use of com.datatorrent.api.Context.OperatorContext in project apex-malhar by apache.
the class JdbcPojoOperatorTest method testJdbcInputOperator.
@Test
public void testJdbcInputOperator() {
JdbcStore store = new JdbcStore();
store.setDatabaseDriver(DB_DRIVER);
store.setDatabaseUrl(URL);
com.datatorrent.api.Attribute.AttributeMap.DefaultAttributeMap attributeMap = new com.datatorrent.api.Attribute.AttributeMap.DefaultAttributeMap();
attributeMap.put(DAG.APPLICATION_ID, APP_ID);
OperatorContext context = mockOperatorContext(OPERATOR_ID, attributeMap);
TestInputOperator inputOperator = new TestInputOperator();
inputOperator.setStore(store);
insertEventsInTable(10);
CollectorTestSink<Object> sink = new CollectorTestSink<>();
inputOperator.outputPort.setSink(sink);
inputOperator.setup(context);
inputOperator.beginWindow(0);
inputOperator.emitTuples();
inputOperator.endWindow();
Assert.assertEquals("rows from db", 10, sink.collectedTuples.size());
}
use of com.datatorrent.api.Context.OperatorContext in project apex-malhar by apache.
the class DeduperTimeBasedPOJOImplTest method testDedup.
@Test
public void testDedup() {
com.datatorrent.api.Attribute.AttributeMap.DefaultAttributeMap attributes = new com.datatorrent.api.Attribute.AttributeMap.DefaultAttributeMap();
attributes.put(DAG.APPLICATION_ID, APP_ID);
attributes.put(DAG.APPLICATION_PATH, applicationPath);
attributes.put(DAG.InputPortMeta.TUPLE_CLASS, TestPojo.class);
OperatorContext context = mockOperatorContext(OPERATOR_ID, attributes);
deduper.setup(context);
deduper.input.setup(new PortContext(attributes, context));
deduper.activate(context);
CollectorTestSink<TestPojo> uniqueSink = new CollectorTestSink<TestPojo>();
TestUtils.setSink(deduper.unique, uniqueSink);
CollectorTestSink<TestPojo> duplicateSink = new CollectorTestSink<TestPojo>();
TestUtils.setSink(deduper.duplicate, duplicateSink);
CollectorTestSink<TestPojo> expiredSink = new CollectorTestSink<TestPojo>();
TestUtils.setSink(deduper.expired, expiredSink);
deduper.beginWindow(0);
long millis = System.currentTimeMillis();
for (int i = 0; i < 100; i++) {
TestPojo pojo = new TestPojo(i, new Date(millis + i));
deduper.input.process(pojo);
}
TestPojo expiredPojo = new TestPojo(100, new Date(millis - 1000 * 60));
deduper.input.process(expiredPojo);
for (int i = 90; i < 200; i++) {
TestPojo pojo = new TestPojo(i, new Date(millis + i));
deduper.input.process(pojo);
}
deduper.handleIdleTime();
deduper.endWindow();
Assert.assertTrue(uniqueSink.collectedTuples.size() == 200);
Assert.assertTrue(duplicateSink.collectedTuples.size() == 10);
Assert.assertTrue(expiredSink.collectedTuples.size() == 1);
deduper.teardown();
}
use of com.datatorrent.api.Context.OperatorContext in project apex-malhar by apache.
the class JdbcNonTransactionalBatchOutputOperatorTest method testAtLeastOnceHalfBatch.
