use of org.apache.hadoop.hive.ql.optimizer.calcite.reloperators.HiveAntiJoin in project hive by apache.
the class JoinVisitor method visit.
@Override
OpAttr visit(RelNode joinRel) throws SemanticException {
// 0. Additional data structures needed for the join optimization
// through Hive
String[] baseSrc = new String[joinRel.getInputs().size()];
String tabAlias = hiveOpConverter.getHiveDerivedTableAlias();
// 1. Convert inputs
OpAttr[] inputs = new OpAttr[joinRel.getInputs().size()];
List<Operator<?>> children = new ArrayList<Operator<?>>(joinRel.getInputs().size());
for (int i = 0; i < inputs.length; i++) {
inputs[i] = hiveOpConverter.dispatch(joinRel.getInput(i));
children.add(inputs[i].inputs.get(0));
baseSrc[i] = inputs[i].tabAlias;
}
// 2. Generate tags
for (int tag = 0; tag < children.size(); tag++) {
ReduceSinkOperator reduceSinkOp = (ReduceSinkOperator) children.get(tag);
reduceSinkOp.getConf().setTag(tag);
}
// 3. Virtual columns
Set<Integer> newVcolsInCalcite = new HashSet<Integer>();
newVcolsInCalcite.addAll(inputs[0].vcolsInCalcite);
if (joinRel instanceof HiveMultiJoin || !((joinRel instanceof Join) && ((((Join) joinRel).isSemiJoin()) || (((Join) joinRel).getJoinType() == JoinRelType.ANTI)))) {
int shift = inputs[0].inputs.get(0).getSchema().getSignature().size();
for (int i = 1; i < inputs.length; i++) {
newVcolsInCalcite.addAll(HiveCalciteUtil.shiftVColsSet(inputs[i].vcolsInCalcite, shift));
shift += inputs[i].inputs.get(0).getSchema().getSignature().size();
}
}
if (LOG.isDebugEnabled()) {
LOG.debug("Translating operator rel#" + joinRel.getId() + ":" + joinRel.getRelTypeName() + " with row type: [" + joinRel.getRowType() + "]");
}
// 4. Extract join key expressions from HiveSortExchange
ExprNodeDesc[][] joinExpressions = new ExprNodeDesc[inputs.length][];
for (int i = 0; i < inputs.length; i++) {
joinExpressions[i] = ((HiveSortExchange) joinRel.getInput(i)).getKeyExpressions();
}
// 5. Extract rest of join predicate info. We infer the rest of join condition
// that will be added to the filters (join conditions that are not part of
// the join key)
List<RexNode> joinFilters;
if (joinRel instanceof HiveJoin) {
joinFilters = ImmutableList.of(((HiveJoin) joinRel).getJoinFilter());
} else if (joinRel instanceof HiveMultiJoin) {
joinFilters = ((HiveMultiJoin) joinRel).getJoinFilters();
} else if (joinRel instanceof HiveSemiJoin) {
joinFilters = ImmutableList.of(((HiveSemiJoin) joinRel).getJoinFilter());
} else if (joinRel instanceof HiveAntiJoin) {
joinFilters = ImmutableList.of(((HiveAntiJoin) joinRel).getJoinFilter());
} else {
throw new SemanticException("Can't handle join type: " + joinRel.getClass().getName());
}
List<List<ExprNodeDesc>> filterExpressions = Lists.newArrayList();
for (int i = 0; i < joinFilters.size(); i++) {
List<ExprNodeDesc> filterExpressionsForInput = new ArrayList<ExprNodeDesc>();
if (joinFilters.get(i) != null) {
for (RexNode conj : RelOptUtil.conjunctions(joinFilters.get(i))) {
ExprNodeDesc expr = HiveOpConverterUtils.convertToExprNode(conj, joinRel, null, newVcolsInCalcite);
filterExpressionsForInput.add(expr);
}
}
filterExpressions.add(filterExpressionsForInput);
}
// 6. Generate Join operator
JoinOperator joinOp = genJoin(joinRel, joinExpressions, filterExpressions, children, baseSrc, tabAlias);
// 7. Return result
return new OpAttr(tabAlias, newVcolsInCalcite, joinOp);
}
Aggregations