includes/api/ApiQueryBacklinks.php

   1 <?php
   2
   3 /*
   4  * Created on Oct 16, 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 is a three-in-one module to query:
  33  *   * backlinks  - links pointing to the given page,
  34  *   * embeddedin - what pages transclude the given page within themselves,
  35  *   * imageusage - what pages use the given image
  36  *
  37  * @ingroup API
  38  */
  39 class ApiQueryBacklinks extends ApiQueryGeneratorBase {
  40
  41         private $params, $rootTitle, $contRedirs, $contLevel, $contTitle, $contID, $redirID, $redirect;
  42         private $bl_ns, $bl_from, $bl_table, $bl_code, $bl_title, $bl_sort, $bl_fields, $hasNS;
  43         private $pageMap, $resultArr;
  44
  45         // output element name, database column field prefix, database table
  46         private $backlinksSettings = array (
  47                 'backlinks' => array (
  48                         'code' => 'bl',
  49                         'prefix' => 'pl',
  50                         'linktbl' => 'pagelinks'
  51                 ),
  52                 'embeddedin' => array (
  53                         'code' => 'ei',
  54                         'prefix' => 'tl',
  55                         'linktbl' => 'templatelinks'
  56                 ),
  57                 'imageusage' => array (
  58                         'code' => 'iu',
  59                         'prefix' => 'il',
  60                         'linktbl' => 'imagelinks'
  61                 )
  62         );
  63
  64         public function __construct($query, $moduleName) {
  65                 extract($this->backlinksSettings[$moduleName]);
  66                 $this->resultArr = array();
  67
  68                 parent :: __construct($query, $moduleName, $code);
  69                 $this->bl_ns = $prefix . '_namespace';
  70                 $this->bl_from = $prefix . '_from';
  71                 $this->bl_table = $linktbl;
  72                 $this->bl_code = $code;
  73
  74                 $this->hasNS = $moduleName !== 'imageusage';
  75                 if ($this->hasNS) {
  76                         $this->bl_title = $prefix . '_title';
  77                         $this->bl_sort = "{$this->bl_ns}, {$this->bl_title}, {$this->bl_from}";
  78                         $this->bl_fields = array (
  79                                 $this->bl_ns,
  80                                 $this->bl_title
  81                         );
  82                 } else {
  83                         $this->bl_title = $prefix . '_to';
  84                         $this->bl_sort = "{$this->bl_title}, {$this->bl_from}";
  85                         $this->bl_fields = array (
  86                                 $this->bl_title
  87                         );
  88                 }
  89         }
  90
  91         public function execute() {
  92                 $this->run();
  93         }
  94
  95         public function executeGenerator($resultPageSet) {
  96                 $this->run($resultPageSet);
  97         }
  98
  99         private function prepareFirstQuery($resultPageSet = null) {
 100                 /* SELECT page_id, page_title, page_namespace, page_is_redirect
 101                  * FROM pagelinks, page WHERE pl_from=page_id
 102                  * AND pl_title='Foo' AND pl_namespace=0
 103                  * LIMIT 11 ORDER BY pl_from
 104                  */
 105                 $db = $this->getDB();
 106                 $this->addTables(array($this->bl_table, 'page'));
 107                 $this->addWhere("{$this->bl_from}=page_id");
 108                 if(is_null($resultPageSet))
 109                         $this->addFields(array('page_id', 'page_title', 'page_namespace'));
 110                 else
 111                         $this->addFields($resultPageSet->getPageTableFields());
 112                 $this->addFields('page_is_redirect');
 113                 $this->addWhereFld($this->bl_title, $this->rootTitle->getDBkey());
 114                 if($this->hasNS)
 115                         $this->addWhereFld($this->bl_ns, $this->rootTitle->getNamespace());
 116                 $this->addWhereFld('page_namespace', $this->params['namespace']);
 117                 if(!is_null($this->contID))
 118                         $this->addWhere("{$this->bl_from}>={$this->contID}");
 119                 if($this->params['filterredir'] == 'redirects')
 120                         $this->addWhereFld('page_is_redirect', 1);
 121                 if($this->params['filterredir'] == 'nonredirects')
 122                         $this->addWhereFld('page_is_redirect', 0);
 123                 $this->addOption('LIMIT', $this->params['limit'] + 1);
 124                 $this->addOption('ORDER BY', $this->bl_from);
