Search in sources :

Example 1 with KTableHolder

use of io.confluent.ksql.execution.plan.KTableHolder in project ksql by confluentinc.

the class TableGroupByBuilderBase method build.

public <K> KGroupedTableHolder build(final KTableHolder<K> table, final QueryContext queryContext, final Formats formats, final List<Expression> groupByExpressions) {
    final LogicalSchema sourceSchema = table.getSchema();
    final List<CompiledExpression> groupBy = CodeGenRunner.compileExpressions(groupByExpressions.stream(), "Group By", sourceSchema, buildContext.getKsqlConfig(), buildContext.getFunctionRegistry());
    final ProcessingLogger logger = buildContext.getProcessingLogger(queryContext);
    final GroupByParams params = paramsFactory.build(sourceSchema, groupBy, logger);
    final PhysicalSchema physicalSchema = PhysicalSchema.from(params.getSchema(), formats.getKeyFeatures(), formats.getValueFeatures());
    final Serde<GenericKey> keySerde = buildContext.buildKeySerde(formats.getKeyFormat(), physicalSchema, queryContext);
    final Serde<GenericRow> valSerde = buildContext.buildValueSerde(formats.getValueFormat(), physicalSchema, queryContext);
    final Grouped<GenericKey, GenericRow> grouped = groupedFactory.create(StreamsUtil.buildOpName(queryContext), keySerde, valSerde);
    final KGroupedTable<GenericKey, GenericRow> groupedTable = table.getTable().filter((k, v) -> v != null).groupBy(new TableKeyValueMapper<>(params.getMapper()), grouped);
    return KGroupedTableHolder.of(groupedTable, params.getSchema());
}
Also used : PhysicalSchema(io.confluent.ksql.schema.ksql.PhysicalSchema) KeyValueMapper(org.apache.kafka.streams.kstream.KeyValueMapper) RuntimeBuildContext(io.confluent.ksql.execution.runtime.RuntimeBuildContext) Expression(io.confluent.ksql.execution.expression.tree.Expression) QueryContext(io.confluent.ksql.execution.context.QueryContext) KeyValue(org.apache.kafka.streams.KeyValue) CompiledExpression(io.confluent.ksql.execution.codegen.CompiledExpression) Formats(io.confluent.ksql.execution.plan.Formats) LogicalSchema(io.confluent.ksql.schema.ksql.LogicalSchema) Function(java.util.function.Function) CodeGenRunner(io.confluent.ksql.execution.codegen.CodeGenRunner) Grouped(org.apache.kafka.streams.kstream.Grouped) KGroupedTableHolder(io.confluent.ksql.execution.plan.KGroupedTableHolder) KTableHolder(io.confluent.ksql.execution.plan.KTableHolder) List(java.util.List) KGroupedTable(org.apache.kafka.streams.kstream.KGroupedTable) GenericRow(io.confluent.ksql.GenericRow) Serde(org.apache.kafka.common.serialization.Serde) Objects.requireNonNull(java.util.Objects.requireNonNull) ProcessingLogger(io.confluent.ksql.logging.processing.ProcessingLogger) GenericKey(io.confluent.ksql.GenericKey) ProcessingLogger(io.confluent.ksql.logging.processing.ProcessingLogger) LogicalSchema(io.confluent.ksql.schema.ksql.LogicalSchema) CompiledExpression(io.confluent.ksql.execution.codegen.CompiledExpression) GenericRow(io.confluent.ksql.GenericRow) PhysicalSchema(io.confluent.ksql.schema.ksql.PhysicalSchema) GenericKey(io.confluent.ksql.GenericKey)

Example 2 with KTableHolder

use of io.confluent.ksql.execution.plan.KTableHolder in project ksql by confluentinc.

the class TableSelectBuilder method build.

