feat(triplestore): Update Lucene index on startup (#1362)

* feat: Update Lucene index on startup.

* feat(triplestore): Update Lucene index when an rdfs:label changes.

webapi/src/main/scala/org/knora/webapi/app/ApplicationStateActor.scala

@@ -4,7 +4,7 @@ import akka.actor.{Actor, ActorLogging, ActorSelection, Timers}
 import org.knora.webapi._
 import org.knora.webapi.messages.app.appmessages.AppState.AppState
 import org.knora.webapi.messages.app.appmessages._
-import org.knora.webapi.messages.store.triplestoremessages.{CheckRepositoryRequest, CheckRepositoryResponse, RepositoryStatus}
+import org.knora.webapi.messages.store.triplestoremessages.{CheckRepositoryRequest, CheckRepositoryResponse, RepositoryStatus, SearchIndexUpdateRequest, SparqlUpdateResponse}
 import org.knora.webapi.messages.v2.responder.SuccessResponseV2
 import org.knora.webapi.messages.v2.responder.ontologymessages.LoadOntologiesRequestV2
 import org.knora.webapi.responders.RESPONDER_MANAGER_ACTOR_PATH
@@ -65,15 +65,17 @@ class ApplicationStateActor extends Actor with Timers with ActorLogging {
                 case AppState.Stopped => // do nothing
                 case AppState.StartingUp => self ! SetAppState(AppState.WaitingForRepository)
                 case AppState.WaitingForRepository => self ! CheckRepository() // check DB
-                case AppState.RepositoryReady =>  self ! SetAppState(AppState.CreatingCaches)
+                case AppState.RepositoryReady => self ! SetAppState(AppState.CreatingCaches)
                 case AppState.CreatingCaches => self ! CreateCaches()
-                case AppState.CachesReady => self ! SetAppState(AppState.LoadingOntologies)
-                case AppState.LoadingOntologies if skipOntologies => self ! SetAppState(AppState.OntologiesReady)  // skipping loading of ontologies
+                case AppState.CachesReady => self ! SetAppState(AppState.UpdatingSearchIndex)
+                case AppState.UpdatingSearchIndex => self ! UpdateSearchIndex()
+                case AppState.SearchIndexReady => self ! SetAppState(AppState.LoadingOntologies)
+                case AppState.LoadingOntologies if skipOntologies => self ! SetAppState(AppState.OntologiesReady) // skipping loading of ontologies
                 case AppState.LoadingOntologies if !skipOntologies => self ! LoadOntologies() // load ontologies
                 case AppState.OntologiesReady => self ! SetAppState(AppState.Running)
                 case AppState.Running => printWelcomeMsg()
                 case AppState.MaintenanceMode => // do nothing
-                case value => throw UnsupportedValueException(s"The value: $value is not supported.")
+                case other => throw UnsupportedValueException(s"The value: $other is not supported.")
             }
         }
         case GetAppState() => {
@@ -130,15 +132,25 @@ class ApplicationStateActor extends Actor with Timers with ActorLogging {
             self ! SetAppState(AppState.CachesReady)
         }
 
+        case UpdateSearchIndex() => {
+            storeManager ! SearchIndexUpdateRequest()
+        }
+
+        case SparqlUpdateResponse() => {
+            self ! SetAppState(AppState.SearchIndexReady)
+        }
+
         /* load ontologies request */
         case LoadOntologies() => {
-            responderManager !  LoadOntologiesRequestV2(KnoraSystemInstances.Users.SystemUser)
+            responderManager ! LoadOntologiesRequestV2(KnoraSystemInstances.Users.SystemUser)
         }
 
         /* load ontologies response */
-        case SuccessResponseV2(msg) => {
+        case SuccessResponseV2(_) => {
             self ! SetAppState(AppState.OntologiesReady)
         }
+
+        case other => throw UnexpectedMessageException(s"ApplicationStateActor received an unexpected message $other of type ${other.getClass.getCanonicalName}")
     }
 
     override def postStop(): Unit = {

webapi/src/main/scala/org/knora/webapi/messages/app/appmessages/ApplicationStateMessages.scala

@@ -106,6 +106,11 @@ case class CheckRepository() extends ApplicationStateRequest
   */
 case class CreateCaches() extends ApplicationStateRequest
 
+/**
+  * Message for updating the triplestore's full-text search index. Used only inside the actor itself.
+  */
+case class UpdateSearchIndex() extends ApplicationStateRequest
+
 /**
   * Message for initiating loading of ontologies. Used only inside the actor itself.
   */
@@ -116,6 +121,7 @@ case class LoadOntologies() extends ApplicationStateRequest
   */
 object AppState extends Enumeration {
     type AppState = Value
-    val Stopped, StartingUp, WaitingForRepository, RepositoryReady, CreatingCaches, CachesReady, LoadingOntologies, OntologiesReady, MaintenanceMode, Running = Value
+    val Stopped, StartingUp, WaitingForRepository, RepositoryReady, CreatingCaches, CachesReady,
+    UpdatingSearchIndex, SearchIndexReady, LoadingOntologies, OntologiesReady, MaintenanceMode, Running = Value
 }
 

webapi/src/main/scala/org/knora/webapi/messages/store/triplestoremessages/TriplestoreMessages.scala