 125                 $this->addOption('STRAIGHT_JOIN');
 126         }
 127
 128         private function prepareSecondQuery($resultPageSet = null) {
 129                 /* SELECT page_id, page_title, page_namespace, page_is_redirect, pl_title, pl_namespace
 130                    FROM pagelinks, page WHERE pl_from=page_id
 131                    AND (pl_title='Foo' AND pl_namespace=0) OR (pl_title='Bar' AND pl_namespace=1)
 132                    ORDER BY pl_namespace, pl_title, pl_from LIMIT 11
 133                  */
 134                 $db = $this->getDB();
 135                 $this->addTables(array('page', $this->bl_table));
 136                 $this->addWhere("{$this->bl_from}=page_id");
 137                 if(is_null($resultPageSet))
 138                         $this->addFields(array('page_id', 'page_title', 'page_namespace', 'page_is_redirect'));
 139                 else
 140                         $this->addFields($resultPageSet->getPageTableFields());
 141                 $this->addFields($this->bl_title);
 142                 if($this->hasNS)
 143                         $this->addFields($this->bl_ns);
 144                 // We can't use LinkBatch here because $this->hasNS may be false
 145                 $titleWhere = array();
 146                 foreach($this->redirTitles as $t)
 147                         $titleWhere[] = "{$this->bl_title} = ".$db->addQuotes($t->getDBkey()).
 148                                         ($this->hasNS ? " AND {$this->bl_ns} = '{$t->getNamespace()}'" : "");
 149                 $this->addWhere($db->makeList($titleWhere, LIST_OR));
 150                 $this->addWhereFld('page_namespace', $this->params['namespace']);
 151                 if(!is_null($this->redirID))
 152                 {
 153                         $first = $this->redirTitles[0];
 154                         $title = $db->strencode($first->getDBkey());
 155                         $ns = $first->getNamespace();
 156                         $from = $this->redirID;
 157                         if($this->hasNS)
 158                                 $this->addWhere("{$this->bl_ns} > $ns OR ".
 159                                                 "({$this->bl_ns} = $ns AND ".
 160                                                 "({$this->bl_title} > '$title' OR ".
 161                                                 "({$this->bl_title} = '$title' AND ".
 162                                                 "{$this->bl_from} >= $from)))");
 163                         else
 164                                 $this->addWhere("{$this->bl_title} > '$title' OR ".
 165                                                 "({$this->bl_title} = '$title' AND ".
 166                                                 "{$this->bl_from} >= $from)");
 167
 168                 }
 169                 if($this->params['filterredir'] == 'redirects')
 170                         $this->addWhereFld('page_is_redirect', 1);
 171                 if($this->params['filterredir'] == 'nonredirects')
 172                         $this->addWhereFld('page_is_redirect', 0);
 173                 $this->addOption('LIMIT', $this->params['limit'] + 1);
 174                 $this->addOption('ORDER BY', $this->bl_sort);
 175                 $this->addOption('USE INDEX', array('page' => 'PRIMARY'));
 176         }
 177
 178         private function run($resultPageSet = null) {
 179                 $this->params = $this->extractRequestParams(false);
 180                 $this->redirect = isset($this->params['redirect']) && $this->params['redirect'];
 181                 $userMax = ( $this->redirect ? ApiBase::LIMIT_BIG1/2 : ApiBase::LIMIT_BIG1 );
 182                 $botMax  = ( $this->redirect ? ApiBase::LIMIT_BIG2/2 : ApiBase::LIMIT_BIG2 );
 183                 if( $this->params['limit'] == 'max' ) {
 184                         $this->params['limit'] = $this->getMain()->canApiHighLimits() ? $botMax : $userMax;
 185                         $this->getResult()->addValue( 'limits', $this->getModuleName(), $this->params['limit'] );
 186                 }
 187
 188                 $this->processContinue();
 189                 $this->prepareFirstQuery($resultPageSet);
 190
 191                 $db = $this->getDB();
 192                 $res = $this->select(__METHOD__.'::firstQuery');
 193
 194                 $count = 0;
 195                 $this->pageMap = array(); // Maps ns and title to pageid
 196                 $this->continueStr = null;
 197                 $this->redirTitles = array();
 198                 while ($row = $db->fetchObject($res)) {
 199                         if (++ $count > $this->params['limit']) {
 200                                 // We've reached the one extra which shows that there are additional pages to be had. Stop here...
