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                 $this->profileOut();
 230         }
 231
 232         /**
 233          * Extract all requested fields from the row received from the database
 234          */
 235         public function processDbRow($row) {
 236                 $pageId = intval($row->page_id);
 237
 238                 // Store Title object in various data structures
 239                 $title = Title :: makeTitle($row->page_namespace, $row->page_title);
 240                 $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
 241
 242                 if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
 243                         $this->mPendingRedirectIDs[$pageId] = $title;
 244                 } else {
 245                         $this->mGoodTitles[$pageId] = $title;
 246                 }
 247
 248                 foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
 249                         $fieldValues[$pageId] = $row-> $fieldName;
 250         }
 251
 252         public function finishPageSetGeneration() {
 253                 $this->profileIn();
 254                 $this->resolvePendingRedirects();
 255                 $this->profileOut();
 256         }
 257
 258         /**
 259          * This method populates internal variables with page information
 260          * based on the given array of title strings.
 261          *
 262          * Steps:
 263          * #1 For each title, get data from `page` table
 264          * #2 If page was not found in the DB, store it as missing
 265          *
 266          * Additionally, when resolving redirects:
 267          * #3 If no more redirects left, stop.
 268          * #4 For each redirect, get its links from `pagelinks` table.
 269          * #5 Substitute the original LinkBatch object with the new list
 270          * #6 Repeat from step #1
 271          */
 272         private function initFromTitles($titles) {
 273                 $db = $this->getDB();
 274
 275                 // Get validated and normalized title objects
 276                 $linkBatch = $this->processTitlesStrArray($titles);
 277                 $set = $linkBatch->constructSet('page', $db);
 278
 279                 // Get pageIDs data from the `page` table
 280                 $this->profileDBIn();
 281                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 282                 $this->profileDBOut();
 283
 284                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 285                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);  // process Titles
 286
 287                 // Resolve any found redirects
 288                 $this->resolvePendingRedirects();
 289         }
 290
 291         private function initFromPageIds($pageids) {
 292                 $db = $this->getDB();
 293
 294                 $set = array (
 295                         'page_id' => $pageids
 296                 );
 297
 298                 // Get pageIDs data from the `page` table
 299                 $this->profileDBIn();
 300                 $res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
 301                 $this->profileDBOut();
 302
 303                 $this->initFromQueryResult($db, $res, array_flip($pageids), false);     // process PageIDs
 304
 305                 // Resolve any found redirects
 306                 $this->resolvePendingRedirects();
 307         }
 308
 309         /**
 310          * Iterate through the result of the query on 'page' table,
 311          * and for each row create and store title object and save any extra fields requested.
 312          * @param $db Database
 313          * @param $res DB Query result
 314          * @param $remaining Array of either pageID or ns/title elements (optional).
 315          *        If given, any missing items will go to $mMissingPageIDs and $mMissingTitles
 316          * @param $processTitles bool Must be provided together with $remaining.
 317          *        If true, treat $remaining as an array of [ns][title]
 318          *        If false, treat it as an array of [pageIDs]
 319          * @return Array of redirect IDs (only when resolving redirects)
 320          */
 321         private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
 322                 if (!is_null($remaining) && is_null($processTitles))
 323                         $this->dieDebug('Missing $processTitles parameter when $remaining is provided');
 324
 325                 while ($row = $db->fetchObject($res)) {
 326
 327                         $pageId = intval($row->page_id);
 328
 329                         // Remove found page from the list of remaining items
 330                         if (isset($remaining)) {
 331                                 if ($processTitles)
 332                                         unset ($remaining[$row->page_namespace][$row->page_title]);
 333                                 else
 334                                         unset ($remaining[$pageId]);
 335                         }
 336
 337                         // Store any extra fields requested by modules
 338                         $this->processDbRow($row);
 339                 }
 340                 $db->freeResult($res);
 341
 342                 if(isset($remaining)) {
 343                         // Any items left in the $remaining list are added as missing
 344                         if($processTitles) {
