/**
* Run the ApplyOperation job
* @throws \Exception
*/
public function perform()
{
try {
$this->debugLog("[JOBID " . $this->job->payload['id'] . "] ApplyOperation::perform() start");
$timer = new \Tripod\Timer();
$timer->start();
$this->validateArgs();
$statsConfig = array();
if (isset($this->args['statsConfig'])) {
$statsConfig['statsConfig'] = $this->args['statsConfig'];
}
// set the config to what is received
\Tripod\Mongo\Config::setConfig($this->args[self::TRIPOD_CONFIG_KEY]);
$this->getStat()->increment(MONGO_QUEUE_APPLY_OPERATION_JOB . '.' . SUBJECT_COUNT, count($this->args[self::SUBJECTS_KEY]));
foreach ($this->args[self::SUBJECTS_KEY] as $subject) {
$opTimer = new \Tripod\Timer();
$opTimer->start();
$impactedSubject = $this->createImpactedSubject($subject);
$impactedSubject->update();
$opTimer->stop();
// stat time taken to perform operation for the given subject
$this->getStat()->timer(MONGO_QUEUE_APPLY_OPERATION . '.' . $subject['operation'], $opTimer->result());
}
$timer->stop();
// stat time taken to process job, from time it was picked up
$this->getStat()->timer(MONGO_QUEUE_APPLY_OPERATION_SUCCESS, $timer->result());
$this->debugLog("[JOBID " . $this->job->payload['id'] . "] ApplyOperation::perform() done in {$timer->result()}ms");
} catch (\Exception $e) {
$this->getStat()->increment(MONGO_QUEUE_APPLY_OPERATION_FAIL);
$this->errorLog("Caught exception in " . get_class($this) . ": " . $e->getMessage());
throw $e;
}
}
$tableSpec = \Tripod\Mongo\Config::getInstance()->getTableSpecification($storeName, $tableId);
if (array_key_exists("from", $tableSpec)) {
\Tripod\Mongo\Config::getInstance()->setMongoCursorTimeout(-1);
print "Generating {$tableId}";
$tripod = new \Tripod\Mongo\Driver($tableSpec['from'], $storeName, array('stat' => $stat));
$tTables = $tripod->getTripodTables();
if ($id) {
print " for {$id}....\n";
$tTables->generateTableRows($tableId, $id);
} else {
print " for all tables....\n";
$tTables->generateTableRows($tableId, null, null, $queue);
}
}
}
$t = new \Tripod\Timer();
$t->start();
\Tripod\Mongo\Config::setConfig(json_decode(file_get_contents($configLocation), true));
if (isset($options['s']) || isset($options['storename'])) {
$storeName = isset($options['s']) ? $options['s'] : $options['storename'];
} else {
$storeName = null;
}
if (isset($options['t']) || isset($options['spec'])) {
$tableId = isset($options['t']) ? $options['t'] : $options['spec'];
} else {
$tableId = null;
}
if (isset($options['i']) || isset($options['id'])) {
$id = isset($options['i']) ? $options['i'] : $options['id'];
} else {
/**
* @param string $tableType
* @param string|null $resource
* @param string|null $context
* @param string|null $queueName Queue for background bulk generation
* @return null //@todo: this should be a bool
*/
public function generateTableRows($tableType, $resource = null, $context = null, $queueName = null)
{
$t = new \Tripod\Timer();
$t->start();
$this->temporaryFields = array();
$tableSpec = Config::getInstance()->getTableSpecification($this->storeName, $tableType);
$collection = $this->config->getCollectionForTable($this->storeName, $tableType);
if ($tableSpec == null) {
$this->debugLog("Could not find a table specification for {$tableType}");
return null;
}
// ensure that the ID field, view type, and the impactIndex indexes are correctly set up
$collection->createIndex(array('_id.r' => 1, '_id.c' => 1, '_id.type' => 1), array('background' => 1));
$collection->createIndex(array('_id.type' => 1), array('background' => 1));
$collection->createIndex(array('value.' . _IMPACT_INDEX => 1), array('background' => 1));
// ensure any custom view indexes
