Remove empty ids from source when importing resources (#4931)

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

ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/RunShip.scala

@@ -18,7 +18,7 @@ import ch.epfl.bluebrain.nexus.ship.files.FileProcessor
 import ch.epfl.bluebrain.nexus.ship.organizations.OrganizationProvider
 import ch.epfl.bluebrain.nexus.ship.projects.{OriginalProjectContext, ProjectProcessor}
 import ch.epfl.bluebrain.nexus.ship.resolvers.ResolverProcessor
-import ch.epfl.bluebrain.nexus.ship.resources.{DistributionPatcher, ResourceProcessor, ResourceWiring}
+import ch.epfl.bluebrain.nexus.ship.resources.{ResourceProcessor, ResourceWiring, SourcePatcher}
 import ch.epfl.bluebrain.nexus.ship.schemas.{SchemaProcessor, SchemaWiring}
 import ch.epfl.bluebrain.nexus.ship.views.{BlazegraphViewProcessor, CompositeViewProcessor, ElasticSearchViewProcessor, ViewPatcher}
 import fs2.Stream
@@ -63,8 +63,8 @@ object RunShip {
                     resolverProcessor            = ResolverProcessor(fetchContext, projectMapper, eventLogConfig, eventClock, xas)
                     schemaProcessor              = SchemaProcessor(schemaLog, fetchContext, schemaImports, rcr, projectMapper, eventClock)
                     fileSelf                     = FileSelf(originalProjectContext)(originalBaseUri)
-                    distributionPatcher          = new DistributionPatcher(fileSelf, projectMapper, targetBaseUri)
-                    resourceProcessor            = ResourceProcessor(resourceLog, rcr, projectMapper, fetchContext, distributionPatcher, eventClock)
+                    sourcePatcher                = SourcePatcher(fileSelf, projectMapper, targetBaseUri)
+                    resourceProcessor            = ResourceProcessor(resourceLog, rcr, projectMapper, fetchContext, sourcePatcher, eventClock)
                     viewPatcher                  = new ViewPatcher(config.projectMapping)
                     esViewsProcessor             = ElasticSearchViewProcessor(fetchContext, rcr, projectMapper, viewPatcher, eventLogConfig, eventClock, xas)
                     bgViewsProcessor             = BlazegraphViewProcessor(fetchContext, rcr, projectMapper, viewPatcher, eventLogConfig, eventClock, xas)

ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/resources/ResourceProcessor.scala

@@ -22,7 +22,7 @@ import io.circe.Decoder
 class ResourceProcessor private (
     resources: Resources,
     projectMapper: ProjectMapper,
-    distributionPatcher: DistributionPatcher,
+    sourcePatcher: SourcePatcher,
     clock: EventClock
 ) extends EventProcessor[ResourceEvent] {
 
@@ -48,11 +48,11 @@ class ResourceProcessor private (
 
     event match {
       case e: ResourceCreated       =>
-        distributionPatcher.singleOrArray(e.source).flatMap { patched =>
+        sourcePatcher(e.source).flatMap { patched =>
           resources.create(e.id, project, e.schema.toIdSegment, patched, e.tag)
         }
       case e: ResourceUpdated       =>
-        distributionPatcher.singleOrArray(e.source).flatMap { patched =>
+        sourcePatcher(e.source).flatMap { patched =>
           resources.update(e.id, project, e.schema.toIdSegment.some, cRev, patched, e.tag)
         }
       case e: ResourceSchemaUpdated =>
@@ -88,11 +88,11 @@ object ResourceProcessor {
       rcr: ResolverContextResolution,
       projectMapper: ProjectMapper,
       fetchContext: FetchContext,
-      distributionPatcher: DistributionPatcher,
+      sourcePatcher: SourcePatcher,
       clock: EventClock
   )(implicit jsonLdApi: JsonLdApi): ResourceProcessor = {
     val resources = ResourcesImpl(log, fetchContext, rcr)
-    new ResourceProcessor(resources, projectMapper, distributionPatcher, clock)
+    new ResourceProcessor(resources, projectMapper, sourcePatcher, clock)
   }
 
 }

ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/resources/SourcePatcher.scala

@@ -0,0 +1,30 @@
+package ch.epfl.bluebrain.nexus.ship.resources
+
+import cats.effect.IO
+import ch.epfl.bluebrain.nexus.delta.plugins.storage.FileSelf
+import ch.epfl.bluebrain.nexus.delta.sdk.model.BaseUri
+import ch.epfl.bluebrain.nexus.ship.ProjectMapper
+import ch.epfl.bluebrain.nexus.ship.resources.SourcePatcher.removeEmptyIds
+import io.circe.Json
+import io.circe.optics.JsonPath.root
+
+final class SourcePatcher(distributionPatcher: DistributionPatcher) {
+
+  def apply(json: Json): IO[Json] = distributionPatcher.singleOrArray(json).map(removeEmptyIds)
+
+}
+
+object SourcePatcher {
+
+  private val emptyString = Json.fromString("")
+
+  // Remove empty ids from sources as they are not accepted anymore in payloads
+  // Resources will be keep their ids as we explicitly pass them along the source payload
+  val removeEmptyIds: Json => Json = root.obj.modify {
+    case obj if obj("@id").contains(emptyString) => obj.remove("@id")
+    case obj => obj
+  }
+
+  def apply(fileSelfParser: FileSelf, projectMapper: ProjectMapper, targetBase: BaseUri): SourcePatcher =
+    new SourcePatcher(new DistributionPatcher(fileSelfParser, projectMapper, targetBase))
+}

ship/src/test/scala/ch/epfl/bluebrain/nexus/ship/resources/SourcePatcherSuite.scala

@@ -0,0 +1,23 @@
+package ch.epfl.bluebrain.nexus.ship.resources
+
+import ch.epfl.bluebrain.nexus.testkit.mu.NexusSuite
+
+class SourcePatcherSuite extends NexusSuite {
+
+  test("Removing empty ids does not modify the source when the @id is non empty") {
+    val source = json"""{ "@id": "bob-id", "name": "Bob" }"""
+    assertEquals(SourcePatcher.removeEmptyIds(source), source)
+  }
+
+  test("Removing empty ids does not modify the source when there is no @id") {
+    val source = json"""{ "name": "Bob" }"""
+    assertEquals(SourcePatcher.removeEmptyIds(source), source)
+  }
+
+  test("Removing empty ids removes the @id field when it is empty") {
+    val source = json"""{ "@id": "", "name": "Bob" }"""
+    val expected = json"""{ "name": "Bob" }"""
+    assertEquals(SourcePatcher.removeEmptyIds(source), expected)
+  }
+
+}