bvarghese1 commented on code in PR #25753: URL: https://github.com/apache/flink/pull/25753#discussion_r1964080867
########## flink-table/flink-table-runtime/src/main/java/org/apache/flink/table/runtime/operators/over/NonTimeRangeUnboundedPrecedingFunction.java: ########## @@ -0,0 +1,596 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.flink.table.runtime.operators.over; + +import org.apache.flink.api.common.functions.OpenContext; +import org.apache.flink.api.common.state.MapState; +import org.apache.flink.api.common.state.MapStateDescriptor; +import org.apache.flink.api.common.state.ValueState; +import org.apache.flink.api.common.state.ValueStateDescriptor; +import org.apache.flink.api.common.typeinfo.Types; +import org.apache.flink.api.java.functions.KeySelector; +import org.apache.flink.api.java.tuple.Tuple2; +import org.apache.flink.api.java.typeutils.ListTypeInfo; +import org.apache.flink.api.java.typeutils.TupleTypeInfo; +import org.apache.flink.streaming.api.functions.KeyedProcessFunction; +import org.apache.flink.table.data.RowData; +import org.apache.flink.table.data.utils.JoinedRowData; +import org.apache.flink.table.runtime.dataview.PerKeyStateDataViewStore; +import org.apache.flink.table.runtime.functions.KeyedProcessFunctionWithCleanupState; +import org.apache.flink.table.runtime.generated.AggsHandleFunction; +import org.apache.flink.table.runtime.generated.GeneratedAggsHandleFunction; +import org.apache.flink.table.runtime.generated.GeneratedRecordComparator; +import org.apache.flink.table.runtime.generated.GeneratedRecordEqualiser; +import org.apache.flink.table.runtime.generated.RecordEqualiser; +import org.apache.flink.table.runtime.keyselector.RowDataKeySelector; +import org.apache.flink.table.runtime.typeutils.InternalTypeInfo; +import org.apache.flink.table.types.logical.LogicalType; +import org.apache.flink.types.RowKind; +import org.apache.flink.util.Collector; + +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import java.io.IOException; +import java.util.ArrayList; +import java.util.Comparator; +import java.util.List; + +/** A basic implementation to support non-time range unbounded over aggregate with retract mode. */ +public class NonTimeRangeUnboundedPrecedingFunction<K> + extends KeyedProcessFunctionWithCleanupState<K, RowData, RowData> { + private static final long serialVersionUID = 1L; + + private static final Logger LOG = + LoggerFactory.getLogger(NonTimeRangeUnboundedPrecedingFunction.class); + + private final GeneratedAggsHandleFunction generatedAggsHandler; + private final GeneratedRecordEqualiser generatedRecordEqualiser; + private final GeneratedRecordEqualiser generatedSortKeyEqualiser; + private final GeneratedRecordComparator generatedRecordComparator; + + // The util to compare two rows based on the sort attribute. + private transient Comparator<RowData> sortKeyComparator; + + protected final KeySelector<RowData, RowData> sortKeySelector; + // The record equaliser used to equal RowData. + private transient RecordEqualiser valueEqualiser; + private transient RecordEqualiser sortKeyEqualiser; + + private final LogicalType[] accTypes; + private final LogicalType[] inputFieldTypes; + private final LogicalType[] sortKeyTypes; + protected transient JoinedRowData output; + + // state to hold the Long ID counter + private transient ValueState<Long> idState; + + // state to hold a sorted list each containing a tuple of sort key with an artificial id + // The artificial id acts as the key in the valueMapState + private transient ValueState<List<Tuple2<RowData, List<Long>>>> sortedListState; + // state to hold rows until state ttl expires + private transient MapState<Long, RowData> valueMapState; + // state to hold sortKey and its associated accumulator + private transient MapState<RowData, RowData> accMapState; + + protected transient AggsHandleFunction