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
 113                 $this->addFields( 'page_is_redirect' );
 114                 $this->addWhereFld( $this->bl_title, $this->rootTitle->getDBkey() );
 115
 116                 if ( $this->hasNS )
 117                         $this->addWhereFld( $this->bl_ns, $this->rootTitle->getNamespace() );
 118                 $this->addWhereFld( 'page_namespace', $this->params['namespace'] );
 119
 120                 if ( !is_null( $this->contID ) )
 121                         $this->addWhere( "{$this->bl_from}>={$this->contID}" );
 122
 123                 if ( $this->params['filterredir'] == 'redirects' )
 124                         $this->addWhereFld( 'page_is_redirect', 1 );
 125                 else if ( $this->params['filterredir'] == 'nonredirects' )
 126                         $this->addWhereFld( 'page_is_redirect', 0 );
 127
 128                 $this->addOption( 'LIMIT', $this->params['limit'] + 1 );
 129                 $this->addOption( 'ORDER BY', $this->bl_from );
 130                 $this->addOption( 'STRAIGHT_JOIN' );
 131         }
 132
 133         private function prepareSecondQuery( $resultPageSet = null ) {
 134                 /* SELECT page_id, page_title, page_namespace, page_is_redirect, pl_title, pl_namespace
 135                    FROM pagelinks, page WHERE pl_from=page_id
 136                    AND (pl_title='Foo' AND pl_namespace=0) OR (pl_title='Bar' AND pl_namespace=1)
 137                    ORDER BY pl_namespace, pl_title, pl_from LIMIT 11
 138                  */
 139                 $db = $this->getDB();
 140                 $this->addTables( array( 'page', $this->bl_table ) );
 141                 $this->addWhere( "{$this->bl_from}=page_id" );
 142
 143                 if ( is_null( $resultPageSet ) )
 144                         $this->addFields( array( 'page_id', 'page_title', 'page_namespace', 'page_is_redirect' ) );
 145                 else
 146                         $this->addFields( $resultPageSet->getPageTableFields() );
 147
 148                 $this->addFields( $this->bl_title );
 149                 if ( $this->hasNS )
 150                         $this->addFields( $this->bl_ns );
 151
 152                 // We can't use LinkBatch here because $this->hasNS may be false
 153                 $titleWhere = array();
 154                 foreach ( $this->redirTitles as $t )
 155                         $titleWhere[] = "{$this->bl_title} = " . $db->addQuotes( $t->getDBkey() ) .
 156                                         ( $this->hasNS ? " AND {$this->bl_ns} = '{$t->getNamespace()}'" : "" );
 157                 $this->addWhere( $db->makeList( $titleWhere, LIST_OR ) );
 158                 $this->addWhereFld( 'page_namespace', $this->params['namespace'] );
 159
 160                 if ( !is_null( $this->redirID ) )
 161                 {
 162                         $first = $this->redirTitles[0];
 163                         $title = $db->strencode( $first->getDBkey() );
 164                         $ns = $first->getNamespace();
 165                         $from = $this->redirID;
 166                         if ( $this->hasNS )
 167                                 $this->addWhere( "{$this->bl_ns} > $ns OR " .
 168                                                 "({$this->bl_ns} = $ns AND " .
 169                                                 "({$this->bl_title} > '$title' OR " .
 170                                                 "({$this->bl_title} = '$title' AND " .
 171                                                 "{$this->bl_from} >= $from)))" );
 172                         else
 173                                 $this->addWhere( "{$this->bl_title} > '$title' OR " .
 174                                                 "({$this->bl_title} = '$title' AND " .
