rpsql

A simple heuristic SQL query planner. Can optionally output in a format executable by pythia.

For example:

SELECT
    O.ZIP,
    COUNT(P.COST)
FROM 
    (SELECT L.OKEY AS LKEY FROM LINEITEM L) L,
    PART P,
    ORDERS O
WHERE
    L.LKEY = P.PKEY AND
    L.LKEY = O.OKEY AND
    O.ZIP <> 3800
GROUP BY O.ZIP
ORDER BY COUNT(P.COST) DESC
LIMIT 20;

Creates this plan:

SortLimit (order_by=[Count(P.COST)], limit=20)
AggGroup (grouping=[O.ZIP], agg_field=Count(P.COST))
HashJoin (P.PKEY == L.LKEY)
(build)>Scan (table=PART, file=part.tbl)
(probe)>HashJoin (L.LKEY == O.OKEY)
        (build)>Projection ([L.OKEY])
                Scan (table=LINEITEM, file=lineitem.tbl.bz2)
        (probe)>Filter (O.ZIP != 3800)
                Scan (table=ORDERS, file=order.tbl.bz2)

SQL Support

• Joins
• Predicates (inequalities against constants)
• Grouping and Aggregation
• Sorting and Limits
• Subqueries
• Projection