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