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