/**
* Once the job has been filled with {@link DomainConfiguration}s, performs the following
* operations:
*
* <ol>
* <li>Edit the harvest template to add/remove deduplicator configuration.
* <li>
* </ol>
*
* @param job the job
*/
protected void editJobOrderXml(Job job) {
Document doc = job.getOrderXMLdoc();
if (DEDUPLICATION_ENABLED) {
// Check that the Deduplicator element is present in the
// OrderXMl and enabled. If missing or disabled log a warning
if (!HeritrixTemplate.isDeduplicationEnabledInTemplate(doc)) {
if (log.isWarnEnabled()) {
log.warn(
"Unable to perform deduplication for this job"
+ " as the required DeDuplicator element is "
+ "disabled or missing from template");
}
}
} else {
// Remove deduplicator Element from OrderXML if present
Node xpathNode = doc.selectSingleNode(HeritrixTemplate.DEDUPLICATOR_XPATH);
if (xpathNode != null) {
xpathNode.detach();
job.setOrderXMLDoc(doc);
if (log.isInfoEnabled()) {
log.info("Removed DeDuplicator element because " + "Deduplication is disabled");
}
}
}
}
