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 ( $this->params['filterredir'] == 'nonredirects' && !$this->redirect )
 126                         // bug 22245 - Check for !redirect, as filtering nonredirects, when getting what links to them is contradictory
 127                         $this->addWhereFld( 'page_is_redirect', 0 );
 128
 129                 $this->addOption( 'LIMIT', $this->params['limit'] + 1 );
 130                 $this->addOption( 'ORDER BY', $this->bl_from );
 131                 $this->addOption( 'STRAIGHT_JOIN' );
 132         }
 133
 134         private function prepareSecondQuery( $resultPageSet = null ) {
 135                 /* SELECT page_id, page_title, page_namespace, page_is_redirect, pl_title, pl_namespace
 136                    FROM pagelinks, page WHERE pl_from=page_id
 137                    AND (pl_title='Foo' AND pl_namespace=0) OR (pl_title='Bar' AND pl_namespace=1)
 138                    ORDER BY pl_namespace, pl_title, pl_from LIMIT 11
 139                  */
 140                 $db = $this->getDB();
 141                 $this->addTables( array( 'page', $this->bl_table ) );
 142                 $this->addWhere( "{$this->bl_from}=page_id" );
 143
 144                 if ( is_null( $resultPageSet ) )
 145                         $this->addFields( array( 'page_id', 'page_title', 'page_namespace', 'page_is_redirect' ) );
 146                 else
 147                         $this->addFields( $resultPageSet->getPageTableFields() );
 148
 149                 $this->addFields( $this->bl_title );
 150                 if ( $this->hasNS )
 151                         $this->addFields( $this->bl_ns );
 152
 153                 // We can't use LinkBatch here because $this->hasNS may be false
 154                 $titleWhere = array();
 155                 foreach ( $this->redirTitles as $t )
 156                         $titleWhere[] = "{$this->bl_title} = " . $db->addQuotes( $t->getDBkey() ) .
 157                                         ( $this->hasNS ? " AND {$this->bl_ns} = '{$t->getNamespace()}'" : "" );
 158                 $this->addWhere( $db->makeList( $titleWhere, LIST_OR ) );
 159                 $this->addWhereFld( 'page_namespace', $this->params['namespace'] );
 160
 161                 if ( !is_null( $this->redirID ) )
 162                 {
 163                         $first = $this->redirTitles[0];
 164                         $title = $db->strencode( $first->getDBkey() );
 165                         $ns = $first->getNamespace();
 166                         $from = $this->redirID;
 167                         if ( $this->hasNS )
 168                                 $this->addWhere( "{$this->bl_ns} > $ns OR " .
 169                                                 "({$this->bl_ns} = $ns AND " .
 170                                                 "({$this->bl_title} > '$title' OR " .
 171                                                 "({$this->bl_title} = '$title' AND " .
 172                                                 "{$this->bl_from} >= $from)))" );
 173                         else
 174                                 $this->addWhere( "{$this->bl_title} > '$title' OR " .
 175                                                 "({$this->bl_title} = '$title' AND " .
 176                                                 "{$this->bl_from} >= $from)" );
 177
 178                 }
 179                 if ( $this->params['filterredir'] == 'redirects' )
 180                         $this->addWhereFld( 'page_is_redirect', 1 );
 181                 else if ( $this->params['filterredir'] == 'nonredirects' )
 182                         $this->addWhereFld( 'page_is_redirect', 0 );
 183
 184                 $this->addOption( 'LIMIT', $this->params['limit'] + 1 );
 185                 $this->addOption( 'ORDER BY', $this->bl_sort );
 186                 $this->addOption( 'USE INDEX', array( 'page' => 'PRIMARY' ) );
 187         }
 188
 189         private function run( $resultPageSet = null ) {
 190                 $this->params = $this->extractRequestParams( false );
 191                 $this->redirect = isset( $this->params['redirect'] ) && $this->params['redirect'];
 192                 $userMax = ( $this->redirect ? ApiBase::LIMIT_BIG1 / 2 : ApiBase::LIMIT_BIG1 );
 193                 $botMax  = ( $this->redirect ? ApiBase::LIMIT_BIG2 / 2 : ApiBase::LIMIT_BIG2 );
 194                 if ( $this->params['limit'] == 'max' ) {
 195                         $this->params['limit'] = $this->getMain()->canApiHighLimits() ? $botMax : $userMax;
 196                         $this->getResult()->addValue( 'limits', $this->getModuleName(), $this->params['limit'] );
 197                 }
 198
 199                 $this->processContinue();
 200                 $this->prepareFirstQuery( $resultPageSet );
 201
 202                 $db = $this->getDB();
 203                 $res = $this->select( __METHOD__ . '::firstQuery' );
 204
 205                 $count = 0;
 206                 $this->pageMap = array(); // Maps ns and title to pageid
 207                 $this->continueStr = null;
 208                 $this->redirTitles = array();
 209                 while ( $row = $db->fetchObject( $res ) ) {
 210                         if ( ++ $count > $this->params['limit'] ) {
 211                                 // We've reached the one extra which shows that there are additional pages to be had. Stop here...