 345                                 // The remaining titles in $remaining are non-existant pages
 346                                 foreach ($remaining as $ns => $dbkeys) {
 347                                         foreach ($dbkeys as $dbkey => $nothing) {
 348                                                 $this->mMissingTitles[] = Title :: makeTitle($ns, $dbkey);
 349                                                 $this->mAllPages[$ns][$dbkey] = 0;
 350                                         }
 351                                 }
 352                         }
 353                         else
 354                         {
 355                                 // The remaining pageids do not exist
 356                                 if(empty($this->mMissingPageIDs))
 357                                         $this->mMissingPageIDs = array_keys($remaining);
 358                                 else
 359                                         $this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
 360                         }
 361                 }
 362         }
 363
 364         private function initFromRevIDs($revids) {
 365
 366                 $db = $this->getDB();
 367                 $pageids = array();
 368                 $remaining = array_flip($revids);
 369
 370                 $tables = array('page', 'revision');
 371                 $fields = array('rev_id','rev_page');
 372                 $where = array( 'rev_deleted' => 0, 'rev_id' => $revids );
 373
 374                 // Get pageIDs data from the `page` table
 375                 $this->profileDBIn();
 376                 $res = $db->select( $tables, $fields, $where,  __METHOD__ );
 377                 while ( $row = $db->fetchObject( $res ) ) {
 378                         $revid = intval($row->rev_id);
 379                         $pageid = intval($row->rev_page);
 380                         $this->mGoodRevIDs[$revid] = $pageid;
 381                         $pageids[$pageid] = '';
 382                         unset($remaining[$revid]);
 383                 }
 384                 $db->freeResult( $res );
 385                 $this->profileDBOut();
 386
 387                 $this->mMissingRevIDs = array_keys($remaining);
 388
 389                 // Populate all the page information
 390                 if($this->mResolveRedirects)
 391                         $this->dieDebug('revids may not be used with redirect resolution');
 392                 $pageids = array_keys($pageids);
 393                 if(!empty($pageids))
 394                         $this->initFromPageIds($pageids);
 395         }
 396
 397         private function resolvePendingRedirects() {
 398
 399                 if($this->mResolveRedirects) {
 400                         $db = $this->getDB();
 401                         $pageFlds = $this->getPageTableFields();
 402
 403                         // Repeat until all redirects have been resolved
 404                         // The infinite loop is prevented by keeping all known pages in $this->mAllPages
 405                         while (!empty ($this->mPendingRedirectIDs)) {
 406
 407                                 // Resolve redirects by querying the pagelinks table, and repeat the process
 408                                 // Create a new linkBatch object for the next pass
 409                                 $linkBatch = $this->getRedirectTargets();
 410
 411                                 if ($linkBatch->isEmpty())
 412                                         break;
 413
 414                                 $set = $linkBatch->constructSet('page', $db);
 415                                 if(false === $set)
 416                                         break;
 417
 418                                 // Get pageIDs data from the `page` table
 419                                 $this->profileDBIn();
 420                                 $res = $db->select('page', $pageFlds, $set, __METHOD__);
 421                                 $this->profileDBOut();
 422
 423                                 // Hack: get the ns:titles stored in array(ns => array(titles)) format
 424                                 $this->initFromQueryResult($db, $res, $linkBatch->data, true);
 425                         }
 426                 }
 427         }
 428
 429         private function getRedirectTargets() {
 430
 431                 $linkBatch = new LinkBatch();
 432                 $db = $this->getDB();
 433
 434                 // find redirect targets for all redirect pages
 435                 $this->profileDBIn();
 436                 $res = $db->select('pagelinks', array (
 437                         'pl_from',
 438                         'pl_namespace',
 439                         'pl_title'
 440                 ), array (
 441                         'pl_from' => array_keys($this->mPendingRedirectIDs
 442                 )), __METHOD__);
 443                 $this->profileDBOut();
 444
 445                 while ($row = $db->fetchObject($res)) {
 446
 447                         $plfrom = intval($row->pl_from);
 448
 449                         // Bug 7304 workaround
 450                         // ( http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 )
 451                         // A redirect page may have more than one link.
 452                         // This code will only use the first link returned.
