Merge pull request IQSS#5506 from QualitativeDataRepository/5505-Update_modifyRegistration_API_calls_to_only_update_when_necessary

IQSS-5505 - only update DOI metadata at PIDprovider when it changes

Author: landreev

pom.xml

@@ -550,6 +550,11 @@
             <artifactId>opennlp-tools</artifactId>
             <version>1.9.1</version>
         </dependency>
+        <dependency>
+            <groupId>org.xmlunit</groupId>
+            <artifactId>xmlunit-core</artifactId>
+            <version>2.9.1</version>
+        </dependency>
         <dependency>
           <groupId>com.google.cloud</groupId>
           <artifactId>google-cloud-storage</artifactId>

src/main/java/edu/harvard/iq/dataverse/api/Datasets.java

@@ -694,8 +694,9 @@ public Response updateDatasetPIDMetadata(@Context ContainerRequestContext crc, @
         }
 
         return response(req -> {
-            execCommand(new UpdateDvObjectPIDMetadataCommand(findDatasetOrDie(id), req));
-            List<String> args = Arrays.asList(id);
+            Dataset dataset = findDatasetOrDie(id);
+            execCommand(new UpdateDvObjectPIDMetadataCommand(dataset, req));
+            List<String> args = Arrays.asList(dataset.getIdentifier());
             return ok(BundleUtil.getStringFromBundle("datasets.api.updatePIDMetadata.success.for.single.dataset", args));
         }, getRequestUser(crc));
     }
@@ -707,7 +708,14 @@ public Response updateDatasetPIDMetadataAll(@Context ContainerRequestContext crc
         return response( req -> {
             datasetService.findAll().forEach( ds -> {
                 try {
+                    logger.fine("ReRegistering: " + ds.getId() + " : " + ds.getIdentifier());
+                    if (!ds.isReleased() || (!ds.isIdentifierRegistered() || (ds.getIdentifier() == null))) {
+                        if (ds.isReleased()) {
+                            logger.warning("Dataset id=" + ds.getId() + " is in an inconsistent state (publicationdate but no identifier/identifier not registered");
+                        }
+                    } else {
                     execCommand(new UpdateDvObjectPIDMetadataCommand(findDatasetOrDie(ds.getId().toString()), req));
+                    }
                 } catch (WrappedResponse ex) {
                     Logger.getLogger(Datasets.class.getName()).log(Level.SEVERE, null, ex);
                 }

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

@@ -50,7 +50,7 @@ protected void executeImpl(CommandContext ctxt) throws CommandException {
         PidProvider pidProvider = PidUtil.getPidProvider(target.getGlobalId().getProviderId());
 
         try {
-            Boolean doiRetString = pidProvider.publicizeIdentifier(target);
+            Boolean doiRetString = pidProvider.updateIdentifier(target);
             if (doiRetString) {
                 target.setGlobalIdCreateTime(new Timestamp(new Date().getTime()));
                 ctxt.em().merge(target);
@@ -71,7 +71,7 @@ protected void executeImpl(CommandContext ctxt) throws CommandException {
                             (!(df.getIdentifier() == null || df.getIdentifier().isEmpty()) || // identifier exists, or
                                      canCreatePidsForFiles) // we can create PIDs for files
                     ) {
-                        doiRetString = pidProvider.publicizeIdentifier(df);
+                        doiRetString = pidProvider.updateIdentifier(df);
                         if (doiRetString) {
                             df.setGlobalIdCreateTime(new Timestamp(new Date().getTime()));
                             ctxt.em().merge(df);

src/main/java/edu/harvard/iq/dataverse/pidproviders/AbstractPidProvider.java

@@ -547,4 +547,10 @@ public JsonObject getProviderSpecification() {
         providerSpecification.add("excludedSet", Strings.join(",", excludedSet.toArray()));
         return providerSpecification.build();
     }
+    
+    @Override
+    public boolean updateIdentifier(DvObject dvObject) {
+        //By default, these are the same
+        return publicizeIdentifier(dvObject);
+    }
 }

src/main/java/edu/harvard/iq/dataverse/pidproviders/PidProvider.java

@@ -54,6 +54,8 @@ public interface PidProvider {
 
     boolean publicizeIdentifier(DvObject studyIn);
 
