Set enable/disable deduplication in hopefully the right place this time.

netarchivesuite · Nov 24, 2016 · 047fe3a · 047fe3a
1 parent 1b893ff
commit 047fe3a
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 3 deletions.
diff --git a/...cheduler/src/main/java/dk/netarkivet/harvester/scheduler/jobgen/AbstractJobGenerator.java b/...cheduler/src/main/java/dk/netarkivet/harvester/scheduler/jobgen/AbstractJobGenerator.java
@@ -207,7 +207,7 @@ public boolean canAccept(Job job, DomainConfiguration cfg, DomainConfiguration p
     protected void editJobOrderXml(Job job) {
         HeritrixTemplate doc = job.getOrderXMLdoc();
         doc.enableOrDisableDeduplication(DEDUPLICATION_ENABLED);
-        if (DEDUPLICATION_ENABLED) {
+/*        if (DEDUPLICATION_ENABLED) {
             // Check that the Deduplicator element is present in the
             // OrderXMl and enabled. If missing or disabled log a warning
         	
@@ -220,7 +220,7 @@ protected void editJobOrderXml(Job job) {
         	doc.removeDeduplicatorIfPresent();
         	job.setOrderXMLDoc(doc);
 	        log.info("Removed DeDuplicator element because Deduplication is disabled");   
-        }
+        }*/
     }
 
     /**

diff --git a/...er/harvester-core/src/main/java/dk/netarkivet/harvester/datamodel/H3HeritrixTemplate.java b/...er/harvester-core/src/main/java/dk/netarkivet/harvester/datamodel/H3HeritrixTemplate.java
@@ -531,7 +531,9 @@ public void removeDeduplicatorIfPresent() {
 	}
 
 	@Override public void enableOrDisableDeduplication(boolean enabled) {
-		template = template.replace(DEDUPLICATION_ENABLED_PLACEHOLDER, Boolean.toString(enabled).toLowerCase());
+		final String replacement = Boolean.toString(enabled).toLowerCase();
+		log.debug("Replacing deduplication enabled placeholder {} with {}.", DEDUPLICATION_ENABLED_PLACEHOLDER, replacement);
+		this.template = template.replace(DEDUPLICATION_ENABLED_PLACEHOLDER, replacement);
 	}
 
 	//<property name="metadataItems">

diff --git a/harvester/harvester-core/src/main/java/dk/netarkivet/harvester/datamodel/Job.java b/harvester/harvester-core/src/main/java/dk/netarkivet/harvester/datamodel/Job.java
@@ -282,6 +282,8 @@ public Job(Long harvestID, DomainConfiguration cfg, HeritrixTemplate orderXMLdoc
 
     	setAttributes(cfg.getAttributesAndTypes());
 
+        orderXMLdoc.enableOrDisableDeduplication(Settings.getBoolean(HarvesterSettings.DEDUPLICATION_ENABLED));
+
         status = JobStatus.NEW;
     }