Read license information from licenseNote in premis access condition (#…

…2695) * Read license information from licenseNote in premis access condition Co-Authored-By: Ashley Ray <a.ray@wellcome.org> * Apply auto-formatting rules * remove vestigial test --------- Co-authored-by: Ashley Ray <a.ray@wellcome.org> Co-authored-by: Buildkite on behalf of Wellcome Collection <wellcomedigitalplatform@wellcome.ac.uk>
wellcomecollection · Sep 4, 2024 · 30ecdac · 30ecdac
1 parent d98add7
commit 30ecdac
Show file tree

Hide file tree

Showing 3 changed files with 122 additions and 20 deletions.
diff --git a/...s/src/main/scala/weco/pipeline/transformer/mets/transformers/PremisAccessConditions.scala b/...s/src/main/scala/weco/pipeline/transformer/mets/transformers/PremisAccessConditions.scala
@@ -1,4 +1,5 @@
 package weco.pipeline.transformer.mets.transformers
+import grizzled.slf4j.Logging
 import weco.pipeline.transformer.result.Result
 
 import scala.xml.Node
@@ -18,7 +19,7 @@ case class PremisAccessConditions(
     )
 }
 
-object PremisAccessConditions {
+object PremisAccessConditions extends Logging {
 
   /** The access conditions are encoded a premis elementin the METS. For
     * example:
@@ -51,14 +52,41 @@ object PremisAccessConditions {
     */
   def apply(rightsMd: Node): PremisAccessConditions = {
     val rightsStatement = rightsMd \ "mdWrap" \ "xmlData" \ "rightsStatement"
+
+    val rightsBasis = (rightsStatement \ "rightsBasis").headOption
+
+    // Copyright specific information is held in the copyrightNote element
     val copyrightNoteElem =
       (rightsStatement \ "copyrightInformation" \ "copyrightNote").headOption
+
+    // CC license information is indicated in the licenseNote element
+    val licenseNoteElem =
+      (rightsStatement \ "licenseInformation" \ "licenseNote").headOption
+
+    // Following discussion with a.ray here:
+    // https://wellcome.slack.com/archives/C02ANCYL90E/p1725026316294549
+    val rightsNote = rightsBasis match {
+      case Some(basis) if basis.text == "Copyright" => copyrightNoteElem
+      case Some(basis) if basis.text == "License"   => licenseNoteElem
+
+      // If we don't have a rightsBasis, pick from either copyright or
+      // license preferring copyright
+      case _ =>
+        warn(s"rightsBasis not found in rightsMD!")
+
+        List(
+          copyrightNoteElem,
+          licenseNoteElem
+        ).flatten.headOption
+    }
+
     val useRightsElem =
       ((rightsStatement \ "rightsGranted").filter(
         n => (n \ "act").text == "use"
       ) \ "rightsGrantedNote").headOption
+
     PremisAccessConditions(
-      copyrightNote = copyrightNoteElem.map(_.text),
+      copyrightNote = rightsNote.map(_.text),
       useRightsGrantedNote = useRightsElem.map(_.text)
     )
   }

diff --git a/...st/scala/weco/pipeline/transformer/mets/generators/PremisAccessConditionsGenerators.scala b/...st/scala/weco/pipeline/transformer/mets/generators/PremisAccessConditionsGenerators.scala
@@ -5,24 +5,21 @@ import scala.xml.Elem
 trait PremisAccessConditionsGenerators {
 
