public void resolveNames(Long batchSize) {
StopWatch watchForEntireRun = new StopWatch();
watchForEntireRun.start();
StopWatch watchForBatch = new StopWatch();
watchForBatch.start();
Long count = 0L;
Index<Node> studyIndex = graphService.index().forNodes("studies");
IndexHits<Node> studies = studyIndex.query("title", "*");
for (Node studyNode : studies) {
final Study study1 = new Study(studyNode);
final Iterable<Relationship> specimens = study1.getSpecimens();
for (Relationship collected : specimens) {
Specimen specimen = new Specimen(collected.getEndNode());
final Relationship classifiedAs =
specimen
.getUnderlyingNode()
.getSingleRelationship(RelTypes.CLASSIFIED_AS, Direction.OUTGOING);
if (classifiedAs == null) {
final Relationship describedAs =
specimen
.getUnderlyingNode()
.getSingleRelationship(RelTypes.ORIGINALLY_DESCRIBED_AS, Direction.OUTGOING);
final TaxonNode describedAsTaxon = new TaxonNode(describedAs.getEndNode());
try {
if (taxonFilter.shouldInclude(describedAsTaxon)) {
TaxonNode resolvedTaxon = taxonIndex.getOrCreateTaxon(describedAsTaxon);
if (resolvedTaxon != null) {
specimen.classifyAs(resolvedTaxon);
}
}
} catch (NodeFactoryException e) {
LOG.warn(
"failed to create taxon with name ["
+ describedAsTaxon.getName()
+ "] and id ["
+ describedAsTaxon.getExternalId()
+ "]",
e);
} finally {
count++;
if (count % batchSize == 0) {
watchForBatch.stop();
final long duration = watchForBatch.getTime();
if (duration > 0) {
LOG.info(
"resolved batch of ["
+ batchSize
+ "] names in "
+ getProgressMsg(batchSize, duration));
}
watchForBatch.reset();
watchForBatch.start();
}
}
}
}
}
studies.close();
watchForEntireRun.stop();
LOG.info(
"resolved [" + count + "] names in " + getProgressMsg(count, watchForEntireRun.getTime()));
}
