includes/api/ApiPageSet.php

   1 <?php
   2
   3 /*
   4  * Created on Sep 24, 2006
   5  *
   6  * API for MediaWiki 1.8+
   7  *
   8  * Copyright (C) 2006 Yuri Astrakhan <Firstname><Lastname>@gmail.com
   9  *
  10  * This program is free software; you can redistribute it and/or modify
  11  * it under the terms of the GNU General Public License as published by
  12  * the Free Software Foundation; either version 2 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  18  * GNU General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU General Public License along
  21  * with this program; if not, write to the Free Software Foundation, Inc.,
  22  * 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  23  * http://www.gnu.org/copyleft/gpl.html
  24  */
  25
  26 if (!defined('MEDIAWIKI')) {
  27         // Eclipse helper - will be ignored in production
  28         require_once ('ApiQueryBase.php');
  29 }
  30
  31 /**
  32  * This class contains a list of pages that the client has requested.
  33  * Initially, when the client passes in titles=, pageids=, or revisions= parameter,
  34  * an instance of the ApiPageSet class will normalize titles,
  35  * determine if the pages/revisions exist, and prefetch any additional data page data requested.
  36  *
  37  * When generator is used, the result of the generator will become the input for the
  38  * second instance of this class, and all subsequent actions will go use the second instance
  39  * for all their work.
  40  *
  41  * @addtogroup API
  42  */
  43 class ApiPageSet extends ApiQueryBase {
  44
  45         private $mAllPages; // [ns][dbkey] => page_id or 0 when missing
  46         private $mTitles, $mGoodTitles, $mMissingTitles, $mMissingPageIDs, $mRedirectTitles, $mNormalizedTitles, $mInterwikiTitles;
  47         private $mResolveRedirects, $mPendingRedirectIDs;
  48         private $mGoodRevIDs, $mMissingRevIDs;
  49
  50         private $mRequestedPageFields;
  51
  52         public function __construct($query, $resolveRedirects = false) {
  53                 parent :: __construct($query, __CLASS__);
  54
  55                 $this->mAllPages = array ();
  56                 $this->mTitles = array();
  57                 $this->mGoodTitles = array ();
  58                 $this->mMissingTitles = array ();
  59                 $this->mMissingPageIDs = array ();
  60                 $this->mRedirectTitles = array ();
  61                 $this->mNormalizedTitles = array ();
  62                 $this->mInterwikiTitles = array ();
  63                 $this->mGoodRevIDs = array();
  64                 $this->mMissingRevIDs = array();
  65
  66                 $this->mRequestedPageFields = array ();
  67                 $this->mResolveRedirects = $resolveRedirects;
  68                 if($resolveRedirects)
  69                         $this->mPendingRedirectIDs = array();
  70         }
  71
  72         public function isResolvingRedirects() {
  73                 return $this->mResolveRedirects;
  74         }
  75
  76         public function requestField($fieldName) {
  77                 $this->mRequestedPageFields[$fieldName] = null;
  78         }
  79
  80         public function getCustomField($fieldName) {
  81                 return $this->mRequestedPageFields[$fieldName];
  82         }
  83
  84         /**
  85          * Get fields that modules have requested from the page table
  86          */
  87         public function getPageTableFields() {
  88                 // Ensure we get minimum required fields
  89                 $pageFlds = array (
  90                         'page_id' => null,
  91                         'page_namespace' => null,
  92                         'page_title' => null
  93                 );
  94
  95                 // only store non-default fields
  96                 $this->mRequestedPageFields = array_diff_key($this->mRequestedPageFields, $pageFlds);
  97
  98                 if ($this->mResolveRedirects)
  99                         $pageFlds['page_is_redirect'] = null;
 100
 101                 $pageFlds = array_merge($pageFlds, $this->mRequestedPageFields);
 102                 return array_keys($pageFlds);
 103         }
 104
 105         /**
 106          * All Title objects provided.
 107          * @return array of Title objects
 108          */
 109         public function getTitles() {
 110                 return $this->mTitles;
 111         }
 112
 113         /**
 114          * Returns the number of unique pages (not revisions) in the set.
 115          */
 116         public function getTitleCount() {
 117                 return count($this->mTitles);
 118         }
 119
 120         /**
 121          * Title objects that were found in the database.
 122          * @return array page_id (int) => Title (obj)
 123          */
 124         public function getGoodTitles() {
 125                 return $this->mGoodTitles;
 126         }
 127
 128         /**
 129          * Returns the number of found unique pages (not revisions) in the set.