foreach (Config::getInstance()->getTableSpecifications($this->storeName) as $tSpec) {
if (isset($tSpec['ensureIndexes']) && $tSpec['to_data_source'] == $tableSpec['to_data_source']) {
foreach ($tSpec['ensureIndexes'] as $ensureIndex) {
$this->ensureIndex($collection, $ensureIndex);
}
}
}
// default the context
$contextAlias = $this->getContextAlias($context);
// default collection
$from = isset($tableSpec["from"]) ? $tableSpec["from"] : $this->podName;
$types = array();
if (is_array($tableSpec["type"])) {
foreach ($tableSpec["type"] as $type) {
$types[] = array("rdf:type.u" => $this->labeller->qname_to_alias($type));
$types[] = array("rdf:type.u" => $this->labeller->uri_to_alias($type));
}
} else {
$types[] = array("rdf:type.u" => $this->labeller->qname_to_alias($tableSpec["type"]));
$types[] = array("rdf:type.u" => $this->labeller->uri_to_alias($tableSpec["type"]));
}
$filter = array('$or' => $types);
if (isset($resource)) {
$filter["_id"] = array(_ID_RESOURCE => $this->labeller->uri_to_alias($resource), _ID_CONTEXT => $contextAlias);
}
$docs = $this->config->getCollectionForCBD($this->storeName, $from)->find($filter, array('maxTimeMS' => 1000000));
foreach ($docs as $doc) {
if ($queueName && !$resource) {
$subject = new ImpactedSubject($doc['_id'], OP_TABLES, $this->storeName, $from, array($tableType));
$jobOptions = array();
if ($this->stat || !empty($this->statsConfig)) {
$jobOptions['statsConfig'] = $this->getStatsConfig();
}
$this->getApplyOperation()->createJob(array($subject), $queueName, $jobOptions);
} else {
// set up ID
$generatedRow = array("_id" => array(_ID_RESOURCE => $doc["_id"][_ID_RESOURCE], _ID_CONTEXT => $doc["_id"][_ID_CONTEXT], _ID_TYPE => $tableSpec['_id']));
$value = array('_id' => $doc['_id']);
// everything must go in the value object todo: this is a hang over from map reduce days, engineer out once we have stability on new PHP method for M/R
$this->addIdToImpactIndex($doc['_id'], $value);
// need to add the doc to the impact index to be consistent with views/search etc. this is needed for discovering impacted operations
$this->addFields($doc, $tableSpec, $value);
if (isset($tableSpec['joins'])) {
$this->doJoins($doc, $tableSpec['joins'], $value, $from, $contextAlias);
}
if (isset($tableSpec['counts'])) {
$this->doCounts($doc, $tableSpec['counts'], $value);
}
if (isset($tableSpec['computed_fields'])) {
$this->doComputedFields($tableSpec, $value);
}
// Remove temp fields from document
$generatedRow['value'] = array_diff_key($value, array_flip($this->temporaryFields));
$this->truncatingSave($collection, $generatedRow);
}
}
$t->stop();
$this->timingLog(MONGO_CREATE_TABLE, array('type' => $tableSpec['type'], 'duration' => $t->result(), 'filter' => $filter, 'from' => $from));
$this->getStat()->timer(MONGO_CREATE_TABLE . ".{$tableType}", $t->result());
}
* @param string $subject
* @param array $triples
* @param array $errors
* @param string $podName
* @param string $storeName
*/
function load(\Tripod\Mongo\TriplesUtil $loader, $subject, array $triples, array &$errors, $podName, $storeName)
{
try {
$loader->loadTriplesAbout($subject, $triples, $storeName, $podName);
} catch (Exception $e) {
print "Exception for subject {$subject} failed with message: " . $e->getMessage() . "\n";
$errors[] = $subject;
}
}
$timer = new \Tripod\Timer();
$timer->start();
if ($argc != 4) {
echo "usage: ./loadTriples.php storename podname tripodConfig.json < ntriplesdata\n";
die;
}
array_shift($argv);
$storeName = $argv[0];
$podName = $argv[1];
\Tripod\Mongo\Config::setConfig(json_decode(file_get_contents($argv[2]), true));
$i = 0;
$currentSubject = "";
$triples = array();
$errors = array();
// array of subjects that failed to insert, even after retry...