currFunction; + + public NonTimeRangeUnboundedPrecedingFunction( + long minRetentionTime, + long maxRetentionTime, + GeneratedAggsHandleFunction genAggsHandler, + GeneratedRecordEqualiser genRecordEqualiser, + GeneratedRecordEqualiser genSortKeyEqualiser, + GeneratedRecordComparator genRecordComparator, + LogicalType[] accTypes, + LogicalType[] inputFieldTypes, + LogicalType[] sortKeyTypes, + RowDataKeySelector sortKeySelector) { + super(minRetentionTime, maxRetentionTime); + this.generatedAggsHandler = genAggsHandler; + this.generatedRecordEqualiser = genRecordEqualiser; + this.generatedSortKeyEqualiser = genSortKeyEqualiser; + this.generatedRecordComparator = genRecordComparator; + this.accTypes = accTypes; + this.inputFieldTypes = inputFieldTypes; + this.sortKeyTypes = sortKeyTypes; + this.sortKeySelector = sortKeySelector; + } + + @Override + public void open(OpenContext openContext) throws Exception { + // Initialize agg functions + currFunction = + generatedAggsHandler.newInstance(getRuntimeContext().getUserCodeClassLoader()); + currFunction.open(new PerKeyStateDataViewStore(getRuntimeContext())); + + // Initialize output record + output = new JoinedRowData(); + + // Initialize value/row equaliser + valueEqualiser = + generatedRecordEqualiser.newInstance(getRuntimeContext().getUserCodeClassLoader()); + + // Initialize sortKey equaliser + sortKeyEqualiser = + generatedSortKeyEqualiser.newInstance(getRuntimeContext().getUserCodeClassLoader()); + + // Initialize sort comparator + sortKeyComparator = + generatedRecordComparator.newInstance(getRuntimeContext().getUserCodeClassLoader()); + + // Initialize state to maintain id counter + ValueStateDescriptor<Long> idStateDescriptor = + new ValueStateDescriptor<Long>("idState", Long.class); + idState = getRuntimeContext().getState(idStateDescriptor); + + // Input elements are all binary rows as they came from network + InternalTypeInfo<RowData> inputRowTypeInfo = InternalTypeInfo.ofFields(inputFieldTypes); + InternalTypeInfo<RowData> sortKeyRowTypeInfo = InternalTypeInfo.ofFields(this.sortKeyTypes); + + // Initialize state which maintains a sorted list of tuples(sortKey, List of IDs) + ListTypeInfo<Long> idListTypeInfo = new ListTypeInfo<Long>(Types.LONG); + ListTypeInfo<Tuple2<RowData, List<Long>>> listTypeInfo = + new ListTypeInfo<>(new TupleTypeInfo<>(sortKeyRowTypeInfo, idListTypeInfo)); + ValueStateDescriptor<List<Tuple2<RowData, List<Long>>>> sortedListStateDescriptor = + new ValueStateDescriptor<List<Tuple2<RowData, List<Long>>>>( + "sortedListState", listTypeInfo); + sortedListState = getRuntimeContext().getState(sortedListStateDescriptor); + + // Initialize state which maintains the actual row + MapStateDescriptor<Long, RowData> valueStateDescriptor = + new MapStateDescriptor<Long, RowData>( + "valueMapState", Types.LONG, inputRowTypeInfo); + valueMapState = getRuntimeContext().getMapState(valueStateDescriptor); + + // Initialize accumulator state per row + InternalTypeInfo<RowData> accTypeInfo = InternalTypeInfo.ofFields(accTypes); + MapStateDescriptor<RowData, RowData> accStateDescriptor = + new MapStateDescriptor<RowData, RowData>( + "accMapState", sortKeyRowTypeInfo, accTypeInfo); + accMapState = getRuntimeContext().getMapState(accStateDescriptor); + + initCleanupTimeState("NonTimeUnboundedPrecedingFunctionCleanupTime"); + } + + /** + * Puts an element from the input stream into state. Emits the aggregated value for the newly + * inserted element. For append stream emits updates(UB, UA) for all elements which are present + * after the newly inserted element. For retract stream emits a DELETE for the element and + * thereafter emits updates(UB, UA) for all elements which are present after the retracted + * element. Emits the same aggregated value for all elements with the same sortKey to comply + * with the sql RANGE syntax. + * + * @param