   def rightsMDWith(
-    copyrightInformation: Option[Elem] = None,
-    licenceInformation: Option[Elem] = None,
-    rightsGranted: Seq[Elem] = Nil
-  ): Elem = <mets:rightsMD ID="rightsMD_1">
+                    copyrightInformation: Option[Elem] = None,
+                    licenceInformation: Option[Elem] = None,
+                    rightsGranted: Seq[Elem] = Nil
+                  ): Elem = <mets:rightsMD ID="rightsMD_1">
     <mets:mdWrap MDTYPE="PREMIS:RIGHTS">
-    <mets:xmlData>
-      <premis:rightsStatement xmlns:premis="http://www.loc.gov/premis/v3" xsi:schemaLocation="http://www.loc.gov/premis/v3 http://www.loc.gov/standards/premis/v3/premis.xsd">
-        <premis:rightsStatementIdentifier>
-          <premis:rightsStatementIdentifierType>UUID</premis:rightsStatementIdentifierType>
-          <premis:rightsStatementIdentifierValue>3392668a-4503-462c-ba68-1d17c853f17c</premis:rightsStatementIdentifierValue>
-        </premis:rightsStatementIdentifier>
-        <premis:rightsBasis>???</premis:rightsBasis>
-        {copyrightInformation}
-        {licenceInformation}
-        {rightsGranted}
-      </premis:rightsStatement>
-    </mets:xmlData>
-  </mets:mdWrap>
+      <mets:xmlData>
+        <premis:rightsStatement xmlns:premis="http://www.loc.gov/premis/v3" xsi:schemaLocation="http://www.loc.gov/premis/v3 http://www.loc.gov/standards/premis/v3/premis.xsd">
+          <premis:rightsStatementIdentifier>
+            <premis:rightsStatementIdentifierType>UUID</premis:rightsStatementIdentifierType>
+            <premis:rightsStatementIdentifierValue>3392668a-4503-462c-ba68-1d17c853f17c</premis:rightsStatementIdentifierValue>
+          </premis:rightsStatementIdentifier>
+          <premis:rightsBasis>???</premis:rightsBasis>{copyrightInformation}{licenceInformation}{rightsGranted}
+        </premis:rightsStatement>
+      </mets:xmlData>
+    </mets:mdWrap>
   </mets:rightsMD>
 