$loader = new \Tripod\Mongo\TriplesUtil();
$viewSpec = \Tripod\Mongo\Config::getInstance()->getViewSpecification($storeName, $viewId);
if (array_key_exists("from", $viewSpec)) {
\Tripod\Mongo\Config::getInstance()->setMongoCursorTimeout(-1);
print "Generating {$viewId}";
$tripod = new \Tripod\Mongo\Driver($viewSpec['from'], $storeName, array('stat' => $stat));
$views = $tripod->getTripodViews();
if ($id) {
print " for {$id}....\n";
$views->generateView($viewId, $id, null, $queue);
} else {
print " for all views....\n";
$views->generateView($viewId, null, null, $queue);
}
}
}
$t = new \Tripod\Timer();
$t->start();
\Tripod\Mongo\Config::setConfig(json_decode(file_get_contents($configLocation), true));
if (isset($options['s']) || isset($options['storename'])) {
$storeName = isset($options['s']) ? $options['s'] : $options['storename'];
} else {
$storeName = null;
}
if (isset($options['v']) || isset($options['spec'])) {
$viewId = isset($options['v']) ? $options['v'] : $options['spec'];
} else {
$viewId = null;
}
if (isset($options['i']) || isset($options['id'])) {
$id = isset($options['i']) ? $options['i'] : $options['id'];
} else {
/**
* Run the DiscoverImpactedSubjects job
* @throws \Exception
*/
public function perform()
{
try {
$this->debugLog("[JOBID " . $this->job->payload['id'] . "] DiscoverImpactedSubjects::perform() start");
$timer = new \Tripod\Timer();
$timer->start();
$this->validateArgs();
// set the config to what is received
\Tripod\Mongo\Config::setConfig($this->args[self::TRIPOD_CONFIG_KEY]);
$statsConfig = array();
if (isset($this->args['statsConfig'])) {
$statsConfig['statsConfig'] = $this->args['statsConfig'];
}
$tripod = $this->getTripod($this->args[self::STORE_NAME_KEY], $this->args[self::POD_NAME_KEY], $statsConfig);
$operations = $this->args[self::OPERATIONS_KEY];
$subjectsAndPredicatesOfChange = $this->args[self::CHANGES_KEY];
$subjectCount = 0;
foreach ($operations as $op) {
/** @var \Tripod\Mongo\Composites\IComposite $composite */
$composite = $tripod->getComposite($op);
$modifiedSubjects = $composite->getImpactedSubjects($subjectsAndPredicatesOfChange, $this->args[self::CONTEXT_ALIAS_KEY]);
if (!empty($modifiedSubjects)) {
/* @var $subject \Tripod\Mongo\ImpactedSubject */
foreach ($modifiedSubjects as $subject) {
$subjectCount++;
$subjectTimer = new \Tripod\Timer();
$subjectTimer->start();
if (isset($this->args[self::QUEUE_KEY]) || count($subject->getSpecTypes()) == 0) {
$queueName = isset($this->args[self::QUEUE_KEY]) ? $this->args[self::QUEUE_KEY] : Config::getApplyQueueName();
$this->addSubjectToQueue($subject, $queueName);
} else {
$specsGroupedByQueue = array();
foreach ($subject->getSpecTypes() as $specType) {
$spec = null;
switch ($subject->getOperation()) {
case OP_VIEWS:
$spec = Config::getInstance()->getViewSpecification($this->args[self::STORE_NAME_KEY], $specType);
break;
case OP_TABLES:
$spec = Config::getInstance()->getTableSpecification($this->args[self::STORE_NAME_KEY], $specType);
break;
case OP_SEARCH:
$spec = Config::getInstance()->getSearchDocumentSpecification($this->args[self::STORE_NAME_KEY], $specType);
break;
}
if (!$spec || !isset($spec['queue'])) {
if (!$spec) {
$spec = array();
}
$spec['queue'] = Config::getApplyQueueName();
}
if (!isset($specsGroupedByQueue[$spec['queue']])) {
$specsGroupedByQueue[$spec['queue']] = array();
}
$specsGroupedByQueue[$spec['queue']][] = $specType;
}
foreach ($specsGroupedByQueue as $queueName => $specs) {
$queuedSubject = new \Tripod\Mongo\ImpactedSubject($subject->getResourceId(), $subject->getOperation(), $subject->getStoreName(), $subject->getPodName(), $specs);
$this->addSubjectToQueue($queuedSubject, $queueName);
}
}
$subjectTimer->stop();
// stat time taken to discover impacted subjects for the given subject of change
$this->getStat()->timer(MONGO_QUEUE_DISCOVER_SUBJECT, $subjectTimer->result());
}
if (!empty($this->subjectsGroupedByQueue)) {
