Deduplicating shapes before Shacl validation (#5021)

* Deduplicating shapes before Shacl validation

---------

Co-authored-by: Simon Dumas <[email protected]>

delta/rdf/src/main/scala/ch/epfl/bluebrain/nexus/delta/rdf/graph/Graph.scala

@@ -60,6 +60,8 @@ final case class Graph private (rootNode: IriOrBNode, value: DatasetGraph) { sel
     */
   def isEmpty: Boolean = value.isEmpty
 
+  def getDefaultGraphSize: Int = value.getDefaultGraph.size()
+
   /**
     * Returns a subgraph retaining all the triples that satisfy the provided predicate.
     */

delta/sdk/src/main/scala/ch/epfl/bluebrain/nexus/delta/sdk/schemas/model/Schema.scala

@@ -2,6 +2,7 @@ package ch.epfl.bluebrain.nexus.delta.sdk.schemas.model
 
 import cats.data.NonEmptyList
 import cats.effect.IO
+import ch.epfl.bluebrain.nexus.delta.rdf.IriOrBNode
 import ch.epfl.bluebrain.nexus.delta.rdf.IriOrBNode.Iri
 import ch.epfl.bluebrain.nexus.delta.rdf.Triple.Triple
 import ch.epfl.bluebrain.nexus.delta.rdf.Vocabulary.{contexts, nxv, owl}
@@ -61,9 +62,17 @@ final case class Schema(
   lazy val ontologies: Graph = graph(types => types.contains(owl.Ontology) && !types.contains(nxv.Schema))
 
   private def graph(filteredTypes: Set[Iri] => Boolean): Graph = {
-    implicit val api: JsonLdApi = JsonLdJavaApi.lenient
-    val filtered                = expanded.filter(expanded => expanded.cursor.getTypes.exists(filteredTypes))
-    val triples                 = filtered.map(_.toGraph.toOption.get).foldLeft(Set.empty[Triple])((acc, g) => acc ++ g.triples)
+    implicit val api: JsonLdApi                         = JsonLdJavaApi.lenient
+    val init: (Set[IriOrBNode], Vector[ExpandedJsonLd]) = (Set.empty[IriOrBNode], Vector.empty[ExpandedJsonLd])
+    val (_, filtered)                                   = expanded.foldLeft(init) {
+      case ((seen, acc), expanded)
+          if !seen.contains(expanded.rootId) && expanded.cursor.getTypes.exists(filteredTypes) =>
+        val updatedSeen = seen + expanded.rootId
+        val updatedAcc  = acc :+ expanded
+        updatedSeen -> updatedAcc
+      case ((seen, acc), _) => seen -> acc
+    }
+    val triples                                         = filtered.map(_.toGraph.toOption.get).foldLeft(Set.empty[Triple])((acc, g) => acc ++ g.triples)
     Graph.empty(id).add(triples)
   }
 

delta/sdk/src/test/scala/ch/epfl/bluebrain/nexus/delta/sdk/schemas/model/SchemaSuite.scala

@@ -0,0 +1,32 @@
+package ch.epfl.bluebrain.nexus.delta.sdk.schemas.model
+
+import cats.data.NonEmptyList
+import cats.effect.IO
+import ch.epfl.bluebrain.nexus.delta.rdf.jsonld.ExpandedJsonLd
+import ch.epfl.bluebrain.nexus.delta.rdf.syntax._
+import ch.epfl.bluebrain.nexus.delta.sdk.model.Tags
+import ch.epfl.bluebrain.nexus.delta.sdk.utils.Fixtures
+import ch.epfl.bluebrain.nexus.delta.sourcing.model.ProjectRef
+import ch.epfl.bluebrain.nexus.testkit.mu.NexusSuite
+
+class SchemaSuite extends NexusSuite with Fixtures {
+
+  private val project = ProjectRef.unsafe("org", "proj")
+
+  test("Extract as a graph the content of the schema, removing the duplicates") {
+    for {
+      entitySource             <- loader.jsonContentOf("schemas/entity.json")
+      entityExpanded           <- ExpandedJsonLd(jsonContentOf("schemas/entity-expanded.json"))
+      entityExpandedGraphSize  <- IO.fromEither(entityExpanded.toGraph.map(_.getDefaultGraphSize))
+      entityCompacted          <- entityExpanded.toCompacted(entitySource.topContextValueOrEmpty)
+      licenseExpanded          <- ExpandedJsonLd(jsonContentOf("schemas/license-expanded.json"))
+      licenseExpandedGraphSize <- IO.fromEither(licenseExpanded.toGraph.map(_.getDefaultGraphSize))
+    } yield {
+      val id        = iri"https://neuroshapes.org/commons/entity"
+      val expandeds = NonEmptyList.of(entityExpanded, licenseExpanded, entityExpanded)
+      val schema    = Schema(id, project, Tags.empty, entitySource, entityCompacted, expandeds)
+      assertEquals(schema.shapes.getDefaultGraphSize, entityExpandedGraphSize + licenseExpandedGraphSize)
+    }
+  }
+
+}