use of org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression in project asterixdb by apache.
the class PushAggregateIntoNestedSubplanRule method pushSubplanAsAggIntoNestedSubplan.
private boolean pushSubplanAsAggIntoNestedSubplan(Mutable<ILogicalOperator> subplanOpRef, AbstractOperatorWithNestedPlans nspOp, LogicalVariable varFromNestedAgg, Map<LogicalVariable, Integer> nspAggVars, Map<LogicalVariable, AbstractOperatorWithNestedPlans> nspWithAgg, Map<LogicalVariable, Integer> nspAggVarToPlanIndex, IOptimizationContext context) throws AlgebricksException {
SubplanOperator subplan = (SubplanOperator) subplanOpRef.getValue();
// only free var can be varFromNestedAgg
HashSet<LogicalVariable> freeVars = new HashSet<>();
OperatorPropertiesUtil.getFreeVariablesInSubplans(subplan, freeVars);
for (LogicalVariable vFree : freeVars) {
if (!vFree.equals(varFromNestedAgg)) {
return false;
}
}
List<ILogicalPlan> plans = subplan.getNestedPlans();
if (plans.size() > 1) {
return false;
}
ILogicalPlan p = plans.get(0);
if (p.getRoots().size() > 1) {
return false;
}
Mutable<ILogicalOperator> opRef = p.getRoots().get(0);
AbstractLogicalOperator op = (AbstractLogicalOperator) opRef.getValue();
if (op.getOperatorTag() != LogicalOperatorTag.AGGREGATE) {
return false;
}
AggregateOperator aggInSubplanOp = (AggregateOperator) op;
LogicalVariable unnestVar = null;
boolean pushableNestedSubplan = false;
while (op.getInputs().size() == 1) {
opRef = op.getInputs().get(0);
op = (AbstractLogicalOperator) opRef.getValue();
switch(op.getOperatorTag()) {
case ASSIGN:
break;
case UNNEST:
UnnestOperator unnest = (UnnestOperator) op;
if (unnest.getPositionalVariable() != null) {
// TODO currently subplan with both accumulating and running aggregate is not supported.
return false;
}
ILogicalExpression expr = unnest.getExpressionRef().getValue();
if (expr.getExpressionTag() != LogicalExpressionTag.FUNCTION_CALL) {
return false;
}
AbstractFunctionCallExpression fun = (AbstractFunctionCallExpression) expr;
if (fun.getFunctionIdentifier() != BuiltinFunctions.SCAN_COLLECTION) {
return false;
}
ILogicalExpression arg0 = fun.getArguments().get(0).getValue();
if (arg0.getExpressionTag() != LogicalExpressionTag.VARIABLE) {
return false;
}
VariableReferenceExpression varExpr = (VariableReferenceExpression) arg0;
if (!varExpr.getVariableReference().equals(varFromNestedAgg)) {
return false;
}
opRef = op.getInputs().get(0);
op = (AbstractLogicalOperator) opRef.getValue();
if (op.getOperatorTag() != LogicalOperatorTag.NESTEDTUPLESOURCE) {
return false;
}
pushableNestedSubplan = true;
unnestVar = unnest.getVariable();
break;
default:
return false;
}
}
if (!pushableNestedSubplan) {
return false;
}
for (int i = 0; i < nspOp.getNestedPlans().size(); i++) {
Mutable<ILogicalOperator> nspAggRef = nspOp.getNestedPlans().get(i).getRoots().get(0);
AggregateOperator nspAgg = (AggregateOperator) nspAggRef.getValue();
Mutable<ILogicalOperator> nspAggChildRef = nspAgg.getInputs().get(0);
LogicalVariable listifyVar = findListifiedVariable(nspAgg, varFromNestedAgg);
if (listifyVar == null) {
continue;
}
OperatorManipulationUtil.substituteVarRec(aggInSubplanOp, unnestVar, listifyVar, true, context);
nspAgg.getVariables().addAll(aggInSubplanOp.getVariables());