foreach ($this->subjectsGroupedByQueue as $queueName => $subjects) {
$this->getApplyOperation()->createJob($subjects, $queueName, $statsConfig);
}
$this->subjectsGroupedByQueue = array();
}
}
}
// stat time taken to process item, from time it was created (queued)
$timer->stop();
$this->getStat()->timer(MONGO_QUEUE_DISCOVER_SUCCESS, $timer->result());
$this->debugLog("[JOBID " . $this->job->payload['id'] . "] DiscoverImpactedSubjects::perform() done in {$timer->result()}ms");
$this->getStat()->increment(MONGO_QUEUE_DISCOVER_JOB . '.' . SUBJECT_COUNT, $subjectCount);
} catch (\Exception $e) {
$this->getStat()->increment(MONGO_QUEUE_DISCOVER_FAIL);
$this->errorLog("Caught exception in " . get_class($this) . ": " . $e->getMessage());
throw $e;
}
}
/**
* Given a specific $viewId, generates a single view for the $resource
* @param string $viewId
* @param string|null $resource
* @param string|null $context
* @param string|null $queueName Queue for background bulk generation
* @throws \Tripod\Exceptions\ViewException
* @return array
*/
public function generateView($viewId, $resource = null, $context = null, $queueName = null)
{
$contextAlias = $this->getContextAlias($context);
$viewSpec = Config::getInstance()->getViewSpecification($this->storeName, $viewId);
if ($viewSpec == null) {
$this->debugLog("Could not find a view specification for {$resource} with viewId '{$viewId}'");
return null;
} else {
$t = new \Tripod\Timer();
$t->start();
$from = $this->getFromCollectionForViewSpec($viewSpec);
$collection = $this->config->getCollectionForView($this->storeName, $viewId);
if (!isset($viewSpec['joins'])) {
throw new \Tripod\Exceptions\ViewException('Could not find any joins in view specification - usecase better served with select()');
}
// ensure that the ID field, view type, and the impactIndex indexes are correctly set up
$collection->createIndex(array('_id.r' => 1, '_id.c' => 1, '_id.type' => 1), array('background' => 1));
$collection->createIndex(array('_id.type' => 1), array('background' => 1));
$collection->createIndex(array('value.' . _IMPACT_INDEX => 1), array('background' => 1));
// ensure any custom view indexes
if (isset($viewSpec['ensureIndexes'])) {
foreach ($viewSpec['ensureIndexes'] as $ensureIndex) {
$collection->createIndex($ensureIndex, array('background' => 1));
}
}
$types = array();
// this is used to filter the CBD table to speed up the view creation
if (is_array($viewSpec["type"])) {
foreach ($viewSpec["type"] as $type) {
$types[] = array("rdf:type.u" => $this->labeller->qname_to_alias($type));
$types[] = array("rdf:type.u" => $this->labeller->uri_to_alias($type));
}
} else {
$types[] = array("rdf:type.u" => $this->labeller->qname_to_alias($viewSpec["type"]));
$types[] = array("rdf:type.u" => $this->labeller->uri_to_alias($viewSpec["type"]));
}
$filter = array('$or' => $types);
if (isset($resource)) {
$resourceAlias = $this->labeller->uri_to_alias($resource);
$filter["_id"] = array(_ID_RESOURCE => $resourceAlias, _ID_CONTEXT => $contextAlias);
}
$docs = $this->config->getCollectionForCBD($this->storeName, $from)->find($filter, array('maxTimeMS' => \Tripod\Mongo\Config::getInstance()->getMongoCursorTimeout()));
foreach ($docs as $doc) {
if ($queueName && !$resource) {
$subject = new ImpactedSubject($doc['_id'], OP_VIEWS, $this->storeName, $from, array($viewId));
$jobOptions = array();
if ($this->stat || !empty($this->statsConfig)) {
$jobOptions['statsConfig'] = $this->getStatsConfig();
}
$this->getApplyOperation()->createJob(array($subject), $queueName, $jobOptions);
} else {
// set up ID
$id = array("_id" => array(_ID_RESOURCE => $doc["_id"][_ID_RESOURCE], _ID_CONTEXT => $doc["_id"][_ID_CONTEXT], _ID_TYPE => $viewSpec['_id']));
$generatedView = $id;
$value = array();
// everything must go in the value object todo: this is a hang over from map reduce days, engineer out once we have stability on new PHP method for M/R
$value[_GRAPHS] = array();