 212                                 // Continue string preserved in case the redirect query doesn't pass the limit
 213                                 $this->continueStr = $this->getContinueStr( $row->page_id );
 214                                 break;
 215                         }
 216
 217                         if ( is_null( $resultPageSet ) )
 218                                 $this->extractRowInfo( $row );
 219                         else
 220                         {
 221                                 $this->pageMap[$row->page_namespace][$row->page_title] = $row->page_id;
 222                                 if ( $row->page_is_redirect )
 223                                         $this->redirTitles[] = Title::makeTitle( $row->page_namespace, $row->page_title );
 224
 225                                 $resultPageSet->processDbRow( $row );
 226                         }
 227                 }
 228                 $db->freeResult( $res );
 229
 230                 if ( $this->redirect && count( $this->redirTitles ) )
 231                 {
 232                         $this->resetQueryParams();
 233                         $this->prepareSecondQuery( $resultPageSet );
 234                         $res = $this->select( __METHOD__ . '::secondQuery' );
 235                         $count = 0;
 236                         while ( $row = $db->fetchObject( $res ) )
 237                         {
 238                                 if ( ++$count > $this->params['limit'] )
 239                                 {
 240                                         // We've reached the one extra which shows that there are additional pages to be had. Stop here...
 241                                         // We need to keep the parent page of this redir in
 242                                         if ( $this->hasNS )
 243                                                 $parentID = $this->pageMap[$row-> { $this->bl_ns } ][$row-> { $this->bl_title } ];
 244                                         else
 245                                                 $parentID = $this->pageMap[NS_IMAGE][$row-> { $this->bl_title } ];
 246                                         $this->continueStr = $this->getContinueRedirStr( $parentID, $row->page_id );
 247                                         break;
 248                                 }
 249
 250                                 if ( is_null( $resultPageSet ) )
 251                                         $this->extractRedirRowInfo( $row );
 252                                 else
 253                                         $resultPageSet->processDbRow( $row );
 254                         }
 255                         $db->freeResult( $res );
 256                 }
 257                 if ( is_null( $resultPageSet ) ) {
 258                         // Try to add the result data in one go and pray that it fits
 259                         $fit = $this->getResult()->addValue( 'query', $this->getModuleName(), array_values( $this->resultArr ) );
 260                         if ( !$fit )
 261                         {
 262                                 // It didn't fit. Add elements one by one until the
 263                                 // result is full.
 264                                 foreach ( $this->resultArr as $pageID => $arr )
 265                                 {
 266                                         // Add the basic entry without redirlinks first
 267                                         $fit = $this->getResult()->addValue(
 268                                                 array( 'query', $this->getModuleName() ),
 269                                                 null, array_diff_key( $arr, array( 'redirlinks' => '' ) ) );
 270                                         if ( !$fit )
 271                                         {
 272                                                 $this->continueStr = $this->getContinueStr( $pageID );
 273                                                 break;
 274                                         }
 275
 276                                         $hasRedirs = false;
 277                                         foreach ( (array)@$arr['redirlinks'] as $key => $redir )
 278                                         {
 279                                                 $fit = $this->getResult()->addValue(
 280                                                         array( 'query', $this->getModuleName(), $pageID, 'redirlinks' ),
 281                                                         $key, $redir );
 282                                                 if ( !$fit )
 283                                                 {
 284                                                         $this->continueStr = $this->getContinueRedirStr( $pageID, $redir['pageid'] );
 285                                                         break;
 286                                                 }
 287                                                 $hasRedirs = true;
 288                                         }
 289                                         if ( $hasRedirs )
 290                                                 $this->getResult()->setIndexedTagName_internal(
 291                                                         array( 'query', $this->getModuleName(), $pageID, 'redirlinks' ),
