// pRS-cRS
@Override
public Object process(ReduceSinkOperator cRS, ReduceSinkDeduplicateProcCtx dedupCtx)
throws SemanticException {
ReduceSinkOperator pRS =
CorrelationUtilities.findPossibleParent(
cRS, ReduceSinkOperator.class, dedupCtx.trustScript());
if (pRS != null && merge(cRS, pRS, dedupCtx.minReducer())) {
CorrelationUtilities.replaceReduceSinkWithSelectOperator(cRS, dedupCtx.getPctx(), dedupCtx);
pRS.getConf().setDeduplicated(true);
return true;
}
return false;
}
// pRS-cRS-cGBY
@Override
public Object process(
ReduceSinkOperator cRS, GroupByOperator cGBY, ReduceSinkDeduplicateProcCtx dedupCtx)
throws SemanticException {
Operator<?> start = CorrelationUtilities.getStartForGroupBy(cRS, dedupCtx);
ReduceSinkOperator pRS =
CorrelationUtilities.findPossibleParent(
start, ReduceSinkOperator.class, dedupCtx.trustScript());
if (pRS != null && merge(cRS, pRS, dedupCtx.minReducer())) {
if (dedupCtx.getPctx().getConf().getBoolVar(HiveConf.ConfVars.HIVEGROUPBYSKEW)) {
return false;
}
CorrelationUtilities.removeReduceSinkForGroupBy(cRS, cGBY, dedupCtx.getPctx(), dedupCtx);
pRS.getConf().setDeduplicated(true);
return true;
}
return false;
}
// pRS-pJOIN-cRS-cGBY
@Override
public Object process(
ReduceSinkOperator cRS, GroupByOperator cGBY, ReduceSinkDeduplicateProcCtx dedupCtx)
throws SemanticException {
Operator<?> start = CorrelationUtilities.getStartForGroupBy(cRS, dedupCtx);
JoinOperator pJoin =
CorrelationUtilities.findPossibleParent(
start, JoinOperator.class, dedupCtx.trustScript());
if (pJoin != null && merge(cRS, pJoin, dedupCtx.minReducer())) {
pJoin.getConf().setFixedAsSorted(true);
CorrelationUtilities.removeReduceSinkForGroupBy(cRS, cGBY, dedupCtx.getPctx(), dedupCtx);
ReduceSinkOperator pRS =
CorrelationUtilities.findPossibleParent(
pJoin, ReduceSinkOperator.class, dedupCtx.trustScript());
if (pRS != null) {
pRS.getConf().setDeduplicated(true);
}
return true;
}
return false;
}