$buildImpactIndex = true;
if (isset($viewSpec['ttl'])) {
$buildImpactIndex = false;
$value[_EXPIRES] = \Tripod\Mongo\DateUtil::getMongoDate($this->getExpirySecFromNow($viewSpec['ttl']) * 1000);
} else {
$value[_IMPACT_INDEX] = array($doc['_id']);
}
$this->doJoins($doc, $viewSpec['joins'], $value, $from, $contextAlias, $buildImpactIndex);
// add top level properties
$value[_GRAPHS][] = $this->extractProperties($doc, $viewSpec, $from);
$generatedView['value'] = $value;
$collection->replaceOne($id, $generatedView, ['upsert' => true]);
}
}
$t->stop();
$this->timingLog(MONGO_CREATE_VIEW, array('view' => $viewSpec['type'], 'duration' => $t->result(), 'filter' => $filter, 'from' => $from));
$this->getStat()->timer(MONGO_CREATE_VIEW . ".{$viewId}", $t->result());
}
}
$spec = \Tripod\Mongo\Config::getInstance()->getSearchDocumentSpecification($storeName, $specId);
if (array_key_exists("from", $spec)) {
\Tripod\Mongo\Config::getInstance()->setMongoCursorTimeout(-1);
print "Generating {$specId}";
$tripod = new \Tripod\Mongo\Driver($spec['from'], $storeName, array('stat' => $stat));
$search = $tripod->getSearchIndexer();
if ($id) {
print " for {$id}....\n";
$search->generateSearchDocuments($specId, $id, null, $queue);
} else {
print " for all tables....\n";
$search->generateSearchDocuments($specId, null, null, $queue);
}
}
}
$t = new \Tripod\Timer();
$t->start();
\Tripod\Mongo\Config::setConfig(json_decode(file_get_contents($configLocation), true));
if (isset($options['s']) || isset($options['storename'])) {
$storeName = isset($options['s']) ? $options['s'] : $options['storename'];
} else {
$storeName = null;
}
if (isset($options['d']) || isset($options['spec'])) {
$specId = isset($options['d']) ? $options['t'] : $options['spec'];
} else {
$specId = null;
}
if (isset($options['i']) || isset($options['id'])) {
$id = isset($options['i']) ? $options['i'] : $options['id'];
} else {
<?php
require_once dirname(__FILE__) . '/common.inc.php';
require_once dirname(dirname(dirname(__FILE__))) . '/src/tripod.inc.php';
if ($argc != 2 && $argc != 3 && $argc != 4) {
echo "usage: php ensureIndexes.php tripodConfig.json [storeName] [forceReindex (default is false)]\n";
die;
}
array_shift($argv);
\Tripod\Mongo\Config::setConfig(json_decode(file_get_contents($argv[0]), true));
$storeName = isset($argv[1]) ? $argv[1] : null;
$forceReindex = isset($argv[2]) && $argv[2] == "true" ? true : false;
\Tripod\Mongo\Config::getInstance()->setMongoCursorTimeout(-1);
$ei = new \Tripod\Mongo\IndexUtils();
$t = new \Tripod\Timer();
$t->start();
print "About to start indexing on {$storeName}...\n";
$ei->ensureIndexes($forceReindex, $storeName);
$t->stop();
print "Indexing complete, took {$t->result()} seconds\n";
/**
* @param array $query
* @param string $type
* @param Collection|null $collection
* @param array $includeProperties
* @param int $cursorSize
* @return MongoGraph
*/
protected function fetchGraph($query, $type, $collection = null, $includeProperties = array(), $cursorSize = 101)
{
$graph = new MongoGraph();
$t = new \Tripod\Timer();
$t->start();
if ($collection == null) {
$collection = $this->collection;
$collectionName = $collection->getCollectionName();
} else {
$collectionName = $collection->getCollectionName();
}
if (empty($includeProperties)) {
$cursor = $collection->find($query);
} else {
$fields = array();
foreach ($includeProperties as $property) {
$fields[$this->labeller->uri_to_alias($property)] = true;
}
$cursor = $collection->find($query, array('projection' => $fields, 'batchSize' => $cursorSize));
}
$ttlExpiredResources = false;
$retries = 1;
$exception = null;
$cursorSuccess = false;
do {
try {
foreach ($cursor as $result) {
// handle MONGO_VIEWS that have expired due to ttl. These are expired
// on read (lazily) rather than on write
if ($type == MONGO_VIEW && array_key_exists(_EXPIRES, $result['value'])) {
// if expires < current date, regenerate view..
