use of io.confluent.ksql.serde.KeyFormat in project ksql by confluentinc.
the class JoinNode method resolveKeyFormats.
/**
* Determines the key format of the join.
*
* <p>For now, the left key format is the preferred join key format unless the
* right source is not already being repartitioned and the left source is.
*
* @see <a href="https://github.com/confluentinc/ksql/blob/master/design-proposals/klip-33-key-format.md">KLIP-33</a>
*/
public void resolveKeyFormats() {
final KeyFormat joinKeyFormat = getPreferredKeyFormat().orElseGet(this::getDefaultSourceKeyFormat);
setKeyFormat(joinKeyFormat);
}
use of io.confluent.ksql.serde.KeyFormat in project ksql by confluentinc.
the class MetaStoreFixture method getNewMetaStore.
public static MutableMetaStore getNewMetaStore(final FunctionRegistry functionRegistry, final ValueFormat valueFormat) {
final MutableMetaStore metaStore = new MetaStoreImpl(functionRegistry);
final KeyFormat keyFormat = KeyFormat.nonWindowed(FormatInfo.of(FormatFactory.KAFKA.name()), SerdeFeatures.of());
final LogicalSchema test1Schema = LogicalSchema.builder().keyColumn(ColumnName.of("COL0"), SqlTypes.BIGINT).valueColumn(ColumnName.of("COL1"), SqlTypes.STRING).valueColumn(ColumnName.of("COL2"), SqlTypes.STRING).valueColumn(ColumnName.of("COL3"), SqlTypes.DOUBLE).valueColumn(ColumnName.of("COL4"), SqlTypes.array(SqlTypes.DOUBLE)).valueColumn(ColumnName.of("COL5"), SqlTypes.map(SqlTypes.STRING, SqlTypes.DOUBLE)).headerColumn(ColumnName.of("HEAD"), Optional.empty()).build();
final KsqlTopic ksqlTopic0 = new KsqlTopic("test0", keyFormat, valueFormat);
final KsqlStream<?> ksqlStream0 = new KsqlStream<>("sqlexpression", SourceName.of("TEST0"), test1Schema, Optional.empty(), false, ksqlTopic0, false);
metaStore.putSource(ksqlStream0, false);
final KsqlTopic ksqlTopic1 = new KsqlTopic("test1", keyFormat, valueFormat);
final KsqlStream<?> ksqlStream1 = new KsqlStream<>("sqlexpression", SourceName.of("TEST1"), test1Schema, Optional.empty(), false, ksqlTopic1, false);
metaStore.putSource(ksqlStream1, false);
final LogicalSchema test2Schema = LogicalSchema.builder().keyColumn(ColumnName.of("COL0"), SqlTypes.BIGINT).valueColumn(ColumnName.of("COL1"), SqlTypes.STRING).valueColumn(ColumnName.of("COL2"), SqlTypes.STRING).valueColumn(ColumnName.of("COL3"), SqlTypes.DOUBLE).valueColumn(ColumnName.of("COL4"), SqlTypes.BOOLEAN).build();
final KsqlTopic ksqlTopic2 = new KsqlTopic("test2", keyFormat, valueFormat);
final KsqlTable<String> ksqlTable = new KsqlTable<>("sqlexpression", SourceName.of("TEST2"), test2Schema, Optional.empty(), false, ksqlTopic2, false);
metaStore.putSource(ksqlTable, false);
final SqlType addressSchema = SqlTypes.struct().field("NUMBER", SqlTypes.BIGINT).field("STREET", SqlTypes.STRING).field("CITY", SqlTypes.STRING).field("STATE", SqlTypes.STRING).field("ZIPCODE", SqlTypes.BIGINT).build();
final SqlType categorySchema = SqlTypes.struct().field("ID", SqlTypes.BIGINT).field("NAME", SqlTypes.STRING).build();
final SqlType itemInfoSchema = SqlTypes.struct().field("ITEMID", SqlTypes.BIGINT).field("NAME", SqlTypes.STRING).field("CATEGORY", categorySchema).build();
