FEATURE: Skip specific node type or namespace from indexing

Author: dfeyer

Classes/Indexer/NodeIndexer.php

@@ -22,6 +22,7 @@
 use Flowpack\ElasticSearch\ContentRepositoryAdaptor\Indexer\Error\MalformedBulkRequestError;
 use Flowpack\ElasticSearch\ContentRepositoryAdaptor\Mapping\NodeTypeMappingBuilder;
 use Flowpack\ElasticSearch\ContentRepositoryAdaptor\Service\ErrorHandlingService;
+use Flowpack\ElasticSearch\ContentRepositoryAdaptor\Service\NodeTypeIndexingConfiguration;
 use Flowpack\ElasticSearch\Domain\Model\Document as ElasticSearchDocument;
 use Flowpack\ElasticSearch\Domain\Model\Index;
 use Flowpack\ElasticSearch\Transfer\Exception\ApiException;
@@ -121,6 +122,12 @@ class NodeIndexer extends AbstractNodeIndexer implements BulkNodeIndexerInterfac
      */
     protected $bulkProcessing = false;
 
+    /**
+     * @var NodeTypeIndexingConfiguration
+     * @Flow\Inject
+     */
+    protected $nodeTypeIndexingConfiguration;
+
     /**
      * Returns the index name to be used for indexing, with optional indexNamePostfix appended.
      *
@@ -170,6 +177,11 @@ public function getIndex()
      */
     public function indexNode(NodeInterface $node, $targetWorkspaceName = null)
     {
+        if ($this->nodeTypeIndexingConfiguration->isIndexable($node->getNodeType()) === false) {
+            $this->logger->log(sprintf('NodeIndexer - Node "%s" (%s) skipped, Node Type is not allowed in the index.', $node->getContextPath(), $node->getNodeType()), LOG_DEBUG, null, 'ElasticSearch (CR)');
+            return;
+        }
+
         $indexer = function (NodeInterface $node, $targetWorkspaceName = null) {
             $contextPath = $node->getContextPath();
 

Classes/Mapping/NodeTypeMappingBuilder.php

@@ -11,6 +11,7 @@
  * source code.
  */
 
+use Flowpack\ElasticSearch\ContentRepositoryAdaptor\Service\NodeTypeIndexingConfiguration;
 use Flowpack\ElasticSearch\Domain\Model\Index;
 use Flowpack\ElasticSearch\Domain\Model\Mapping;
 use Flowpack\ElasticSearch\Mapping\MappingCollection;
@@ -50,6 +51,12 @@ class NodeTypeMappingBuilder
      */
     protected $configurationManager;
 
+    /**
+     * @var NodeTypeIndexingConfiguration
+     * @Flow\Inject
+     */
+    protected $nodeTypeIndexingConfiguration;
+
     /**
      * Called by the Flow object framework after creating the object and resolving all dependencies.
      *
@@ -92,6 +99,10 @@ public function buildMappingInformation(Index $index)
                 continue;
             }
 
+            if ($this->nodeTypeIndexingConfiguration->isIndexable($nodeType) === false) {
+                continue;
+            }
+
             $type = $index->findType(self::convertNodeTypeNameToMappingName($nodeTypeName));
             $mapping = new Mapping($type);
             $fullConfiguration = $nodeType->getFullConfiguration();

Classes/Service/NodeTypeIndexingConfiguration.php

@@ -0,0 +1,61 @@
+<?php
+namespace Flowpack\ElasticSearch\ContentRepositoryAdaptor\Service;
+
+/*
+ * This file is part of the Flowpack.ElasticSearch.ContentRepositoryAdaptor package.
+ *
+ * (c) Contributors of the Neos Project - www.neos.io
+ *
+ * This package is Open Source Software. For the full copyright and license
+ * information, please view the LICENSE file which was distributed with this
+ * source code.
+ */
+
+use Flowpack\ElasticSearch\ContentRepositoryAdaptor\Exception;
+use Neos\ContentRepository\Domain\Model\NodeType;
+use Neos\Flow\Annotations as Flow;
+
+/**
+ * @Flow\Scope("singleton")
+ */
+final class NodeTypeIndexingConfiguration
+{
+    /**
+     * @var array
+     * @Flow\InjectConfiguration(path="configuration")
+     */
+    protected $settings;
+
+    /**
+     * @param NodeType $nodeType
+     * @return bool
+     * @throws Exception
+     */
+    public function isIndexable(NodeType $nodeType)
+    {
+        if (!isset($this->settings['nodeTypes'])) {
+            return true;
+        }
+
+        if (!\is_array($this->settings['nodeTypes'])) {
+            throw new Exception('Check your configuration at indexingConfiguration.nodeTypes, this path must be an array', 1504721629);
+        }
+
+        $settings = $this->settings['nodeTypes'];
+
+        if (isset($settings[$nodeType->getName()]['indexed'])) {
+            return $settings[$nodeType->getName()]['indexed'];
+        }
+
+        $nodeTypeParts = \explode(':', $nodeType->getName());
+        $namespace = reset($nodeTypeParts) . ':*';
+        if (isset($settings[$namespace]['indexed'])) {
+            return $settings[$namespace]['indexed'];
+        }
+        if (isset($settings['*']['indexed'])) {
+            return $settings['*']['indexed'];
+        }
+
+        return false;
+    }
+}

Configuration/Settings.yaml

@@ -2,6 +2,10 @@
 Flowpack:
   ElasticSearch:
     ContentRepositoryAdaptor:
+      configuration:
+        nodeTypes:
+          '*':
+            indexed: true
       driver:
         version: 1.x
         mapping:

README.md

@@ -66,7 +66,36 @@ changes the index alias.
 ./flow nodeindex:cleanup
 ```
 
+### Advanced Configuration
+
+By default the indexing process all Node Types, but you can change this configuration in your ```Settings.yaml```:
+
+```
+Flowpack:
+  ElasticSearch:
+    ContentRepositoryAdaptor:
+      configuration:
+        nodeTypes:
+          '*':
+            indexed: true
+          'Neos.Neos:FallbackNode':
+            indexed: false
+          'Neos.Neos:Shortcut':
+            indexed: false
+          'Neos.Neos:ContentCollection':
+            indexed: false
+```
+
+You need to configure explicitly the Node Types (this feature does not check the Super Type configuration). You 
+can use a special notation to configuration a full name space, ```Neos.NodeTypes:*``` will be applied for all node
+types in the given namespace. The most specific configuration is used in this order: 
+
+- Node Type name (```Neos.Neos:Shortcut```)
+- Full namespace notation (```Neos.Neos:*```)
+- Catch all (```*```)
+
 ### Advanced Index Settings
+
 If you need advanced settings you can define them in your *Settings.yaml*:
 
 Example is from the Documentation of the used *Flowpack.ElasticSearch* Package