 175                                                 "{$this->bl_from} >= $from)" );
 176
 177                 }
 178                 if ( $this->params['filterredir'] == 'redirects' )
 179                         $this->addWhereFld( 'page_is_redirect', 1 );
 180                 else if ( $this->params['filterredir'] == 'nonredirects' )
 181                         $this->addWhereFld( 'page_is_redirect', 0 );
 182
 183                 $this->addOption( 'LIMIT', $this->params['limit'] + 1 );
 184                 $this->addOption( 'ORDER BY', $this->bl_sort );
 185                 $this->addOption( 'USE INDEX', array( 'page' => 'PRIMARY' ) );
 186         }
 187
 188         private function run( $resultPageSet = null ) {
 189                 $this->params = $this->extractRequestParams( false );
 190                 $this->redirect = isset( $this->params['redirect'] ) && $this->params['redirect'];
 191                 $userMax = ( $this->redirect ? ApiBase::LIMIT_BIG1 / 2 : ApiBase::LIMIT_BIG1 );
 192                 $botMax  = ( $this->redirect ? ApiBase::LIMIT_BIG2 / 2 : ApiBase::LIMIT_BIG2 );
 193                 if ( $this->params['limit'] == 'max' ) {
 194                         $this->params['limit'] = $this->getMain()->canApiHighLimits() ? $botMax : $userMax;
 195                         $this->getResult()->addValue( 'limits', $this->getModuleName(), $this->params['limit'] );
 196                 }
 197
 198                 $this->processContinue();
 199                 $this->prepareFirstQuery( $resultPageSet );
 200
 201                 $db = $this->getDB();
 202                 $res = $this->select( __METHOD__ . '::firstQuery' );
 203
 204                 $count = 0;
 205                 $this->pageMap = array(); // Maps ns and title to pageid
 206                 $this->continueStr = null;
 207                 $this->redirTitles = array();
 208                 while ( $row = $db->fetchObject( $res ) ) {
 209                         if ( ++ $count > $this->params['limit'] ) {
 210                                 // We've reached the one extra which shows that there are additional pages to be had. Stop here...
 211                                 // Continue string preserved in case the redirect query doesn't pass the limit
 212                                 $this->continueStr = $this->getContinueStr( $row->page_id );
 213                                 break;
 214                         }
 215
 216                         if ( is_null( $resultPageSet ) )
 217                                 $this->extractRowInfo( $row );
 218                         else
 219                         {
 220                                 $this->pageMap[$row->page_namespace][$row->page_title] = $row->page_id;
 221                                 if ( $row->page_is_redirect )
 222                                         $this->redirTitles[] = Title::makeTitle( $row->page_namespace, $row->page_title );
 223                                 $resultPageSet->processDbRow( $row );
 224                         }
 225                 }
 226                 $db->freeResult( $res );
 227
 228                 if ( $this->redirect && count( $this->redirTitles ) )
 229                 {
 230                         $this->resetQueryParams();
 231                         $this->prepareSecondQuery( $resultPageSet );
 232                         $res = $this->select( __METHOD__ . '::secondQuery' );
 233                         $count = 0;
 234                         while ( $row = $db->fetchObject( $res ) )
 235                         {
 236                                 if ( ++$count > $this->params['limit'] )
 237                                 {
 238                                         // We've reached the one extra which shows that there are additional pages to be had. Stop here...
 239                                         // We need to keep the parent page of this redir in
 240                                         if ( $this->hasNS )
 241                                                 $parentID = $this->pageMap[$row-> { $this->bl_ns } ][$row-> { $this->bl_title } ];
 242                                         else
 243                                                 $parentID = $this->pageMap[NS_IMAGE][$row-> { $this->bl_title } ];
 244                                         $this->continueStr = $this->getContinueRedirStr( $parentID, $row->page_id );
 245                                         break;
 246                                 }
 247
 248                                 if ( is_null( $resultPageSet ) )
 249                                         $this->extractRedirRowInfo( $row );
 250                                 else
 251                                         $resultPageSet->processDbRow( $row );
 252                         }
 253                         $db->freeResult( $res );
 254                 }
 255                 if ( is_null( $resultPageSet ) ) {
 256                         // Try to add the result data in one go and pray that it fits
 257                         $fit = $this->getResult()->addValue( 'query', $this->getModuleName(), array_values( $this->resultArr ) );
 258                         if ( !$fit )
 259                         {
 260                                 // It didn't fit. Add elements one by one until the
 261                                 // result is full.
