includes/api/ApiQueryCategoryMembers.php

   1 <?php
   2 /**
   3  * API for MediaWiki 1.8+
   4  *
   5  * Created on June 14, 2007
   6  *
   7  * Copyright © 2006 Yuri Astrakhan <Firstname><Lastname>@gmail.com
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License along
  20  * with this program; if not, write to the Free Software Foundation, Inc.,
  21  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  22  * http://www.gnu.org/copyleft/gpl.html
  23  *
  24  * @file
  25  */
  26
  27 if ( !defined( 'MEDIAWIKI' ) ) {
  28         // Eclipse helper - will be ignored in production
  29         require_once( "ApiQueryBase.php" );
  30 }
  31
  32 /**
  33  * A query module to enumerate pages that belong to a category.
  34  *
  35  * @ingroup API
  36  */
  37 class ApiQueryCategoryMembers extends ApiQueryGeneratorBase {
  38
  39         public function __construct( $query, $moduleName ) {
  40                 parent::__construct( $query, $moduleName, 'cm' );
  41         }
  42
  43         public function execute() {
  44                 $this->run();
  45         }
  46
  47         public function getCacheMode( $params ) {
  48                 return 'public';
  49         }
  50
  51         public function executeGenerator( $resultPageSet ) {
  52                 $this->run( $resultPageSet );
  53         }
  54
  55         private function run( $resultPageSet = null ) {
  56                 $params = $this->extractRequestParams();
  57
  58                 $categoryTitle = Title::newFromText( $params['title'] );
  59
  60                 if ( is_null( $categoryTitle ) || $categoryTitle->getNamespace() != NS_CATEGORY ) {
  61                         $this->dieUsage( 'The category name you entered is not valid', 'invalidcategory' );
  62                 }
  63
  64                 $prop = array_flip( $params['prop'] );
  65                 $fld_ids = isset( $prop['ids'] );
  66                 $fld_title = isset( $prop['title'] );
  67                 $fld_sortkey = isset( $prop['sortkey'] );
  68                 $fld_sortkeyprefix = isset( $prop['sortkeyprefix'] );
  69                 $fld_timestamp = isset( $prop['timestamp'] );
  70                 $fld_type = isset( $prop['type'] );
  71
  72                 if ( is_null( $resultPageSet ) ) {
  73                         $this->addFields( array( 'cl_from', 'cl_sortkey', 'cl_type', 'page_namespace', 'page_title' ) );
  74                         $this->addFieldsIf( 'page_id', $fld_ids );
  75                         $this->addFieldsIf( 'cl_sortkey_prefix', $fld_sortkeyprefix );
  76                 } else {
  77                         $this->addFields( $resultPageSet->getPageTableFields() ); // will include page_ id, ns, title
  78                         $this->addFields( array( 'cl_from', 'cl_sortkey', 'cl_type' ) );
  79                 }
  80
  81                 $this->addFieldsIf( 'cl_timestamp', $fld_timestamp || $params['sort'] == 'timestamp' );
  82
  83                 $this->addTables( array( 'page', 'categorylinks' ) );   // must be in this order for 'USE INDEX'
  84
  85                 $this->addWhereFld( 'cl_to', $categoryTitle->getDBkey() );
  86                 $queryTypes = $params['type'];
  87                 $contWhere = false;
  88
  89                 // Scanning large datasets for rare categories sucks, and I already told
  90                 // how to have efficient subcategory access :-) ~~~~ (oh well, domas)
  91                 global $wgMiserMode;
  92                 $miser_ns = array();
  93                 if ( $wgMiserMode ) {
  94                         $miser_ns = $params['namespace'];
  95                 } else {
  96                         $this->addWhereFld( 'page_namespace', $params['namespace'] );
  97                 }
  98
  99                 $dir = $params['dir'] == 'asc' ? 'newer' : 'older';
 100
 101                 if ( $params['sort'] == 'timestamp' ) {
 102                         $this->addWhereRange( 'cl_timestamp',
 103                                 $dir,
 104                                 $params['start'],
 105                                 $params['end'] );
 106
 107                         $this->addOption( 'USE INDEX', 'cl_timestamp' );
 108                 } else {
 109                         if ( $params['continue'] ) {
 110                                 $cont = explode( '|', $params['continue'], 3 );
 111                                 if ( count( $cont ) != 3 ) {
 112                                         $this->dieUsage( 'Invalid continue param. You should pass the original value returned '.
