Merge branch 'main' into Aggregation-updates

wellcomecollection · Dec 2, 2024 · 2a29a74 · 2a29a74
2 parents 5fab5a7 + cd74c2b
commit 2a29a74
Show file tree

Hide file tree

Showing 4 changed files with 47 additions and 17 deletions.
diff --git a/pipeline/terraform/modules/stack/elastic_pipeline.tf b/pipeline/terraform/modules/stack/elastic_pipeline.tf
@@ -132,6 +132,10 @@ module "pipeline_indices" {
 locals {
   indices = module.pipeline_indices.index_names
   service_index_permissions = {
+    read_only = {
+      read  = ["*"]
+      write = []
+    }
     transformer = {
       read  = []
       write = [local.indices.source]

diff --git a/...rmer/transformer_calm/src/main/scala/weco/pipeline/transformer/calm/CalmTransformer.scala b/...rmer/transformer_calm/src/main/scala/weco/pipeline/transformer/calm/CalmTransformer.scala
@@ -154,7 +154,7 @@ object CalmTransformer
       format = Some(CalmFormat(record)),
       collectionPath = Some(collectionPath),
       referenceNumber = collectionPath.label.map(ReferenceNumber(_)),
-      subjects = subjects(record),
+      subjects = CalmSubjects(record),
       languages = languages,
       items = CalmItems(record),
       contributors = CalmContributors(record),
@@ -259,17 +259,4 @@ object CalmTransformer
         )
     }
   }
-
-  def subjects(record: CalmRecord): List[Subject[IdState.Unminted]] =
-    record
-      .getList("Subject")
-      .map {
-        label =>
-          val normalisedLabel =
-            NormaliseText(label, whitelist = NormaliseText.none)
-          Subject(
-            label = normalisedLabel,
-            concepts = List(Concept(normalisedLabel))
-          )
-      }
 }
diff --git a/...former_calm/src/main/scala/weco/pipeline/transformer/calm/transformers/CalmSubjects.scala b/...former_calm/src/main/scala/weco/pipeline/transformer/calm/transformers/CalmSubjects.scala
@@ -0,0 +1,33 @@
+package weco.pipeline.transformer.calm.transformers
+
+import weco.catalogue.internal_model.identifiers.IdState
+import weco.catalogue.internal_model.work.{Concept, Subject}
+import weco.catalogue.source_model.calm.CalmRecord
+import weco.pipeline.transformer.calm.NormaliseText
+import weco.pipeline.transformer.calm.models.CalmRecordOps
+import weco.pipeline.transformer.identifiers.LabelDerivedIdentifiers
+
+object CalmSubjects extends CalmRecordOps with LabelDerivedIdentifiers {
+  def apply(record: CalmRecord): List[Subject[IdState.Unminted]] =
+    record.getList("Subject").map {
+      label =>
+        val normalisedLabel =
+          NormaliseText(label, whitelist = NormaliseText.none)
+
+        val labelDerivedId = identifierFromText(
+          label = normalisedLabel,
+          ontologyType = "Concept"
+        )
+
+        Subject(
+          label = normalisedLabel,
+          concepts = List(
+            Concept(
+              id = labelDerivedId,
+              label = normalisedLabel
+            )
+          ),
+          id = labelDerivedId
+        )
+    }
+}
diff --git a/.../transformer_calm/src/test/scala/weco/pipeline/transformer/calm/CalmTransformerTest.scala b/.../transformer_calm/src/test/scala/weco/pipeline/transformer/calm/CalmTransformerTest.scala
@@ -253,13 +253,19 @@ class CalmTransformerTest
       "Subject" -> "<i>anatomy</i>",
       "CatalogueStatus" -> "Catalogued"
     )
+
     CalmTransformer(
       record,
       version
     ).right.get.data.subjects should contain theSameElementsAs List(
-      Subject("anatomy", List(Concept("anatomy"))),
-      Subject("botany", List(Concept("botany")))
-    )
+      "anatomy",
+      "botany"
+    ).map((strippedSubject: String) => {
+      val id = labelDerivedConceptIdentifier(strippedSubject)
+      val concepts = List(Concept(id, strippedSubject))
+
+      Subject(id, strippedSubject, concepts)
+    })
   }
 
   it("finds a single language") {