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 negative when missing
  46         private $mTitles, $mGoodTitles, $mMissingTitles, $mInvalidTitles;
  47         private $mMissingPageIDs, $mRedirectTitles;
  48         private $mNormalizedTitles, $mInterwikiTitles;
  49         private $mResolveRedirects, $mPendingRedirectIDs;
  50         private $mGoodRevIDs, $mMissingRevIDs;
  51         private $mFakePageId;
  52
  53         private $mRequestedPageFields;
  54
  55         public function __construct($query, $resolveRedirects = false) {
  56                 parent :: __construct($query, __CLASS__);
  57
  58                 $this->mAllPages = array ();
  59                 $this->mTitles = array();
  60                 $this->mGoodTitles = array ();
  61                 $this->mMissingTitles = array ();
  62                 $this->mInvalidTitles = array ();
  63                 $this->mMissingPageIDs = array ();
  64                 $this->mRedirectTitles = array ();
  65                 $this->mNormalizedTitles = array ();
  66                 $this->mInterwikiTitles = array ();
  67                 $this->mGoodRevIDs = array();
  68                 $this->mMissingRevIDs = array();
  69
  70                 $this->mRequestedPageFields = array ();
  71                 $this->mResolveRedirects = $resolveRedirects;
  72                 if($resolveRedirects)
  73                         $this->mPendingRedirectIDs = array();
  74
  75                 $this->mFakePageId = -1;
  76         }
  77
  78         public function isResolvingRedirects() {
  79                 return $this->mResolveRedirects;
  80         }
  81
  82         public function requestField($fieldName) {
  83                 $this->mRequestedPageFields[$fieldName] = null;
  84         }
  85
  86         public function getCustomField($fieldName) {
  87                 return $this->mRequestedPageFields[$fieldName];
  88         }
  89
  90         /**
  91          * Get fields that modules have requested from the page table
  92          */
  93         public function getPageTableFields() {
  94                 // Ensure we get minimum required fields
  95                 $pageFlds = array (
  96                         'page_id' => null,
  97                         'page_namespace' => null,
  98                         'page_title' => null
  99                 );
 100
 101                 // only store non-default fields
 102                 $this->mRequestedPageFields = array_diff_key($this->mRequestedPageFields, $pageFlds);
 103
 104                 if ($this->mResolveRedirects)
 105                         $pageFlds['page_is_redirect'] = null;
 106
 107                 $pageFlds = array_merge($pageFlds, $this->mRequestedPageFields);
 108                 return array_keys($pageFlds);
 109         }
 110
 111         /**
 112          * Returns an array [ns][dbkey] => page_id for all requested titles.
 113          * page_id is a unique negative number in case title was not found.
 114          * Invalid titles will also have negative page IDs and will be in namespace 0
 115          */
 116         public function getAllTitlesByNamespace() {
 117                 return $this->mAllPages;
 118         }
 119
 120         /**
 121          * All Title objects provided.
 122          * @return array of Title objects
 123          */
 124         public function getTitles() {
 125                 return $this->mTitles;
 126         }
 127
 128         /**
 129          * Returns the number of unique pages (not revisions) in the set.
 130          */
 131         public function getTitleCount() {
 132                 return count($this->mTitles);
 133         }
 134
 135         /**
 136          * Title objects that were found in the database.
 137          * @return array page_id (int) => Title (obj)
 138          */
 139         public function getGoodTitles() {
 140                 return $this->mGoodTitles;
 141         }
 142
 143         /**
 144          * Returns the number of found unique pages (not revisions) in the set.
 145          */
 146         public function getGoodTitleCount() {
 147                 return count($this->mGoodTitles);
 148         }
 149
 150         /**
 151          * Title objects that were NOT found in the database.