@SuppressWarnings("unchecked")
public static <K> KTableHolder<K> build(final KTableHolder<K> table, final TableSelect<K> step, final RuntimeBuildContext buildContext, final Optional<Formats> formats, final MaterializedFactory materializedFactory) {
    final LogicalSchema sourceSchema = table.getSchema();
    final QueryContext queryContext = step.getProperties().getQueryContext();
    final Selection<K> selection = Selection.of(sourceSchema, step.getKeyColumnNames(), step.getSelectExpressions(), buildContext.getKsqlConfig(), buildContext.getFunctionRegistry());
    final SelectValueMapper<K> selectMapper = selection.getMapper();
    final ProcessingLogger logger = buildContext.getProcessingLogger(queryContext);
    final Named selectName = Named.as(StreamsUtil.buildOpName(queryContext));
    final Optional<MaterializationInfo.Builder> matBuilder = table.getMaterializationBuilder();
    final boolean forceMaterialize = !matBuilder.isPresent();
    final Serde<K> keySerde;
    final Serde<GenericRow> valSerde;
    if (formats.isPresent()) {
        final Formats materializationFormat = formats.get();
        final PhysicalSchema physicalSchema = PhysicalSchema.from(selection.getSchema(), materializationFormat.getKeyFeatures(), materializationFormat.getValueFeatures());
        keySerde = (Serde<K>) buildContext.buildKeySerde(materializationFormat.getKeyFormat(), physicalSchema, queryContext);
        valSerde = buildContext.buildValueSerde(materializationFormat.getValueFormat(), physicalSchema, queryContext);
        if (forceMaterialize) {
            final Stacker stacker = Stacker.of(step.getProperties().getQueryContext());
            final String stateStoreName = StreamsUtil.buildOpName(stacker.push(PROJECT_OP).getQueryContext());
            final Materialized<K, GenericRow, KeyValueStore<Bytes, byte[]>> materialized = materializedFactory.create(keySerde, valSerde, stateStoreName);
            final KTable<K, GenericRow> transFormedTable = table.getTable().transformValues(() -> new KsTransformer<>(selectMapper.getTransformer(logger)), materialized);
            return KTableHolder.materialized(transFormedTable, selection.getSchema(), table.getExecutionKeyFactory(), MaterializationInfo.builder(stateStoreName, selection.getSchema()));
        }
    } else {
        keySerde = null;
        valSerde = null;
    }
    final KTable<K, GenericRow> transFormedTable = table.getTable().transformValues(() -> new KsTransformer<>(selectMapper.getTransformer(logger)), Materialized.with(keySerde, valSerde), selectName);
    final Optional<MaterializationInfo.Builder> materialization = matBuilder.map(b -> b.map(pl -> (KsqlTransformer<Object, GenericRow>) selectMapper.getTransformer(pl), selection.getSchema(), queryContext));
    return table.withTable(transFormedTable, selection.getSchema()).withMaterialization(materialization);
}
Also used : TableSelect(io.confluent.ksql.execution.plan.TableSelect) PhysicalSchema(io.confluent.ksql.schema.ksql.PhysicalSchema) KTable(org.apache.kafka.streams.kstream.KTable) RuntimeBuildContext(io.confluent.ksql.execution.runtime.RuntimeBuildContext) KsqlTransformer(io.confluent.ksql.execution.transform.KsqlTransformer) QueryContext(io.confluent.ksql.execution.context.QueryContext) MaterializationInfo(io.confluent.ksql.execution.materialization.MaterializationInfo) Formats(io.confluent.ksql.execution.plan.Formats) LogicalSchema(io.confluent.ksql.schema.ksql.LogicalSchema) Bytes(org.apache.kafka.common.utils.Bytes) KTableHolder(io.confluent.ksql.execution.plan.KTableHolder) SelectValueMapper(io.confluent.ksql.execution.transform.select.SelectValueMapper) KsTransformer(io.confluent.ksql.execution.streams.transform.KsTransformer) GenericRow(io.confluent.ksql.GenericRow) Stacker(io.confluent.ksql.execution.context.QueryContext.Stacker) Serde(org.apache.kafka.common.serialization.Serde) Named(org.apache.kafka.streams.kstream.Named) KeyValueStore(org.apache.kafka.streams.state.KeyValueStore) ProcessingLogger(io.confluent.ksql.logging.processing.ProcessingLogger) Materialized(org.apache.kafka.streams.kstream.Materialized) Optional(java.util.Optional) Selection(io.confluent.ksql.execution.transform.select.Selection) ProcessingLogger(io.confluent.ksql.logging.processing.ProcessingLogger) Named(org.apache.kafka.streams.kstream.Named) Stacker(io.confluent.ksql.execution.context.QueryContext.Stacker) LogicalSchema(io.confluent.ksql.schema.ksql.LogicalSchema) KeyValueStore(org.apache.kafka.streams.state.KeyValueStore) QueryContext(io.confluent.ksql.execution.context.QueryContext) Formats(io.confluent.ksql.execution.plan.Formats) GenericRow(io.confluent.ksql.GenericRow) PhysicalSchema(io.confluent.ksql.schema.ksql.PhysicalSchema) KsqlTransformer(io.confluent.ksql.execution.transform.KsqlTransformer)