 130          */
 131         public function getGoodTitleCount() {
 132                 return count($this->mGoodTitles);
 133         }
 134
 135         /**
 136          * Title objects that were NOT found in the database.
 137          * @return array of Title objects
 138          */
 139         public function getMissingTitles() {
 140                 return $this->mMissingTitles;
 141         }
 142
 143         /**
 144          * Page IDs that were not found in the database
 145          * @return array of page IDs
 146          */
 147         public function getMissingPageIDs() {
 148                 return $this->mMissingPageIDs;
 149         }
 150
 151         /**
 152          * Get a list of redirects when doing redirect resolution
 153          * @return array prefixed_title (string) => prefixed_title (string)
 154          */
 155         public function getRedirectTitles() {
 156                 return $this->mRedirectTitles;
 157         }
 158
 159         /**
 160          * Get a list of title normalizations - maps the title given
 161          * with its normalized version.
 162          * @return array raw_prefixed_title (string) => prefixed_title (string)
 163          */
 164         public function getNormalizedTitles() {
 165                 return $this->mNormalizedTitles;
 166         }
 167
 168         /**
 169          * Get a list of interwiki titles - maps the title given
 170          * with to the interwiki prefix.
 171          * @return array raw_prefixed_title (string) => interwiki_prefix (string)
 172          */
 173         public function getInterwikiTitles() {
 174                 return $this->mInterwikiTitles;
 175         }
 176
 177         /**
 178          * Get the list of revision IDs (requested with revids= parameter)
 179          * @return array revID (int) => pageID (int)
 180          */
 181         public function getRevisionIDs() {
 182                 return $this->mGoodRevIDs;
 183         }
 184
 185         /**
 186          * Revision IDs that were not found in the database
 187          * @return array of revision IDs
 188          */
 189         public function getMissingRevisionIDs() {
 190                 return $this->mMissingRevIDs;
 191         }
 192
 193         /**
 194          * Returns the number of revisions (requested with revids= parameter)
 195          */
 196         public function getRevisionCount() {
 197                 return count($this->getRevisionIDs());
 198         }
 199
 200         /**
 201          * Populate from the request parameters
 202          */
 203         public function execute() {
 204                 $this->profileIn();
 205                 $titles = $pageids = $revids = null;
 206                 extract($this->extractRequestParams());
 207
 208                 // Only one of the titles/pageids/revids is allowed at the same time
 209                 $dataSource = null;
 210                 if (isset ($titles))
 211                         $dataSource = 'titles';
 212                 if (isset ($pageids)) {
 213                         if (isset ($dataSource))
 214                                 $this->dieUsage("Cannot use 'pageids' at the same time as '$dataSource'", 'multisource');
 215                         $dataSource = 'pageids';
 216                 }
 217                 if (isset ($revids)) {
 218                         if (isset ($dataSource))
 219                                 $this->dieUsage("Cannot use 'revids' at the same time as '$dataSource'", 'multisource');
 220                         $dataSource = 'revids';
 221                 }
 222
 223                 switch ($dataSource) {
 224                         case 'titles' :
 225                                 $this->initFromTitles($titles);
 226                                 break;
 227                         case 'pageids' :
 228                                 $this->initFromPageIds($pageids);
 229                                 break;
 230                         case 'revids' :
 231                                 if($this->mResolveRedirects)
 232                                         $this->dieUsage('revids may not be used with redirect resolution', 'params');
 233                                 $this->initFromRevIDs($revids);
 234                                 break;
 235                         default :
 236                                 // Do nothing - some queries do not need any of the data sources.