 152          * The array's index will be negative for each item
 153          * @return array of Title objects
 154          */
 155         public function getMissingTitles() {
 156                 return $this->mMissingTitles;
 157         }
 158
 159         /**
 160          * Titles that were deemed invalid by Title::newFromText()
 161          * The array's index will be unique and negative for each item
 162          * @return array of strings (not Title objects)
 163          */
 164         public function getInvalidTitles() {
 165                 return $this->mInvalidTitles;
 166         }
 167
 168         /**
 169          * Page IDs that were not found in the database
 170          * @return array of page IDs
 171          */
 172         public function getMissingPageIDs() {
 173                 return $this->mMissingPageIDs;
 174         }
 175
 176         /**
 177          * Get a list of redirects when doing redirect resolution
 178          * @return array prefixed_title (string) => prefixed_title (string)
 179          */
 180         public function getRedirectTitles() {
 181                 return $this->mRedirectTitles;
 182         }
 183
 184         /**
 185          * Get a list of title normalizations - maps the title given
 186          * with its normalized version.
 187          * @return array raw_prefixed_title (string) => prefixed_title (string)
 188          */
 189         public function getNormalizedTitles() {
 190                 return $this->mNormalizedTitles;
 191         }
 192
 193         /**
 194          * Get a list of interwiki titles - maps the title given
 195          * with to the interwiki prefix.
 196          * @return array raw_prefixed_title (string) => interwiki_prefix (string)
 197          */
 198         public function getInterwikiTitles() {
 199                 return $this->mInterwikiTitles;
 200         }
 201
 202         /**
 203          * Get the list of revision IDs (requested with revids= parameter)
 204          * @return array revID (int) => pageID (int)
 205          */
 206         public function getRevisionIDs() {
 207                 return $this->mGoodRevIDs;
 208         }
 209
 210         /**
 211          * Revision IDs that were not found in the database
 212          * @return array of revision IDs
 213          */
 214         public function getMissingRevisionIDs() {
 215                 return $this->mMissingRevIDs;
 216         }
 217
 218         /**
 219          * Returns the number of revisions (requested with revids= parameter)
 220          */
 221         public function getRevisionCount() {
 222                 return count($this->getRevisionIDs());
 223         }
 224
 225         /**
 226          * Populate from the request parameters
 227          */
 228         public function execute() {
 229                 $this->profileIn();
 230                 $titles = $pageids = $revids = null;
 231                 extract($this->extractRequestParams());
 232
 233                 // Only one of the titles/pageids/revids is allowed at the same time
 234                 $dataSource = null;
 235                 if (isset ($titles))
 236                         $dataSource = 'titles';
 237                 if (isset ($pageids)) {
 238                         if (isset ($dataSource))
 239                                 $this->dieUsage("Cannot use 'pageids' at the same time as '$dataSource'", 'multisource');
 240                         $dataSource = 'pageids';
 241                 }
 242                 if (isset ($revids)) {
 243                         if (isset ($dataSource))
 244                                 $this->dieUsage("Cannot use 'revids' at the same time as '$dataSource'", 'multisource');
 245                         $dataSource = 'revids';
 246                 }
 247
 248                 switch ($dataSource) {
 249                         case 'titles' :
 250                                 $this->initFromTitles($titles);
 251                                 break;
 252                         case 'pageids' :
 253                                 $this->initFromPageIds($pageids);
 254                                 break;
 255                         case 'revids' :
 256                                 if($this->mResolveRedirects)
 257                                         $this->dieUsage('revids may not be used with redirect resolution', 'params');
 258                                 $this->initFromRevIDs($revids);
 259                                 break;
 260                         default :
 261                                 // Do nothing - some queries do not need any of the data sources.