 113                                                 'by the previous query', '_badcontinue'
 114                                         );
 115                                 }
 116
 117                                 // Remove the types to skip from $queryTypes
 118                                 $contTypeIndex = array_search( $cont[0], $queryTypes );
 119                                 $queryTypes = array_slice( $queryTypes, $contTypeIndex );
 120
 121                                 // Add a WHERE clause for sortkey and from
 122                                 // pack( "H*", $foo ) is used to convert hex back to binary
 123                                 $escSortkey = $this->getDB()->addQuotes( pack( "H*", $cont[1] ) );
 124                                 $from = intval( $cont[2] );
 125                                 $op = $dir == 'newer' ? '>' : '<';
 126                                 // $contWhere is used further down
 127                                 $contWhere = "cl_sortkey $op $escSortkey OR " .
 128                                         "(cl_sortkey = $escSortkey AND " .
 129                                         "cl_from $op= $from)";
 130
 131                         } else {
 132                                 // The below produces ORDER BY cl_sortkey, cl_from, possibly with DESC added to each of them
 133                                 $this->addWhereRange( 'cl_sortkey',
 134                                         $dir,
 135                                         $params['startsortkey'],
 136                                         $params['endsortkey'] );
 137                                 $this->addWhereRange( 'cl_from', $dir, null, null );
 138                         }
 139                         $this->addOption( 'USE INDEX', 'cl_sortkey' );
 140                 }
 141
 142                 $this->addWhere( 'cl_from=page_id' );
 143
 144                 $limit = $params['limit'];
 145                 $this->addOption( 'LIMIT', $limit + 1 );
 146
 147                 if ( $params['sort'] == 'sortkey' ) {
 148                         // Run a separate SELECT query for each value of cl_type.
 149                         // This is needed because cl_type is an enum, and MySQL has
 150                         // inconsistencies between ORDER BY cl_type and
 151                         // WHERE cl_type >= 'foo' making proper paging impossible
 152                         // and unindexed.
 153                         $rows = array();
 154                         $first = true;
 155                         foreach ( $queryTypes as $type ) {
 156                                 $extraConds = array( 'cl_type' => $type );
 157                                 if ( $first && $contWhere ) {
 158                                         // Continuation condition. Only added to the
 159                                         // first query, otherwise we'll skip things
 160                                         $extraConds[] = $contWhere;
 161                                 }
 162                                 $res = $this->select( __METHOD__, array( 'where' => $extraConds ) );
 163                                 $rows = array_merge( $rows, iterator_to_array( $res ) );
 164                                 if ( count( $rows ) >= $limit + 1 ) {
 165                                         break;
 166                                 }
 167                                 $first = false;
 168                         }
 169                 } else {
 170                         // Sorting by timestamp
 171                         // No need to worry about per-type queries because we
 172                         // aren't sorting or filtering by type anyway
 173                         $res = $this->select( __METHOD__ );
 174                         $rows = iterator_to_array( $res );
 175                 }
 176                 $count = 0;
 177                 foreach ( $rows as $row ) {
 178                         if ( ++ $count > $limit ) {
 179                                 // We've reached the one extra which shows that there are additional pages to be had. Stop here...
