBatch->constructSet('page', $db);
 313
 314                 // Get pageIDs data from the `page` table
 315                 $this->profileDBIn();
 316                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 317                 $this->profileDBOut();
 318
 319                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 320                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);  // process Titles
 321
 322                 // Resolve any found redirects
 323                 $this->resolvePendingRedirects();
 324         }
 325
 326         private function initFromPageIds($pageids) {
 327                 if(empty($pageids))
 328                         return;
 329
 330                 $set = array (
 331                         'page_id' => $pageids
 332                 );
 333
 334                 $db = $this->getDB();
 335
 336                 // Get pageIDs data from the `page` table
 337                 $this->profileDBIn();
 338                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 339                 $this->profileDBOut();
 340
 341                 $this->initFromQueryResult($db, $res, array_flip($pageids), false);     // process PageIDs
 342
 343                 // Resolve any found redirects
 344                 $this->resolvePendingRedirects();
 345         }
 346
 347         /**
 348          * Iterate through the result of the query on 'page' table,
 349          * and for each row create and store title object and save any extra fields requested.
 350          * @param $db Database
 351          * @param $res DB Query result
 352          * @param $remaining Array of either pageID or ns/title elements (optional).
 353          *        If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
 354          * @param $processTitles bool Must be provided together with $remaining.
 355          *        If true, treat $remaining as an array of [ns][title]
 356          *        If false, treat it as an array of [pageIDs]
 357          * @return Array of redirect IDs (only when resolving redirects)
 358          */
 359         private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
 360                 if (!is_null($remaining) && is_null($processTitles))
 361                         ApiBase :: dieDebug(__METHOD__, 'Missing $processTitles parameter when $remaining is provided');
 362
 363                 while ($row = $db->fetchObject($res)) {
 364
 365                         $pageId = intval($row->page_id);
 366
 367                         // Remove found page from the list of remaining items
 368                         if (isset($remaining)) {
 369                                 if ($processTitles)
 370                                         unset ($remaining[$row->page_namespace][$row->page_title]);
 371                                 else
 372                                         unset ($remaining[$pageId]);
 373                         }
 374
 375                         // Store any extra fields requested by modules
 376                         $this->processDbRow($row);
 377                 }
 378                 $db->freeResult($res);
 379
 380                 if(isset($remaining)) {
 381                         // Any items left in the $remaining list are added as missing
 382                         if($processTitles) {
 383                                 // The remaining titles in $remaining are non-existant pages
 384                                 foreach ($remaining as $ns => $dbkeys) {
 385                                         foreach ( $dbkeys as $dbkey => $unused ) {
 386                                                 $title = Title :: makeTitle($ns, $dbkey);
 387                                                 $this->mMissingTitles[] = $title;
 388                                                 $this->mAllPages[$ns][$dbkey] = 0;
 389                                                 $this->mTitles[] = $title;
 390                                         }
 391                                 }
 392                         }
 393                         else
 394                         {
 395                                 // The remaining pageids do not exist
 396                                 if(empty($this->mMissingPageIDs))
 397                                         $this->mMissingPageIDs = array_keys($remaining);
 398                                 else
 399                                         $this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
 400                         }
 401                 }
 402         }
 403
 404         private function initFromRevIDs($revids) {
 405
 406                 if(empty($revids))
 407                         return;
 408
 409                 $db = $this->getDB();
 410                 $pageids = array();
 411                 $remaining = array_flip($revids);
 412
 413                 $tables = array('revision');
 414                 $fields = array('rev_id','rev_page');
 415                 $where = array('rev_deleted' => 0, 'rev_id' => $revids);
 416
 417                 // Get pageIDs data from the `page` table
 418                 $this->profileDBIn();
 419                 $res = $db->select( $tables, $fields, $where,  __METHOD__ );
 420                 while ( $row = $db->fetchObject( $res ) ) {
 421                         $revid = intval($row->rev_id);
 422                         $pageid = intval($row->rev_page);
 423                         $this->mGoodRevIDs[$revid] = $pageid;
 424                         $pageids[$pageid] = '';
 425                         unset($remaining[$revid]);
 426                 }
 427                 $db->freeResult( $res );
 428                 $this->profileDBOut();
 429
 430                 $this->mMissingRevIDs = array_keys($remaining);
 431
 432                 // Populate all the page information
 433                 if($this->mResolveRedirects)
 434                         ApiBase :: dieDebug(__METHOD__, 'revids may not be used with redirect resolution');
 435                 $this->initFromPageIds(array_keys($pageids));
 436         }
 437
 438         private function resolvePendingRedirects() {
 439
 440                 if($this->mResolveRedirects) {
 441                         $db = $this->getDB();
 442                         $pageFlds = $this->getPageTableFields();
 443
 444                         // Repeat until all redirects have been resolved
 445                         // The infinite loop is prevented by keeping all known pages in $this->mAllPages
 446                         while (!empty ($this->mPendingRedirectIDs)) {
 447
 448                                 // Resolve redirects by querying the pagelinks table, and repeat the process
 449                                 // Create a new linkBatch object for the next pass
 450                                 $linkBatch = $this->getRedirectTargets();
 451
 452                                 if ($linkBatch->isEmpty())
 453                                         break;
 454
 455                                 $set = $linkBatch->constructSet('page', $db);
 456                                 if(false === $set)
 457                                         break;
 458
 459                                 // Get pageIDs data from the `page` table
 460                                 $this->profileDBIn();
 461                                 $res = $db->select('page', $pageFlds, $set, __METHOD__);
 462                                 $this->profileDBOut();
 463
 464                                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 465                                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);
 466                         }
 467                 }
 468         }
 469
 470         private function getRedirectTargets() {
 471
 472                 $linkBatch = new LinkBatch();
 473                 $db = $this->getDB();
 474
 475                 // find redirect targets for all redirect pages
 476                 $this->profileDBIn();
 477                 $res = $db->select('pagelinks', array (
 478                         'pl_from',
 479                         'pl_namespace',
 480                         'pl_title'
 481                 ), array (
 482                         'pl_from' => array_keys($this->mPendingRedirectIDs
 483                 )), __METHOD__);
 484                 $this->profileDBOut();
 485
 486                 while ($row = $db->fetchObject($res)) {
 487
 488                         $plfrom = intval($row->pl_from);
 489
 490                         // Bug 7304 workaround
 491                         // ( http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 )
 492                         // A redirect page may have more than one link.