$currentDate = \Tripod\Mongo\DateUtil::getMongoDate();
if ($result['value'][_EXPIRES]->__toString() < $currentDate) {
// regenerate!
$this->generateView($result['_id']['type'], $result['_id']['r']);
}
}
$graph->add_tripod_array($result);
}
$cursorSuccess = true;
} catch (\Exception $e) {
self::getLogger()->error("CursorException attempt " . $retries . ". Retrying...:" . $e->getMessage());
sleep(1);
$retries++;
$exception = $e;
}
} while ($retries <= Config::CONNECTION_RETRIES && $cursorSuccess === false);
if ($cursorSuccess === false) {
self::getLogger()->error("CursorException failed after " . $retries . " attempts (MAX:" . Config::CONNECTION_RETRIES . "): " . $e->getMessage());
throw new \Exception($exception);
}
if ($ttlExpiredResources) {
// generate views and retry...
$this->debugLog("One or more view had exceeded TTL was regenerated - request again...");
$graph = $this->fetchGraph($query, $type, $collection);
}
$t->stop();
$this->timingLog($type, array('duration' => $t->result(), 'query' => $query, 'collection' => $collectionName));
if ($type == MONGO_VIEW) {
if (array_key_exists("_id.type", $query)) {
$this->getStat()->timer("{$type}.{$query["_id.type"]}", $t->result());
} else {
if (array_key_exists("_id", $query) && array_key_exists("type", $query["_id"])) {
$this->getStat()->timer("{$type}.{$query["_id"]["type"]}", $t->result());
}
}
} else {
$this->getStat()->timer("{$type}.{$collectionName}", $t->result());
}
return $graph;
}
/**
* Selects $fields from the result set determined by $query.
* Returns an array of all results, each array element is a CBD graph, keyed by r
* @param array $query
* @param array $fields array of fields, in the same format as prescribed by MongoPHP
* @param null $sortBy
* @param null $limit
* @param int $offset
* @param null $context
* @return array MongoGraphs, keyed by subject
*/
public function select($query, $fields, $sortBy = null, $limit = null, $offset = 0, $context = null)
{
$t = new \Tripod\Timer();
$t->start();
$contextAlias = $this->getContextAlias($context);
// make sure context is represented - but not at the expense of $ operands queries failing
if (array_key_exists('_id', $query) && is_array($query["_id"])) {
if (!array_key_exists(_ID_CONTEXT, $query['_id']) && array_key_exists(_ID_RESOURCE, $query['_id'])) {
// add context
$query["_id"][_ID_CONTEXT] = $contextAlias;
} else {
// check query does not have a $ operand
foreach ($query["_id"] as $key => $queryProps) {
if (substr($key, 0, 1) == '$' && is_array($queryProps)) {
foreach ($queryProps as $index => $queryProp) {
if (is_array($queryProp) && array_key_exists(_ID_RESOURCE, $queryProp)) {
$queryProp[_ID_CONTEXT] = $contextAlias;
$query["_id"][$key][$index] = $queryProp;
}
}
}
}
}
} else {
if (!array_key_exists('_id', $query)) {
// this query did not have _id referenced at all - just add an _id.c clause
$query["_id." . _ID_CONTEXT] = $contextAlias;
}
}
$findOptions = array('projection' => $fields);
if (!empty($limit)) {
$findOptions['skip'] = (int) $offset;
$findOptions['limit'] = (int) $limit;
}
if (isset($sortBy)) {
$findOptions['sort'] = $sortBy;
}
$results = $this->collection->find($query, $findOptions);
$t->stop();
$this->timingLog(MONGO_SELECT, array('duration' => $t->result(), 'query' => $query));
$this->getStat()->timer(MONGO_SELECT . ".{$this->podName}", $t->result());
$rows = array();
$count = $this->collection->count($query);
foreach ($results as $doc) {
$row = array();
foreach ($doc as $key => $value) {
if ($key == _ID_KEY || $key === _VERSION) {
$row[$key] = $value;
} else {
if (array_key_exists(VALUE_LITERAL, $value)) {
$row[$key] = $value[VALUE_LITERAL];
} else {
if (array_key_exists(VALUE_URI, $value)) {
$row[$key] = $value[VALUE_URI];
} else {
$row[$key] = array();
// possible array of values
foreach ($value as $v) {
