Struct RangePlanRewriter

pub struct RangePlanRewriter {
    table_provider: DfTableSourceProvider,
    query_ctx: QueryContextRef,
}

In order to implement RangeSelect query like avg(field_0) RANGE '5m' FILL NULL, All RangeSelect query items are converted into udf scalar function in sql parse stage, with format like range_fn(avg(field_0), .....). range_fn contains all the parameters we need to execute RangeSelect. In order to correctly execute the query process of range select, we need to modify the query plan generated by datafusion. We need to recursively find the entire LogicalPlan, and find all range_fn scalar udf contained in the project plan, collecting info we need to generate RangeSelect Query LogicalPlan and rewrite th original LogicalPlan.

Fields

table_provider: DfTableSourceProvider

query_ctx: QueryContextRef

Implementations

impl RangePlanRewriter

pub fn new( table_provider: DfTableSourceProvider, query_ctx: QueryContextRef, ) -> Self

pub async fn rewrite(&mut self, plan: LogicalPlan) -> Result<LogicalPlan>

fn rewrite_logical_plan<'life0, 'life_self, 'async_recursion>( &'life_self mut self, plan: &'life0 LogicalPlan, ) -> Pin<Box<dyn Future<Output = Result<Option<LogicalPlan>>> + Send + 'async_recursion>>

where 'life0: 'async_recursion, 'life_self: 'async_recursion,

async fn get_index_by( &mut self, schema: &Arc<DFSchema>, ) -> Result<(Expr, Vec<Expr>)>

this function use to find the time_index column and row columns from input schema, return (time_index, [row_columns]) to the rewriter. If the user does not explicitly use the by keyword to indicate time series, [row_columns] will be use as default time series