Example 3 with KTableHolder

use of io.confluent.ksql.execution.plan.KTableHolder in project ksql by confluentinc.

the class TableSuppressBuilder method build.

@VisibleForTesting
@SuppressWarnings("unchecked")
<K> KTableHolder<K> build(final KTableHolder<K> table, final TableSuppress<K> step, final RuntimeBuildContext buildContext, final ExecutionKeyFactory<K> executionKeyFactory, final PhysicalSchemaFactory physicalSchemaFactory, final BiFunction<Serde<K>, Serde<GenericRow>, Materialized> materializedFactory) {
    final PhysicalSchema physicalSchema = physicalSchemaFactory.create(table.getSchema(), step.getInternalFormats().getKeyFeatures(), step.getInternalFormats().getValueFeatures());
    final QueryContext queryContext = QueryContext.Stacker.of(step.getProperties().getQueryContext()).push(SUPPRESS_OP_NAME).getQueryContext();
    final Serde<K> keySerde = executionKeyFactory.buildKeySerde(step.getInternalFormats().getKeyFormat(), physicalSchema, queryContext);
    final Serde<GenericRow> valueSerde = buildContext.buildValueSerde(step.getInternalFormats().getValueFormat(), physicalSchema, queryContext);
    final Materialized<K, GenericRow, KeyValueStore<Bytes, byte[]>> materialized = materializedFactory.apply(keySerde, valueSerde);
    final Suppressed.StrictBufferConfig strictBufferConfig;
    final long maxBytes = buildContext.getKsqlConfig().getLong(KsqlConfig.KSQL_SUPPRESS_BUFFER_SIZE_BYTES);
    if (maxBytes < 0) {
        strictBufferConfig = Suppressed.BufferConfig.unbounded();
    } else {
        strictBufferConfig = Suppressed.BufferConfig.maxBytes(maxBytes).shutDownWhenFull();
    }
    /* This is a dummy transformValues() call, we do this to ensure that the correct materialized
    with the correct key and val serdes is passed on when we call suppress
     */
    final KTable<K, GenericRow> suppressed = table.getTable().transformValues((() -> new KsTransformer<>((k, v, ctx) -> v)), materialized).suppress((Suppressed<? super K>) Suppressed.untilWindowCloses(strictBufferConfig).withName(SUPPRESS_OP_NAME));
    return table.withTable(suppressed, table.getSchema());
}
Also used : PhysicalSchema(io.confluent.ksql.schema.ksql.PhysicalSchema) KTable(org.apache.kafka.streams.kstream.KTable) TableSuppress(io.confluent.ksql.execution.plan.TableSuppress) RuntimeBuildContext(io.confluent.ksql.execution.runtime.RuntimeBuildContext) QueryContext(io.confluent.ksql.execution.context.QueryContext) BiFunction(java.util.function.BiFunction) Suppressed(org.apache.kafka.streams.kstream.Suppressed) KsqlConfig(io.confluent.ksql.util.KsqlConfig) ExecutionKeyFactory(io.confluent.ksql.execution.plan.ExecutionKeyFactory) LogicalSchema(io.confluent.ksql.schema.ksql.LogicalSchema) Bytes(org.apache.kafka.common.utils.Bytes) KTableHolder(io.confluent.ksql.execution.plan.KTableHolder) KsTransformer(io.confluent.ksql.execution.streams.transform.KsTransformer) GenericRow(io.confluent.ksql.GenericRow) Serde(org.apache.kafka.common.serialization.Serde) KeyValueStore(org.apache.kafka.streams.state.KeyValueStore) Materialized(org.apache.kafka.streams.kstream.Materialized) VisibleForTesting(com.google.common.annotations.VisibleForTesting) SerdeFeatures(io.confluent.ksql.serde.SerdeFeatures) KeyValueStore(org.apache.kafka.streams.state.KeyValueStore) QueryContext(io.confluent.ksql.execution.context.QueryContext) GenericRow(io.confluent.ksql.GenericRow) PhysicalSchema(io.confluent.ksql.schema.ksql.PhysicalSchema) Suppressed(org.apache.kafka.streams.kstream.Suppressed) VisibleForTesting(com.google.common.annotations.VisibleForTesting)