$row[$key][] = array_key_exists(VALUE_LITERAL, $v) ? $v[VALUE_LITERAL] : $v[VALUE_URI];
}
}
}
}
}
$rows[] = $row;
}
$result = array("head" => array("count" => $count, "offset" => $offset, "limit" => $limit), "results" => $rows);
return $result;
}
/**
* @param array $rdfTypes
* @param string $resource
* @param string $context
* @return array
* @throws \Exception
*/
public function generateSearchDocumentsBasedOnRdfTypes(array $rdfTypes, $resource, $context)
{
if (empty($resource)) {
throw new \Exception("Resource must be specified");
}
if (empty($context)) {
throw new \Exception("Context must be specified");
}
// this is what is returned
$generatedSearchDocuments = array();
$timer = new \Tripod\Timer();
$timer->start();
foreach ($rdfTypes as $rdfType) {
$specs = Config::getInstance()->getSearchDocumentSpecifications($this->storeName, $rdfType);
if (empty($specs)) {
continue;
}
// no point doing anything else if there is no spec for the type
foreach ($specs as $searchSpec) {
$generatedSearchDocuments[] = $this->generateSearchDocumentBasedOnSpecId($searchSpec['_id'], $resource, $context);
}
}
$timer->stop();
//echo "\n\tTook " . $timer->result() . " ms to generate search documents\n";
return $generatedSearchDocuments;
}
/**
* @param string $searchDocumentType
* @param string|null $resourceUri
* @param string|null $context
* @param string|null $queueName
*/
public function generateSearchDocuments($searchDocumentType, $resourceUri = null, $context = null, $queueName = null)
{
$t = new \Tripod\Timer();
$t->start();
// default the context
$contextAlias = $this->getContextAlias($context);
$spec = \Tripod\Mongo\Config::getInstance()->getSearchDocumentSpecification($this->getStoreName(), $searchDocumentType);
if ($resourceUri) {
$this->generateAndIndexSearchDocuments($resourceUri, $contextAlias, $spec['from'], $searchDocumentType);
return;
}
// default collection
$from = isset($spec["from"]) ? $spec["from"] : $this->podName;
$types = array();
if (is_array($spec["type"])) {
foreach ($spec["type"] as $type) {
$types[] = array("rdf:type.u" => $this->labeller->qname_to_alias($type));
$types[] = array("rdf:type.u" => $this->labeller->uri_to_alias($type));
}
} else {
$types[] = array("rdf:type.u" => $this->labeller->qname_to_alias($spec["type"]));
$types[] = array("rdf:type.u" => $this->labeller->uri_to_alias($spec["type"]));
}
$filter = array('$or' => $types);
if (isset($resource)) {
$filter["_id"] = array(_ID_RESOURCE => $this->labeller->uri_to_alias($resource), _ID_CONTEXT => $contextAlias);
}
$docs = $this->config->getCollectionForCBD($this->getStoreName(), $from)->find($filter, array('maxTimeMS' => $this->config->getMongoCursorTimeout()));
foreach ($docs as $doc) {
if ($queueName && !$resourceUri) {
$subject = new ImpactedSubject($doc['_id'], OP_SEARCH, $this->storeName, $from, array($searchDocumentType));
$jobOptions = array();
if ($this->stat || !empty($this->statsConfig)) {
$jobOptions['statsConfig'] = $this->getStatsConfig();
}
$this->getApplyOperation()->createJob(array($subject), $queueName, $jobOptions);
} else {
$this->generateAndIndexSearchDocuments($doc[_ID_KEY][_ID_RESOURCE], $doc[_ID_KEY][_ID_CONTEXT], $spec['from'], $searchDocumentType);
}
}
$t->stop();
$this->timingLog(MONGO_CREATE_TABLE, array('type' => $spec['type'], 'duration' => $t->result(), 'filter' => $filter, 'from' => $from));
$this->getStat()->timer(MONGO_CREATE_SEARCH_DOC . ".{$searchDocumentType}", $t->result());
}
/**
* Processes each subject synchronously
* @param array $subjectsAndPredicatesOfChange
* @param string $contextAlias
*/
protected function processSyncOperations(array $subjectsAndPredicatesOfChange, $contextAlias)
{
foreach ($this->getSyncOperations() as $op) {
/** @var \Tripod\Mongo\Composites\IComposite $composite */
$composite = $this->tripod->getComposite($op);
$opSubjects = $composite->getImpactedSubjects($subjectsAndPredicatesOfChange, $contextAlias);