 493                         // This code will only use the first link returned.
 494                         if (isset ($this->mPendingRedirectIDs[$plfrom])) { // remove line when bug 7304 is fixed
 495
 496                                 $titleStrFrom = $this->mPendingRedirectIDs[$plfrom]->getPrefixedText();
 497                                 $titleStrTo = Title :: makeTitle($row->pl_namespace, $row->pl_title)->getPrefixedText();
 498                                 unset ($this->mPendingRedirectIDs[$plfrom]); // remove line when bug 7304 is fixed
 499
 500                                 // Avoid an infinite loop by checking if we have already processed this target
 501                                 if (!isset ($this->mAllPages[$row->pl_namespace][$row->pl_title])) {
 502                                         $linkBatch->add($row->pl_namespace, $row->pl_title);
 503                                 }
 504                         } else {
 505                                 // This redirect page has more than one link.
 506                                 // This is very slow, but safer until bug 7304 is resolved
 507                                 $title = Title :: newFromID($plfrom);
 508                                 $titleStrFrom = $title->getPrefixedText();
 509
 510                                 $article = new Article($title);
 511                                 $text = $article->getContent();
 512                                 $titleTo = Title :: newFromRedirect($text);
 513                                 $titleStrTo = $titleTo->getPrefixedText();
 514
 515                                 if (is_null($titleStrTo))
 516                                         ApiBase :: dieDebug(__METHOD__, 'Bug7304 workaround: redir target from {$title->getPrefixedText()} not found');
 517
 518                                 // Avoid an infinite loop by checking if we have already processed this target
 519                                 if (!isset ($this->mAllPages[$titleTo->getNamespace()][$titleTo->getDBkey()])) {
 520                                         $linkBatch->addObj($titleTo);
 521                                 }
 522                         }
 523
 524                         $this->mRedirectTitles[$titleStrFrom] = $titleStrTo;
 525                 }
 526                 $db->freeResult($res);
 527
 528                 // All IDs must exist in the page table
 529                 if (!empty($this->mPendingRedirectIDs[$plfrom]))
 530                         ApiBase :: dieDebug(__METHOD__, 'Invalid redirect IDs were found');
 531
 532                 return $linkBatch;
 533         }
 534
 535         /**
 536          * Given an array of title strings, convert them into Title objects.
 537          * This method validates access rights for the title,
 538          * and appends normalization values to the output.
 539          *
 540          * @return LinkBatch of title objects.
 541          */
 542         private function processTitlesStrArray($titles) {
 543
 544                 $linkBatch = new LinkBatch();
 545
 546                 foreach ($titles as $titleString) {
 547                         $titleObj = Title :: newFromText($titleString);
 548
 549                         // Validation
 550                         if (!$titleObj)
 551                                 $this->dieUsage("bad title $titleString", 'invalidtitle');
 552                         if ($titleObj->getNamespace() < 0)
 553                                 $this->dieUsage("No support for special page $titleString has been implemented", 'unsupportednamespace');
 554                         if (!$titleObj->userCanRead())
 555                                 $this->dieUsage("No read permission for $titleString", 'titleaccessdenied');
 556
 557                         $linkBatch->addObj($titleObj);
 558
 559                         // Make sure we remember the original title that was given to us
 560                         // This way the caller can correlate new titles with the originally requested,
 561                         // i.e. namespace is localized or capitalization is different
 562                         if ($titleString !== $titleObj->getPrefixedText()) {
 563                                 $this->mNormalizedTitles[$titleString] = $titleObj->getPrefixedText();
 564                         }
 565                 }
 566
 567                 return $linkBatch;
 568         }
 569
 570         protected function getAllowedParams() {
 571                 return array (
 572                         'titles' => array (
 573                                 ApiBase :: PARAM_ISMULTI => true
 574                         ),
 575                         'pageids' => array (
 576                                 ApiBase :: PARAM_TYPE => 'integer',
 577                                 ApiBase :: PARAM_ISMULTI => true
 578                         ),
 579                         'revids' => array (
 580                                 ApiBase :: PARAM_TYPE => 'integer',
 581                                 ApiBase :: PARAM_ISMULTI => true
 582                         )
 583                 );
 584         }
 585
 586         protected function getParamDescription() {
 587                 return array (
 588                         'titles' => 'A list of titles to work on',
 589                         'pageids' => 'A list of page IDs to work on',
 590                         'revids' => 'A list of revision IDs to work on'
 591                 );
 592         }
 593
 594         public function getVersion() {
 595                 return __CLASS__ . ': $Id$';
 596         }
 597 }
 598 ?>