Additional changes needed for the optimized "embargo publication date…

…" aggregate. #9763

src/main/java/edu/harvard/iq/dataverse/Dataset.java

@@ -692,15 +692,19 @@ public Timestamp getCitationDate() {
         Timestamp citationDate = null;
         //Only calculate if this dataset doesn't use an alternate date field for publication date
         if (citationDateDatasetFieldType == null) {
+            // @todo: remove this commented-out code once/if the PR passes review - L.A. 
             //List<DatasetVersion> versions = this.versions;
             // TODo - is this ever not version 1.0 (or draft if not published yet)
             //DatasetVersion oldest = versions.get(versions.size() - 1);
+            // - I believe the answer is yes, the oldest versions will always be 
+            // either 1.0 or draft - L.A. 
             citationDate = super.getPublicationDate();
             if (embargoCitationDate != null) {
                 if (citationDate.compareTo(embargoCitationDate) < 0) {
                     return embargoCitationDate;
                 }
             }
+            // @todo: remove this commented-out code once/if the PR passes review - L.A. 
             /*if (oldest.isPublished()) {
                 List<FileMetadata> fms = oldest.getFileMetadatas();
                 for (FileMetadata fm : fms) {

src/main/java/edu/harvard/iq/dataverse/engine/command/impl/FinalizeDatasetPublicationCommand.java

@@ -10,6 +10,7 @@
 import edu.harvard.iq.dataverse.DatasetVersionUser;
 import edu.harvard.iq.dataverse.Dataverse;
 import edu.harvard.iq.dataverse.DvObject;
+import edu.harvard.iq.dataverse.Embargo;
 import edu.harvard.iq.dataverse.UserNotification;
 import edu.harvard.iq.dataverse.authorization.Permission;
 import edu.harvard.iq.dataverse.authorization.users.AuthenticatedUser;
@@ -117,9 +118,37 @@ public Dataset execute(CommandContext ctxt) throws CommandException {
         // is this the first publication of the dataset?
         if (theDataset.getPublicationDate() == null) {
             theDataset.setReleaseUser((AuthenticatedUser) getUser());
-        }
-        if ( theDataset.getPublicationDate() == null ) {
+
             theDataset.setPublicationDate(new Timestamp(new Date().getTime()));
+
+            // if there are any embargoed files in this version, we will save 
+            // the latest availability date as the "embargoCitationDate" for future 
+            // reference (if the files are not available yet, as of publishing of 
+            // the dataset, this date will be used as the "ciatation date" of the dataset, 
+            // instead of the publicatonDate, in compliance with the DataCite 
+            // best practices). 
+            // the code below replicates the logic that used to be in the method 
+            // Dataset.getCitationDate() that calculated this adjusted date in real time.
+
+            Timestamp latestEmbargoDate = null; 
+            for (DataFile dataFile : theDataset.getFiles()) {
+                // this is the first version of the dataset that is being published. 
+                // therefore we can iterate through .getFiles() instead of obtaining
+                // the DataFiles by going through the FileMetadatas in the current version.
+                Embargo embargo = dataFile.getEmbargo();
+                if (embargo != null) {
+                    // "dataAvailable" is not nullable in the Embargo class, no need for a null check
+                    Timestamp embargoDate = Timestamp.valueOf(embargo.getDateAvailable().atStartOfDay());
+                    if (latestEmbargoDate == null || latestEmbargoDate.compareTo(embargoDate) < 0) {
+                        latestEmbargoDate = embargoDate;
+                    }
+                }
+            }
+            // the above loop could be easily replaced with a database query; 
+            // but we iterate through .getFiles() elsewhere in the command, when 
+            // updating and/or registering the files, so it should not result in 
+            // an extra performance hit. 
+            theDataset.setEmbargoCitationDate(latestEmbargoDate);
         } 
 
         //Clear any external status

src/main/resources/db/migration/V6.0.0.1__9763-embargocitationdate.sql

@@ -0,0 +1,14 @@
+-- An aggregated timestamp which is the latest of the availability dates of any embargoed files in the first published version, if present 
+ALTER TABLE dataset ADD COLUMN IF NOT EXISTS embargoCitationDate timestamp without time zone;
+-- ... and an update query that will populate this column for all the published datasets with embargoed files in the first released version:
+UPDATE dataset SET embargocitationdate=o.embargocitationdate
+FROM (SELECT d.id, MAX(e.dateavailable) AS embargocitationdate
+FROM embargo e, dataset d, datafile f, datasetversion v, filemetadata m
+WHERE v.dataset_id = d.id
+AND v.versionstate = 'RELEASED'
+AND v.versionnumber = 1
+AND v.minorversionnumber = 0
+AND f.embargo_id = e.id
+AND m.datasetversion_id = v.id
+AND m.datafile_id = f.id GROUP BY d.id) o WHERE o.id = dataset.id;
+-- (the query follows the logic that used to be in the method Dataset.getCitationDate() that calculated this adjusted date in real time).