Module df_optimizer

Datafusion optimizer for flow plan

Structs

AvgExpandRule 🔒

CheckGroupByRule 🔒

This rule check all group by exprs, and make sure they are also in select clause in a aggr query

ExpandAvgRewriter 🔒

rewrite avg(<expr>) function into CASE WHEN count(<expr>) !=0 THEN cast(sum((<expr>) AS avg_return_type)/count((<expr>) ELSE 0

FindColumn 🔒

Find all column names in a plan

TumbleExpand

This is a placeholder for tumble_start and tumble_end function, so that datafusion can recognize them as scalar function

TumbleExpandRule 🔒

expand tumble in aggr expr to tumble_start and tumble_end with column name like window_start

Functions

apply_df_optimizer

check_group_by_analyzer 🔒

make sure everything in group by’s expr is in select

expand_avg_analyzer 🔒

expand avg(<expr>) function into cast(sum((<expr>) AS f64)/count((<expr>)

expand_tumble_analyzer 🔒

expand tumble in aggr expr to tumble_start and tumble_end, also expand related alias and column ref

put_aggr_to_proj_analyzer 🔒

lift aggr’s composite aggr_expr to outer proj, and leave aggr only with simple direct aggr expr i.e.

sql_to_flow_plan

To reuse existing code for parse sql, the sql is first parsed into a datafusion logical plan, then to a substrait plan, and finally to a flow plan.