/** * Helper function for reading all data for from a given property table * (specified by an SMWSQLStore3Table object), based on certain * restrictions. The function can filter data based on the subject (1) * or on the property it belongs to (2) -- but one of those must be * done. The Boolean $issubject is true for (1) and false for (2). * * In case (1), the first two parameters are taken to refer to a * subject; in case (2) they are taken to refer to a property. In any * case, the retrieval is limited to the specified $proptable. The * parameters are an internal $id (of a subject or property), and an * $object (being an DIWikiPage or SMWDIProperty). Moreover, when * filtering by property, it is assumed that the given $proptable * belongs to the property: if it is a table with fixed property, it * will not be checked that this is the same property as the one that * was given in $object. * * In case (1), the result in general is an array of pairs (arrays of * size 2) consisting of a property key (string), and DB keys (array if * many, string if one) from which a datvalue object for this value can * be built. It is possible that some of the DB keys are based on * internal objects; these will be represented by similar result arrays * of (recursive calls of) fetchSemanticData(). * * In case (2), the result is simply an array of DB keys (array) * without the property keys. Container objects will be encoded with * nested arrays like in case (1). * * @todo Maybe share DB handler; asking for it seems to take quite some * time and we do not want to change it in one call. * * @param integer $id * @param SMWDataItem $object * @param TableDefinition $propTable * @param boolean $isSubject * @param SMWRequestOptions $requestOptions * * @return array */ private function fetchSemanticData($id, SMWDataItem $object = null, TableDefinition $propTable, $isSubject = true, SMWRequestOptions $requestOptions = null) { // stop if there is not enough data: // properties always need to be given as object, // subjects at least if !$proptable->idsubject if ($id == 0 || is_null($object) && (!$isSubject || !$propTable->usesIdSubject())) { return array(); } $result = array(); $db = $this->store->getConnection(); $diHandler = $this->store->getDataItemHandlerForDIType($propTable->getDiType()); // *** First build $from, $select, and $where for the DB query ***// $from = $db->tableName($propTable->getName()); // always use actual table $select = ''; $where = ''; if ($isSubject) { // restrict subject, select property $where .= $propTable->usesIdSubject() ? 's_id=' . $db->addQuotes($id) : 's_title=' . $db->addQuotes($object->getDBkey()) . ' AND s_namespace=' . $db->addQuotes($object->getNamespace()); if (!$propTable->isFixedPropertyTable()) { // select property name $from .= ' INNER JOIN ' . $db->tableName(SMWSql3SmwIds::tableName) . ' AS p ON p_id=p.smw_id'; $select .= 'p.smw_title as prop'; } // else: fixed property, no select needed } elseif (!$propTable->isFixedPropertyTable()) { // restrict property only $where .= 'p_id=' . $db->addQuotes($id); } $valuecount = 0; // Don't use DISTINCT for value of one subject: $usedistinct = !$isSubject; $valueField = $diHandler->getIndexField(); $labelField = $diHandler->getLabelField(); $fields = $diHandler->getFetchFields(); foreach ($fields as $fieldname => $typeid) { // select object column(s) if ($typeid == 'p') { // get data from ID table $from .= ' INNER JOIN ' . $db->tableName(SMWSql3SmwIds::tableName) . " AS o{$valuecount} ON {$fieldname}=o{$valuecount}.smw_id"; $select .= ($select !== '' ? ',' : '') . "{$fieldname} AS id{$valuecount}" . ",o{$valuecount}.smw_title AS v{$valuecount}" . ",o{$valuecount}.smw_namespace AS v" . ($valuecount + 1) . ",o{$valuecount}.smw_iw AS v" . ($valuecount + 2) . ",o{$valuecount}.smw_sortkey AS v" . ($valuecount + 3) . ",o{$valuecount}.smw_subobject AS v" . ($valuecount + 4); if ($valueField == $fieldname) { $valueField = "o{$valuecount}.smw_sortkey"; } if ($labelField == $fieldname) { $labelField = "o{$valuecount}.smw_sortkey"; } $valuecount += 4; } else { $select .= ($select !== '' ? ',' : '') . "{$fieldname} AS v{$valuecount}"; } $valuecount += 1; } if (!$isSubject) { // Apply sorting/string matching; only with given property $where .= $this->store->getSQLConditions($requestOptions, $valueField, $labelField, $where !