 180                                 // TODO: Security issue - if the user has no right to view next title, it will still be shown
 181                                 if ( $params['sort'] == 'timestamp' ) {
 182                                         $this->setContinueEnumParameter( 'start', wfTimestamp( TS_ISO_8601, $row->cl_timestamp ) );
 183                                 } else {
 184                                         $sortkey = bin2hex( $row->cl_sortkey );
 185                                         $this->setContinueEnumParameter( 'continue',
 186                                                 "{$row->cl_type}|$sortkey|{$row->cl_from}"
 187                                         );
 188                                 }
 189                                 break;
 190                         }
 191
 192                         // Since domas won't tell anyone what he told long ago, apply
 193                         // cmnamespace here. This means the query may return 0 actual
 194                         // results, but on the other hand it could save returning 5000
 195                         // useless results to the client. ~~~~
 196                         if ( count( $miser_ns ) && !in_array( $row->page_namespace, $miser_ns ) ) {
 197                                 continue;
 198                         }
 199
 200                         if ( is_null( $resultPageSet ) ) {
 201                                 $vals = array();
 202                                 if ( $fld_ids ) {
 203                                         $vals['pageid'] = intval( $row->page_id );
 204                                 }
 205                                 if ( $fld_title ) {
 206                                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 207                                         ApiQueryBase::addTitleInfo( $vals, $title );
 208                                 }
 209                                 if ( $fld_sortkey ) {
 210                                         $vals['sortkey'] = bin2hex( $row->cl_sortkey );
 211                                 }
 212                                 if ( $fld_sortkeyprefix ) {
 213                                         $vals['sortkeyprefix'] = $row->cl_sortkey_prefix;
 214                                 }
 215                                 if ( $fld_type  ) {
 216                                         $vals['type'] = $row->cl_type;
 217                                 }
 218                                 if ( $fld_timestamp ) {
 219                                         $vals['timestamp'] = wfTimestamp( TS_ISO_8601, $row->cl_timestamp );
 220                                 }
 221                                 $fit = $this->getResult()->addValue( array( 'query', $this->getModuleName() ),
 222                                                 null, $vals );
 223                                 if ( !$fit ) {
 224                                         if ( $params['sort'] == 'timestamp' ) {
 225                                                 $this->setContinueEnumParameter( 'start', wfTimestamp( TS_ISO_8601, $row->cl_timestamp ) );
 226                                         } else {
 227                                                 $sortkey = bin2hex( $row->cl_sortkey );
 228                                                 $this->setContinueEnumParameter( 'continue',
 229                                                         "{$row->cl_type}|$sortkey|{$row->cl_from}"
 230                                                 );
 231                                         }
 232                                         break;
 233                                 }
 234                         } else {
 235                                 $resultPageSet->processDbRow( $row );
 236                         }
 237                 }
 238
 239                 if ( is_null( $resultPageSet ) ) {
 240                         $this->getResult()->setIndexedTagName_internal(
 241                                          array( 'query', $this->getModuleName() ), 'cm' );
 242                 }
 243         }
 244
 245         public function getAllowedParams() {
 246                 return array(
 247                         'title' => array(
 248                                 ApiBase::PARAM_TYPE => 'string',
 249                                 ApiBase::PARAM_REQUIRED => true
 250                         ),
 251                         'prop' => array(
 252                                 ApiBase::PARAM_DFLT => 'ids|title',
 253                                 ApiBase::PARAM_ISMULTI => true,
 254                                 ApiBase::PARAM_TYPE => array (
 255                                         'ids',
 256                                         'title',
 257                                         'sortkey',
 258                                         'sortkeyprefix',
 259                                         'type',
 260                                         'timestamp',
 261                                 )
 262                         ),
 263                         'namespace' => array (
 264                                 ApiBase::PARAM_ISMULTI => true,
 265                                 ApiBase::PARAM_TYPE => 'namespace',
 266                         ),
 267                         'type' => array(
 268                                 ApiBase::PARAM_ISMULTI => true,
 269                                 ApiBase::PARAM_DFLT => 'page|subcat|file',
 270                                 ApiBase::PARAM_TYPE => array(
 271                                         'page',
 272                                         'subcat',
 273                                         'file'
 274                                 )
