Module stream 

Sort-Merge Join execution

This module implements the runtime state machine for the Sort-Merge Join operator. It drives two sorted input streams (the streamed side and the buffered side), compares join keys, and produces joined RecordBatches.

Structs

BufferedBatch 🔒

A buffered batch that contains contiguous rows with same join key

BufferedData 🔒

Buffered data contains all buffered batches with one unique join key

JoinedRecordBatches 🔒

Joined batches with attached join filter information

SortMergeJoinStream 🔒

Sort-Merge join stream that consumes streamed and buffered data streams and produces joined output stream.

StreamedBatch 🔒

Represents a record batch from streamed input.

StreamedJoinedChunk 🔒

Represents a chunk of joined data from streamed and buffered side

Enums

BufferedBatchState 🔒

BufferedState 🔒

State of buffered data stream

SortMergeJoinState 🔒

State of SMJ stream

StreamedState 🔒

State of streamed data stream

Functions

create_unmatched_columns 🔒

fetch_right_columns_by_idxs 🔒

Get buffered_indices rows for buffered_data[buffered_batch_idx] by specific column indices

fetch_right_columns_from_batch_by_idxs 🔒

get_corrected_filter_mask 🔒

get_filter_column 🔒

Gets the arrays which join filters are applied on.

is_join_arrays_equal 🔒

A faster version of compare_join_arrays() that only output whether the given two rows are equal

join_arrays 🔒

Get join array refs of given batch and join columns

last_index_for_row 🔒

True if next index refers to either:

produce_buffered_null_batch 🔒