From 3be1701703e530d6e3cd6e3fba0dfb74ba2561f1 Mon Sep 17 00:00:00 2001
From: Jan Robert <15772440+Yalz@users.noreply.github.com>
Date: Mon, 5 Aug 2024 14:40:24 +0200
Subject: [PATCH] feat: RepoSink: delete skolem uris (#670)

Co-authored-by: Yalz <Yalz@users.noreply.github.com>
---
 .../informatievlaanderen/ldes/ldi/RepositorySink.java |  7 +++++++
 .../ldes/ldi/RepositorySinkIT.java                    | 11 ++++++++---
 .../src/test/resources/people/6-updated.nq            |  5 +++++
 .../repository-sink/src/test/resources/people/6.nq    |  6 ++++++
 4 files changed, 26 insertions(+), 3 deletions(-)
 create mode 100644 ldi-core/repository-sink/src/test/resources/people/6-updated.nq
 create mode 100644 ldi-core/repository-sink/src/test/resources/people/6.nq

diff --git a/ldi-core/repository-sink/src/main/java/be/vlaanderen/informatievlaanderen/ldes/ldi/RepositorySink.java b/ldi-core/repository-sink/src/main/java/be/vlaanderen/informatievlaanderen/ldes/ldi/RepositorySink.java
index e1a15e4ac..32b65a8d7 100644
--- a/ldi-core/repository-sink/src/main/java/be/vlaanderen/informatievlaanderen/ldes/ldi/RepositorySink.java
+++ b/ldi-core/repository-sink/src/main/java/be/vlaanderen/informatievlaanderen/ldes/ldi/RepositorySink.java
@@ -14,11 +14,13 @@
 import java.util.Deque;
 import java.util.List;
 import java.util.Set;
+import java.util.regex.Pattern;
 
 /**
  * Component that will write linked data models to a specified triples store or RDF repository
  */
 public class RepositorySink {
+	private static final Pattern SKOLEM_PATTERN = Pattern.compile(".*/.well-known/genid/.*");
 	private final RepositorySinkConnection repositorySinkConnection;
 
 	public RepositorySink(String hostUrl, String repositoryId, String namedGraph) {
@@ -75,6 +77,11 @@ private Set<Resource> getAllSubjectsFromModel(org.eclipse.rdf4j.model.Model mode
 					subjectStack.push(bnode);
 					subjects.add(bnode);
 				}
+				if (SKOLEM_PATTERN.matcher(object.stringValue()).matches()) {
+					Resource bnode = (Resource) object;
+					subjectStack.push(bnode);
+					subjects.add(bnode);
+				}
 			});
 
 		}
diff --git a/ldi-core/repository-sink/src/test/java/be/vlaanderen/informatievlaanderen/ldes/ldi/RepositorySinkIT.java b/ldi-core/repository-sink/src/test/java/be/vlaanderen/informatievlaanderen/ldes/ldi/RepositorySinkIT.java
index 1da54b9f4..1e65f5437 100644
--- a/ldi-core/repository-sink/src/test/java/be/vlaanderen/informatievlaanderen/ldes/ldi/RepositorySinkIT.java
+++ b/ldi-core/repository-sink/src/test/java/be/vlaanderen/informatievlaanderen/ldes/ldi/RepositorySinkIT.java
@@ -77,19 +77,24 @@ void when_DeleteEntities_Then_EntitiesRemovedFromStore(String namedGraph) throws
 	void when_UpdateEntities_Then_OldTriplesRemoved(String namedGraph) throws Exception {
 		repositorySink = new RepositorySink(subject, LOCAL_REPOSITORY_ID, namedGraph);
 
-		final List<String> testFiles = IntStream.range(1, 6).mapToObj("src/test/resources/people/%d.nq"::formatted).toList();
+		final List<String> testFiles = IntStream.rangeClosed(1, 6).mapToObj("src/test/resources/people/%d.nq"::formatted).toList();
 		populateAndCheckRepository(testFiles, namedGraph);
 
 		List<org.apache.jena.rdf.model.Model> models = List.of(RDFParser.source("people/5-updated.nq").toModel());
 		repositorySink.process(models);
 
+		models = List.of(RDFParser.source("people/6-updated.nq").toModel());
+		repositorySink.process(models);
+
 		List<Statement> statements = repositorySink.getRepositoryConnection()
 				.getStatements(null, null, null).stream().toList();
 
 		assertThat(statements)
-				.hasSize(21)
+				.hasSize(26)
 				.anyMatch(statement -> statement.getObject().stringValue().equals("CHANGED"))
-				.noneMatch(statement -> statement.getObject().stringValue().equals("Taylor"));
+				.noneMatch(statement -> statement.getObject().stringValue().equals("Taylor"))
+				.noneMatch(statement -> statement.getObject().stringValue().equals("Twift"))
+				.noneMatch(statement -> statement.getPredicate().stringValue().equals("http://www.w3.org/2001/vcard-rdf/3.0#Nickname"));
 	}
 
 	@Test
diff --git a/ldi-core/repository-sink/src/test/resources/people/6-updated.nq b/ldi-core/repository-sink/src/test/resources/people/6-updated.nq
new file mode 100644
index 000000000..1305f827e
--- /dev/null
+++ b/ldi-core/repository-sink/src/test/resources/people/6-updated.nq
@@ -0,0 +1,5 @@
+<http://somewhere/SaylorTwift/> <http://www.w3.org/2001/vcard-rdf/3.0#FN> "Saylor Swiffer" .
+<http://somewhere/SaylorTwift/> <http://www.w3.org/2001/vcard-rdf/3.0#N> <http://somewhere/.well-known/genid/3> .
+<http://somewhere/.well-known/genid/3> <http://www.w3.org/2001/vcard-rdf/3.0#Family> "Saylor" .
+<http://somewhere/.well-known/genid/3> <http://www.w3.org/2001/vcard-rdf/3.0#Given> <http://somewhere/.well-known/genid/4> .
+<http://somewhere/.well-known/genid/4> <http://www.w3.org/2001/vcard-rdf/3.0#Last> "Swiffer" .
diff --git a/ldi-core/repository-sink/src/test/resources/people/6.nq b/ldi-core/repository-sink/src/test/resources/people/6.nq
new file mode 100644
index 000000000..aaa4bf10e
--- /dev/null
+++ b/ldi-core/repository-sink/src/test/resources/people/6.nq
@@ -0,0 +1,6 @@
+<http://somewhere/SaylorTwift/> <http://www.w3.org/2001/vcard-rdf/3.0#FN> "Saylor Twift" .
+<http://somewhere/SaylorTwift/> <http://www.w3.org/2001/vcard-rdf/3.0#N> <http://somewhere/.well-known/genid/1> .
+<http://somewhere/.well-known/genid/1> <http://www.w3.org/2001/vcard-rdf/3.0#Family> "Saylor" .
+<http://somewhere/.well-known/genid/1> <http://www.w3.org/2001/vcard-rdf/3.0#Given> <http://somewhere/.well-known/genid/2> .
+<http://somewhere/.well-known/genid/2> <http://www.w3.org/2001/vcard-rdf/3.0#Nickname> "1Saylor" .
+<http://somewhere/.well-known/genid/2> <http://www.w3.org/2001/vcard-rdf/3.0#Last> "Twift" .
\ No newline at end of file