@Test
public void testAtLeastOnceHalfBatch() {
JdbcOperatorTest.cleanTable();
Random random = new Random();
TestOutputOperator outputOperator = createOperator(ProcessingMode.AT_LEAST_ONCE);
outputOperator.beginWindow(0);
for (int batchCounter = 0; batchCounter < BATCH_SIZE; batchCounter++) {
outputOperator.input.put(new TestEvent(random.nextInt()));
}
outputOperator.endWindow();
Assert.assertEquals("Commit window id ", 0, outputOperator.getStore().getCommittedWindowId(APP_ID, OPERATOR_ID));
Assert.assertEquals("Batch should be written", BATCH_SIZE, outputOperator.getNumOfEventsInStore(outputOperator.getStore().connection));
outputOperator.beginWindow(1);
for (int batchCounter = 0; batchCounter < HALF_BATCH_SIZE; batchCounter++) {
outputOperator.input.put(new TestEvent(random.nextInt()));
}
Assert.assertEquals("Commit window id ", 0, outputOperator.getStore().getCommittedWindowId(APP_ID, OPERATOR_ID));
Assert.assertEquals("Batch should be written", BATCH_SIZE, outputOperator.getNumOfEventsInStore(outputOperator.getStore().connection));
outputOperator.getStore().disconnect();
// //
com.datatorrent.api.Attribute.AttributeMap.DefaultAttributeMap attributeMap = new com.datatorrent.api.Attribute.AttributeMap.DefaultAttributeMap();
attributeMap.put(OperatorContext.PROCESSING_MODE, ProcessingMode.AT_LEAST_ONCE);
attributeMap.put(OperatorContext.ACTIVATION_WINDOW_ID, 0L);
attributeMap.put(DAG.APPLICATION_ID, APP_ID);
OperatorContext context = mockOperatorContext(OPERATOR_ID, attributeMap);
outputOperator.setup(context);
Assert.assertEquals("Commit window id ", 0, outputOperator.getStore().getCommittedWindowId(APP_ID, OPERATOR_ID));
Assert.assertEquals("Batch should be written", BATCH_SIZE, outputOperator.getNumOfEventsInStore(outputOperator.getStore().connection));
outputOperator.beginWindow(0);
for (int batchCounter = 0; batchCounter < BATCH_SIZE; batchCounter++) {
outputOperator.input.put(new TestEvent(random.nextInt()));
}
outputOperator.endWindow();
Assert.assertEquals("Commit window id ", 0, outputOperator.getStore().getCommittedWindowId(APP_ID, OPERATOR_ID));
Assert.assertEquals("Batch should be written", BATCH_SIZE, outputOperator.getNumOfEventsInStore(outputOperator.getStore().connection));
outputOperator.beginWindow(1);
for (int batchCounter = 0; batchCounter < HALF_BATCH_SIZE; batchCounter++) {
outputOperator.input.put(new TestEvent(random.nextInt()));
}
outputOperator.endWindow();
Assert.assertEquals("Commit window id ", 1, outputOperator.getStore().getCommittedWindowId(APP_ID, OPERATOR_ID));
Assert.assertEquals("Batch should be written", 2 * BATCH_SIZE, outputOperator.getNumOfEventsInStore(outputOperator.getStore().connection));
}
use of com.datatorrent.api.Context.OperatorContext in project apex-malhar by apache.
the class JdbcPojoPollableOpeartorTest method testDBPoller.
@Test
public void testDBPoller() throws Exception {
insertEvents(10, true, 0);
JdbcStore store = new JdbcStore();
store.setDatabaseDriver(DB_DRIVER);
store.setDatabaseUrl(URL);
List<FieldInfo> fieldInfos = getFieldInfos();
Attribute.AttributeMap.DefaultAttributeMap portAttributes = new Attribute.AttributeMap.DefaultAttributeMap();
portAttributes.put(Context.PortContext.TUPLE_CLASS, TestPOJOEvent.class);
TestPortContext tpc = new TestPortContext(portAttributes);
JdbcPOJOPollInputOperator inputOperator = new JdbcPOJOPollInputOperator();