Example 4 with KTableHolder

use of io.confluent.ksql.execution.plan.KTableHolder in project ksql by confluentinc.

the class TableFilterBuilder method build.

static <K> KTableHolder<K> build(final KTableHolder<K> table, final TableFilter<K> step, final RuntimeBuildContext buildContext, final SqlPredicateFactory sqlPredicateFactory) {
    final SqlPredicate predicate = sqlPredicateFactory.create(step.getFilterExpression(), table.getSchema(), buildContext.getKsqlConfig(), buildContext.getFunctionRegistry());
    final ProcessingLogger processingLogger = buildContext.getProcessingLogger(step.getProperties().getQueryContext());
    final Stacker stacker = Stacker.of(step.getProperties().getQueryContext());
    final KTable<K, GenericRow> filtered = table.getTable().transformValues(() -> new KsTransformer<>(predicate.getTransformer(processingLogger)), Named.as(StreamsUtil.buildOpName(stacker.push(PRE_PROCESS_OP).getQueryContext()))).filter((k, v) -> v.isPresent(), Named.as(StreamsUtil.buildOpName(stacker.push(FILTER_OP).getQueryContext()))).mapValues(Optional::get, Named.as(StreamsUtil.buildOpName(stacker.push(POST_PROCESS_OP).getQueryContext())));
    return table.withTable(filtered, table.getSchema()).withMaterialization(table.getMaterializationBuilder().map(b -> b.filter(predicate::getTransformer, step.getProperties().getQueryContext())));
}
Also used : GenericRow(io.confluent.ksql.GenericRow) SqlPredicate(io.confluent.ksql.execution.transform.sqlpredicate.SqlPredicate) KTableHolder(io.confluent.ksql.execution.plan.KTableHolder) TableFilter(io.confluent.ksql.execution.plan.TableFilter) KTable(org.apache.kafka.streams.kstream.KTable) RuntimeBuildContext(io.confluent.ksql.execution.runtime.RuntimeBuildContext) KsTransformer(io.confluent.ksql.execution.streams.transform.KsTransformer) GenericRow(io.confluent.ksql.GenericRow) Stacker(io.confluent.ksql.execution.context.QueryContext.Stacker) Named(org.apache.kafka.streams.kstream.Named) ProcessingLogger(io.confluent.ksql.logging.processing.ProcessingLogger) Optional(java.util.Optional) ProcessingLogger(io.confluent.ksql.logging.processing.ProcessingLogger) Optional(java.util.Optional) SqlPredicate(io.confluent.ksql.execution.transform.sqlpredicate.SqlPredicate) Stacker(io.confluent.ksql.execution.context.QueryContext.Stacker)

Example 5 with KTableHolder

use of io.confluent.ksql.execution.plan.KTableHolder in project ksql by confluentinc.

the class QueryBuilder method buildTransientQuery.

