Struct RowCursorStream 

pub struct RowCursorStream {
    converter: RowConverter,
    column_expressions: Vec<Arc<dyn PhysicalExpr>>,
    streams: FusedStreams,
    reservation: MemoryReservation,
    rows: ReusableRows,
}

A PartitionedStream that wraps a set of SendableRecordBatchStream and computes RowValues based on the provided PhysicalSortExpr Note: the stream returns an error if the consumer buffers more than one RowValues (i.e. holds on to two RowValues from the same partition at the same time).

Fields

converter: RowConverter

Converter to convert output of physical expressions

column_expressions: Vec<Arc<dyn PhysicalExpr>>

The physical expressions to sort by

streams: FusedStreams

Input streams

reservation: MemoryReservation

Tracks the memory used by converter

rows: ReusableRows

Allocated rows for each partition, we keep two to allow for buffering one in the consumer of the stream

Implementations

impl RowCursorStream

pub fn try_new( schema: &Schema, expressions: &LexOrdering, streams: Vec<SendableRecordBatchStream>, reservation: MemoryReservation, ) -> Result<Self>

fn convert_batch( &mut self, batch: &RecordBatch, stream_idx: usize, ) -> Result<RowValues>

Trait Implementations

impl Debug for RowCursorStream

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartitionedStream for RowCursorStream

type Output = Result<(RowValues, RecordBatch), DataFusionError>

fn partitions(&self) -> usize

Returns the number of partitions

fn poll_next( &mut self, cx: &mut Context<'_>, stream_idx: usize, ) -> Poll<Option<Self::Output>>