 201                                 // Continue string preserved in case the redirect query doesn't pass the limit
 202                                 $this->continueStr = $this->getContinueStr($row->page_id);
 203                                 break;
 204                         }
 205
 206                         if (is_null($resultPageSet))
 207                                 $this->extractRowInfo($row);
 208                         else
 209                         {
 210                                 $this->pageMap[$row->page_namespace][$row->page_title] = $row->page_id;
 211                                 if($row->page_is_redirect)
 212                                         $this->redirTitles[] = Title::makeTitle($row->page_namespace, $row->page_title);
 213                                 $resultPageSet->processDbRow($row);
 214                         }
 215                 }
 216                 $db->freeResult($res);
 217
 218                 if($this->redirect && count($this->redirTitles))
 219                 {
 220                         $this->resetQueryParams();
 221                         $this->prepareSecondQuery($resultPageSet);
 222                         $res = $this->select(__METHOD__.'::secondQuery');
 223                         $count = 0;
 224                         while($row = $db->fetchObject($res))
 225                         {
 226                                 if(++$count > $this->params['limit'])
 227                                 {
 228                                         // We've reached the one extra which shows that there are additional pages to be had. Stop here...
 229                                         // We need to keep the parent page of this redir in
 230                                         if($this->hasNS)
 231                                                 $parentID = $this->pageMap[$row->{$this->bl_ns}][$row->{$this->bl_title}];
 232                                         else
 233                                                 $parentID = $this->pageMap[NS_IMAGE][$row->{$this->bl_title}];
 234                                         $this->continueStr = $this->getContinueRedirStr($parentID, $row->page_id);
 235                                         break;
 236                                 }
 237
 238                                 if(is_null($resultPageSet))
 239                                         $this->extractRedirRowInfo($row);
 240                                 else
 241                                         $resultPageSet->processDbRow($row);
 242                         }
 243                         $db->freeResult($res);
 244                 }
 245                 if (is_null($resultPageSet)) {
 246                         // Try to add the result data in one go and pray that it fits
 247                         $fit = $this->getResult()->addValue('query', $this->getModuleName(), array_values($this->resultArr));
 248                         if(!$fit)
 249                         {
 250                                 // It didn't fit. Add elements one by one until the
 251                                 // result is full.