input The input value. + * @param ctx A {@link Context} that allows querying the timestamp of the element and getting + * TimerService for registering timers and querying the time. The context is only valid + * during the invocation of this method, do not store it. + * @param out The collector for returning result values. + * @throws Exception + */ + @Override + public void processElement( + RowData input, + KeyedProcessFunction<K, RowData, RowData>.Context ctx, + Collector<RowData> out) + throws Exception { + // register state-cleanup timer + registerProcessingCleanupTimer(ctx, ctx.timerService().currentProcessingTime()); + + RowKind rowKind = input.getRowKind(); + + switch (rowKind) { + case INSERT: + case UPDATE_AFTER: + insertIntoSortedList(input, out); + break; + + case DELETE: + case UPDATE_BEFORE: + removeFromSortedList(input, out); + break; + } + + // Reset acc state since we can have out of order inserts into the ordered list + currFunction.resetAccumulators(); + currFunction.cleanup(); + } + + /** + * Adds a new element(input) to a sortedList. The sortedList contains a list of Tuple<SortKey, + * List<Long Ids>> Extracts the inputSortKey from the input and compares it with every element + * in the sortedList If an sortKey already exists in the sortedList for the input, add the id to + * the list of ids and update the sortedList Otherwise find the right position in the sortedList + * and add a new entry in the middle. If no matching sortKey is found, add a new entry in the + * end of the sortedList. After and element is inserted, an INSERT/UPDATE_AFTER event is emitted + * for the newly inserted element, and for all subsequent elements an UPDATE_BEFORE and + * UPDATE_AFTER event is emitted based on the previous and newly aggregated values. Updates are + * skipped if the previously accumulated value is the same as the newly accumulated value to + * save on network bandwidth. + * + * @param input The input value. + * @param out The collector for returning result values. + * @throws Exception + */ + private void insertIntoSortedList(RowData input, Collector<RowData> out) throws Exception { + Long id = getNextId(); + List<Tuple2<RowData, List<Long>>> sortedList = getSortedList(); + RowKind origRowKind = input.getRowKind(); + input.setRowKind(RowKind.INSERT); + RowData inputSortKey = sortKeySelector.getKey(input); + int i; + boolean isInserted = false; + for (i = 0; i < sortedList.size(); i++) { + RowData curSortKey = sortedList.get(i).f0; + List<Long> ids = new ArrayList<>(sortedList.get(i).f1); + if (sortKeyComparator.compare(curSortKey, inputSortKey) == 0) { + // Insert at the (i)th location by adding the new id to the list of ids + ids.add(id); + sortedList.set(i, new Tuple2<>(curSortKey, ids)); + setAccumulatorOfPrevRow(sortedList, i - 1); + reAccumulateIdsAfterInsert(ids, input); + emitUpdatesForIds( + ids, + ids.size() - 1, + accMapState.get(curSortKey), + currFunction.getValue(), + origRowKind, + input, + out); + isInserted = true; + i++; + break; + } else if (sortKeyComparator.compare(curSortKey, inputSortKey) > 0) { + // Insert in the middle + sortedList.add(i, new Tuple2<>(inputSortKey, List.of(id))); + setAccumulatorOfPrevRow(sortedList, i - 1); + currFunction.accumulate(input); + collectInsertOrUpdateAfter(out, input, origRowKind); + isInserted = true; + i++; + break; + } + } + + if (!isInserted) { + // Insert at the end + sortedList.add(new Tuple2<>(inputSortKey, List.of(id))); + setAccumulatorOfPrevRow(sortedList, sortedList.size() - 2); + currFunction.accumulate(input); + collectInsertOrUpdateAfter(out, input, origRowKind); + i++; + } + + addToState(id, input, inputSortKey, currFunction.getAccumulators(), sortedList); + + processRemainingElements(sortedList, i, out); + } + + /** + * @return the next id after reading from the idState + * @throws IOException + */ + private Long getNextId() throws IOException { + Long id = idState.value(); + if (id == null) { + id = 0L; + } + return id; + } + + /** + * @return the sortedList containing sortKeys and Ids with the same sortKey + * @throws IOException + */ + private List<Tuple2<RowData, List<Long>>> getSortedList() throws IOException { + List<Tuple2<RowData, List<Long>>> sortedList = sortedListState.value(); + if (sortedList == null) { + sortedList = new ArrayList<>(); + } + return sortedList; + } + + /** + * Helper method to set the accumulator based on the prevIndex + * + * @param sortedList + * @param prevIndex + * @throws Exception + */ + private void setAccumulatorOfPrevRow( + List<Tuple2<RowData, List<Long>>> sortedList, int prevIndex) throws Exception { + if (prevIndex < 0) { + currFunction.createAccumulators(); + } else { + RowData prevAcc = accMapState.get(sortedList.get(prevIndex).f0); + if (prevAcc == null) { + // TODO: Throw exception since this could due to state cleaning + currFunction.createAccumulators(); + } else { + currFunction.setAccumulators(prevAcc); + } + } + } + + /** + * Helper method to re-accumulate the aggregated value for all ids after an id was inserted to + * the end of the ids list. + * + * @param ids + * @param input + * @throws Exception + */ + private void reAccumulateIdsAfterInsert(List<Long> ids, RowData input) throws Exception { + // Update acc value for all ids + for (int j = 0; j < ids.size(); j++) { + if (j == ids.size() - 1) { + // Update acc for newly inserted value + currFunction.accumulate(input); + } else { + // Get rowValue for id and re-accumulate it + RowData value = valueMapState.get(ids.get(j)); + currFunction.accumulate(value); + } + } + } + + /** + * Emits updates for all ids. The id belonging to idx location either emits a DELETE or + * INSERT/UPDATE_AFTER, depending on whether the id is being added to ids or removed from ids. + * All other ids except the id at position idx emit old and new aggregated values in the form of + * UPDATE_BEFORE and UPDATE_AFTER respectively. + * + * @param ids + * @param idx + * @param prevAcc + * @param currAcc + * @param rowKind + * @param input + * @param out + * @throws Exception + */ + private void emitUpdatesForIds( + List<Long> ids, + int idx, + RowData prevAcc, + RowData currAcc, + RowKind rowKind, + RowData input, + Collector<RowData> out) + throws Exception { + for (int j = 0; j < ids.size(); j++) { + if (j == idx) { + if (rowKind == RowKind.DELETE) { + // TODO: change to collectDelete + collectUpdateBefore(out, input, prevAcc); + } else { + // For insert, emit either Insert or UpdateAfter + collectInsertOrUpdateAfter(out, input, rowKind); + } + } else { + if (prevAcc.equals(currAcc)) { + // Previous accumulator is the same as the current accumulator + // Skip sending downstream updates in such cases to reduce network traffic + LOG.info("Prev accumulator is same as curr accumulator. Skipping updates."); + continue; + } + RowData value = valueMapState.get(ids.get(j)); + collectUpdateBefore(out, value, prevAcc); + collectUpdateAfter(out, value, currAcc); + } + } + } + + private void addToState( + Long id, + RowData input, + RowData inputSortKey, + RowData accumulators, + List<Tuple2<RowData, List<Long>>> sortedList) + throws Exception { + valueMapState.put(id, input); + accMapState.put(inputSortKey, accumulators); + sortedListState.update(sortedList); + idState.update(++id); + } + + /** + * Process all sort keys starting from startPos location. Calculates the new agg value for all + * ids belonging to the same sort key. Finally, emits the old and new aggregated values for the + * ids if the newly aggregated value is different from the previously aggregated values. + * + * @param sortedList + * @param startPos + * @param out + * @throws Exception + */ + private void processRemainingElements( + List<Tuple2<RowData, List<Long>>> sortedList, int