 237                                 break;
 238                 }
 239                 $this->profileOut();
 240         }
 241
 242         /**
 243          * Initialize PageSet from a list of Titles
 244          */
 245         public function populateFromTitles($titles) {
 246                 $this->profileIn();
 247                 $this->initFromTitles($titles);
 248                 $this->profileOut();
 249         }
 250
 251         /**
 252          * Initialize PageSet from a list of Page IDs
 253          */
 254         public function populateFromPageIDs($pageIDs) {
 255                 $this->profileIn();
 256                 $pageIDs = array_map('intval', $pageIDs); // paranoia
 257                 $this->initFromPageIds($pageIDs);
 258                 $this->profileOut();
 259         }
 260
 261         /**
 262          * Initialize PageSet from a rowset returned from the database
 263          */
 264         public function populateFromQueryResult($db, $queryResult) {
 265                 $this->profileIn();
 266                 $this->initFromQueryResult($db, $queryResult);
 267                 $this->profileOut();
 268         }
 269
 270         /**
 271          * Initialize PageSet from a list of Revision IDs
 272          */
 273         public function populateFromRevisionIDs($revIDs) {
 274                 $this->profileIn();
 275                 $revIDs = array_map('intval', $revIDs); // paranoia
 276                 $this->initFromRevIDs($revIDs);
 277                 $this->profileOut();
 278         }
 279
 280         /**
 281          * Extract all requested fields from the row received from the database
 282          */
 283         public function processDbRow($row) {
 284
 285                 // Store Title object in various data structures
 286                 $title = Title :: makeTitle($row->page_namespace, $row->page_title);
 287
 288                 // skip any pages that user has no rights to read
 289                 if ($title->userCanRead()) {
 290
 291                         $pageId = intval($row->page_id);
 292                         $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
 293                         $this->mTitles[] = $title;
 294
 295                         if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
 296                                 $this->mPendingRedirectIDs[$pageId] = $title;
 297                         } else {
 298                                 $this->mGoodTitles[$pageId] = $title;
 299                         }
 300
 301                         foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
 302                                 $fieldValues[$pageId] = $row-> $fieldName;
 303                 }
 304         }
 305
 306         public function finishPageSetGeneration() {
 307                 $this->profileIn();
 308                 $this->resolvePendingRedirects();
 309                 $this->profileOut();
 310         }
 311
 312         /**
 313          * This method populates internal variables with page information
 314          * based on the given array of title strings.
 315          *
 316          * Steps:
 317          * #1 For each title, get data from `page` table
 318          * #2 If page was not found in the DB, store it as missing
 319          *
 320          * Additionally, when resolving redirects:
 321          * #3 If no more redirects left, stop.
 322          * #4 For each redirect, get its links from `pagelinks` table.
 323          * #5 Substitute the original LinkBatch object with the new list
 324          * #6 Repeat from step #1
 325          */
 326         private function initFromTitles($titles) {
 327
 328                 // Get validated and normalized title objects
 329                 $linkBatch = $this->processTitlesArray($titles);
 330                 if($linkBatch->isEmpty())
 331                         return;
 332
 333                 $db = $this->getDB();
 334                 $set = $linkBatch->constructSet('page', $db);
 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                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 342                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);  // process Titles
 343
 344                 // Resolve any found redirects
 345                 $this->resolvePendingRedirects();
 346         }
 347
 348         private function initFromPageIds($pageids) {
 349                 if(empty($pageids))
 350                         return;
 351
 352                 $set = array (
 353                         'page_id' => $pageids
 354                 );
 355
 356                 $db = $this->getDB();
 357
 358                 // Get pageIDs data from the `page` table
 359                 $this->profileDBIn();
 360                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 361                 $this->profileDBOut();
 362
 363                 $this->initFromQueryResult($db, $res, array_flip($pageids), false);     // process PageIDs
 364
 365                 // Resolve any found redirects
 366                 $this->resolvePendingRedirects();
 367         }
 368
 369         /**
 370          * Iterate through the result of the query on 'page' table,
 371          * and for each row create and store title object and save any extra fields requested.
 372          * @param $db Database
 373          * @param $res DB Query result
 374          * @param $remaining Array of either pageID or ns/title elements (optional).
 375          *        If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
 376          * @param $processTitles bool Must be provided together with $remaining.
