public function testCanConstructSearchTableRebuilder()
{
$connection = $this->getMockBuilder('\\SMW\\MediaWiki\\Database')->disableOriginalConstructor()->getMock();
$this->store->expects($this->atLeastOnce())->method('getConnection')->will($this->returnValue($connection));
$instance = new FulltextSearchTableFactory();
$this->assertInstanceOf('\\SMW\\SQLStore\\QueryEngine\\Fulltext\\SearchTableRebuilder', $instance->newSearchTableRebuilder($this->store));
}
/**
* @see Job::run
*
* @since 2.5
*/
public function run()
{
$fulltextSearchTableFactory = new FulltextSearchTableFactory();
$textByChangeUpdater = $fulltextSearchTableFactory->newTextByChangeUpdater(ApplicationFactory::getInstance()->getStore('\\SMW\\SQLStore\\SQLStore'));
$textByChangeUpdater->pushUpdatesFromJobParameters($this->params);
Hooks::run('SMW::Job::AfterSearchTableUpdateComplete', array($this));
return true;
}
/**
* @see Maintenance::execute
*/
public function execute()
{
if (!defined('SMW_VERSION') || !$GLOBALS['smwgSemanticsEnabled']) {
$this->output("You need to have SMW enabled in order to use this maintenance script!\n\n");
exit;
}
$applicationFactory = ApplicationFactory::getInstance();
$maintenanceFactory = $applicationFactory->newMaintenanceFactory();
$fulltextSearchTableFactory = new FulltextSearchTableFactory();
// Only the SQLStore is supported
$searchTableRebuilder = $fulltextSearchTableFactory->newSearchTableRebuilder($applicationFactory->getStore('\\SMW\\SQLStore\\SQLStore'));
$searchTableRebuilder->reportVerbose($this->hasOption('v'));
$this->reportMessage("\nThe script rebuilds the search index from property tables that\n" . "support a fulltext search. Any change of the index rules (altered\n" . "stopwords, new stemmer etc.) and/or a newly added or altered table\n" . "requires to run this script again to ensure that the index complies\n" . "with the rules set forth by the DB or Sanitizer.\n\n");
$searchTable = $searchTableRebuilder->getSearchTable();
$textSanitizer = $fulltextSearchTableFactory->newTextSanitizer();
foreach ($textSanitizer->getVersions() as $key => $value) {
$this->reportMessage("\r" . sprintf("%-35s%s", "- {$key}", $value) . "\n");
}
$this->reportMessage("\nThe following properties are exempted from the fulltext search index.\n");
$exemptionList = '';
foreach ($searchTable->getPropertyExemptionList() as $prop) {
$exemptionList .= ($exemptionList === '' ? '' : ', ') . $prop;
if (strlen($exemptionList) > 60) {
$this->reportMessage("\n- " . $exemptionList);
$exemptionList = '';
}
}
$this->reportMessage("\n- " . $exemptionList . "\n\n");
$this->reportMessage("The entire index table is going to be purged first and \n" . "it may take a moment before the rebuild is completed due to\n" . "dependencies on table content and selected options.\n");
if (!$this->hasOption('quick')) {
$this->reportMessage("\n" . 'Abort the rebuild with control-c in the next five seconds ... ');
wfCountDown(5);
}
$maintenanceHelper = $maintenanceFactory->newMaintenanceHelper();
$maintenanceHelper->initRuntimeValues();
// Need to instantiate an extra object here since we cannot make this class itself
// into a MessageReporter since the maintenance script does not load the interface in time.
$reporter = MessageReporterFactory::getInstance()->newObservableMessageReporter();
$reporter->registerReporterCallback(array($this, 'reportMessage'));
$searchTableRebuilder->setMessageReporter($reporter);
$result = $searchTableRebuilder->run();
if ($result && $this->hasOption('report-runtime')) {
$this->reportMessage("\n" . $maintenanceHelper->transformRuntimeValuesForOutput() . "\n");
}
if ($this->hasOption('with-maintenance-log')) {
$maintenanceLogger = $maintenanceFactory->newMaintenanceLogger('RebuildFulltextSearchTableLogger');
$maintenanceLogger->log($maintenanceHelper->transformRuntimeValuesForOutput());
}
$maintenanceHelper->reset();
return $result;
}
private function addFulltextSearchCondition($query, $comparator, &$value)
{
// Remove remaining ~ from the search string
$value = str_replace('~', '', $value);
$valueMatchConditionBuilder = $this->fulltextSearchTableFactory->newValueMatchConditionBuilderByType($this->querySegmentListBuilder->getStore());
if (!$valueMatchConditionBuilder->isEnabled() || !$valueMatchConditionBuilder->hasMinTokenLength($value)) {
return false;
}
$query->joinTable = $valueMatchConditionBuilder->getTableName();
$query->joinfield = "{$query->alias}.s_id";
$query->components = array();
$query->where = $valueMatchConditionBuilder->getWhereCondition(new ValueDescription(new DIBlob($value), null, $comparator), $query->alias);
return $query;
}
/**
* Given an Description that is just a conjunction or disjunction of
* ValueDescription objects, create and return a plain WHERE condition
* string for it.
