private void addLink(
Study study,
Map<Integer, Term> stageForNode,
Map<Integer, String> taxonForNode,
LabeledCSVParser links,
Location location)
throws StudyImporterException, NodeFactoryException {
Integer consumerNodeID = Integer.parseInt(links.getValueByLabel("ConsumerNodeID"));
Integer resourceNodeID = Integer.parseInt(links.getValueByLabel("ResourceNodeID"));
String linkType = links.getValueByLabel("LinkType");
InteractType interactType =
interactionMapping.get(StringUtils.trim(StringUtils.lowerCase(linkType)));
if (interactType == null) {
throw new StudyImporterException(
"failed to map interaction type ["
+ linkType
+ "] in line ["
+ links.lastLineNumber()
+ "]");
}
Specimen consumer = nodeFactory.createSpecimen(study, taxonForNode.get(consumerNodeID));
consumer.setLifeStage(stageForNode.get(consumerNodeID));
consumer.setExternalId(getNamespace() + ":NodeID:" + consumerNodeID);
consumer.caughtIn(location);
String resourceName = taxonForNode.get(resourceNodeID);
Specimen resource = nodeFactory.createSpecimen(study, resourceName);
resource.setLifeStage(stageForNode.get(resourceNodeID));
resource.setExternalId(getNamespace() + ":NodeID:" + resourceNodeID);
resource.caughtIn(location);
consumer.interactsWith(resource, interactType);
}
@Test
public void writeSiteMapWithNames() throws StudyImporterException, IOException {
final PropertyEnricher taxonEnricher =
new PropertyEnricher() {
@Override
public Map<String, String> enrich(Map<String, String> properties) {
Taxon taxon = new TaxonImpl();
TaxonUtil.mapToTaxon(properties, taxon);
if ("H**o sapiens".equals(taxon.getName())) {
taxon.setExternalId("homoSapiensId");
taxon.setPath("one two three");
} else if ("Canis lupus".equals(taxon.getName())) {
taxon.setExternalId("canisLupusId");
taxon.setPath("four five six");
}
return TaxonUtil.taxonToMap(taxon);
}
@Override
public void shutdown() {}
};
taxonIndex = ExportTestUtil.taxonIndexWithEnricher(taxonEnricher, getGraphDb());
Study study = nodeFactory.getOrCreateStudy("title", "source", "citation 123");
final Specimen human = nodeFactory.createSpecimen(study, "H**o sapiens");
final Specimen dog = nodeFactory.createSpecimen(study, "Canis familiaris");
human.ate(dog);
resolveNames();
final File baseDirNames = createBaseDir("target/sitemap/names");
final GraphExporter siteMapForNames = new ExporterSiteMapForNames();
siteMapForNames.export(getGraphDb(), baseDirNames.getAbsolutePath());
assertSiteMap(
baseDirNames,
"http://www.globalbioticinteractions.org/?interactionType=interactsWith&sourceTaxon=H**o%20sapiens",
"https://globi.s3.amazonaws.com/snapshot/target/data/sitemap/names/sitemap.xml.gz");
}
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()));
}