inputOperator.setStore(store);
inputOperator.setTableName(TABLE_POJO_NAME);
inputOperator.setKey("id");
inputOperator.setFieldInfos(fieldInfos);
inputOperator.setFetchSize(100);
inputOperator.setBatchSize(100);
inputOperator.setPartitionCount(2);
Collection<com.datatorrent.api.Partitioner.Partition<AbstractJdbcPollInputOperator<Object>>> newPartitions = inputOperator.definePartitions(new ArrayList<Partitioner.Partition<AbstractJdbcPollInputOperator<Object>>>(), null);
int operatorId = 0;
for (com.datatorrent.api.Partitioner.Partition<AbstractJdbcPollInputOperator<Object>> partition : newPartitions) {
Attribute.AttributeMap.DefaultAttributeMap partitionAttributeMap = new Attribute.AttributeMap.DefaultAttributeMap();
partitionAttributeMap.put(DAG.APPLICATION_ID, APP_ID);
partitionAttributeMap.put(Context.DAGContext.APPLICATION_PATH, dir);
OperatorContext partitioningContext = mockOperatorContext(operatorId++, partitionAttributeMap);
JdbcPOJOPollInputOperator parition = (JdbcPOJOPollInputOperator) partition.getPartitionedInstance();
parition.outputPort.setup(tpc);
parition.setScheduledExecutorService(mockscheduler);
parition.setup(partitioningContext);
parition.activate(partitioningContext);
}
Iterator<com.datatorrent.api.Partitioner.Partition<AbstractJdbcPollInputOperator<Object>>> itr = newPartitions.iterator();
// First partition is for range queries,last is for polling queries
JdbcPOJOPollInputOperator firstInstance = (JdbcPOJOPollInputOperator) itr.next().getPartitionedInstance();
CollectorTestSink<Object> sink1 = new CollectorTestSink<>();
firstInstance.outputPort.setSink(sink1);
firstInstance.beginWindow(0);
firstInstance.pollRecords();
try {
firstInstance.pollRecords();
// non-poller partition
Assert.fail("expected closed connection");
} catch (Exception e) {
// expected
}
firstInstance.emitTuples();
firstInstance.endWindow();
Assert.assertEquals("rows from db", 5, sink1.collectedTuples.size());
for (Object tuple : sink1.collectedTuples) {
TestPOJOEvent pojoEvent = (TestPOJOEvent) tuple;
Assert.assertTrue("date", pojoEvent.getStartDate() instanceof Date);
Assert.assertTrue("date", pojoEvent.getId() < 5);
}
JdbcPOJOPollInputOperator secondInstance = (JdbcPOJOPollInputOperator) itr.next().getPartitionedInstance();
CollectorTestSink<Object> sink2 = new CollectorTestSink<>();
secondInstance.outputPort.setSink(sink2);
secondInstance.beginWindow(0);
secondInstance.pollRecords();
secondInstance.emitTuples();
secondInstance.endWindow();
Assert.assertEquals("rows from db", 5, sink2.collectedTuples.size());
for (Object tuple : sink2.collectedTuples) {
TestPOJOEvent pojoEvent = (TestPOJOEvent) tuple;
Assert.assertTrue("date", pojoEvent.getId() < 10 && pojoEvent.getId() >= 5);
}
insertEvents(4, false, 10);
JdbcPOJOPollInputOperator thirdInstance = (JdbcPOJOPollInputOperator) itr.next().getPartitionedInstance();
CollectorTestSink<Object> sink3 = new CollectorTestSink<>();
thirdInstance.outputPort.setSink(sink3);
thirdInstance.beginWindow(0);
thirdInstance.pollRecords();
thirdInstance.pollRecords();
thirdInstance.emitTuples();
thirdInstance.endWindow();
Assert.assertEquals("rows from db", 4, sink3.collectedTuples.size());
}
use of com.datatorrent.api.Context.OperatorContext in project apex-malhar by apache.
the class JdbcPojoPollableOpeartorTest method testDBPollerExtraField.
