/**
* 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");
}
}
}
}
/**
* Tests that:
*
* <ol>
* <li>The given domain configuration and job are not null.
* <li>The job does not already contain the given domain configuration.
* <li>The domain configuration has the same order xml name as the first inserted domain config.
* </ol>
*
* @param job a given Job
* @param cfg a given DomainConfiguration
* @return true, if the given DomainConfiguration can be inserted into the given job
*/
private boolean checkAddDomainConfInvariant(Job job, DomainConfiguration cfg) {
ArgumentNotValid.checkNotNull(job, "job");
ArgumentNotValid.checkNotNull(cfg, "cfg");
// check if domain in DomainConfiguration cfg is not already in this job
// domainName is used as key in domainConfigurationMap
if (job.getDomainConfigurationMap().containsKey(cfg.getDomainName())) {
if (log.isDebugEnabled()) {
log.debug("Job already has a configuration for Domain '" + cfg.getDomainName() + "'.");
}
return false;
}
// check if template is same as this job.
String orderXMLname = job.getOrderXMLName();
if (!orderXMLname.equals(cfg.getOrderXmlName())) {
if (log.isDebugEnabled()) {
log.debug(
"This Job only accept configurations "
+ "using the harvest template '"
+ orderXMLname
+ "'. This configuration uses the harvest template '"
+ cfg.getOrderXmlName()
+ "'.");
}
return false;
}
return true;
}
/**
* Instantiates a new job.
*
* @param cfg the {@link DomainConfiguration} being processed
* @param harvest the {@link HarvestDefinition} being processed
* @return an instance of {@link Job}
*/
public static Job getNewJob(HarvestDefinition harvest, DomainConfiguration cfg) {
HarvestChannelDAO harvestChannelDao = HarvestChannelDAO.getInstance();
HarvestChannel channel = harvestChannelDao.getChannelForHarvestDefinition(harvest.getOid());
if (channel == null) {
log.info(
"No channel mapping registered for harvest id "
+ harvest.getOid()
+ ", will use default.");
channel = harvestChannelDao.getDefaultChannel(harvest.isSnapShot());
}
if (harvest.isSnapShot()) {
return Job.createSnapShotJob(
harvest.getOid(),
channel,
cfg,
harvest.getMaxCountObjects(),
harvest.getMaxBytes(),
((FullHarvest) harvest).getMaxJobRunningTime(),
harvest.getNumEvents());
}
return Job.createJob(harvest.getOid(), channel, cfg, harvest.getNumEvents());
}