 262                                 break;
 263                 }
 264                 $this->profileOut();
 265         }
 266
 267         /**
 268          * Initialize PageSet from a list of Titles
 269          */
 270         public function populateFromTitles($titles) {
 271                 $this->profileIn();
 272                 $this->initFromTitles($titles);
 273                 $this->profileOut();
 274         }
 275
 276         /**
 277          * Initialize PageSet from a list of Page IDs
 278          */
 279         public function populateFromPageIDs($pageIDs) {
 280                 $this->profileIn();
 281                 $this->initFromPageIds($pageIDs);
 282                 $this->profileOut();
 283         }
 284
 285         /**
 286          * Initialize PageSet from a rowset returned from the database
 287          */
 288         public function populateFromQueryResult($db, $queryResult) {
 289                 $this->profileIn();
 290                 $this->initFromQueryResult($db, $queryResult);
 291                 $this->profileOut();
 292         }
 293
 294         /**
 295          * Initialize PageSet from a list of Revision IDs
 296          */
 297         public function populateFromRevisionIDs($revIDs) {
 298                 $this->profileIn();
 299                 $revIDs = array_map('intval', $revIDs); // paranoia
 300                 $this->initFromRevIDs($revIDs);
 301                 $this->profileOut();
 302         }
 303
 304         /**
 305          * Extract all requested fields from the row received from the database
 306          */
 307         public function processDbRow($row) {
 308
 309                 // Store Title object in various data structures
 310                 $title = Title :: makeTitle($row->page_namespace, $row->page_title);
 311
 312                 $pageId = intval($row->page_id);
 313                 $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
 314                 $this->mTitles[] = $title;
 315
 316                 if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
 317                         $this->mPendingRedirectIDs[$pageId] = $title;
 318                 } else {
 319                         $this->mGoodTitles[$pageId] = $title;
 320                 }
 321
 322                 foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
 323                         $fieldValues[$pageId] = $row-> $fieldName;
 324         }
 325
 326         public function finishPageSetGeneration() {
 327                 $this->profileIn();
 328                 $this->resolvePendingRedirects();
 329                 $this->profileOut();
 330         }
 331
 332         /**
 333          * This method populates internal variables with page information
 334          * based on the given array of title strings.
 335          *
 336          * Steps:
 337          * #1 For each title, get data from `page` table
 338          * #2 If page was not found in the DB, store it as missing
 339          *
 340          * Additionally, when resolving redirects:
 341          * #3 If no more redirects left, stop.
 342          * #4 For each redirect, get its links from `pagelinks` table.
 343          * #5 Substitute the original LinkBatch object with the new list
 344          * #6 Repeat from step #1
 345          */
 346         private function initFromTitles($titles) {
 347
 348                 // Get validated and normalized title objects
 349                 $linkBatch = $this->processTitlesArray($titles);
 350                 if($linkBatch->isEmpty())
 351                         return;
 352
 353                 $db = $this->getDB();
 354                 $set = $linkBatch->constructSet('page', $db);
 355
 356                 // Get pageIDs data from the `page` table
 357                 $this->profileDBIn();
 358                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 359                 $this->profileDBOut();
 360
 361                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 362                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);  // process Titles
 363
 364                 // Resolve any found redirects
 365                 $this->resolvePendingRedirects();
 366         }
 367
 368         private function initFromPageIds($pageids) {
 369                 if(empty($pageids))
 370                         return;
 371
 372                 $pageids = array_map('intval', $pageids); // paranoia
 373                 $set = array (
 374                         'page_id' => $pageids
 375                 );
 376
 377                 $db = $this->getDB();
 378
 379                 // Get pageIDs data from the `page` table
 380                 $this->profileDBIn();
 381                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 382                 $this->profileDBOut();
 383
 384                 $remaining = array_flip($pageids);
 385                 $this->initFromQueryResult($db, $res, $remaining, false);       // process PageIDs
 386
 387                 // Resolve any found redirects
 388                 $this->resolvePendingRedirects();
 389         }
 390
 391         /**
 392          * Iterate through the result of the query on 'page' table,
 393          * and for each row create and store title object and save any extra fields requested.
 394          * @param $db Database
 395          * @param $res DB Query result
 396          * @param $remaining Array of either pageID or ns/title elements (optional).
 397          *        If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
 398          * @param $processTitles bool Must be provided together with $remaining.