final LogicalSchema ordersSchema = LogicalSchema.builder().keyColumn(ColumnName.of("ORDERTIME"), SqlTypes.BIGINT).valueColumn(ColumnName.of("ORDERID"), SqlTypes.BIGINT).valueColumn(ColumnName.of("ITEMID"), SqlTypes.STRING).valueColumn(ColumnName.of("ITEMINFO"), itemInfoSchema).valueColumn(ColumnName.of("ORDERUNITS"), SqlTypes.INTEGER).valueColumn(ColumnName.of("ARRAYCOL"), SqlTypes.array(SqlTypes.DOUBLE)).valueColumn(ColumnName.of("MAPCOL"), SqlTypes.map(SqlTypes.STRING, SqlTypes.DOUBLE)).valueColumn(ColumnName.of("ADDRESS"), addressSchema).valueColumn(ColumnName.of("TIMESTAMPCOL"), SqlTypes.TIMESTAMP).valueColumn(ColumnName.of("TIMECOL"), SqlTypes.TIME).valueColumn(ColumnName.of("DATECOL"), SqlTypes.DATE).valueColumn(ColumnName.of("BYTESCOL"), SqlTypes.BYTES).build();
final KsqlTopic ksqlTopicOrders = new KsqlTopic("orders_topic", keyFormat, valueFormat);
final KsqlStream<?> ksqlStreamOrders = new KsqlStream<>("sqlexpression", SourceName.of("ORDERS"), ordersSchema, Optional.empty(), false, ksqlTopicOrders, false);
metaStore.putSource(ksqlStreamOrders, false);
final LogicalSchema testTable3 = LogicalSchema.builder().keyColumn(ColumnName.of("COL0"), SqlTypes.BIGINT).valueColumn(ColumnName.of("COL1"), SqlTypes.STRING).valueColumn(ColumnName.of("COL2"), SqlTypes.STRING).valueColumn(ColumnName.of("COL3"), SqlTypes.DOUBLE).valueColumn(ColumnName.of("COL4"), SqlTypes.BOOLEAN).build();
final KsqlTopic ksqlTopic3 = new KsqlTopic("test3", keyFormat, valueFormat);
final KsqlTable<String> ksqlTable3 = new KsqlTable<>("sqlexpression", SourceName.of("TEST3"), testTable3, Optional.empty(), false, ksqlTopic3, false);
metaStore.putSource(ksqlTable3, false);
final SqlType nestedOrdersSchema = SqlTypes.struct().field("ORDERTIME", SqlTypes.BIGINT).field("ORDERID", SqlTypes.BIGINT).field("ITEMID", SqlTypes.STRING).field("ITEMINFO", itemInfoSchema).field("ORDERUNITS", SqlTypes.INTEGER).field("ARRAYCOL", SqlTypes.array(SqlTypes.DOUBLE)).field("MAPCOL", SqlTypes.map(SqlTypes.STRING, SqlTypes.DOUBLE)).field("ADDRESS", addressSchema).build();
final LogicalSchema nestedArrayStructMapSchema = LogicalSchema.builder().keyColumn(ColumnName.of("K"), SqlTypes.STRING).valueColumn(ColumnName.of("ARRAYCOL"), SqlTypes.array(itemInfoSchema)).valueColumn(ColumnName.of("MAPCOL"), SqlTypes.map(SqlTypes.STRING, itemInfoSchema)).valueColumn(ColumnName.of("NESTED_ORDER_COL"), nestedOrdersSchema).valueColumn(ColumnName.of("ITEM"), itemInfoSchema).build();
final KsqlTopic nestedArrayStructMapTopic = new KsqlTopic("NestedArrayStructMap_topic", keyFormat, valueFormat);
final KsqlStream<?> nestedArrayStructMapOrders = new KsqlStream<>("sqlexpression", SourceName.of("NESTED_STREAM"), nestedArrayStructMapSchema, Optional.empty(), false, nestedArrayStructMapTopic, false);
metaStore.putSource(nestedArrayStructMapOrders, false);
final KsqlTopic ksqlTopic4 = new KsqlTopic("test4", keyFormat, valueFormat);
final KsqlStream<?> ksqlStream4 = new KsqlStream<>("sqlexpression4", SourceName.of("TEST4"), test1Schema, Optional.empty(), false, ksqlTopic4, false);
metaStore.putSource(ksqlStream4, false);
final LogicalSchema sensorReadingsSchema = LogicalSchema.builder().keyColumn(ColumnName.of("ID"), SqlTypes.BIGINT).valueColumn(ColumnName.of("SENSOR_NAME"), SqlTypes.STRING).valueColumn(ColumnName.of("ARR1"), SqlTypes.array(SqlTypes.BIGINT)).valueColumn(ColumnName.of("ARR2"), SqlTypes.array(SqlTypes.STRING)).build();