== ''); } else { $valueField = ''; } // *** Now execute the query and read the results ***// $res = $db->select($from, $select, $where, __METHOD__, $usedistinct ? $this->store->getSQLOptions($requestOptions, $valueField) + array('DISTINCT') : $this->store->getSQLOptions($requestOptions, $valueField)); foreach ($res as $row) { $valueHash = ''; if ($isSubject) { // use joined or predefined property name $propertykey = $propTable->isFixedPropertyTable() ? $propTable->getFixedProperty() : $row->prop; $valueHash = $propertykey; } // Use enclosing array only for results with many values: if ($valuecount > 1) { $valuekeys = array(); for ($i = 0; $i < $valuecount; $i += 1) { // read the value fields from the current row $fieldname = "v{$i}"; $valuekeys[] = $row->{$fieldname}; } } else { $valuekeys = $row->v0; } // #Issue 615 // If the iw field contains a redirect marker then remove it if (isset($valuekeys[2]) && $valuekeys[2] === SMW_SQL3_SMWREDIIW) { $valuekeys[2] = ''; } // The valueHash prevents from inserting duplicate entries of the same content $valueHash = $valuecount > 1 ? md5($valueHash . implode('#', $valuekeys)) : md5($valueHash . $valuekeys); // Filter out any accidentally retrieved internal things (interwiki starts with ":"): if ($valuecount < 3 || implode('', $fields) != 'p' || $valuekeys[2] === '' || $valuekeys[2][0] != ':') { if (isset($result[$valueHash])) { $this->store->getLogger()->log(__METHOD__, "found duplicate for {$propertykey} " . (is_array($valuekeys) ? implode(',', $valuekeys) : $valuekeys)); } $result[$valueHash] = $isSubject ? array($propertykey, $valuekeys) : $valuekeys; } } $db->freeResult($res); return $result; }
/** * Using a preprocessed internal query description referenced by $rootid, * compute the proper result instance output for the given query. * @todo The SQL standard requires us to select all fields by which we sort, leading * to wrong results regarding the given limit: the user expects limit to be applied to * the number of distinct pages, but we can use DISTINCT only to whole rows. Thus, if * rows contain sortfields, then pages with multiple values for that field are distinct * and appear multiple times in the result. Filtering duplicates in post processing * would still allow such duplicates to push aside wanted values, leading to less than * "limit" results although there would have been "limit" really distinct results. For * this reason, we select sortfields only for POSTGRES. MySQL is able to perform what * we want here. It would be nice if we could eliminate the bug in POSTGRES as well. * * @param Query $query * @param integer $rootid * * @return QueryResult */ private function getInstanceQueryResult(Query $query, $rootid) { global $wgDBtype; $db = $this->store->getConnection(); $qobj = $this->querySegments[$rootid]; if ($qobj->joinfield === '') { // empty result, no query needed $result = new QueryResult($query->getDescription()->getPrintrequests(), $query, array(), $this->store, false); return $result; } $sql_options = $this->getSQLOptions($query, $rootid); // Selecting those is required in standard SQL (but MySQL does not require it). $sortfields = implode($qobj->sortfields, ','); $res = $db->select($db->tableName($qobj->joinTable) . " AS {$qobj->alias}" . $qobj->from, "DISTINCT {$qobj->alias}.smw_id AS id,{$qobj->alias}.smw_title AS t,{$qobj->alias}.smw_namespace AS ns,{$qobj->alias}.smw_iw AS iw,{$qobj->alias}.smw_subobject AS so,{$qobj->alias}.smw_sortkey AS sortkey" . ($wgDBtype == 'postgres' ? ($sortfields ? ',' : '') . $sortfields : ''), $qobj->where, __METHOD__, $sql_options); $qr = array(); $count = 0; // the number of fetched results ( != number of valid results in array $qr) $missedCount = 0; $dataItemCache = array(); $logToTable = array(); $hasFurtherResults = false; $prs = $query->getDescription()->getPrintrequests(); $diHandler = $this->store->getDataItemHandlerForDIType(DataItem::TYPE_WIKIPAGE); while ($count < $query->getLimit() && ($row = $db->fetchObject($res))) { if ($row->iw === '' || $row->iw[0] != ':') { // Catch exception for non-existing predefined properties that // still registered within non-updated pages (@see bug 48711) try { $dataItem = $diHandler->dataItemFromDBKeys(array($row->t, intval($row->ns), $row->iw, '', $row->so)); } catch (InvalidPredefinedPropertyException $e) { $logToTable[$row->t] = "issue creating a {$row->t} dataitem from a database row"; $this->store->getLogger()->log(__METHOD__, $e->getMessage()); $dataItem = ''; } if ($dataItem instanceof DIWikiPage && !isset($dataItemCache[$dataItem->getHash()])) { $count++; $dataItemCache[$dataItem->getHash()] = true; $qr[] = $dataItem; // These IDs are usually needed for displaying the page (esp. if more property values are displayed): $this->store->smwIds->setCache($row->t, $row->ns, $row->iw, $row->so, $row->id, $row->sortkey); } else { $missedCount++; $logToTable[$row->t] = "skip result for {$row->t} existing cache entry / query " . $query->getHash(); } } else { $missedCount++; $logToTable[$row->t] = "skip result for {$row->t} due to an internal `{$row->iw}` pointer / query " . $query->getHash(); } } if ($db->fetchObject($res)) { $count++; } if ($logToTable !== array()) { foreach ($logToTable as $key => $entry) { $this->store->getLogger()->logToTable('sqlstore-query-execution', 'query performer', $key, $entry); } } if ($count > $query->getLimit() || $count + $missedCount > $query->getLimit()) { $hasFurtherResults = true; } $db->freeResult($res); $result = new QueryResult($prs, $query, $qr, $this->store, $hasFurtherResults); return $result; }