For ISSN loading, use the table id to iterate over the existing list …

…of issns and process them just once
ORCID · Jul 31, 2023 · 0ba9124 · 0ba9124
1 parent 771ad25
commit 0ba9124
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 16 deletions.
diff --git a/orcid-persistence/src/main/java/org/orcid/persistence/dao/GroupIdRecordDao.java b/orcid-persistence/src/main/java/org/orcid/persistence/dao/GroupIdRecordDao.java
@@ -1,6 +1,5 @@
 package org.orcid.persistence.dao;
 
-import java.util.Date;
 import java.util.List;
 
 import org.orcid.persistence.jpa.entities.GroupIdRecordEntity;
@@ -17,5 +16,5 @@ public interface GroupIdRecordDao extends GenericDao<GroupIdRecordEntity, Long>
 
     boolean duplicateExists(Long putCode, String groupId);
 
-    List<GroupIdRecordEntity> getIssnRecordsNotModifiedSince(int batchSize, Date start);
+    List<GroupIdRecordEntity> getIssnRecordsSortedById(int batchSize, long initialId);
 }
diff --git a/orcid-persistence/src/main/java/org/orcid/persistence/dao/impl/GroupIdRecordDaoImpl.java b/orcid-persistence/src/main/java/org/orcid/persistence/dao/impl/GroupIdRecordDaoImpl.java
@@ -1,6 +1,5 @@
 package org.orcid.persistence.dao.impl;
 
-import java.util.Date;
 import java.util.List;
 
 import javax.persistence.Query;
@@ -9,8 +8,8 @@
 import org.orcid.persistence.dao.GroupIdRecordDao;
 import org.orcid.persistence.jpa.entities.GroupIdRecordEntity;
 
-public class GroupIdRecordDaoImpl extends GenericDaoImpl<GroupIdRecordEntity, Long> implements GroupIdRecordDao {
-
+public class GroupIdRecordDaoImpl extends GenericDaoImpl<GroupIdRecordEntity, Long> implements GroupIdRecordDao {    
+    
     public GroupIdRecordDaoImpl() {
         super(GroupIdRecordEntity.class);
     }
@@ -71,13 +70,13 @@ public boolean duplicateExists(Long putCode, String groupId) {
         }
         Long result = query.getSingleResult();
         return (result != null && result > 0);
-    }
-
-    @SuppressWarnings("unchecked")
+    }   
+
     @Override
-    public List<GroupIdRecordEntity> getIssnRecordsNotModifiedSince(int pageSize, Date date) {
-        Query query = entityManager.createNativeQuery("SELECT * FROM group_id_record g LEFT OUTER JOIN invalid_issn_group_id_record p ON g.id = p.id where p.id IS NULL AND g.group_id like 'issn:%' and g.last_modified < :date", GroupIdRecordEntity.class);
-        query.setParameter("date", date);
+    public List<GroupIdRecordEntity> getIssnRecordsSortedById(int batchSize, long initialId) {
+        Query query = entityManager.createNativeQuery("SELECT * FROM group_id_record g LEFT OUTER JOIN invalid_issn_group_id_record p ON g.id = p.id where p.id IS NULL AND g.group_id like 'issn:%' and g.id > :initialId order by g.id", GroupIdRecordEntity.class);
+        query.setParameter("initialId", initialId);
+        query.setMaxResults(batchSize);        
         return query.getResultList();
     }
 }
diff --git a/orcid-scheduler-web/src/main/java/org/orcid/scheduler/loader/source/issn/IssnLoadSource.java b/orcid-scheduler-web/src/main/java/org/orcid/scheduler/loader/source/issn/IssnLoadSource.java
@@ -1,6 +1,5 @@
 package org.orcid.scheduler.loader.source.issn;
 
-import java.util.Date;
 import java.util.List;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
@@ -29,7 +28,7 @@ public class IssnLoadSource {
 
     private static Pattern issnGroupTypePattern = Pattern.compile("^issn:(\\d{4}-\\d{3}[\\dXx])$");
 
-    @Value("${org.orcid.scheduler.issnLoadSource.batchSize:50}")
+    @Value("${org.orcid.scheduler.issnLoadSource.batchSize:5000}")
     private int batchSize;
 
     @Value("${org.orcid.scheduler.issnLoadSource.waitBetweenBatches:30000}")
@@ -69,8 +68,10 @@ public void loadIssn(String issnSource) {
     }
 
     private void updateIssnGroupIdRecords() {
-        Date start = new Date();
-        List<GroupIdRecordEntity> issnEntities = groupIdRecordDaoReadOnly.getIssnRecordsNotModifiedSince(batchSize, start);
+        Long nextBatchStartId = 0L;
+        // Get the first batch of issn's
+        LOG.info("Loading batch of ISSN's, starting id: " + nextBatchStartId + " batch size: " + batchSize);
+        List<GroupIdRecordEntity> issnEntities = groupIdRecordDaoReadOnly.getIssnRecordsSortedById(batchSize, nextBatchStartId);
         int batchCount = 0;
         int total = 0;
         while (!issnEntities.isEmpty()) {
@@ -104,8 +105,13 @@ private void updateIssnGroupIdRecords() {
                     // TODO Auto-generated catch block
                     LOG.warn("Exception while pausing the issn loader", e);                    
                 }
+
+                if (issnEntity.getId() > nextBatchStartId) {
+                    nextBatchStartId = issnEntity.getId();
+                }
             }
-            issnEntities = groupIdRecordDaoReadOnly.getIssnRecordsNotModifiedSince(batchSize, start);
+            LOG.info("Loading batch of ISSN's, starting id: " + nextBatchStartId);
+            issnEntities = groupIdRecordDaoReadOnly.getIssnRecordsSortedById(batchSize, nextBatchStartId);
         }
     }