 252                                 foreach($this->resultArr as $pageID => $arr)
 253                                 {
 254                                         // Add the basic entry without redirlinks first
 255                                         $fit = $this->getResult()->addValue(
 256                                                 array('query', $this->getModuleName()),
 257                                                 null, array_diff_key($arr, array('redirlinks' => '')));
 258                                         if(!$fit)
 259                                         {
 260                                                 $this->continueStr = $this->getContinueStr($pageID);
 261                                                 break;
 262                                         }
 263
 264                                         $hasRedirs = false;
 265                                         foreach((array)@$arr['redirlinks'] as $key => $redir)
 266                                         {
 267                                                 $fit = $this->getResult()->addValue(
 268                                                         array('query', $this->getModuleName(), $pageID, 'redirlinks'),
 269                                                         $key, $redir);
 270                                                 if(!$fit)
 271                                                 {
 272                                                         $this->continueStr = $this->getContinueRedirStr($pageID, $redir['pageid']);
 273                                                         break;
 274                                                 }
 275                                                 $hasRedirs = true;
 276                                         }
 277                                         if($hasRedirs)
 278                                                 $this->getResult()->setIndexedTagName_internal(
 279                                                         array('query', $this->getModuleName(), $pageID, 'redirlinks'),
 280                                                         $this->bl_code);
 281                                         if(!$fit)
 282                                                 break;
 283                                 }
 284                         }
 285
 286                         $this->getResult()->setIndexedTagName_internal(
 287                                          array('query', $this->getModuleName()),
 288                                          $this->bl_code);
 289                 }
 290                 if(!is_null($this->continueStr))
 291                         $this->setContinueEnumParameter('continue', $this->continueStr);
 292         }
 293
 294         private function extractRowInfo($row) {
 295                 $this->pageMap[$row->page_namespace][$row->page_title] = $row->page_id;
 296                 $t = Title::makeTitle($row->page_namespace, $row->page_title);
 297                 $a = array('pageid' => intval($row->page_id));
 298                 ApiQueryBase::addTitleInfo($a, $t);
 299                 if($row->page_is_redirect)
 300                 {
 301                         $a['redirect'] = '';
 302                         $this->redirTitles[] = $t;
 303                 }
 304                 // Put all the results in an array first
 305                 $this->resultArr[$a['pageid']] = $a;
 306         }
 307
 308         private function extractRedirRowInfo($row)
 309         {
 310                 $a['pageid'] = intval($row->page_id);
 311                 ApiQueryBase::addTitleInfo($a, Title::makeTitle($row->page_namespace, $row->page_title));
 312                 if($row->page_is_redirect)
 313                         $a['redirect'] = '';
 314                 $ns = $this->hasNS ? $row->{$this->bl_ns} : NS_FILE;
 315                 $parentID = $this->pageMap[$ns][$row->{$this->bl_title}];
 316                 // Put all the results in an array first
 317                 $this->resultArr[$parentID]['redirlinks'][] = $a;
 318                 $this->getResult()->setIndexedTagName($this->resultArr[$parentID]['redirlinks'], $this->bl_code);
 319         }
 320
 321         protected function processContinue() {
 322                 if (!is_null($this->params['continue']))
 323                         $this->parseContinueParam();
 324                 else {
 325                         if ( $this->params['title'] !== "" ) {
 326                                 $title = Title::newFromText( $this->params['title'] );
 327                                 if ( !$title ) {
 328                                         $this->dieUsageMsg(array('invalidtitle', $this->params['title']));
 329                                 } else {
 330                                         $this->rootTitle = $title;
 331                                 }
 332                         } else {
 333                                 $this->dieUsageMsg(array('missingparam', 'title'));
 334                         }
 335                 }
 336
 337                 // only image titles are allowed for the root in imageinfo mode
 338                 if (!$this->hasNS && $this->rootTitle->getNamespace() !== NS_FILE)
 339                         $this->dieUsage("The title for {$this->getModuleName()} query must be an image", 'bad_image_title');
 340         }
 341
 342         protected function parseContinueParam() {
 343                 $continueList = explode('|', $this->params['continue']);
 344                 // expected format:
 345                 // ns | key | id1 [| id2]
 346                 // ns+key: root title
 347                 // id1: first-level page ID to continue from
 348                 // id2: second-level page ID to continue from
 349
 350                 // null stuff out now so we know what's set and what isn't
 351                 $this->rootTitle = $this->contID = $this->redirID = null;
 352                 $rootNs = intval($continueList[0]);
 353                 if($rootNs === 0 && $continueList[0] !== '0')
 354                         // Illegal continue parameter
 355                         $this->dieUsage("Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue");
 356                 $this->rootTitle = Title::makeTitleSafe($rootNs, $continueList[1]);
 357                 if(!$this->rootTitle)
 358                         $this->dieUsage("Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue");
 359                 $contID = intval($continueList[2]);
 360                 if($contID === 0 && $continueList[2] !== '0')
 361                         $this->dieUsage("Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue");
 362                 $this->contID = $contID;
 363                 $redirID = intval(@$continueList[3]);
 364                 if($redirID === 0 && @$continueList[3] !== '0')
 365                         // This one isn't required
 366                         return;
 367                 $this->redirID = $redirID;
 368
 369         }
 370
 371         protected function getContinueStr($lastPageID) {
 372                 return $this->rootTitle->getNamespace() .
