$this->run( $resultPageSet );
}
+ /**
+ * @param $resultPageSet ApiPageSet
+ * @return void
+ */
private function run( $resultPageSet = null ) {
$params = $this->extractRequestParams();
$fld_type = isset( $prop['type'] );
if ( is_null( $resultPageSet ) ) {
- $this->addFields( array( 'cl_from', 'page_namespace', 'page_title' ) );
+ $this->addFields( array( 'cl_from', 'cl_sortkey', 'cl_type', 'page_namespace', 'page_title' ) );
$this->addFieldsIf( 'page_id', $fld_ids );
$this->addFieldsIf( 'cl_sortkey_prefix', $fld_sortkeyprefix );
- $this->addFieldsIf( 'cl_sortkey', $fld_sortkey );
} else {
$this->addFields( $resultPageSet->getPageTableFields() ); // will include page_ id, ns, title
- $this->addFields( array( 'cl_from', 'cl_sortkey' ) );
+ $this->addFields( array( 'cl_from', 'cl_sortkey', 'cl_type' ) );
}
$this->addFieldsIf( 'cl_timestamp', $fld_timestamp || $params['sort'] == 'timestamp' );
- $this->addFieldsIf( 'cl_type', $fld_type );
$this->addTables( array( 'page', 'categorylinks' ) ); // must be in this order for 'USE INDEX'
$this->addWhereFld( 'cl_to', $categoryTitle->getDBkey() );
- $this->addWhereFld( 'cl_type', $params['type'] );
+ $queryTypes = $params['type'];
+ $contWhere = false;
// Scanning large datasets for rare categories sucks, and I already told
// how to have efficient subcategory access :-) ~~~~ (oh well, domas)
$this->addOption( 'USE INDEX', 'cl_timestamp' );
} else {
- $this->addOption( 'ORDER BY', 'cl_type' );
-
- $this->addWhereRange( 'cl_sortkey',
- $dir,
- $params['startsortkey'],
- $params['endsortkey'] );
+ if ( $params['continue'] ) {
+ $cont = explode( '|', $params['continue'], 3 );
+ if ( count( $cont ) != 3 ) {
+ $this->dieUsage( 'Invalid continue param. You should pass the original value returned '.
+ 'by the previous query', '_badcontinue'
+ );
+ }
- $this->addWhereRange( 'cl_from', $dir, null, null );
+ // Remove the types to skip from $queryTypes
+ $contTypeIndex = array_search( $cont[0], $queryTypes );
+ $queryTypes = array_slice( $queryTypes, $contTypeIndex );
+
+ // Add a WHERE clause for sortkey and from
+ // pack( "H*", $foo ) is used to convert hex back to binary
+ $escSortkey = $this->getDB()->addQuotes( pack( "H*", $cont[1] ) );
+ $from = intval( $cont[2] );
+ $op = $dir == 'newer' ? '>' : '<';
+ // $contWhere is used further down
+ $contWhere = "cl_sortkey $op $escSortkey OR " .
+ "(cl_sortkey = $escSortkey AND " .
+ "cl_from $op= $from)";
+ // The below produces ORDER BY cl_sortkey, cl_from, possibly with DESC added to each of them
+ $this->addWhereRange( 'cl_sortkey', $dir, null, null );
+ $this->addWhereRange( 'cl_from', $dir, null, null );
+ } else {
+ // The below produces ORDER BY cl_sortkey, cl_from, possibly with DESC added to each of them
+ $this->addWhereRange( 'cl_sortkey',
+ $dir,
+ $params['startsortkey'],
+ $params['endsortkey'] );
+ $this->addWhereRange( 'cl_from', $dir, null, null );
+ }
$this->addOption( 'USE INDEX', 'cl_sortkey' );
}
- $this->setContinuation( $params['continue'], $params['dir'] );
-
$this->addWhere( 'cl_from=page_id' );
$limit = $params['limit'];
$this->addOption( 'LIMIT', $limit + 1 );
+ if ( $params['sort'] == 'sortkey' ) {
+ // Run a separate SELECT query for each value of cl_type.