*
* @param $query
* @param ValueDescription $description
* @param DataItemHandler $diHandler for that table
* @param string $operator SQL operator "AND" or "OR"
*/
private function mapValueDescription($query, ValueDescription $description, DataItemHandler $diHandler, $operator)
{
$where = '';
$dataItem = $description->getDataItem();
$db = $this->querySegmentListBuilder->getStore()->getConnection('mw.db.queryengine');
$valueMatchConditionBuilder = $this->fulltextSearchTableFactory->newValueMatchConditionBuilderByType($this->querySegmentListBuilder->getStore());
// TODO Better get the handle from the property type
// Some comparators (e.g. LIKE) could use DI values of
// a different type; we care about the property table, not
// about the value
// Do not support smw_id joined data for now.
$indexField = $diHandler->getIndexField();
//Hack to get to the field used as index
$keys = $diHandler->getWhereConds($dataItem);
$value = $keys[$indexField];
// See if the getSQLCondition method exists and call it if this is the case.
// Invoked by SMAreaValueDescription, SMGeoCoordsValueDescription
if (method_exists($description, 'getSQLCondition')) {
$fields = $diHandler->getTableFields();
$where = $description->getSQLCondition($query->alias, array_keys($fields), $this->querySegmentListBuilder->getStore()->getConnection(DB_SLAVE));
}
if ($where == '' && $valueMatchConditionBuilder->canApplyFulltextSearchMatchCondition($description)) {
$query->joinTable = $valueMatchConditionBuilder->getTableName();
$query->sortIndexField = $valueMatchConditionBuilder->getSortIndexField($query->alias);
$query->components = array();
$where = $valueMatchConditionBuilder->getWhereCondition($description, $query->alias);
} elseif ($where == '') {
$comparator = $this->comparatorMapper->mapComparator($description, $value);
$where = "{$query->alias}.{$indexField}{$comparator}" . $db->addQuotes($value);
}
if ($where !== '') {
if ($query->where && substr($query->where, -1) != '(') {
$query->where .= " {$operator} ";
}
$query->where .= "({$where})";
}
}
private function registerHooksForInternalUse(ApplicationFactory $applicationFactory, DeferredRequestDispatchManager $deferredRequestDispatchManager)
{
$queryDependencyLinksStoreFactory = new QueryDependencyLinksStoreFactory();
$queryDependencyLinksStore = $queryDependencyLinksStoreFactory->newQueryDependencyLinksStore($applicationFactory->getStore());
$transitionalDiffStore = $applicationFactory->singleton('TransitionalDiffStore');
/**
* @see https://www.semantic-mediawiki.org/wiki/Hooks#SMW::SQLStore::AfterDataUpdateComplete
*/
$this->handlers['SMW::SQLStore::AfterDataUpdateComplete'] = function ($store, $semanticData, $compositePropertyTableDiffIterator) use($queryDependencyLinksStoreFactory, $queryDependencyLinksStore, $deferredRequestDispatchManager, $transitionalDiffStore) {
// When in commandLine mode avoid deferred execution and run a process
// within the same transaction
$deferredRequestDispatchManager->isCommandLineMode($store->getOptions()->has('isCommandLineMode') ? $store->getOptions()->get('isCommandLineMode') : $GLOBALS['wgCommandLineMode']);
$queryDependencyLinksStore->setStore($store);
$subject = $semanticData->getSubject();
$slot = $transitionalDiffStore->createSlotFrom($compositePropertyTableDiffIterator);
$queryDependencyLinksStore->pruneOutdatedTargetLinks($subject, $compositePropertyTableDiffIterator);
$entityIdListRelevanceDetectionFilter = $queryDependencyLinksStoreFactory->newEntityIdListRelevanceDetectionFilter($store, $compositePropertyTableDiffIterator);
$jobParameters = $queryDependencyLinksStore->buildParserCachePurgeJobParametersFrom($entityIdListRelevanceDetectionFilter);
$deferredRequestDispatchManager->scheduleParserCachePurgeJobWith($subject->getTitle(), $jobParameters);
$fulltextSearchTableFactory = new FulltextSearchTableFactory();
$textByChangeUpdater = $fulltextSearchTableFactory->newTextByChangeUpdater($store);
$textByChangeUpdater->pushUpdates($compositePropertyTableDiffIterator, $deferredRequestDispatchManager, $slot);
// Since we cannot predict as to when the slot is used and by whom,
// schedule a job to ensure to be the last in-line to clean-up
// any remaining slots for this transaction
$deferredRequestDispatchManager->scheduleChronologyPurgeJobWith($subject->getTitle(), array('slot:id' => $slot));
return true;
};
/**
* @see https://www.semantic-mediawiki.org/wiki/Hooks#SMW::Store::AfterQueryResultLookupComplete
*/
$this->handlers['SMW::Store::AfterQueryResultLookupComplete'] = function ($store, &$result) use($queryDependencyLinksStore, $applicationFactory) {
$queryDependencyLinksStore->setStore($store);
$queryDependencyLinksStore->doUpdateDependenciesFrom($result);
$applicationFactory->singleton('CachedQueryResultPrefetcher')->recordStats();
return true;
};
}