 292                                                         $this->bl_code );
 293                                         if ( !$fit )
 294                                                 break;
 295                                 }
 296                         }
 297
 298                         $this->getResult()->setIndexedTagName_internal(
 299                                          array( 'query', $this->getModuleName() ),
 300                                          $this->bl_code );
 301                 }
 302                 if ( !is_null( $this->continueStr ) )
 303                         $this->setContinueEnumParameter( 'continue', $this->continueStr );
 304         }
 305
 306         private function extractRowInfo( $row ) {
 307                 $this->pageMap[$row->page_namespace][$row->page_title] = $row->page_id;
 308                 $t = Title::makeTitle( $row->page_namespace, $row->page_title );
 309                 $a = array( 'pageid' => intval( $row->page_id ) );
 310                 ApiQueryBase::addTitleInfo( $a, $t );
 311                 if ( $row->page_is_redirect )
 312                 {
 313                         $a['redirect'] = '';
 314                         $this->redirTitles[] = $t;
 315                 }
 316                 // Put all the results in an array first
 317                 $this->resultArr[$a['pageid']] = $a;
 318         }
 319
 320         private function extractRedirRowInfo( $row )
 321         {
 322                 $a['pageid'] = intval( $row->page_id );
 323                 ApiQueryBase::addTitleInfo( $a, Title::makeTitle( $row->page_namespace, $row->page_title ) );
 324                 if ( $row->page_is_redirect )
 325                         $a['redirect'] = '';
 326                 $ns = $this->hasNS ? $row-> { $this->bl_ns } : NS_FILE;
 327                 $parentID = $this->pageMap[$ns][$row-> { $this->bl_title } ];
 328                 // Put all the results in an array first
 329                 $this->resultArr[$parentID]['redirlinks'][] = $a;
 330                 $this->getResult()->setIndexedTagName( $this->resultArr[$parentID]['redirlinks'], $this->bl_code );
 331         }
 332
 333         protected function processContinue() {
 334                 if ( !is_null( $this->params['continue'] ) )
 335                         $this->parseContinueParam();
 336                 else {
 337                         if ( $this->params['title'] !== "" ) {
 338                                 $title = Title::newFromText( $this->params['title'] );
 339                                 if ( !$title ) {
 340                                         $this->dieUsageMsg( array( 'invalidtitle', $this->params['title'] ) );
 341                                 } else {
 342                                         $this->rootTitle = $title;
 343                                 }
 344                         } else {
 345                                 $this->dieUsageMsg( array( 'missingparam', 'title' ) );
 346                         }
 347                 }
 348
 349                 // only image titles are allowed for the root in imageinfo mode
 350                 if ( !$this->hasNS && $this->rootTitle->getNamespace() !== NS_FILE )
 351                         $this->dieUsage( "The title for {$this->getModuleName()} query must be an image", 'bad_image_title' );
 352         }
 353
 354         protected function parseContinueParam() {
 355                 $continueList = explode( '|', $this->params['continue'] );
 356                 // expected format:
 357                 // ns | key | id1 [| id2]
 358                 // ns+key: root title
 359                 // id1: first-level page ID to continue from
 360                 // id2: second-level page ID to continue from
 361
 362                 // null stuff out now so we know what's set and what isn't
 363                 $this->rootTitle = $this->contID = $this->redirID = null;
 364                 $rootNs = intval( $continueList[0] );
 365                 if ( $rootNs === 0 && $continueList[0] !== '0' )
 366                         // Illegal continue parameter
 367                         $this->dieUsage( "Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue" );
 368                 $this->rootTitle = Title::makeTitleSafe( $rootNs, $continueList[1] );
 369
 370                 if ( !$this->rootTitle )
 371                         $this->dieUsage( "Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue" );
 372                 $contID = intval( $continueList[2] );
 373
 374                 if ( $contID === 0 && $continueList[2] !== '0' )
 375                         $this->dieUsage( "Invalid continue param. You should pass the original value returned by the previous query", "_badcontinue" );
 376                 $this->contID = $contID;
 377                 $redirID = intval( @$continueList[3] );
 378
 379                 if ( $redirID === 0 && @$continueList[3] !== '0' )
 380                         // This one isn't required
 381                         return;
 382                 $this->redirID = $redirID;
 383
 384         }
 385
 386         protected function getContinueStr( $lastPageID ) {
 387                 return $this->rootTitle->getNamespace() .
 388                 '|' . $this->rootTitle->getDBkey() .