 275                         ),
 276                         'continue' => null,
 277                         'limit' => array(
 278                                 ApiBase::PARAM_TYPE => 'limit',
 279                                 ApiBase::PARAM_DFLT => 10,
 280                                 ApiBase::PARAM_MIN => 1,
 281                                 ApiBase::PARAM_MAX => ApiBase::LIMIT_BIG1,
 282                                 ApiBase::PARAM_MAX2 => ApiBase::LIMIT_BIG2
 283                         ),
 284                         'sort' => array(
 285                                 ApiBase::PARAM_DFLT => 'sortkey',
 286                                 ApiBase::PARAM_TYPE => array(
 287                                         'sortkey',
 288                                         'timestamp'
 289                                 )
 290                         ),
 291                         'dir' => array(
 292                                 ApiBase::PARAM_DFLT => 'asc',
 293                                 ApiBase::PARAM_TYPE => array(
 294                                         'asc',
 295                                         'desc'
 296                                 )
 297                         ),
 298                         'start' => array(
 299                                 ApiBase::PARAM_TYPE => 'timestamp'
 300                         ),
 301                         'end' => array(
 302                                 ApiBase::PARAM_TYPE => 'timestamp'
 303                         ),
 304                         'startsortkey' => null,
 305                         'endsortkey' => null,
 306                 );
 307         }
 308
 309         public function getParamDescription() {
 310                 global $wgMiserMode;
 311                 $p = $this->getModulePrefix();
 312                 $desc = array(
 313                         'title' => 'Which category to enumerate (required). Must include Category: prefix',
 314                         'prop' => array(
 315                                 'What pieces of information to include',
 316                                 ' ids           - Adds the page ID',
 317                                 ' title         - Adds the title and namespace ID of the page',
 318                                 ' sortkey       - Adds the sortkey used for sorting in the category (hexadecimal string)',
 319                                 ' sortkeyprefix - Adds the sortkey prefix used for sorting in the category (human-readable part of the sortkey)',
 320                                 ' type          - Adds the type that the page has been categorised as (page, subcat or file)',
 321                                 ' timestamp     - Adds the timestamp of when the page was included',
 322                         ),
 323                         'namespace' => 'Only include pages in these namespaces',
 324                         'type' => "What type of category members to include. Ignored when {$p}sort=timestamp is set",
 325                         'sort' => 'Property to sort by',
 326                         'dir' => 'In which direction to sort',
 327                         'start' => "Timestamp to start listing from. Can only be used with {$p}sort=timestamp",
 328                         'end' => "Timestamp to end listing at. Can only be used with {$p}sort=timestamp",
 329                         'startsortkey' => "Sortkey to start listing from. Can only be used with {$p}sort=sortkey",
 330                         'endsortkey' => "Sortkey to end listing at. Can only be used with {$p}sort=sortkey",
 331                         'continue' => 'For large categories, give the value retured from previous query',
 332                         'limit' => 'The maximum number of pages to return.',
 333                 );
 334                 if ( $wgMiserMode ) {
 335                         $desc['namespace'] = array(
 336                                 $desc['namespace'],
 337                                 'NOTE: Due to $wgMiserMode, using this may result in fewer than "limit" results',
 338                                 'returned before continuing; in extreme cases, zero results may be returned.',
 339                                 'Note that you can use cmtype=subcat or cmtype=file instead of cmnamespace=14 or 6.',
 340                         );
 341                 }
 342                 return $desc;
 343         }
 344
 345         public function getDescription() {
 346                 return 'List all pages in a given category';
 347         }
 348
 349         public function getPossibleErrors() {
 350                 return array_merge( parent::getPossibleErrors(), array(
 351                         array( 'code' => 'notitle', 'info' => 'The cmtitle parameter is required' ),
 352                         array( 'code' => 'invalidcategory', 'info' => 'The category name you entered is not valid' ),
 353                         array( 'code' => 'badcontinue', 'info' => 'Invalid continue param. You should pass the original value returned by the previous query' ),
 354                 ) );
 355         }
 356
 357         protected function getExamples() {
 358                 return array(
 359                         'Get first 10 pages in [[Category:Physics]]:',
 360                         '  api.php?action=query&list=categorymembers&cmtitle=Category:Physics',
 361                         'Get page info about first 10 pages in [[Category:Physics]]:',
 362                         '  api.php?action=query&generator=categorymembers&gcmtitle=Category:Physics&prop=info',
 363                 );
 364         }
 365
 366         public function getVersion() {
 367                 return __CLASS__ . ': $Id$';
 368         }
 369 }