 399          *        If true, treat $remaining as an array of [ns][title]
 400          *        If false, treat it as an array of [pageIDs]
 401          * @return Array of redirect IDs (only when resolving redirects)
 402          */
 403         private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
 404                 if (!is_null($remaining) && is_null($processTitles))
 405                         ApiBase :: dieDebug(__METHOD__, 'Missing $processTitles parameter when $remaining is provided');
 406
 407                 while ($row = $db->fetchObject($res)) {
 408
 409                         $pageId = intval($row->page_id);
 410
 411                         // Remove found page from the list of remaining items
 412                         if (isset($remaining)) {
 413                                 if ($processTitles)
 414                                         unset ($remaining[$row->page_namespace][$row->page_title]);
 415                                 else
 416                                         unset ($remaining[$pageId]);
 417                         }
 418
 419                         // Store any extra fields requested by modules
 420                         $this->processDbRow($row);
 421                 }
 422                 $db->freeResult($res);
 423
 424                 if(isset($remaining)) {
 425                         // Any items left in the $remaining list are added as missing
 426                         if($processTitles) {
 427                                 // The remaining titles in $remaining are non-existant pages
 428                                 foreach ($remaining as $ns => $dbkeys) {
 429                                         foreach ( $dbkeys as $dbkey => $unused ) {
 430                                                 $title = Title :: makeTitle($ns, $dbkey);
 431                                                 $this->mAllPages[$ns][$dbkey] = $this->mFakePageId;
 432                                                 $this->mMissingTitles[$this->mFakePageId] = $title;
 433                                                 $this->mFakePageId--;
 434                                                 $this->mTitles[] = $title;
 435                                         }
 436                                 }
 437                         }
 438                         else
 439                         {
 440                                 // The remaining pageids do not exist
 441                                 if(empty($this->mMissingPageIDs))
 442                                         $this->mMissingPageIDs = array_keys($remaining);
 443                                 else
 444                                         $this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
 445                         }
 446                 }
 447         }
 448
 449         private function initFromRevIDs($revids) {
 450
 451                 if(empty($revids))
 452                         return;
 453
 454                 $db = $this->getDB();
 455                 $pageids = array();
 456                 $remaining = array_flip($revids);
 457
 458                 $tables = array('revision');
 459                 $fields = array('rev_id','rev_page');
 460                 $where = array('rev_deleted' => 0, 'rev_id' => $revids);
 461
 462                 // Get pageIDs data from the `page` table
 463                 $this->profileDBIn();
 464                 $res = $db->select( $tables, $fields, $where,  __METHOD__ );
 465                 while ( $row = $db->fetchObject( $res ) ) {
 466                         $revid = intval($row->rev_id);
 467                         $pageid = intval($row->rev_page);
 468                         $this->mGoodRevIDs[$revid] = $pageid;
 469                         $pageids[$pageid] = '';
 470                         unset($remaining[$revid]);
 471                 }
 472                 $db->freeResult( $res );
 473                 $this->profileDBOut();
 474
 475                 $this->mMissingRevIDs = array_keys($remaining);
 476
 477                 // Populate all the page information
 478                 if($this->mResolveRedirects)
 479                         ApiBase :: dieDebug(__METHOD__, 'revids may not be used with redirect resolution');
 480                 $this->initFromPageIds(array_keys($pageids));
 481         }
 482
 483         private function resolvePendingRedirects() {
 484
 485                 if($this->mResolveRedirects) {
 486                         $db = $this->getDB();
 487                         $pageFlds = $this->getPageTableFields();
 488
 489                         // Repeat until all redirects have been resolved
 490                         // The infinite loop is prevented by keeping all known pages in $this->mAllPages
 491                         while (!empty ($this->mPendingRedirectIDs)) {
 492
 493                                 // Resolve redirects by querying the pagelinks table, and repeat the process
 494                                 // Create a new linkBatch object for the next pass
 495                                 $linkBatch = $this->getRedirectTargets();
 496
 497                                 if ($linkBatch->isEmpty())
 498                                         break;
 499
 500                                 $set = $linkBatch->constructSet('page', $db);
 501                                 if(false === $set)
 502                                         break;
 503
 504                                 // Get pageIDs data from the `page` table
 505                                 $this->profileDBIn();
 506                                 $res = $db->select('page', $pageFlds, $set, __METHOD__);
 507                                 $this->profileDBOut();
 508
 509                                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 510                                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);
 511                         }
 512                 }
 513         }
 514