 262                                 foreach ( $this->resultArr as $pageID => $arr )
 263                                 {
 264                                         // Add the basic entry without redirlinks first
 265                                         $fit = $this->getResult()->addValue(
 266                                                 array( 'query', $this->getModuleName() ),
 267                                                 null, array_diff_key( $arr, array( 'redirlinks' => '' ) ) );
 268                                         if ( !$fit )
 269                                         {
 270                                                 $this->continueStr = $this->getContinueStr( $pageID );
 271                                                 break;
 272                                         }
 273
 274                                         $hasRedirs = false;
 275                                         foreach ( (array)@$arr['redirlinks'] as $key => $redir )
 276                                         {
 277                                                 $fit = $this->getResult()->addValue(
 278                                                         array( 'query', $this->getModuleName(), $pageID, 'redirlinks' ),
 279                                                         $key, $redir );
 280                                                 if ( !$fit )
 281                                                 {
 282                                                         $this->continueStr = $this->getContinueRedirStr( $pageID, $redir['pageid'] );
 283                                                         break;
 284                                                 }
 285                                                 $hasRedirs = true;
 286                                         }
 287                                         if ( $hasRedirs )
 288                                                 $this->getResult()->setIndexedTagName_internal(
 289                                                         array( 'query', $this->getModuleName(), $pageID, 'redirlinks' ),
 290                                                         $this->bl_code );
 291                                         if ( !$fit )
 292                                                 break;
 293                                 }
 294                         }
 295
 296                         $this->getResult()->setIndexedTagName_internal(
 297                                          array( 'query', $this->getModuleName() ),
 298                                          $this->bl_code );
 299                 }
 300                 if ( !is_null( $this->continueStr ) )
 301                         $this->setContinueEnumParameter( 'continue', $this->continueStr );
 302         }
 303
 304         private function extractRowInfo( $row ) {
 305                 $this->pageMap[$row->page_namespace][$row->page_title] = $row->page_id;
 306                 $t = Title::makeTitle( $row->page_namespace, $row->page_title );
 307                 $a = array( 'pageid' => intval( $row->page_id ) );
 308                 ApiQueryBase::addTitleInfo( $a, $t );
 309                 if ( $row->page_is_redirect )
 310                 {
 311                         $a['redirect'] = '';
 312                         $this->redirTitles[] = $t;
 313                 }
 314                 // Put all the results in an array first
 315                 $this->resultArr[$a['pageid']] = $a;
 316         }
 317
 318         private function extractRedirRowInfo( $row )
 319         {
 320                 $a['pageid'] = intval( $row->page_id );
 321                 ApiQueryBase::addTitleInfo( $a, Title::makeTitle( $row->page_namespace, $row->page_title ) );
 322                 if ( $row->page_is_redirect )
 323                         $a['redirect'] = '';
 324                 $ns = $this->hasNS ? $row-> { $this->bl_ns } : NS_FILE;
 325                 $parentID = $this->pageMap[$ns][$row-> { $this->bl_title } ];
 326                 // Put all the results in an array first
 327                 $this->resultArr[$parentID]['redirlinks'][] = $a;
 328                 $this->getResult()->setIndexedTagName( $this->resultArr[$parentID]['redirlinks'], $this->bl_code );
 329         }
 330
 331         protected function processContinue() {
 332                 if ( !is_null( $this->params['continue'] ) )
 333                         $this->parseContinueParam();
 334                 else {
 335                         if ( $this->params['title'] !== "" ) {
 336                                 $title = Title::newFromText( $this->params['title'] );
 337                                 if ( !$title ) {
 338                                         $this->dieUsageMsg( array( 'invalidtitle', $this->params['title'] ) );
 339                                 } else {
 340                                         $this->rootTitle = $title;
 341                                 }
 342                         } else {
 343                                 $this->dieUsageMsg( array( 'missingparam', 'title' ) );
 344                         }
 345                 }
 346
 347                 // only image titles are allowed for the root in imageinfo mode
 348                 if ( !$this->hasNS && $this->rootTitle->getNamespace() !== NS_FILE )
 349                         $this->dieUsage( "The title for {$this->getModuleName()} query must be an image", 'bad_image_title' );
 350         }
 351
 352         protected function parseContinueParam() {
 353                 $continueList = explode( '|', $this->params['continue'] );
 354                 // expected format:
 355                 // ns | key | id1 [| id2]
 356                 // ns+key: root title
 357                 // id1: first-level page ID to continue from
 358                 // id2: second-level page ID to continue from
 359
 360                 // null stuff out now so we know what's set and what isn't
 361                 $this->rootTitle = $this->contID = $this->redirID = null;
 362                 $rootNs = intval( $continueList[0] );
 363                 if ( $rootNs === 0 && $continueList[0] !== '0' )
 364                         // Illegal continue parameter
 365                         $this->dieUsage( "Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue" );
 366                 $this->rootTitle = Title::makeTitleSafe( $rootNs, $continueList[1] );
 367
 368                 if ( !$this->rootTitle )
 369                         $this->dieUsage( "Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue" );
 370                 $contID = intval( $continueList[2] );
 371
 372                 if ( $contID === 0 && $continueList[2] !== '0' )
 373                         $this->dieUsage( "Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue" );
 374                 $this->contID = $contID;
 375                 $redirID = intval( @$continueList[3] );
 376
 377                 if ( $redirID === 0 && @$continueList[3] !== '0' )
 378                         // This one isn't required
 379                         return;
 380                 $this->redirID = $redirID;
 381
 382         }
 383
 384         protected function getContinueStr( $lastPageID ) {
 385                 return $this->rootTitle->getNamespace() .
