Struct mentat_query_projector::CombinedProjection

pub struct CombinedProjection {
    pub sql_projection: Projection,
    pub pre_aggregate_projection: Option<Projection>,
    pub datalog_projector: Box<Projector>,
    pub distinct: bool,
    pub group_by_cols: Vec<GroupBy>,
}

Combines the things you need to turn a query into SQL and turn its results into QueryResults: SQL-related projection information (DISTINCT, columns, etc.) and a Datalog projector that turns SQL into structures.

Fields

sql_projection: Projection

A SQL projection, mapping columns mentioned in the body of the query to columns in the output.

pre_aggregate_projection: Option<Projection>

If a query contains aggregates, we need to generate a nested subquery: an inner query that returns our distinct variable bindings (and any :with vars), and an outer query that applies aggregation. That's so we can put DISTINCT in the inner query and apply aggregation afterwards -- SELECT DISTINCT count(foo) counts then uniques, and we need the opposite to implement Datalog distinct semantics. If this is the case, sql_projection will be the outer query's projection list, and pre_aggregate_projection will be the inner. If the query doesn't use aggregation, this field will be None.

datalog_projector: Box<Projector>

A Datalog projection. This consumes rows of the appropriate shape (as defined by the SQL projection) to yield one of the four kinds of Datalog query result.

distinct: bool

True if this query requires the SQL query to include DISTINCT.

group_by_cols: Vec<GroupBy>

Trait Implementations