 377          *        If true, treat $remaining as an array of [ns][title]
 378          *        If false, treat it as an array of [pageIDs]
 379          * @return Array of redirect IDs (only when resolving redirects)
 380          */
 381         private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
 382                 if (!is_null($remaining) && is_null($processTitles))
 383                         ApiBase :: dieDebug(__METHOD__, 'Missing $processTitles parameter when $remaining is provided');
 384
 385                 while ($row = $db->fetchObject($res)) {
 386
 387                         $pageId = intval($row->page_id);
 388
 389                         // Remove found page from the list of remaining items
 390                         if (isset($remaining)) {
 391                                 if ($processTitles)
 392                                         unset ($remaining[$row->page_namespace][$row->page_title]);
 393                                 else
 394                                         unset ($remaining[$pageId]);
 395                         }
 396
 397                         // Store any extra fields requested by modules
 398                         $this->processDbRow($row);
 399                 }
 400                 $db->freeResult($res);
 401
 402                 if(isset($remaining)) {
 403                         // Any items left in the $remaining list are added as missing
 404                         if($processTitles) {
 405                                 // The remaining titles in $remaining are non-existant pages
 406                                 foreach ($remaining as $ns => $dbkeys) {
 407                                         foreach ( $dbkeys as $dbkey => $unused ) {
 408                                                 $title = Title :: makeTitle($ns, $dbkey);
 409                                                 $this->mMissingTitles[] = $title;
 410                                                 $this->mAllPages[$ns][$dbkey] = 0;
 411                                                 $this->mTitles[] = $title;
 412                                         }
 413                                 }
 414                         }
 415                         else
 416                         {
 417                                 // The remaining pageids do not exist
 418                                 if(empty($this->mMissingPageIDs))
 419                                         $this->mMissingPageIDs = array_keys($remaining);
 420                                 else
 421                                         $this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
 422                         }
 423                 }
 424         }
 425
 426         private function initFromRevIDs($revids) {
 427
 428                 if(empty($revids))
 429                         return;
 430
 431                 $db = $this->getDB();
 432                 $pageids = array();
 433                 $remaining = array_flip($revids);
 434
 435                 $tables = array('revision');
 436                 $fields = array('rev_id','rev_page');
 437                 $where = array('rev_deleted' => 0, 'rev_id' => $revids);
 438
 439                 // Get pageIDs data from the `page` table
 440                 $this->profileDBIn();
 441                 $res = $db->select( $tables, $fields, $where,  __METHOD__ );
 442                 while ( $row = $db->fetchObject( $res ) ) {
 443                         $revid = intval($row->rev_id);
 444                         $pageid = intval($row->rev_page);
 445                         $this->mGoodRevIDs[$revid] = $pageid;
 446                         $pageids[$pageid] = '';
 447                         unset($remaining[$revid]);
 448                 }
 449                 $db->freeResult( $res );
 450                 $this->profileDBOut();
 451
 452                 $this->mMissingRevIDs = array_keys($remaining);
 453
 454                 // Populate all the page information
 455                 if($this->mResolveRedirects)
 456                         ApiBase :: dieDebug(__METHOD__, 'revids may not be used with redirect resolution');
 457                 $this->initFromPageIds(array_keys($pageids));
 458         }
 459
 460         private function resolvePendingRedirects() {
 461
 462                 if($this->mResolveRedirects) {
 463                         $db = $this->getDB();
 464                         $pageFlds = $this->getPageTableFields();
 465
 466                         // Repeat until all redirects have been resolved
 467                         // The infinite loop is prevented by keeping all known pages in $this->mAllPages
 468                         while (!empty ($this->mPendingRedirectIDs)) {
 469
 470                                 // Resolve redirects by querying the pagelinks table, and repeat the process
 471                                 // Create a new linkBatch object for the next pass
 472                                 $linkBatch = $this->getRedirectTargets();
 473
 474                                 if ($linkBatch->isEmpty())
 475                                         break;
 476
 477                                 $set = $linkBatch->constructSet('page', $db);
 478                                 if(false === $set)
 479                                         break;
 480
 481                                 // Get pageIDs data from the `page` table
 482                                 $this->profileDBIn();
 483                                 $res = $db->select('page', $pageFlds, $set, __METHOD__);
 484                                 $this->profileDBOut();
 485
 486                                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 487                                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);
 488                         }
 489                 }
 490         }
 491
 492         private function getRedirectTargets() {
 493
 494                 $linkBatch = new LinkBatch();
 495                 $db = $this->getDB();
 496
 497                 // find redirect targets for all redirect pages
 498                 $this->profileDBIn();
 499                 $res = $db->select('pagelinks', array (
 500                         'pl_from',
 501                         'pl_namespace',
 502                         'pl_title'
 503                 ), array (
 504                         'pl_from' => array_keys($this->mPendingRedirectIDs
 505                 )), __METHOD__);
 506                 $this->profileDBOut();
 507
 508                 while ($row = $db->fetchObject($res)) {
 509
 510                         $plfrom = intval($row->pl_from);
 511
 512                         // Bug 7304 workaround
 513                         // ( http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 )
 514                         // A redirect page may have more than one link.
 515                         // This code will only use the first link returned.
