Module stream_join_utils 

This file contains common subroutines for symmetric hash join related functionality, used both in join calculations and optimization rules.

Structs

PruningJoinHashMap

The PruningJoinHashMap is similar to a regular JoinHashMap, but with the capability of pruning elements in an efficient manner. This structure is particularly useful for cases where it’s necessary to remove elements from the map based on their buffer order.

SortedFilterExpr

The SortedFilterExpr object represents a sorted filter expression. It contains the following information: The origin expression, the filter expression, an interval encapsulating expression bounds, and a stable index identifying the expression in the expression DAG.

StreamJoinMetrics

Metrics for HashJoinExec

StreamJoinSideMetrics

Functions

build_filter_input_order

This function is used to build the filter expression based on the sort order of input columns.

calculate_filter_expr_intervals

Calculate the filter expression intervals.

check_filter_expr_contains_sort_information 🔒

combine_two_batches

convert_filter_columns 🔒

Convert a physical expression into a filter expression using the given column mapping information.

convert_sort_expr_with_filter_schema

This function analyzes PhysicalSortExpr graphs with respect to output orderings (sorting) properties. This is necessary since monotonically increasing and/or decreasing expressions are required when using join filter expressions for data pruning purposes.

get_pruning_anti_indices

Get the anti join indices from the visited hash set.

get_pruning_semi_indices

This method creates a boolean buffer from the visited rows hash set and the indices of the pruned record batch slice.

map_origin_col_to_filter_col

Create a one to one mapping from main columns to filter columns using filter column indices. A column index looks like:

prepare_sorted_exprs

Prepares and sorts expressions based on a given filter, left and right schemas, and sort expressions.

record_visited_indices

Records the visited indices from the input PrimitiveArray of type T into the given hash set visited. This function will insert the indices (offset by offset) into the visited hash set.

update_filter_expr_interval

This is a subroutine of the function calculate_filter_expr_intervals. It constructs the current interval using the given batch and updates the filter expression (i.e. sorted_expr) with this interval.

update_sorted_exprs_with_node_indices 🔒

Updates sorted filter expressions with corresponding node indices from the expression interval graph.