 389                 '|' . $lastPageID;
 390         }
 391
 392         protected function getContinueRedirStr( $lastPageID, $lastRedirID ) {
 393                 return $this->getContinueStr( $lastPageID ) . '|' . $lastRedirID;
 394         }
 395
 396         public function getAllowedParams() {
 397                 $retval =  array (
 398                         'title' => null,
 399                         'continue' => null,
 400                         'namespace' => array (
 401                                 ApiBase :: PARAM_ISMULTI => true,
 402                                 ApiBase :: PARAM_TYPE => 'namespace'
 403                         ),
 404                         'filterredir' => array(
 405                                 ApiBase :: PARAM_DFLT => 'all',
 406                                 ApiBase :: PARAM_TYPE => array(
 407                                         'all',
 408                                         'redirects',
 409                                         'nonredirects'
 410                                 )
 411                         ),
 412                         'limit' => array (
 413                                 ApiBase :: PARAM_DFLT => 10,
 414                                 ApiBase :: PARAM_TYPE => 'limit',
 415                                 ApiBase :: PARAM_MIN => 1,
 416                                 ApiBase :: PARAM_MAX => ApiBase :: LIMIT_BIG1,
 417                                 ApiBase :: PARAM_MAX2 => ApiBase :: LIMIT_BIG2
 418                         )
 419                 );
 420                 if ( $this->getModuleName() == 'embeddedin' )
 421                         return $retval;
 422                 $retval['redirect'] = false;
 423                 return $retval;
 424         }
 425
 426         public function getParamDescription() {
 427                 $retval = array (
 428                         'title' => 'Title to search.',
 429                         'continue' => 'When more results are available, use this to continue.',
 430                         'namespace' => 'The namespace to enumerate.',
 431                 );
 432                 if ( $this->getModuleName() != 'embeddedin' )
 433                         return array_merge( $retval, array(
 434                                 'redirect' => 'If linking page is a redirect, find all pages that link to that redirect as well. Maximum limit is halved.',
 435                                 'filterredir' => "How to filter for redirects. If set to nonredirects when {$this->bl_code}redirect is enabled, this is only applied to the second level",
 436                                 '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)."
 437                         ) );
 438                 return array_merge( $retval, array(
 439                         'filterredir' => 'How to filter for redirects',
 440                         'limit' => 'How many total pages to return.'
 441                 ) );
 442         }
 443
 444         public function getDescription() {
 445                 switch ( $this->getModuleName() ) {
 446                         case 'backlinks' :
 447                                 return 'Find all pages that link to the given page';
 448                         case 'embeddedin' :
 449                                 return 'Find all pages that embed (transclude) the given title';
 450                         case 'imageusage' :
 451                                 return 'Find all pages that use the given image title.';
 452                         default :
 453                                 ApiBase :: dieDebug( __METHOD__, 'Unknown module name' );
 454                 }
 455         }
 456
 457         public function getPossibleErrors() {
 458                 return array_merge( parent::getPossibleErrors(), array(
 459                         array( 'invalidtitle', 'title' ),
 460                         array( 'missingparam', 'title' ),
 461                         array( 'code' => 'bad_image_title', 'info' => "The title for {$this->getModuleName()} query must be an image" ),
 462                         array( 'code' => '_badcontinue', 'info' => 'Invalid continue param. You should pass the original value returned by the previous query' ),
 463                 ) );
 464         }
 465
 466         protected function getExamples() {
 467                 static $examples = array (
 468                         'backlinks' => array (
 469                                 "api.php?action=query&list=backlinks&bltitle=Main%20Page",
 470                                 "api.php?action=query&generator=backlinks&gbltitle=Main%20Page&prop=info"
 471                         ),
 472                         'embeddedin' => array (
 473                                 "api.php?action=query&list=embeddedin&eititle=Template:Stub",
 474                                 "api.php?action=query&generator=embeddedin&geititle=Template:Stub&prop=info"
 475                         ),
 476                         'imageusage' => array (
 477                                 "api.php?action=query&list=imageusage&iutitle=File:Albert%20Einstein%20Head.jpg",
 478                                 "api.php?action=query&generator=imageusage&giutitle=File:Albert%20Einstein%20Head.jpg&prop=info"
 479                         )
 480                 );
 481
 482                 return $examples[$this->getModuleName()];
 483         }
 484
 485         public function getVersion() {
 486                 return __CLASS__ . ': $Id$';
 487         }
 488 }