/**
* (Re-) indexes an object to the ElasticSearch index, no matter if the change is actually required.
*
* @param object $object
* @param string $signalInformation Signal information, if called from a signal
* @param \Flowpack\ElasticSearch\Domain\Model\Client $client
*
* @return void
*/
public function indexObject($object, $signalInformation = null, Client $client = null)
{
$type = $this->getIndexTypeForObject($object, $client);
if ($type === null) {
return null;
}
$data = $this->getIndexablePropertiesAndValuesFromObject($object);
$id = $this->persistenceManager->getIdentifierByObject($object);
$document = new Document($type, $data, $id);
$document->store();
}
/**
* index this node, and add it to the current bulk request.
*
* @param NodeInterface $node
* @param string $targetWorkspaceName In case this is triggered during publishing, a workspace name will be passed in
* @return void
* @throws \TYPO3\TYPO3CR\Search\Exception\IndexingException
*/
public function indexNode(NodeInterface $node, $targetWorkspaceName = NULL)
{
$contextPath = $node->getContextPath();
if ($targetWorkspaceName !== NULL) {
$contextPath = str_replace($node->getContext()->getWorkspace()->getName(), $targetWorkspaceName, $contextPath);
}
$contextPathHash = sha1($contextPath);
$nodeType = $node->getNodeType();
$mappingType = $this->getIndex()->findType(NodeTypeMappingBuilder::convertNodeTypeNameToMappingName($nodeType));
// Remove document with the same contextPathHash but different NodeType, required after NodeType change
$this->getIndex()->request('DELETE', '/_query', array(), json_encode(['query' => ['bool' => ['must' => ['ids' => ['values' => [$contextPathHash]]], 'must_not' => ['term' => ['_type' => str_replace('.', '/', $node->getNodeType()->getName())]]]]]));
if ($node->isRemoved()) {
// TODO: handle deletion from the fulltext index as well
$mappingType->deleteDocumentById($contextPathHash);
$this->logger->log(sprintf('NodeIndexer: Removed node %s from index (node flagged as removed). ID: %s', $contextPath, $contextPathHash), LOG_DEBUG, NULL, 'ElasticSearch (CR)');
return;
}
$logger = $this->logger;
$fulltextIndexOfNode = array();
$nodePropertiesToBeStoredInIndex = $this->extractPropertiesAndFulltext($node, $fulltextIndexOfNode, function ($propertyName) use($logger, $contextPathHash) {
$logger->log(sprintf('NodeIndexer (%s) - Property "%s" not indexed because no configuration found.', $contextPathHash, $propertyName), LOG_DEBUG, NULL, 'ElasticSearch (CR)');
});
$document = new ElasticSearchDocument($mappingType, $nodePropertiesToBeStoredInIndex, $contextPathHash);
$documentData = $document->getData();
if ($targetWorkspaceName !== NULL) {
$documentData['__workspace'] = $targetWorkspaceName;
}
$dimensionCombinations = $node->getContext()->getDimensions();
if (is_array($dimensionCombinations)) {
$documentData['__dimensionCombinations'] = $dimensionCombinations;
}
if ($this->isFulltextEnabled($node)) {
if ($this->isFulltextRoot($node)) {
// for fulltext root documents, we need to preserve the "__fulltext" field. That's why we use the
// "update" API instead of the "index" API, with a custom script internally; as we
// shall not delete the "__fulltext" part of the document if it has any.
$this->currentBulkRequest[] = array(array('update' => array('_type' => $document->getType()->getName(), '_id' => $document->getId())), array('script' => '
fulltext = (ctx._source.containsKey("__fulltext") ? ctx._source.__fulltext : new LinkedHashMap());
fulltextParts = (ctx._source.containsKey("__fulltextParts") ? ctx._source.__fulltextParts : new LinkedHashMap());
ctx._source = newData;
ctx._source.__fulltext = fulltext;
ctx._source.__fulltextParts = fulltextParts
', 'params' => array('newData' => $documentData), 'upsert' => $documentData, 'lang' => 'groovy'));
} else {
// non-fulltext-root documents can be indexed as-they-are
$this->currentBulkRequest[] = array(array('index' => array('_type' => $document->getType()->getName(), '_id' => $document->getId())), $documentData);
}
$this->updateFulltext($node, $fulltextIndexOfNode, $targetWorkspaceName);
}
$this->logger->log(sprintf('NodeIndexer: Added / updated node %s. ID: %s', $contextPath, $contextPathHash), LOG_DEBUG, NULL, 'ElasticSearch (CR)');
}
/**
* index this node, and add it to the current bulk request.
*
* @param NodeInterface $node
* @param string $targetWorkspaceName In case this is triggered during publishing, a workspace name will be passed in
* @return void
* @throws \TYPO3\TYPO3CR\Search\Exception\IndexingException
*/
public function indexNode(NodeInterface $node, $targetWorkspaceName = null)
{
$indexer = function (NodeInterface $node, $targetWorkspaceName = null) {
$contextPath = $node->getContextPath();
if ($this->settings['indexAllWorkspaces'] === false) {
// we are only supposed to index the live workspace.