 386                 '|' . $this->rootTitle->getDBkey() .
 387                 '|' . $lastPageID;
 388         }
 389
 390         protected function getContinueRedirStr( $lastPageID, $lastRedirID ) {
 391                 return $this->getContinueStr( $lastPageID ) . '|' . $lastRedirID;
 392         }
 393
 394         public function getAllowedParams() {
 395                 $retval =  array (
 396                         'title' => null,
 397                         'continue' => null,
 398                         'namespace' => array (
 399                                 ApiBase :: PARAM_ISMULTI => true,
 400                                 ApiBase :: PARAM_TYPE => 'namespace'
 401                         ),
 402                         'filterredir' => array(
 403                                 ApiBase :: PARAM_DFLT => 'all',
 404                                 ApiBase :: PARAM_TYPE => array(
 405                                         'all',
 406                                         'redirects',
 407                                         'nonredirects'
 408                                 )
 409                         ),
 410                         'limit' => array (
 411                                 ApiBase :: PARAM_DFLT => 10,
 412                                 ApiBase :: PARAM_TYPE => 'limit',
 413                                 ApiBase :: PARAM_MIN => 1,
 414                                 ApiBase :: PARAM_MAX => ApiBase :: LIMIT_BIG1,
 415                                 ApiBase :: PARAM_MAX2 => ApiBase :: LIMIT_BIG2
 416                         )
 417                 );
 418                 if ( $this->getModuleName() == 'embeddedin' )
 419                         return $retval;
 420                 $retval['redirect'] = false;
 421                 return $retval;
 422         }
 423
 424         public function getParamDescription() {
 425                 $retval = array (
 426                         'title' => 'Title to search.',
 427                         'continue' => 'When more results are available, use this to continue.',
 428                         'namespace' => 'The namespace to enumerate.',
 429                         'filterredir' => 'How to filter for redirects'
 430                 );
 431                 if ( $this->getModuleName() != 'embeddedin' )
 432                         return array_merge( $retval, array(
 433                                 'redirect' => 'If linking page is a redirect, find all pages that link to that redirect as well. Maximum limit is halved.',
 434                                 'limit' => "How many total pages to return. If {$this->bl_code}redirect is enabled, limit applies to each level separately (which means you may get up to 2 * limit results)."
 435                         ) );
 436                 return array_merge( $retval, array(
 437                         'limit' => "How many total pages to return."
 438                 ) );
 439         }
 440
 441         public function getDescription() {
 442                 switch ( $this->getModuleName() ) {
 443                         case 'backlinks' :
 444                                 return 'Find all pages that link to the given page';
 445                         case 'embeddedin' :
 446                                 return 'Find all pages that embed (transclude) the given title';
 447                         case 'imageusage' :
 448                                 return 'Find all pages that use the given image title.';
 449                         default :
 450                                 ApiBase :: dieDebug( __METHOD__, 'Unknown module name' );
 451                 }
 452         }
 453
 454         protected function getExamples() {
 455                 static $examples = array (
 456                         'backlinks' => array (
 457                                 "api.php?action=query&list=backlinks&bltitle=Main%20Page",
 458                                 "api.php?action=query&generator=backlinks&gbltitle=Main%20Page&prop=info"
 459                         ),
 460                         'embeddedin' => array (
 461                                 "api.php?action=query&list=embeddedin&eititle=Template:Stub",
 462                                 "api.php?action=query&generator=embeddedin&geititle=Template:Stub&prop=info"
 463                         ),
 464                         'imageusage' => array (
 465                                 "api.php?action=query&list=imageusage&iutitle=File:Albert%20Einstein%20Head.jpg",
 466                                 "api.php?action=query&generator=imageusage&giutitle=File:Albert%20Einstein%20Head.jpg&prop=info"
 467                         )
 468                 );
 469
 470                 return $examples[$this->getModuleName()];
 471         }
 472
 473         public function getVersion() {
 474                 return __CLASS__ . ': $Id$';
 475         }
 476 }