 515         private function getRedirectTargets() {
 516                 $lb = new LinkBatch();
 517                 $db = $this->getDB();
 518
 519                 $this->profileDBIn();
 520                 $res = $db->select('redirect', array(
 521                                 'rd_from',
 522                                 'rd_namespace',
 523                                 'rd_title'
 524                         ), array('rd_from' => array_keys($this->mPendingRedirectIDs)),
 525                         __METHOD__
 526                 );
 527                 $this->profileDBOut();
 528
 529                 while($row = $db->fetchObject($res))
 530                 {
 531                         $rdfrom = intval($row->rd_from);
 532                         $from = $this->mPendingRedirectIDs[$rdfrom]->getPrefixedText();
 533                         $to = Title::makeTitle($row->rd_namespace, $row->rd_title)->getPrefixedText();
 534                         unset($this->mPendingRedirectIDs[$rdfrom]);
 535                         if(!isset($this->mAllPages[$row->rd_namespace][$row->rd_title]))
 536                                 $lb->add($row->rd_namespace, $row->rd_title);
 537                         $this->mRedirectTitles[$from] = $to;
 538                 }
 539                 $db->freeResult($res);
 540                 if(!empty($this->mPendingRedirectIDs))
 541                 {
 542                         # We found pages that aren't in the redirect table
 543                         # Add them
 544                         foreach($this->mPendingRedirectIDs as $id => $title)
 545                         {
 546                                 $article = new Article($title);
 547                                 $rt = $article->insertRedirect();
 548                                 if(!$rt)
 549                                         # What the hell. Let's just ignore this
 550                                         continue;
 551                                 $lb->addObj($rt);
 552                                 $this->mRedirectTitles[$title->getPrefixedText()] = $rt->getPrefixedText();
 553                                 unset($this->mPendingRedirectIDs[$id]);
 554                         }
 555                         $this->getMain()->scheduleCommit();
 556                 }
 557                 return $lb;
 558         }
 559
 560         /**
 561          * Given an array of title strings, convert them into Title objects.
 562          * Alternativelly, an array of Title objects may be given.
 563          * This method validates access rights for the title,
 564          * and appends normalization values to the output.
 565          *
 566          * @return LinkBatch of title objects.
 567          */
 568         private function processTitlesArray($titles) {
 569
 570                 $linkBatch = new LinkBatch();
 571
 572                 foreach ($titles as $title) {
 573
 574                         $titleObj = is_string($title) ? Title :: newFromText($title) : $title;
 575                         if (!$titleObj)
 576                         {
 577                                 # Handle invalid titles gracefully
 578                                 $this->mAllpages[0][$title] = $this->mFakePageId;
 579                                 $this->mInvalidTitles[$this->mFakePageId] = $title;
 580                                 $this->mFakePageId--;
 581                                 continue; // There's nothing else we can do
 582                         }
 583                         $iw = $titleObj->getInterwiki();
 584                         if (!empty($iw)) {
 585                                 // This title is an interwiki link.
 586                                 $this->mInterwikiTitles[$titleObj->getPrefixedText()] = $iw;
 587                         } else {
 588
 589                                 // Validation
 590                                 if ($titleObj->getNamespace() < 0)
 591                                         $this->dieUsage("No support for special pages has been implemented", 'unsupportednamespace');
 592
 593                                 $linkBatch->addObj($titleObj);
 594                         }
 595
 596                         // Make sure we remember the original title that was given to us
 597                         // This way the caller can correlate new titles with the originally requested,
 598                         // i.e. namespace is localized or capitalization is different
 599                         if (is_string($title) && $title !== $titleObj->getPrefixedText()) {
 600                                 $this->mNormalizedTitles[$title] = $titleObj->getPrefixedText();
 601                         }
 602                 }
 603
 604                 return $linkBatch;
 605         }
 606
 607         protected function getAllowedParams() {
 608                 return array (
 609                         'titles' => array (
 610                                 ApiBase :: PARAM_ISMULTI => true
 611                         ),
 612                         'pageids' => array (
 613                                 ApiBase :: PARAM_TYPE => 'integer',
 614                                 ApiBase :: PARAM_ISMULTI => true
 615                         ),
 616                         'revids' => array (
 617                                 ApiBase :: PARAM_TYPE => 'integer',
 618                                 ApiBase :: PARAM_ISMULTI => true
 619                         )
 620                 );
 621         }
 622
 623         protected function getParamDescription() {
 624                 return array (
 625                         'titles' => 'A list of titles to work on',
 626                         'pageids' => 'A list of page IDs to work on',
 627                         'revids' => 'A list of revision IDs to work on'
 628                 );
 629         }
 630
 631         public function getVersion() {
 632                 return __CLASS__ . ': $Id$';
 633         }
 634 }