 373                 '|' . $this->rootTitle->getDBkey() .
 374                 '|' . $lastPageID;
 375         }
 376
 377         protected function getContinueRedirStr($lastPageID, $lastRedirID) {
 378                 return $this->getContinueStr($lastPageID) . '|' . $lastRedirID;
 379         }
 380
 381         public function getAllowedParams() {
 382                 $retval =  array (
 383                         'title' => null,
 384                         'continue' => null,
 385                         'namespace' => array (
 386                                 ApiBase :: PARAM_ISMULTI => true,
 387                                 ApiBase :: PARAM_TYPE => 'namespace'
 388                         ),
 389                         'filterredir' => array(
 390                                 ApiBase :: PARAM_DFLT => 'all',
 391                                 ApiBase :: PARAM_TYPE => array(
 392                                         'all',
 393                                         'redirects',
 394                                         'nonredirects'
 395                                 )
 396                         ),
 397                         'limit' => array (
 398                                 ApiBase :: PARAM_DFLT => 10,
 399                                 ApiBase :: PARAM_TYPE => 'limit',
 400                                 ApiBase :: PARAM_MIN => 1,
 401                                 ApiBase :: PARAM_MAX => ApiBase :: LIMIT_BIG1,
 402                                 ApiBase :: PARAM_MAX2 => ApiBase :: LIMIT_BIG2
 403                         )
 404                 );
 405                 if($this->getModuleName() == 'embeddedin')
 406                         return $retval;
 407                 $retval['redirect'] = false;
 408                 return $retval;
 409         }
 410
 411         public function getParamDescription() {
 412                 $retval = array (
 413                         'title' => 'Title to search.',
 414                         'continue' => 'When more results are available, use this to continue.',
 415                         'namespace' => 'The namespace to enumerate.',
 416                         'filterredir' => 'How to filter for redirects'
 417                 );
 418                 if($this->getModuleName() != 'embeddedin')
 419                         return array_merge($retval, array(
 420                                 'redirect' => 'If linking page is a redirect, find all pages that link to that redirect as well. Maximum limit is halved.',
 421                                 'limit' => "How many total pages to return. If {$this->bl_code}redirect is enabled, limit applies to each level separately (which means you may get up to 2 * limit results)."
 422                         ));
 423                 return array_merge($retval, array(
 424                         'limit' => "How many total pages to return."
 425                 ));
 426         }
 427
 428         public function getDescription() {
 429                 switch ($this->getModuleName()) {
 430                         case 'backlinks' :
 431                                 return 'Find all pages that link to the given page';
 432                         case 'embeddedin' :
 433                                 return 'Find all pages that embed (transclude) the given title';
 434                         case 'imageusage' :
 435                                 return 'Find all pages that use the given image title.';
 436                         default :
 437                                 ApiBase :: dieDebug(__METHOD__, 'Unknown module name');
 438                 }
 439         }
 440
 441         protected function getExamples() {
 442                 static $examples = array (
 443                         'backlinks' => array (
 444                                 "api.php?action=query&list=backlinks&bltitle=Main%20Page",
 445                                 "api.php?action=query&generator=backlinks&gbltitle=Main%20Page&prop=info"
 446                         ),
 447                         'embeddedin' => array (
 448                                 "api.php?action=query&list=embeddedin&eititle=Template:Stub",
 449                                 "api.php?action=query&generator=embeddedin&geititle=Template:Stub&prop=info"
 450                         ),
 451                         'imageusage' => array (
 452                                 "api.php?action=query&list=imageusage&iutitle=File:Albert%20Einstein%20Head.jpg",
 453                                 "api.php?action=query&generator=imageusage&giutitle=File:Albert%20Einstein%20Head.jpg&prop=info"
 454                         )
 455                 );
 456
 457                 return $examples[$this->getModuleName()];
 458         }
 459
 460         public function getVersion() {
 461                 return __CLASS__ . ': $Id$';
 462         }
 463 }