+    boolean updateIdentifier(DvObject dvObject);
+    
     boolean isGlobalIdUnique(GlobalId globalId);
 
     String getUrlPrefix();

src/main/java/edu/harvard/iq/dataverse/pidproviders/doi/datacite/DOIDataCiteRegisterService.java

@@ -23,6 +23,12 @@
 import edu.harvard.iq.dataverse.pidproviders.AbstractPidProvider;
 import edu.harvard.iq.dataverse.pidproviders.doi.XmlMetadataTemplate;
 
+import org.xmlunit.builder.DiffBuilder;
+import org.xmlunit.builder.Input;
+import org.xmlunit.builder.Input.Builder;
+import org.xmlunit.diff.Diff;
+import org.xmlunit.diff.Difference;
+
 /**
  *
  * @author luopc
@@ -69,6 +75,35 @@ public String registerIdentifier(String identifier, Map<String, String> metadata
 
         return retString;
     }
+    
+    
+    public String reRegisterIdentifier(String identifier, Map<String, String> metadata, DvObject dvObject) throws IOException {
+        String retString = "";
+        String numericIdentifier = identifier.substring(identifier.indexOf(":") + 1);
+        String xmlMetadata = getMetadataFromDvObject(identifier, metadata, dvObject);
+        String target = metadata.get("_target");
+        String currentMetadata = client.getMetadata(numericIdentifier);
+        Diff myDiff = DiffBuilder.compare(xmlMetadata)
+                .withTest(currentMetadata).ignoreWhitespace().checkForSimilar()
+                .build();
+
+        if (myDiff.hasDifferences()) {
+            for(Difference d : myDiff.getDifferences()) {
+            
+              logger.fine(d.toString());
+            }
+            retString = "metadata:\\r" + client.postMetadata(xmlMetadata) + "\\r";
+        }
+        if (!target.equals(client.getUrl(numericIdentifier))) {
+            logger.info("Updating target URL to " +  target);
+            client.postUrl(numericIdentifier, target);
+            retString = retString + "url:\\r" + target;
+
+        }
+
+        return retString;
+    }
+
 
     public String deactivateIdentifier(String identifier, Map<String, String> metadata, DvObject dvObject) throws IOException {
         String retString = "";

src/main/java/edu/harvard/iq/dataverse/pidproviders/doi/datacite/DataCiteDOIProvider.java

@@ -313,7 +313,31 @@ String getPidStatus(DvObject dvObject) {
         return status;
     }
 
-
 
+    @Override
+    public boolean updateIdentifier(DvObject dvObject) {
+        logger.log(Level.FINE,"updateIdentifierStatus");
+        if(dvObject.getIdentifier() == null || dvObject.getIdentifier().isEmpty() ){
+            dvObject = generatePid(dvObject);
+        }
+        String identifier = getIdentifier(dvObject);
+        Map<String, String> metadata = getUpdateMetadata(dvObject);
+        metadata.put("_status", "public");
+        metadata.put("datacite.publicationyear", generateYear(dvObject));
+        metadata.put("_target", getTargetUrl(dvObject));
+        try {
+            String updated = doiDataCiteRegisterService.reRegisterIdentifier(identifier, metadata, dvObject);
+            if(updated.length()!=0) {
+                logger.info(identifier + "updated: " + updated );
+                return true;
+            } else {
+                logger.info("No updated needed for " + identifier);
+                return false; //No update needed
+            }
+        } catch (Exception e) {
+            logger.log(Level.WARNING, "updateIdentifier failed: " + e.getMessage(), e);
+            return false;
+        }
+    }
 
 }