+ // This is needed because cl_type is an enum, and MySQL has
+ // inconsistencies between ORDER BY cl_type and
+ // WHERE cl_type >= 'foo' making proper paging impossible
+ // and unindexed.
+ $rows = array();
+ $first = true;
+ foreach ( $queryTypes as $type ) {
+ $extraConds = array( 'cl_type' => $type );
+ if ( $first && $contWhere ) {
+ // Continuation condition. Only added to the
+ // first query, otherwise we'll skip things
+ $extraConds[] = $contWhere;
+ }
+ $res = $this->select( __METHOD__, array( 'where' => $extraConds ) );
+ $rows = array_merge( $rows, iterator_to_array( $res ) );
+ if ( count( $rows ) >= $limit + 1 ) {
+ break;
+ }
+ $first = false;
+ }
+ } else {
+ // Sorting by timestamp
+ // No need to worry about per-type queries because we
+ // aren't sorting or filtering by type anyway
+ $res = $this->select( __METHOD__ );
+ $rows = iterator_to_array( $res );
+ }
+
+ $result = $this->getResult();
$count = 0;
- $lastFrom = null;
- $res = $this->select( __METHOD__ );
- foreach ( $res as $row ) {
+ foreach ( $rows as $row ) {
if ( ++ $count > $limit ) {
// We've reached the one extra which shows that there are additional pages to be had. Stop here...
// TODO: Security issue - if the user has no right to view next title, it will still be shown
if ( $params['sort'] == 'timestamp' ) {
$this->setContinueEnumParameter( 'start', wfTimestamp( TS_ISO_8601, $row->cl_timestamp ) );
} else {
- $this->setContinueEnumParameter( 'continue', $lastFrom );
+ $sortkey = bin2hex( $row->cl_sortkey );
+ $this->setContinueEnumParameter( 'continue',
+ "{$row->cl_type}|$sortkey|{$row->cl_from}"
+ );
}
break;
}
ApiQueryBase::addTitleInfo( $vals, $title );
}
if ( $fld_sortkey ) {
- $vals['sortkey'] = $row->cl_sortkey;
+ $vals['sortkey'] = bin2hex( $row->cl_sortkey );
}
if ( $fld_sortkeyprefix ) {
$vals['sortkeyprefix'] = $row->cl_sortkey_prefix;
if ( $fld_timestamp ) {
$vals['timestamp'] = wfTimestamp( TS_ISO_8601, $row->cl_timestamp );
}
- $fit = $this->getResult()->addValue( array( 'query', $this->getModuleName() ),
+ $fit = $result->addValue( array( 'query', $this->getModuleName() ),
null, $vals );
if ( !$fit ) {
if ( $params['sort'] == 'timestamp' ) {
$this->setContinueEnumParameter( 'start', wfTimestamp( TS_ISO_8601, $row->cl_timestamp ) );
} else {
- $this->setContinueEnumParameter( 'continue', $lastFrom );
+ $sortkey = bin2hex( $row->cl_sortkey );
+ $this->setContinueEnumParameter( 'continue',
+ "{$row->cl_type}|$sortkey|{$row->cl_from}"
+ );
}
break;
}
} else {
$resultPageSet->processDbRow( $row );
}
- $lastFrom = $row->cl_from; // detect duplicate sortkeys
}
if ( is_null( $resultPageSet ) ) {
- $this->getResult()->setIndexedTagName_internal(
+ $result->setIndexedTagName_internal(
array( 'query', $this->getModuleName() ), 'cm' );
}
}
- /**
- * Add DB WHERE clause to continue previous query based on 'continue' parameter
- */
- private function setContinuation( $continue, $dir ) {
- if ( is_null( $continue ) ) {
- return; // This is not a continuation request
- }
-
- $encFrom = $this->getDB()->addQuotes( intval( $continue ) );
-
- $op = ( $dir == 'desc' ? '<' : '>' );
-
- $this->addWhere( "cl_from $op $encFrom" );
- }
-
public function getAllowedParams() {
return array(
'title' => array(
global $wgMiserMode;
$p = $this->getModulePrefix();
$desc = array(
- 'title' => 'Which category to enumerate (required). Must include Category: prefix. Cannot be used together with cmpageid',
- 'pageid' => 'Page ID of the category to enumerate. Cannot be used together with cmtitle',
+ 'title' => "Which category to enumerate (required). Must include Category: prefix. Cannot be used together with {$p}pageid",
