Interface VectorAggregator

All Known Implementing Classes:
CardinalityVectorAggregator, CountVectorAggregator, DoubleAnyVectorAggregator, DoubleFirstLastVectorAggregator, DoubleFirstVectorAggregator, DoubleLastVectorAggregator, DoubleMaxVectorAggregator, DoubleMeanVectorAggregator, DoubleMinVectorAggregator, DoubleSumVectorAggregator, FilteredVectorAggregator, FirstLastVectorAggregator, FloatAnyVectorAggregator, FloatFirstLastVectorAggregator, FloatFirstVectorAggregator, FloatLastVectorAggregator, FloatMaxVectorAggregator, FloatMinVectorAggregator, FloatSumVectorAggregator, HyperUniquesVectorAggregator, LongAnyVectorAggregator, LongConstantVectorAggregator, LongFirstLastVectorAggregator, LongFirstVectorAggregator, LongLastVectorAggregator, LongMaxVectorAggregator, LongMinVectorAggregator, LongSumVectorAggregator, NilVectorAggregator, NoopVectorAggregator, NullableNumericVectorAggregator, NumericAnyVectorAggregator, SingleStringFirstDimensionVectorAggregator, SingleStringFirstLastDimensionVectorAggregator, SingleStringLastDimensionVectorAggregator, StringAnyVectorAggregator, StringFirstLastVectorAggregator, StringFirstVectorAggregator, StringLastVectorAggregator, SuppressedAggregatorFactory.SuppressedVectorAggregator

public interface VectorAggregator
An object that can aggregate metrics into a ByteBuffer, from vectorized column selectors. Its aggregation-related methods (namely, "aggregate" and "get") do not take the actual input values to aggregate, because it is assumed that the VectorAggregator was given something that it can use to get at the current batch of data. None of the methods in this class are annotated with CalledFromHotLoop because vectorized query engines do not use monomorphic-processing-style specialization. Unlike Aggregator, VectorAggregators are never used by multiple threads at once. Implementations are not required to be thread safe.
  • Method Details

    • init

      void init(ByteBuffer buf, int position)
    • aggregate

      void aggregate(ByteBuffer buf, int position, int startRow, int endRow)
      Aggregate a range of rows into a single aggregation slot. Implementations must not change the position, limit or mark of the given buffer
      Parameters:
      buf - byte buffer storing the byte array representation of the aggregate
      position - offset within the byte buffer at which the current aggregate value is stored
      startRow - first row of the range within the current batch to aggregate (inclusive)
      endRow - end row of the range (exclusive)
    • aggregate

      void aggregate(ByteBuffer buf, int numRows, int[] positions, @Nullable int[] rows, int positionOffset)
      Aggregate a list of rows ("rows") into a list of aggregation slots ("positions"). Implementations must not change the position, limit or mark of the given buffer
      Parameters:
      buf - byte buffer storing the byte array representation of the aggregate
      numRows - number of rows to aggregate
      positions - array of aggregate value positions within the buffer; must be at least as long as "numRows"
      rows - array of row numbers within the current row batch; must be at least as long as "numRows". If null, the aggregator will aggregate rows from 0 (inclusive) to numRows (exclusive).
      positionOffset - an offset to apply to each value from "positions"
    • get

      @Nullable Object get(ByteBuffer buf, int position)
    • relocate

      default void relocate(int oldPosition, int newPosition, ByteBuffer oldBuffer, ByteBuffer newBuffer)
    • close

      void close()
      Release any resources used by the aggregator. The aggregator may be reused after this call, by calling init(ByteBuffer, int) followed by other methods as normal. This call would be more properly named "reset", but we use the name "close" to improve compatibility with existing aggregator implementations in extensions.