   lazy val emptyRightsMD = rightsMDWith(None, None, Nil)
@@ -54,4 +51,64 @@ trait PremisAccessConditionsGenerators {
       </mets:xmlData>
     </mets:mdWrap>
   </mets:rightsMD>
+
+  val openCCBYNCRightsMD = <mets:rightsMD ID="rightsMD_1">
+    <mets:mdWrap MDTYPE="PREMIS:RIGHTS">
+      <mets:xmlData>
+        <premis:rightsStatement xmlns:premis="http://www.loc.gov/premis/v3" xsi:schemaLocation="http://www.loc.gov/premis/v3 http://www.loc.gov/standards/premis/v3/premis.xsd">
+          <premis:rightsStatementIdentifier>
+            <premis:rightsStatementIdentifierType>UUID</premis:rightsStatementIdentifierType>
+            <premis:rightsStatementIdentifierValue>a93acc1b-7978-4107-80f7-a4bdf5787a85</premis:rightsStatementIdentifierValue>
+          </premis:rightsStatementIdentifier>
+          <premis:rightsBasis>License</premis:rightsBasis>
+          <premis:licenseInformation>
+            <premis:licenseTerms/>
+            <premis:licenseNote>CC-BY-NC</premis:licenseNote>
+          </premis:licenseInformation>
+          <premis:rightsGranted>
+            <premis:act>use</premis:act>
+            <premis:rightsGrantedNote>Open</premis:rightsGrantedNote>
+          </premis:rightsGranted>
+          <premis:linkingObjectIdentifier>
+            <premis:linkingObjectIdentifierType>UUID</premis:linkingObjectIdentifierType>
+            <premis:linkingObjectIdentifierValue>433cbf46-4ce1-421d-bf4f-2bc13f9450a7</premis:linkingObjectIdentifierValue>
+          </premis:linkingObjectIdentifier>
+        </premis:rightsStatement>
+      </mets:xmlData>
+    </mets:mdWrap>
+  </mets:rightsMD>
+
+  // This should never happen in practice, as we should always have a rightsBasis
+  // but we fail gracefully if it does, preferring copyrightInformation
+  val openMixedCCBYNCWithCopyrightRightsMD = <mets:rightsMD ID="rightsMD_1">
+    <mets:mdWrap MDTYPE="PREMIS:RIGHTS">
+      <mets:xmlData>
+        <premis:rightsStatement xmlns:premis="http://www.loc.gov/premis/v3" xsi:schemaLocation="http://www.loc.gov/premis/v3 http://www.loc.gov/standards/premis/v3/premis.xsd">
+          <premis:rightsStatementIdentifier>
+            <premis:rightsStatementIdentifierType>UUID</premis:rightsStatementIdentifierType>
+            <premis:rightsStatementIdentifierValue>a93acc1b-7978-4107-80f7-a4bdf5787a85</premis:rightsStatementIdentifierValue>
+          </premis:rightsStatementIdentifier>
+          <premis:licenseInformation>
+            <premis:licenseTerms/>
+            <premis:licenseNote>CC-BY-NC</premis:licenseNote>
+          </premis:licenseInformation>
+          <premis:copyrightInformation>
+            <premis:copyrightStatus>copyrighted</premis:copyrightStatus>
+            <premis:copyrightJurisdiction>UK</premis:copyrightJurisdiction>
+            <premis:copyrightStatusDeterminationDate/>
+            <premis:copyrightNote>In copyright</premis:copyrightNote>
+          </premis:copyrightInformation>
+          <premis:rightsGranted>
+            <premis:act>use</premis:act>
+            <premis:rightsGrantedNote>Open</premis:rightsGrantedNote>
+          </premis:rightsGranted>
+          <premis:linkingObjectIdentifier>
+            <premis:linkingObjectIdentifierType>UUID</premis:linkingObjectIdentifierType>
+            <premis:linkingObjectIdentifierValue>433cbf46-4ce1-421d-bf4f-2bc13f9450a7</premis:linkingObjectIdentifierValue>
+          </premis:linkingObjectIdentifier>
+        </premis:rightsStatement>
+      </mets:xmlData>
+    </mets:mdWrap>
+  </mets:rightsMD>
 }
+
diff --git a/...c/test/scala/weco/pipeline/transformer/mets/transformers/PremisAccessConditionsTest.scala b/...c/test/scala/weco/pipeline/transformer/mets/transformers/PremisAccessConditionsTest.scala
@@ -19,12 +19,14 @@ class PremisAccessConditionsTest
           Some("Open")
         ).parse.right.get.accessStatus.get shouldBe AccessStatus.Open
       }
+
       it("translates the copyrightNote into a Licence") {
         PremisAccessConditions(
           Some("In Copyright"),
           None
         ).parse.right.get.licence.get shouldBe License.InCopyright
       }
+
       it("has no access conditions if none are given") {
         val conditions = PremisAccessConditions(
           None,
@@ -50,13 +52,28 @@ class PremisAccessConditionsTest
     }
 
     describe("extracting values from a rightsMD section") {
-      it("pulls out the copyrightNote for the licence") {
+      it("pulls out the copyrightNote where rightsBasis is 'Copyright'") {
         PremisAccessConditions(
           openInCopyrightRightsMD
         ).copyrightNote shouldBe Some(
           "In copyright"
         )
       }
+
+      it("pulls out the licenceNote where rightsBasis is 'License'") {
+        PremisAccessConditions(
+          openCCBYNCRightsMD
+        ).copyrightNote shouldBe Some("CC-BY-NC")
+      }
+
+      it("pulls out the copyrightNote where rightsBasis is not specified") {
+        PremisAccessConditions(
+          openMixedCCBYNCWithCopyrightRightsMD
+        ).copyrightNote shouldBe Some(
+          "In copyright"
+        )
+      }
+
       it("pulls out the rightsGrantedNote for the access status") {
         PremisAccessConditions(
           openInCopyrightRightsMD