@@ -227,6 +227,14 @@ case class CheckRepositoryRequest() extends TriplestoreRequest
   */
 case class CheckRepositoryResponse(repositoryStatus: RepositoryStatus, msg: String)
 
+/**
+  * Updates the triplestore's full-text search index.
+  *
+  * @param subjectIri if a subject has changed, update the index for that subject. Otherwise, updates
+  *                   the index to add any subjects not yet indexed.
+  */
+case class SearchIndexUpdateRequest(subjectIri: Option[String] = None) extends TriplestoreRequest
+
 
 //////////////////////////////////////////////////////////////////////////////////////////////////////////////////
 // Components of messages

webapi/src/main/scala/org/knora/webapi/responders/v2/ResourcesResponderV2.scala

@@ -341,6 +341,16 @@ class ResourcesResponderV2(responderData: ResponderData) extends ResponderWithSt
 
                     case None => ()
                 }
+
+                // If the resource's label was changed, update the full-text search index.
+                _ <- updateResourceMetadataRequestV2.maybeLabel match {
+                    case Some(_) =>
+                        for {
+                            _ <- (storeManager ? SearchIndexUpdateRequest(Some(updateResourceMetadataRequestV2.resourceIri))).mapTo[SparqlUpdateResponse]
+                        } yield ()
+
+                    case None => FastFuture.successful(())
+                }
             } yield SuccessResponseV2("Resource metadata updated")
         }
 

webapi/src/main/scala/org/knora/webapi/store/triplestore/http/HttpTriplestoreConnector.scala

@@ -128,6 +128,7 @@ class HttpTriplestoreConnector extends Actor with ActorLogging {
         case InsertTriplestoreContent(rdfDataObjects) => try2Message(sender(), insertDataIntoTriplestore(rdfDataObjects), log)
         case HelloTriplestore(msg) if msg == triplestoreType => sender ! HelloTriplestore(triplestoreType)
         case CheckRepositoryRequest() => try2Message(sender(), checkRepository(), log)
+        case SearchIndexUpdateRequest(subjectIri) => try2Message(sender(), updateLuceneIndex(subjectIri), log)
         case other => sender ! Status.Failure(UnexpectedMessageException(s"Unexpected message $other of type ${other.getClass.getCanonicalName}"))
     }
 
@@ -346,6 +347,29 @@ class HttpTriplestoreConnector extends Actor with ActorLogging {
         } yield response
     }
 
+    /**
+      * Updates the Lucene full-text search index.
+      */
+    private def updateLuceneIndex(subjectIri: Option[IRI] = None): Try[SparqlUpdateResponse] = {
+        val indexUpdateSparqlString = subjectIri match {
+            case Some(definedSubjectIri) =>
+                // A subject's content has changed. Update the index for that subject.
+                s"""PREFIX luc: <http://www.ontotext.com/owlim/lucene#>
+                  |INSERT DATA { luc:fullTextSearchIndex luc:addToIndex <$definedSubjectIri> . }
+                """.stripMargin
+
+            case None =>
+                // Add new subjects to the index.
+                """PREFIX luc: <http://www.ontotext.com/owlim/lucene#>
+                  |INSERT DATA { luc:fullTextSearchIndex luc:updateIndex _:b1 . }
+                """.stripMargin
+        }
+
+        for {
+            _ <- getTriplestoreHttpResponse(indexUpdateSparqlString, isUpdate = true)
+        } yield SparqlUpdateResponse()
+    }
+
     /**
       * Performs a SPARQL update operation.
       *
@@ -361,12 +385,7 @@ class HttpTriplestoreConnector extends Actor with ActorLogging {
 
             // If we're using GraphDB, update the full-text search index.
             _ = if (triplestoreType == TriplestoreTypes.HttpGraphDBSE | triplestoreType == TriplestoreTypes.HttpGraphDBFree) {
-                val indexUpdateSparqlString =
-                    """
-                        PREFIX luc: <http://www.ontotext.com/owlim/lucene#>
-                        INSERT DATA { luc:fullTextSearchIndex luc:updateIndex _:b1 . }
-                    """
-                getTriplestoreHttpResponse(indexUpdateSparqlString, isUpdate = true)
+                updateLuceneIndex()
             }
         } yield SparqlUpdateResponse()
     }
@@ -458,12 +477,7 @@ class HttpTriplestoreConnector extends Actor with ActorLogging {
 
             if (triplestoreType == TriplestoreTypes.HttpGraphDBSE || triplestoreType == TriplestoreTypes.HttpGraphDBFree) {
                 /* need to update the lucene index */
-                val indexUpdateSparqlString =
-                    """
-                        PREFIX luc: <http://www.ontotext.com/owlim/lucene#>
-                        INSERT DATA { luc:fullTextSearchIndex luc:updateIndex _:b1 . }
-                    """
-                getTriplestoreHttpResponse(indexUpdateSparqlString, isUpdate = true)
+                updateLuceneIndex()
             }
 
             log.debug("==>> Loading Data End")
@@ -610,7 +624,7 @@ class HttpTriplestoreConnector extends Actor with ActorLogging {
                 }
 
                 val took = System.currentTimeMillis() - start
-                log.info(s"[${statusCode}] GraphDB Query took: ${took}ms")
+                log.info(s"[$statusCode] GraphDB Query took: ${took}ms")
 
                 responseEntityStr
             } finally {