Merge branch 'EMC-399-schema.org-deduplication' into 'develop'

Make schema.org output group organisations for responsible parties See merge request eip/catalogue!810
NERC-CEH · Jan 31, 2025 · fc1db3e · fc1db3e
2 parents a5857e7 + 127d8df
commit fc1db3e
Show file tree

Hide file tree

Showing 3 changed files with 37 additions and 4 deletions.
diff --git a/java/src/main/java/uk/ac/ceh/gateway/catalogue/gemini/GeminiDocument.java b/java/src/main/java/uk/ac/ceh/gateway/catalogue/gemini/GeminiDocument.java
@@ -337,6 +337,23 @@ public List<OnlineResource> getInfoLinks() {
             .collect(Collectors.toCollection(ArrayList::new));
     }
 
+    @JsonIgnore
+    public List<ResponsibleParty> getAuthorPointOfContactWithRORs() {
+        val seenRORs = new HashSet<String>();
+        return getResponsibleParties()
+            .stream()
+            .filter(party -> {
+                val role = party.getRole();
+                val authorOrPOC = role.equalsIgnoreCase("author") || role.equalsIgnoreCase("pointOfContact");
+                if (!authorOrPOC || !party.isRor()) return false;
+                val ror = party.getOrganisationIdentifier();
+                if (seenRORs.contains(ror)) return false;
+                seenRORs.add(ror);
+                return true;
+            })
+            .toList();
+    }
+
     private static @NonNull String convertEmail(@NonNull String email) {
         return email.endsWith("@ceh.ac.uk") ? "enquiries@ceh.ac.uk" : email;
     }

diff --git a/java/src/main/java/uk/ac/ceh/gateway/catalogue/model/ResponsibleParty.java b/java/src/main/java/uk/ac/ceh/gateway/catalogue/model/ResponsibleParty.java
@@ -40,17 +40,17 @@ private ResponsibleParty(
 
     @JsonIgnore
     public boolean isOrcid() {
-        return nameIdentifier.matches("^http(|s)://orcid.org/\\d{4}-\\d{4}-\\d{4}-\\d{3}(X|\\d)$");
+        return nameIdentifier.matches("^https?://orcid\\.org/\\d{4}-\\d{4}-\\d{4}-\\d{3}(X|\\d)$");
     }
 
     @JsonIgnore
     public boolean isIsni() {
-        return nameIdentifier.matches("^https://isni.org/isni/\\d{15}(X|\\d)$");
+        return nameIdentifier.matches("^https://isni\\.org/isni/\\d{15}(X|\\d)$");
     }
 
     @JsonIgnore
     public boolean isRor() {
-        return organisationIdentifier.matches("^https://ror.org/\\w{8,10}$");
+        return organisationIdentifier.matches("^https://ror\\.org/\\w{8,10}$");
     }
 
     public String getRoleDisplayName() {

diff --git a/templates/schema.org/macros.ftl b/templates/schema.org/macros.ftl
@@ -22,7 +22,7 @@
     {
       "@context":"http://schema.org/",
       "@graph": [
-        <@m.schemaDocument docType parts/>
+        <@schemaDocument docType parts/>
       ]
     }
   </#if>
@@ -83,6 +83,7 @@
   <#if funding?has_content>,<@fundDetails/></#if>
   <#if parts?has_content>,<@partDetails parts/></#if>
   <#if downloads?has_content>,<@distributionDetails/></#if>
+  <#if authorPointOfContactWithRORs?has_content>,<@organisationRORs/></#if>
 </#macro>
 
 <#macro alternateTitlesList>
@@ -322,11 +323,15 @@
           </#if>
           <#if contact.organisationName?has_content>
             ,"affiliation":{
+              <#if contact.organisationIdentifier?matches("^https://ror\\.org/\\w{8,10}$")>
+              "@id": "${contact.organisationIdentifier}"
+              <#else>
               "@type":"Organization",
               "name":"${contact.organisationName}"
               <#if contact.organisationIdentifier?has_content>
               ,"identifier":"${contact.organisationIdentifier}"
               </#if>
+              </#if>
             }
           </#if>
         <#else>
@@ -405,3 +410,14 @@
     ],
   </#if>
 </#macro>
+
+<#macro organisationRORs>
+  <#list authorPointOfContactWithRORs as contact>
+  {
+    "@id": "${contact.organisationIdentifier}",
+    "@type": "Organization",
+    "name": "${contact.organisationName}",
+    "identifier": "${contact.organisationIdentifier}"
+  }<#sep>,
+  </#list>
+</#macro>