startPos, Collector<RowData> out) + throws Exception { + // Send updates for remaining sort keys after the inserted/removed idx + for (int i = startPos; i < sortedList.size(); i++) { + Tuple2<RowData, List<Long>> sortKeyAndIds = sortedList.get(i); + RowData curSortKey = sortKeyAndIds.f0; + List<Long> ids = sortKeyAndIds.f1; + // Get previous accumulator + RowData prevAcc = accMapState.get(curSortKey); + + // Calculate new agg value for all ids with the same sortKey + for (int j = 0; j < ids.size(); j++) { + RowData value = valueMapState.get(ids.get(j)); + currFunction.accumulate(value); + } + // Emit old and new agg values for all ids with the same sortKey + for (int j = 0; j < ids.size(); j++) { + RowData value = valueMapState.get(ids.get(j)); + if (prevAcc.equals(currFunction.getAccumulators())) { + // Previous accumulator is the same as the current accumulator. + // This means all the ids will have no change in the accumulated value. + // Skip sending downstream updates in such cases to reduce network traffic + LOG.info("Prev accumulator is same as curr accumulator. Skipping updates."); + break; + } + collectUpdateBefore(out, value, prevAcc); + collectUpdateAfter(out, value, currFunction.getValue()); + } + // update accumulated state for sortKey if prevAcc != currAcc + if (!prevAcc.equals(currFunction.getAccumulators())) { + accMapState.put(curSortKey, currFunction.getAccumulators()); + } + } + } + + private void collectInsertOrUpdateAfter(Collector<RowData> out, RowData value, RowKind rowKind) + throws Exception { + output.setRowKind(rowKind); + output.replace(value, currFunction.getValue()); + out.collect(output); + } + + private void collectUpdateBefore( + Collector<RowData> out, RowData rowValue, RowData prevAccValue) { + output.setRowKind(RowKind.UPDATE_BEFORE); + output.replace(rowValue, prevAccValue); + out.collect(output); + } + + private void collectUpdateAfter( + Collector<RowData> out, RowData rowValue, RowData currAccValue) { + output.setRowKind(RowKind.UPDATE_AFTER); + output.replace(rowValue, currAccValue); + out.collect(output); + } + + private void collectDelete(Collector<RowData> out, RowData rowValue, RowData prevAccValue) { + output.setRowKind(RowKind.DELETE); + output.replace(rowValue, prevAccValue); + out.collect(output); + } + + /** + * Removes the row matching input from the sortedList. Matching is done based on the sortKeys. + * Once matched, the actual row is compared for all ids belonging to the same sortKey. After the + * element is removed, a DELETE event is emitted for the removed element, and for all subsequent + * elements an UPDATE_BEFORE and UPDATE_AFTER event is emitted based on the previous and newly + * aggregated values. Updates are skipped if the previously accumulated value is the same as the + * newly accumulated value to save on network bandwidth. + * + * @param input + * @param out + * @throws Exception + */ + private void removeFromSortedList(RowData input, Collector<RowData> out) throws Exception { + int i; + input.setRowKind(RowKind.INSERT); + RowData inputSortKey = sortKeySelector.getKey(input); + List<Tuple2<RowData, List<Long>>> sortedList = getSortedList(); + + for (i = 0; i < sortedList.size(); i++) { + RowData curSortKey = sortedList.get(i).f0; + List<Long> ids = new ArrayList<>(sortedList.get(i).f1); + if (sortKeyEqualiser.equals(curSortKey, inputSortKey)) { + setAccumulatorOfPrevRow(sortedList, i - 1); + // Accumulate all ids except id which needs to be removed + int removeIndex = reAccumulateIdsAndGetRemoveIndex(ids, input); + if (removeIndex == -1) { + // TODO: can this ever happen? may be during/after state cleaning? Review Comment: Added a log and a metric when data is not returned by state. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@flink.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