final KsqlTopic ksqlTopicSensorReadings = new KsqlTopic("sensor_readings_topic", keyFormat, valueFormat);
final KsqlStream<?> ksqlStreamSensorReadings = new KsqlStream<>("sqlexpression", SourceName.of("SENSOR_READINGS"), sensorReadingsSchema, Optional.empty(), false, ksqlTopicSensorReadings, false);
metaStore.putSource(ksqlStreamSensorReadings, false);
final LogicalSchema testTable5 = LogicalSchema.builder().keyColumn(ColumnName.of("A"), SqlTypes.BOOLEAN).valueColumn(ColumnName.of("B"), SqlTypes.BOOLEAN).valueColumn(ColumnName.of("C"), SqlTypes.BOOLEAN).valueColumn(ColumnName.of("D"), SqlTypes.BOOLEAN).valueColumn(ColumnName.of("E"), SqlTypes.BOOLEAN).valueColumn(ColumnName.of("F"), SqlTypes.BOOLEAN).valueColumn(ColumnName.of("G"), SqlTypes.BOOLEAN).build();
final KsqlTopic ksqlTopic5 = new KsqlTopic("test5", keyFormat, valueFormat);
final KsqlTable<String> ksqlTable5 = new KsqlTable<>("sqlexpression", SourceName.of("TEST5"), testTable5, Optional.empty(), false, ksqlTopic5, false);
metaStore.putSource(ksqlTable5, false);
return metaStore;
}
use of io.confluent.ksql.serde.KeyFormat in project ksql by confluentinc.
the class ScalablePushRegistryTest method setUp.
@Before
public void setUp() {
when(ksqlTopic.getKafkaTopicName()).thenReturn(TOPIC);
when(kafkaConsumerFactory.create(any(), any(), any(), any(), any(), any())).thenReturn(kafkaConsumer);
catchupCoordinator = new TestCatchupCoordinator();
latestConsumer = new TestLatestConsumer(TOPIC, false, SCHEMA, kafkaConsumer, catchupCoordinator, assignment -> {
}, ksqlConfig, Clock.systemUTC());
latestConsumer2 = new TestLatestConsumer(TOPIC, false, SCHEMA, kafkaConsumer, catchupCoordinator, assignment -> {
}, ksqlConfig, Clock.systemUTC());
catchupConsumer = new TestCatchupConsumer(TOPIC, false, SCHEMA, kafkaConsumer, () -> latestConsumer, catchupCoordinator, pushOffsetRange, Clock.systemUTC(), pq -> {
});
when(latestConsumerFactory.create(any(), anyBoolean(), any(), any(), any(), any(), any(), any())).thenReturn(latestConsumer, latestConsumer2);
when(catchupConsumerFactory.create(any(), anyBoolean(), any(), any(), any(), any(), any(), any(), anyLong(), any())).thenReturn(catchupConsumer);
when(ksqlTopic.getKeyFormat()).thenReturn(keyFormat);
when(keyFormat.isWindowed()).thenReturn(false);
realExecutorService = Executors.newFixedThreadPool(2);
doAnswer(a -> {
final Runnable runnable = a.getArgument(0);
startLatestRunnable.set(runnable);
realExecutorService.submit(runnable);
return null;
}).when(executorService).submit(any(Runnable.class));
doAnswer(a -> {
final Runnable runnable = a.getArgument(0);
realExecutorService.submit(runnable);
return null;
}).when(catchupService).submit(any(Runnable.class));
when(processingQueue.getQueryId()).thenReturn(new QueryId("q1"));
when(processingQueue2.getQueryId()).thenReturn(new QueryId("q2"));
registry = new ScalablePushRegistry(locator, SCHEMA, false, ImmutableMap.of(), ksqlTopic, serviceContext, ksqlConfig, SOURCE_APP_ID, kafkaConsumerFactory, latestConsumerFactory, catchupConsumerFactory, executorService, catchupService);
when(ksqlConfig.getInt(KsqlConfig.KSQL_QUERY_PUSH_V2_MAX_CATCHUP_CONSUMERS)).thenReturn(10);
}
use of io.confluent.ksql.serde.KeyFormat in project ksql by confluentinc.