if (!empty($opSubjects)) {
foreach ($opSubjects as $subject) {
/* @var $subject ImpactedSubject */
$t = new \Tripod\Timer();
$t->start();
// Call update on the subject, rather than the composite directly, in case the change was to
// another pod
$subject->update($subject);
$t->stop();
$this->timingLog(MONGO_ON_THE_FLY_MR, array("duration" => $t->result(), "storeName" => $subject->getStoreName(), "podName" => $subject->getPodName(), "resourceId" => $subject->getResourceId()));
$this->getStat()->timer(MONGO_ON_THE_FLY_MR, $t->result());
}
}
}
}
/**
* @param string $q
* @param string $type
* @param array $indices
* @param array $fields
* @param int $limit
* @param int $offset
* @return array|mixed
* @throws \Tripod\Exceptions\SearchException
*/
public function search($q, $type, $indices = array(), $fields = array(), $limit = 10, $offset = 0)
{
if (empty($q)) {
throw new \Tripod\Exceptions\SearchException("You must specify a query");
}
if (empty($type)) {
throw new \Tripod\Exceptions\SearchException("You must specify the search document type to restrict the query to");
}
if (empty($indices)) {
throw new \Tripod\Exceptions\SearchException("You must specify at least one index from the search document specification to query against");
}
if (empty($fields)) {
throw new \Tripod\Exceptions\SearchException("You must specify at least one field from the search document specification to return");
}
if (!is_numeric($limit) || $limit < 0) {
throw new \Tripod\Exceptions\SearchException("Value for limit must be a positive number");
}
if (!is_numeric($offset) || $offset < 0) {
throw new \Tripod\Exceptions\SearchException("Value for offset must be a positive number");
}
$original_terms = explode(" ", trim(strtolower($q)));
$terms = array_values(array_diff($original_terms, $this->stopWords));
// todo: this means if all the words entered were stop words, then use the orginal terms rather than do nothing!
if (empty($terms)) {
$terms = $original_terms;
}
$regexes = array();
foreach ($terms as $t) {
$regexes[] = new Regex("{$t}", '');
}
$query = array();
$query['_id.type'] = $type;
if (count($indices) == 1) {
$searchIndex = $indices[0];
$query[$searchIndex] = array('$all' => $regexes);
} else {
$query['$or'] = array();
foreach ($indices as $searchIndex) {
$query['$or'][] = array("{$searchIndex}" => array('$all' => $regexes));
}
}
$fieldsToReturn = array();
foreach ($fields as $field) {
$fieldsToReturn[$field] = 1;
}
$searchTimer = new \Tripod\Timer();
$searchTimer->start();
$cursor = $this->config->getCollectionForSearchDocument($this->storeName, $type)->find($query, array('projection' => $fieldsToReturn, 'limit' => $limit, 'skip' => $offset));
$searchResults = array();
$searchResults['head'] = array();
$searchResults['head']['count'] = "";
$searchResults['head']['limit'] = $limit;
$searchResults['head']['offset'] = $offset;
$searchResults['head']['duration'] = "";
$searchResults['head']['query'] = $q;
$searchResults['head']['query_terms_used'] = $terms;
$searchResults['results'] = array();
$count = $this->config->getCollectionForSearchDocument($this->storeName, $type)->count($query);
if ($count > 0) {
$searchResults['head']['count'] = $count;
foreach ($cursor as $result) {
// if more than one field has been asked for we need to
// enumerate them in the results returned. However if only one has been
// asked for then results is just set to that single fields value.
if (count($fields) > 1) {
$r = array();
foreach ($fields as $field) {
if (isset($result[$field])) {
$r[$field] = $result[$field];
} else {
$r[$field] = "";
}
}
$searchResults['results'][] = $r;
} else {
$searchResults['results'][] = $result[$fields[0]];
}
}
} else {
$searchResults['head']['count'] = 0;
}
$searchTimer->stop();
$searchResults['head']["duration"] = $searchTimer->result() . ' ms';
return $searchResults;
}