// We need to check the workspace at two occasions; checking the
// $targetWorkspaceName and the workspace name of the node's context as fallback
if ($targetWorkspaceName !== null && $targetWorkspaceName !== 'live') {
return;
}
if ($targetWorkspaceName === null && $node->getContext()->getWorkspaceName() !== 'live') {
return;
}
}
if ($targetWorkspaceName !== null) {
$contextPath = str_replace($node->getContext()->getWorkspace()->getName(), $targetWorkspaceName, $contextPath);
}
$contextPathHash = sha1($contextPath);
$nodeType = $node->getNodeType();
$mappingType = $this->getIndex()->findType(NodeTypeMappingBuilder::convertNodeTypeNameToMappingName($nodeType));
// Remove document with the same contextPathHash but different NodeType, required after NodeType change
$this->logger->log(sprintf('NodeIndexer: Removing node %s from index (if node type changed from %s). ID: %s', $contextPath, $node->getNodeType()->getName(), $contextPathHash), LOG_DEBUG, null, 'ElasticSearch (CR)');
$this->getIndex()->request('DELETE', '/_query', [], json_encode(['query' => ['bool' => ['must' => ['ids' => ['values' => [$contextPathHash]]], 'must_not' => ['term' => ['_type' => NodeTypeMappingBuilder::convertNodeTypeNameToMappingName($node->getNodeType()->getName())]]]]]));
if ($node->isRemoved()) {
// TODO: handle deletion from the fulltext index as well
$mappingType->deleteDocumentById($contextPathHash);
$this->logger->log(sprintf('NodeIndexer: Removed node %s from index (node flagged as removed). ID: %s', $contextPath, $contextPathHash), LOG_DEBUG, null, 'ElasticSearch (CR)');
return;
}
$logger = $this->logger;
$fulltextIndexOfNode = [];
$nodePropertiesToBeStoredInIndex = $this->extractPropertiesAndFulltext($node, $fulltextIndexOfNode, function ($propertyName) use($logger, $contextPathHash) {
$logger->log(sprintf('NodeIndexer (%s) - Property "%s" not indexed because no configuration found.', $contextPathHash, $propertyName), LOG_DEBUG, null, 'ElasticSearch (CR)');
});
$document = new ElasticSearchDocument($mappingType, $nodePropertiesToBeStoredInIndex, $contextPathHash);
$documentData = $document->getData();
if ($targetWorkspaceName !== null) {
$documentData['__workspace'] = $targetWorkspaceName;
}
$dimensionCombinations = $node->getContext()->getDimensions();
if (is_array($dimensionCombinations)) {
$documentData['__dimensionCombinations'] = $dimensionCombinations;
$documentData['__dimensionCombinationHash'] = md5(json_encode($dimensionCombinations));
}
if ($this->isFulltextEnabled($node)) {
if ($this->isFulltextRoot($node)) {
// for fulltext root documents, we need to preserve the "__fulltext" field. That's why we use the
// "update" API instead of the "index" API, with a custom script internally; as we
// shall not delete the "__fulltext" part of the document if it has any.
$this->currentBulkRequest[] = [['update' => ['_type' => $document->getType()->getName(), '_id' => $document->getId()]], ['script' => '
fulltext = (ctx._source.containsKey("__fulltext") ? ctx._source.__fulltext : new LinkedHashMap());
fulltextParts = (ctx._source.containsKey("__fulltextParts") ? ctx._source.__fulltextParts : new LinkedHashMap());
ctx._source = newData;
ctx._source.__fulltext = fulltext;
ctx._source.__fulltextParts = fulltextParts
', 'params' => ['newData' => $documentData], 'upsert' => $documentData, 'lang' => 'groovy']];
} else {
// non-fulltext-root documents can be indexed as-they-are
$this->currentBulkRequest[] = [['index' => ['_type' => $document->getType()->getName(), '_id' => $document->getId()]], $documentData];
}
$this->updateFulltext($node, $fulltextIndexOfNode, $targetWorkspaceName);
}
$this->logger->log(sprintf('NodeIndexer: Added / updated node %s. ID: %s Context: %s', $contextPath, $contextPathHash, json_encode($node->getContext()->getProperties())), LOG_DEBUG, null, 'ElasticSearch (CR)');
};
$dimensionCombinations = $this->contentDimensionCombinator->getAllAllowedCombinations();
$workspaceName = $targetWorkspaceName ?: 'live';
$nodeIdentifier = $node->getIdentifier();
if ($dimensionCombinations !== []) {
foreach ($dimensionCombinations as $combination) {
$context = $this->contextFactory->create(['workspaceName' => $workspaceName, 'dimensions' => $combination]);
$node = $context->getNodeByIdentifier($nodeIdentifier);
if ($node !== null) {
$indexer($node, $targetWorkspaceName);
}
}
} else {
$context = $this->contextFactory->create(['workspaceName' => $workspaceName]);
$node = $context->getNodeByIdentifier($nodeIdentifier);
if ($node !== null) {
$indexer($node, $targetWorkspaceName);
}
}
}