+ 'pageid' => "Page ID of the category to enumerate. Cannot be used together with {$p}title",
'prop' => array(
'What pieces of information to include',
' ids - Adds the page ID',
' title - Adds the title and namespace ID of the page',
- ' sortkey - Adds the sortkey used for the category (note, may be non human readable)',
- ' sortkeyprefix - Adds the sortkey prefix used for the category',
- ' type - Adds the type that the page has been categorised as',
+ ' sortkey - Adds the sortkey used for sorting in the category (hexadecimal string)',
+ ' sortkeyprefix - Adds the sortkey prefix used for sorting in the category (human-readable part of the sortkey)',
+ ' type - Adds the type that the page has been categorised as (page, subcat or file)',
' timestamp - Adds the timestamp of when the page was included',
),
'namespace' => 'Only include pages in these namespaces',
- 'type' => 'What type of category members to include',
+ 'type' => "What type of category members to include. Ignored when {$p}sort=timestamp is set",
'sort' => 'Property to sort by',
'dir' => 'In which direction to sort',
'start' => "Timestamp to start listing from. Can only be used with {$p}sort=timestamp",
'end' => "Timestamp to end listing at. Can only be used with {$p}sort=timestamp",
- 'startsortkey' => "Sortkey to start listing from. Can only be used with {$p}sort=sortkey",
- 'endsortkey' => "Sortkey to end listing at. Can only be used with {$p}sort=sortkey",
+ 'startsortkey' => "Sortkey to start listing from. Must be given in binary format. Can only be used with {$p}sort=sortkey",
+ 'endsortkey' => "Sortkey to end listing at. Must be given in binary format. Can only be used with {$p}sort=sortkey",
'continue' => 'For large categories, give the value retured from previous query',
'limit' => 'The maximum number of pages to return.',
);
+
if ( $wgMiserMode ) {
$desc['namespace'] = array(
$desc['namespace'],
- 'NOTE: Due to $wgMiserMode, using this may result in fewer than "limit" results',
- 'returned before continuing; in extreme cases, zero results may be returned',
+ "NOTE: Due to \$wgMiserMode, using this may result in fewer than \"{$p}limit\" results",
+ 'returned before continuing; in extreme cases, zero results may be returned.',
+ "Note that you can use {$p}type=subcat or {$p}type=file instead of {$p}namespace=14 or 6.",
);
}
return $desc;
}
public function getPossibleErrors() {
- return array_merge( parent::getPossibleErrors(), array(
- array( 'code' => 'cmmissingparam', 'info' => 'One of the parameters title, pageid is required' ),
- array( 'code' => 'cminvalidparammix', 'info' => 'The parameters title, pageid can not be used together' ),
- array( 'code' => 'invalidcategory', 'info' => 'The category name you entered is not valid' ),
- array( 'code' => 'badcontinue', 'info' => 'Invalid continue param. You should pass the original value returned by the previous query' ),
- array( 'nosuchpageid', 'pageid' ),
- ) );
+ return array_merge( parent::getPossibleErrors(),
+ $this->getRequireOnlyOneParameterErrorMessages( array( 'title', 'pageid' ) ),
+ array(
+ array( 'code' => 'invalidcategory', 'info' => 'The category name you entered is not valid' ),
+ array( 'code' => 'badcontinue', 'info' => 'Invalid continue param. You should pass the original value returned by the previous query' ),
+ array( 'nosuchpageid', 'pageid' ),
+ )
+ );
}
- protected function getExamples() {
+ public function getExamples() {
return array(
'Get first 10 pages in [[Category:Physics]]:',
' api.php?action=query&list=categorymembers&cmtitle=Category:Physics',
);
}
+ public function getHelpUrls() {
+ return 'http://www.mediawiki.org/wiki/API:Categorymembers';
+ }
+
public function getVersion() {
return __CLASS__ . ': $Id$';
}