 516                         if (isset ($this->mPendingRedirectIDs[$plfrom])) { // remove line when bug 7304 is fixed
 517
 518                                 $titleStrFrom = $this->mPendingRedirectIDs[$plfrom]->getPrefixedText();
 519                                 $titleStrTo = Title :: makeTitle($row->pl_namespace, $row->pl_title)->getPrefixedText();
 520                                 unset ($this->mPendingRedirectIDs[$plfrom]); // remove line when bug 7304 is fixed
 521
 522                                 // Avoid an infinite loop by checking if we have already processed this target
 523                                 if (!isset ($this->mAllPages[$row->pl_namespace][$row->pl_title])) {
 524                                         $linkBatch->add($row->pl_namespace, $row->pl_title);
 525                                 }
 526                         } else {
 527                                 // This redirect page has more than one link.
 528                                 // This is very slow, but safer until bug 7304 is resolved
 529                                 $title = Title :: newFromID($plfrom);
 530                                 $titleStrFrom = $title->getPrefixedText();
 531
 532                                 $article = new Article($title);
 533                                 $text = $article->getContent();
 534                                 $titleTo = Title :: newFromRedirect($text);
 535                                 $titleStrTo = $titleTo->getPrefixedText();
 536
 537                                 if (is_null($titleStrTo))
 538                                         ApiBase :: dieDebug(__METHOD__, 'Bug7304 workaround: redir target from {$title->getPrefixedText()} not found');
 539
 540                                 // Avoid an infinite loop by checking if we have already processed this target
 541                                 if (!isset ($this->mAllPages[$titleTo->getNamespace()][$titleTo->getDBkey()])) {
 542                                         $linkBatch->addObj($titleTo);
 543                                 }
 544                         }
 545
 546                         $this->mRedirectTitles[$titleStrFrom] = $titleStrTo;
 547                 }
 548                 $db->freeResult($res);
 549
 550                 // All IDs must exist in the page table
 551                 if (!empty($this->mPendingRedirectIDs[$plfrom]))
 552                         ApiBase :: dieDebug(__METHOD__, 'Invalid redirect IDs were found');
 553
 554                 return $linkBatch;
 555         }
 556
 557         /**
 558          * Given an array of title strings, convert them into Title objects.
 559          * Alternativelly, an array of Title objects may be given.
 560          * This method validates access rights for the title,
 561          * and appends normalization values to the output.
 562          *
 563          * @return LinkBatch of title objects.
 564          */
 565         private function processTitlesArray($titles) {
 566
 567                 $linkBatch = new LinkBatch();
 568
 569                 foreach ($titles as $title) {
 570
 571                         $titleObj = is_string($title) ? Title :: newFromText($title) : $title;
 572                         if (!$titleObj)
 573                                 $this->dieUsage("bad title $titleString", 'invalidtitle');
 574
 575                         $iw = $titleObj->getInterwiki();
 576                         if (!empty($iw)) {
 577                                 // This title is an interwiki link.
 578                                 $this->mInterwikiTitles[$titleObj->getPrefixedText()] = $iw;
 579                         } else {
 580
 581                                 // Validation
 582                                 if ($titleObj->getNamespace() < 0)
 583                                         $this->dieUsage("No support for special page $titleString has been implemented", 'unsupportednamespace');
 584                                 if (!$titleObj->userCanRead())
 585                                         $this->dieUsage("No read permission for $titleString", 'titleaccessdenied');
 586
 587                                 $linkBatch->addObj($titleObj);
 588                         }
 589
 590                         // Make sure we remember the original title that was given to us
 591                         // This way the caller can correlate new titles with the originally requested,
 592                         // i.e. namespace is localized or capitalization is different
 593                         if (is_string($title) && $title !== $titleObj->getPrefixedText()) {
 594                                 $this->mNormalizedTitles[$title] = $titleObj->getPrefixedText();
 595                         }
 596                 }
 597
 598                 return $linkBatch;
 599         }
 600
 601         protected function getAllowedParams() {
 602                 return array (
 603                         'titles' => array (
 604                                 ApiBase :: PARAM_ISMULTI => true
 605                         ),
 606                         'pageids' => array (
 607                                 ApiBase :: PARAM_TYPE => 'integer',
 608                                 ApiBase :: PARAM_ISMULTI => true
 609                         ),
 610                         'revids' => array (
 611                                 ApiBase :: PARAM_TYPE => 'integer',
 612                                 ApiBase :: PARAM_ISMULTI => true
 613                         )
 614                 );
 615         }
 616
 617         protected function getParamDescription() {
 618                 return array (
 619                         'titles' => 'A list of titles to work on',
 620                         'pageids' => 'A list of page IDs to work on',
 621                         'revids' => 'A list of revision IDs to work on'
 622                 );
 623         }
 624
 625         public function getVersion() {
 626                 return __CLASS__ . ': $Id$';
 627         }
 628 }
 629