 453                         if (isset ($this->mPendingRedirectIDs[$plfrom])) { // remove line when bug 7304 is fixed
 454
 455                                 $titleStrFrom = $this->mPendingRedirectIDs[$plfrom]->getPrefixedText();
 456                                 $titleStrTo = Title :: makeTitle($row->pl_namespace, $row->pl_title)->getPrefixedText();
 457                                 unset ($this->mPendingRedirectIDs[$plfrom]); // remove line when bug 7304 is fixed
 458
 459                                 // Avoid an infinite loop by checking if we have already processed this target
 460                                 if (!isset ($this->mAllPages[$row->pl_namespace][$row->pl_title])) {
 461                                         $linkBatch->add($row->pl_namespace, $row->pl_title);
 462                                 }
 463                         } else {
 464                                 // This redirect page has more than one link.
 465                                 // This is very slow, but safer until bug 7304 is resolved
 466                                 $title = Title :: newFromID($plfrom);
 467                                 $titleStrFrom = $title->getPrefixedText();
 468
 469                                 $article = new Article($title);
 470                                 $text = $article->getContent();
 471                                 $titleTo = Title :: newFromRedirect($text);
 472                                 $titleStrTo = $titleTo->getPrefixedText();
 473
 474                                 if (is_null($titleStrTo))
 475                                         ApiBase :: dieDebug(__METHOD__, 'Bug7304 workaround: redir target from {$title->getPrefixedText()} not found');
 476
 477                                 // Avoid an infinite loop by checking if we have already processed this target
 478                                 if (!isset ($this->mAllPages[$titleTo->getNamespace()][$titleTo->getDBkey()])) {
 479                                         $linkBatch->addObj($titleTo);
 480                                 }
 481                         }
 482
 483                         $this->mRedirectTitles[$titleStrFrom] = $titleStrTo;
 484                 }
 485                 $db->freeResult($res);
 486
 487                 // All IDs must exist in the page table
 488                 if (!empty($this->mPendingRedirectIDs[$plfrom]))
 489                         $this->dieDebug('Invalid redirect IDs were found');
 490
 491                 return $linkBatch;
 492         }
 493
 494         /**
 495          * Given an array of title strings, convert them into Title objects.
 496          * This method validates access rights for the title,
 497          * and appends normalization values to the output.
 498          *
 499          * @return LinkBatch of title objects.
 500          */
 501         private function processTitlesStrArray($titles) {
 502
 503                 $linkBatch = new LinkBatch();
 504
 505                 foreach ($titles as $titleString) {
 506                         $titleObj = Title :: newFromText($titleString);
 507
 508                         // Validation
 509                         if (!$titleObj)
 510                                 $this->dieUsage("bad title $titleString", 'invalidtitle');
 511                         if ($titleObj->getNamespace() < 0)
 512                                 $this->dieUsage("No support for special page $titleString has been implemented", 'unsupportednamespace');
 513                         if (!$titleObj->userCanRead())
 514                                 $this->dieUsage("No read permission for $titleString", 'titleaccessdenied');
 515
 516                         $linkBatch->addObj($titleObj);
 517
 518                         // Make sure we remember the original title that was given to us
 519                         // This way the caller can correlate new titles with the originally requested,
 520                         // i.e. namespace is localized or capitalization is different
 521                         if ($titleString !== $titleObj->getPrefixedText()) {
 522                                 $this->mNormalizedTitles[$titleString] = $titleObj->getPrefixedText();
 523                         }
 524                 }
 525
 526                 return $linkBatch;
 527         }
 528
 529         protected function getAllowedParams() {
 530                 return array (
 531                         'titles' => array (
 532                                 ApiBase :: PARAM_ISMULTI => true
 533                         ),
 534                         'pageids' => array (
 535                                 ApiBase :: PARAM_TYPE => 'integer',
 536                                 ApiBase :: PARAM_ISMULTI => true
 537                         ),
 538                         'revids' => array (
 539                                 ApiBase :: PARAM_TYPE => 'integer',
 540                                 ApiBase :: PARAM_ISMULTI => true
 541                         )
 542                 );
 543         }
 544
 545         protected function getParamDescription() {
 546                 return array (
 547                         'titles' => 'A list of titles to work on',
 548                         'pageids' => 'A list of page IDs to work on',
 549                         'revids' => 'A list of revision IDs to work on'
 550                 );
 551         }
 552
 553         public function getVersion() {
 554                 return __CLASS__ . ': $Id$';
 555         }
 556 }
 557 ?>