@SuppressWarnings("ParameterNumber")
TransientQueryMetadata buildTransientQuery(final String statementText, final QueryId queryId, final Set<SourceName> sources, final ExecutionStep<?> physicalPlan, final String planSummary, final LogicalSchema schema, final OptionalInt limit, final Optional<WindowInfo> windowInfo, final boolean excludeTombstones, final QueryMetadata.Listener listener, final StreamsBuilder streamsBuilder, final Optional<ImmutableMap<TopicPartition, Long>> endOffsets, final MetricCollectors metricCollectors) {
    final KsqlConfig ksqlConfig = config.getConfig(true);
    final String applicationId = QueryApplicationId.build(ksqlConfig, false, queryId);
    final RuntimeBuildContext runtimeBuildContext = buildContext(applicationId, queryId, streamsBuilder);
    final Map<String, Object> streamsProperties = buildStreamsProperties(applicationId, Optional.of(queryId), metricCollectors, config.getConfig(true), processingLogContext);
    final Object buildResult = buildQueryImplementation(physicalPlan, runtimeBuildContext);
    final TransientQueryQueue queue = buildTransientQueryQueue(buildResult, limit, excludeTombstones, endOffsets);
    final Topology topology = streamsBuilder.build(PropertiesUtil.asProperties(streamsProperties));
    final TransientQueryMetadata.ResultType resultType = buildResult instanceof KTableHolder ? windowInfo.isPresent() ? ResultType.WINDOWED_TABLE : ResultType.TABLE : ResultType.STREAM;
    return new TransientQueryMetadata(statementText, schema, sources, planSummary, queue, queryId, applicationId, topology, kafkaStreamsBuilder, streamsProperties, config.getOverrides(), ksqlConfig.getLong(KSQL_SHUTDOWN_TIMEOUT_MS_CONFIG), ksqlConfig.getInt(KsqlConfig.KSQL_QUERY_ERROR_MAX_QUEUE_SIZE), resultType, ksqlConfig.getLong(KsqlConfig.KSQL_QUERY_RETRY_BACKOFF_INITIAL_MS), ksqlConfig.getLong(KsqlConfig.KSQL_QUERY_RETRY_BACKOFF_MAX_MS), listener);
}
Also used : RuntimeBuildContext(io.confluent.ksql.execution.runtime.RuntimeBuildContext) KTableHolder(io.confluent.ksql.execution.plan.KTableHolder) KsqlConfig(io.confluent.ksql.util.KsqlConfig) Topology(org.apache.kafka.streams.Topology) NamedTopology(org.apache.kafka.streams.processor.internals.namedtopology.NamedTopology) TransientQueryMetadata(io.confluent.ksql.util.TransientQueryMetadata)

Aggregations

KTableHolder (io.confluent.ksql.execution.plan.KTableHolder)8 RuntimeBuildContext (io.confluent.ksql.execution.runtime.RuntimeBuildContext)5 GenericRow (io.confluent.ksql.GenericRow)4 QueryContext (io.confluent.ksql.execution.context.QueryContext)4 LogicalSchema (io.confluent.ksql.schema.ksql.LogicalSchema)4 GenericKey (io.confluent.ksql.GenericKey)3 Stacker (io.confluent.ksql.execution.context.QueryContext.Stacker)3 Formats (io.confluent.ksql.execution.plan.Formats)3 KsTransformer (io.confluent.ksql.execution.streams.transform.KsTransformer)3 ProcessingLogger (io.confluent.ksql.logging.processing.ProcessingLogger)3 PhysicalSchema (io.confluent.ksql.schema.ksql.PhysicalSchema)3 KsqlConfig (io.confluent.ksql.util.KsqlConfig)3 Serde (org.apache.kafka.common.serialization.Serde)3 KTable (org.apache.kafka.streams.kstream.KTable)3 KeyFormat (io.confluent.ksql.serde.KeyFormat)2 WindowInfo (io.confluent.ksql.serde.WindowInfo)2 Optional (java.util.Optional)2 Bytes (org.apache.kafka.common.utils.Bytes)2 Materialized (org.apache.kafka.streams.kstream.Materialized)2 Named (org.apache.kafka.streams.kstream.Named)2