com.bigdata.rdf.sparql.ast.optimizers

Class ASTOrderByAggregateFlatteningOptimizer

java.lang.Object

com.bigdata.rdf.sparql.ast.optimizers.ASTOrderByAggregateFlatteningOptimizer

All Implemented Interfaces:

IASTOptimizer

public class ASTOrderByAggregateFlatteningOptimizer
extends Object
implements IASTOptimizer

This compulsory AST transformation (not an optional optimizer!) enforces the correct treatment of aggregates in ORDER BY clauses, according to the SPARQL semantic, under the asumption that MemorySortOp does not have to deal with aggregates. In a nutshell, this is done by introducing aliases for the aggregate expressions and thus pushing the computation of the aggregates to where they can already be processed. Simple example. Consider this query (sparql11-order-02.rq): SELECT ?type WHERE { ?subj a ?type } GROUP BY ?type ORDER BY (count(?subj)) It contains aggregate count(?subj) in ORDER BY. The idea is to rewrite it into this query: SELECT ?type (count(?subj) as ?cnt)[excludeFromProjection] WHERE { ?subj a ?type } GROUP BY ?type ORDER BY ?cnt Here ?cnt is an auxiliary alias, i.e, a fresh variable (UUIDs are used as fresh variable names in the implementation). This query can be computed even if the sorting does not support aggregates. Note that the rewritten query is not completely equivalent to the original one: it will assign ?cnt, which does not occur in the original query. To rectify this, some variables in a projection are designated as excluded from the projection outputs: see the label "[excludeFromProjection]" by ?cnt in the rewritten query (pseudo-SPARQL). More complex example. The original query is: PREFIX ex: SELECT ?x ?y WHERE { ?x ex:r ?y . ?y ex:q ?z } GROUP BY ?x ?y ORDER BY DESC(max(?z)) ?x (count(?z)) DESC(?y) The rewritten query is: PREFIX ex: SELECT ?x ?y (max(?z) AS ?maxz)[excludeFromProjection] (count(?z) AS ?cntz)[excludeFromProjection] WHERE { ?x ex:r ?y . ?y ex:q ?z } GROUP BY ?x ?y ORDER BY DESC(?maxz) ?x ?cntz DESC(?y) Here ?maxz and ?cntz are the introduced auxiliary aliases for the aggregates.

Author:

Alexandre Riazanov

Constructor Summary

Constructors

Constructor and Description
ASTOrderByAggregateFlatteningOptimizer()

Method Summary

Methods

Modifier and Type	Method and Description
QueryNodeWithBindingSet	optimize(AST2BOpContext context, QueryNodeWithBindingSet input) Optimize the AST.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

ASTOrderByAggregateFlatteningOptimizer

public ASTOrderByAggregateFlatteningOptimizer()

Method Detail

optimize

public QueryNodeWithBindingSet optimize(AST2BOpContext context,
                               QueryNodeWithBindingSet input)

Description copied from interface: IASTOptimizer

Optimize the AST.

Specified by:

optimize in interface IASTOptimizer

Parameters:

context - The evaluation context.

input - The input to the optimizer, consisting of a queryNode and input binding set.

Returns:

An instance of type QueryNodeWithBindingSet.