@Test
public void testDBPollerExtraField() throws Exception {
insertEvents(10, true, 0);
JdbcStore store = new JdbcStore();
store.setDatabaseDriver(DB_DRIVER);
store.setDatabaseUrl(URL);
List<FieldInfo> fieldInfos = getFieldInfos();
Attribute.AttributeMap.DefaultAttributeMap portAttributes = new Attribute.AttributeMap.DefaultAttributeMap();
portAttributes.put(Context.PortContext.TUPLE_CLASS, TestPOJOEvent.class);
TestPortContext tpc = new TestPortContext(portAttributes);
JdbcPOJOPollInputOperator inputOperator = new JdbcPOJOPollInputOperator();
inputOperator.setStore(store);
inputOperator.setTableName(TABLE_POJO_NAME);
inputOperator.setColumnsExpression("ID,STARTDATE,STARTTIME,STARTTIMESTAMP");
inputOperator.setKey("id");
inputOperator.setFieldInfos(fieldInfos);
inputOperator.setFetchSize(100);
inputOperator.setBatchSize(100);
inputOperator.setPartitionCount(2);
Collection<com.datatorrent.api.Partitioner.Partition<AbstractJdbcPollInputOperator<Object>>> newPartitions = inputOperator.definePartitions(new ArrayList<Partitioner.Partition<AbstractJdbcPollInputOperator<Object>>>(), null);
int operatorId = 0;
for (com.datatorrent.api.Partitioner.Partition<AbstractJdbcPollInputOperator<Object>> partition : newPartitions) {
Attribute.AttributeMap.DefaultAttributeMap partitionAttributeMap = new Attribute.AttributeMap.DefaultAttributeMap();
partitionAttributeMap.put(DAG.APPLICATION_ID, APP_ID);
partitionAttributeMap.put(Context.DAGContext.APPLICATION_PATH, dir);
OperatorContext partitioningContext = mockOperatorContext(operatorId++, partitionAttributeMap);
JdbcPOJOPollInputOperator parition = (JdbcPOJOPollInputOperator) partition.getPartitionedInstance();
parition.outputPort.setup(tpc);
parition.setScheduledExecutorService(mockscheduler);
parition.setup(partitioningContext);
parition.activate(partitioningContext);
}
Iterator<com.datatorrent.api.Partitioner.Partition<AbstractJdbcPollInputOperator<Object>>> itr = newPartitions.iterator();
// First partition is for range queries,last is for polling queries
JdbcPOJOPollInputOperator firstInstance = (JdbcPOJOPollInputOperator) itr.next().getPartitionedInstance();
CollectorTestSink<Object> sink1 = new CollectorTestSink<>();
firstInstance.outputPort.setSink(sink1);
firstInstance.beginWindow(0);
Assert.assertFalse(firstInstance.ps.isClosed());
firstInstance.pollRecords();
Assert.assertTrue(firstInstance.ps.isClosed());
firstInstance.emitTuples();
firstInstance.endWindow();
Assert.assertEquals("rows from db", 5, sink1.collectedTuples.size());
for (Object tuple : sink1.collectedTuples) {
TestPOJOEvent pojoEvent = (TestPOJOEvent) tuple;
Assert.assertTrue("date", pojoEvent.getStartDate() instanceof Date);
Assert.assertTrue("date", pojoEvent.getId() < 5);
}
JdbcPOJOPollInputOperator secondInstance = (JdbcPOJOPollInputOperator) itr.next().getPartitionedInstance();
CollectorTestSink<Object> sink2 = new CollectorTestSink<>();
secondInstance.outputPort.setSink(sink2);
secondInstance.beginWindow(0);
secondInstance.pollRecords();
secondInstance.emitTuples();
secondInstance.endWindow();
Assert.assertEquals("rows from db", 5, sink2.collectedTuples.size());
for (Object tuple : sink2.collectedTuples) {
TestPOJOEvent pojoEvent = (TestPOJOEvent) tuple;
Assert.assertTrue("date", pojoEvent.getId() < 10 && pojoEvent.getId() >= 5);
}
insertEvents(4, false, 10);
JdbcPOJOPollInputOperator thirdInstance = (JdbcPOJOPollInputOperator) itr.next().getPartitionedInstance();
CollectorTestSink<Object> sink3 = new CollectorTestSink<>();
thirdInstance.outputPort.setSink(sink3);
thirdInstance.beginWindow(0);
thirdInstance.pollRecords();
thirdInstance.emitTuples();
thirdInstance.endWindow();
Assert.assertEquals("rows from db", 4, sink3.collectedTuples.size());
}
Aggregations