the class SchemaKGroupedTable method aggregate.
@Override
public SchemaKTable<GenericKey> aggregate(final List<ColumnName> nonAggregateColumns, final List<FunctionCall> aggregations, final Optional<WindowExpression> windowExpression, final FormatInfo valueFormat, final Stacker contextStacker) {
if (windowExpression.isPresent()) {
throw new KsqlException("Windowing not supported for table aggregations.");
}
final List<String> unsupportedFunctionNames = aggregations.stream().map(call -> UdafUtil.resolveAggregateFunction(functionRegistry, call, schema, ksqlConfig)).filter(function -> !(function instanceof TableAggregationFunction)).map(KsqlAggregateFunction::name).map(FunctionName::text).distinct().collect(Collectors.toList());
if (!unsupportedFunctionNames.isEmpty()) {
final String postfix = unsupportedFunctionNames.size() == 1 ? "" : "s";
throw new KsqlException("The aggregation function" + postfix + " " + GrammaticalJoiner.and().join(unsupportedFunctionNames) + " cannot be applied to a table source, only to a stream source.");
}
final TableAggregate step = ExecutionStepFactory.tableAggregate(contextStacker, sourceTableStep, InternalFormats.of(keyFormat, valueFormat), nonAggregateColumns, aggregations);
return new SchemaKTable<>(step, resolveSchema(step), keyFormat, ksqlConfig, functionRegistry);
}
use of io.confluent.ksql.serde.KeyFormat in project ksql by confluentinc.
the class SchemaKStream method groupBy.
public SchemaKGroupedStream groupBy(final FormatInfo valueFormat, final List<Expression> groupByExpressions, final Stacker contextStacker) {
if (!repartitionNeeded(groupByExpressions)) {
return groupByKey(keyFormat, valueFormat, contextStacker);
}
final KeyFormat rekeyedFormat;
if (keyFormat.getFormatInfo().getFormat().equals(NoneFormat.NAME)) {
final FormatInfo defaultFormat = FormatInfo.of(ksqlConfig.getString(KsqlConfig.KSQL_DEFAULT_KEY_FORMAT_CONFIG));
rekeyedFormat = KeyFormat.nonWindowed(defaultFormat, SerdeFeatures.of());
} else {
rekeyedFormat = keyFormat;
}
final KeyFormat sanitizedKeyFormat = SerdeFeaturesFactory.sanitizeKeyFormat(rekeyedFormat, toSqlTypes(groupByExpressions), true);
final StreamGroupBy<K> source = ExecutionStepFactory.streamGroupBy(contextStacker, sourceStep, InternalFormats.of(sanitizedKeyFormat, valueFormat), groupByExpressions);
return new SchemaKGroupedStream(source, resolveSchema(source), sanitizedKeyFormat, ksqlConfig, functionRegistry);
}
Aggregations