use of org.neo4j.unsafe.impl.batchimport.input.Collector in project neo4j by neo4j.
the class ParallelInputEntityDeserializerTest method shouldParseDataInParallel.
@Test
public void shouldParseDataInParallel() throws Exception {
// GIVEN
int entities = 500;
Data<InputNode> data = testData(entities);
Configuration config = new Configuration.Overridden(COMMAS) {
@Override
public int bufferSize() {
return 100;
}
};
IdType idType = ACTUAL;
Collector badCollector = mock(Collector.class);
Groups groups = new Groups();
Set<Thread> observedProcessingThreads = new CopyOnWriteArraySet<>();
int threads = 4;
DeserializerFactory<InputNode> deserializerFactory = (header, chunk, decorator, validator) -> {
observedProcessingThreads.add(Thread.currentThread());
// Make sure there will be 4 different processing threads doing this
boolean allThreadsStarted;
do {
allThreadsStarted = observedProcessingThreads.size() == threads;
} while (!allThreadsStarted);
return new InputEntityDeserializer<>(header, chunk, config.delimiter(), new InputNodeDeserialization(header, chunk, groups, idType.idsAreExternal()), decorator, validator, badCollector);
};
try (ParallelInputEntityDeserializer<InputNode> deserializer = new ParallelInputEntityDeserializer<>(data, defaultFormatNodeFileHeader(), config, idType, threads, threads, deserializerFactory, Validators.<InputNode>emptyValidator(), InputNode.class)) {
// WHEN/THEN
long previousLineNumber = -1;
long previousPosition = -1;
for (long i = 0; i < entities; i++) {
assertTrue(deserializer.hasNext());
InputNode entity = deserializer.next();
assertEquals(i, ((Long) entity.id()).longValue());
assertEquals("name", entity.properties()[0]);
assertTrue(entity.properties()[1].toString().startsWith(i + "-"));
assertTrue(entity.lineNumber() > previousLineNumber);
previousLineNumber = entity.lineNumber();
assertTrue(entity.position() > previousPosition);
previousPosition = entity.position();
}
assertFalse(deserializer.hasNext());
assertEquals(threads, observedProcessingThreads.size());
}
}
Aggregations