nspAgg.getExpressions().addAll(aggInSubplanOp.getExpressions());
for (LogicalVariable v : aggInSubplanOp.getVariables()) {
nspWithAgg.put(v, nspOp);
nspAggVars.put(v, 0);
nspAggVarToPlanIndex.put(v, i);
}
Mutable<ILogicalOperator> opRef1InSubplan = aggInSubplanOp.getInputs().get(0);
if (!opRef1InSubplan.getValue().getInputs().isEmpty()) {
Mutable<ILogicalOperator> opRef2InSubplan = opRef1InSubplan.getValue().getInputs().get(0);
AbstractLogicalOperator op2InSubplan = (AbstractLogicalOperator) opRef2InSubplan.getValue();
if (op2InSubplan.getOperatorTag() != LogicalOperatorTag.NESTEDTUPLESOURCE) {
List<Mutable<ILogicalOperator>> nspInpList = nspAgg.getInputs();
nspInpList.clear();
nspInpList.add(opRef1InSubplan);
while (true) {
opRef2InSubplan = opRef1InSubplan.getValue().getInputs().get(0);
op2InSubplan = (AbstractLogicalOperator) opRef2InSubplan.getValue();
if (op2InSubplan.getOperatorTag() == LogicalOperatorTag.UNNEST) {
List<Mutable<ILogicalOperator>> opInpList = opRef1InSubplan.getValue().getInputs();
opInpList.clear();
opInpList.add(nspAggChildRef);
break;
}
opRef1InSubplan = opRef2InSubplan;
if (opRef1InSubplan.getValue().getInputs().isEmpty()) {
throw new IllegalStateException("PushAggregateIntoNestedSubplanRule: could not find UNNEST.");
}
}
}
}
subplanOpRef.setValue(subplan.getInputs().get(0).getValue());
OperatorPropertiesUtil.typeOpRec(nspAggRef, context);
}
return true;
}
use of org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression in project asterixdb by apache.
the class PushFieldAccessRule method propagateFieldAccessRec.
@SuppressWarnings("unchecked")
private boolean propagateFieldAccessRec(Mutable<ILogicalOperator> opRef, IOptimizationContext context, String finalAnnot) throws AlgebricksException {
AssignOperator access = (AssignOperator) opRef.getValue();
Mutable<ILogicalOperator> opRef2 = access.getInputs().get(0);
AbstractLogicalOperator op2 = (AbstractLogicalOperator) opRef2.getValue();
// rewritten into index search.
if (op2.getOperatorTag() == LogicalOperatorTag.PROJECT || context.checkAndAddToAlreadyCompared(access, op2) && !(op2.getOperatorTag() == LogicalOperatorTag.SELECT && isAccessToIndexedField(access, context))) {
return false;
}
Object annotation = op2.getAnnotations().get(IS_MOVABLE);
if (annotation != null && !((Boolean) annotation)) {
return false;
}
if (tryingToPushThroughSelectionWithSameDataSource(access, op2)) {
return false;
}
if (testAndModifyRedundantOp(access, op2)) {
propagateFieldAccessRec(opRef2, context, finalAnnot);
return true;
}
List<LogicalVariable> usedInAccess = new LinkedList<>();
VariableUtilities.getUsedVariables(access, usedInAccess);
List<LogicalVariable> produced2 = new LinkedList<>();
if (op2.getOperatorTag() == LogicalOperatorTag.GROUP) {
VariableUtilities.getLiveVariables(op2, produced2);
} else {
VariableUtilities.getProducedVariables(op2, produced2);
}
boolean pushItDown = false;
List<LogicalVariable> inter = new ArrayList<>(usedInAccess);
if (inter.isEmpty()) {
// ground value
return false;
}
inter.retainAll(produced2);
if (inter.isEmpty()) {
pushItDown = true;
} else if (op2.getOperatorTag() == LogicalOperatorTag.GROUP) {
GroupByOperator g = (GroupByOperator) op2;
List<Pair<LogicalVariable, LogicalVariable>> varMappings = new ArrayList<>();
for (Pair<LogicalVariable, Mutable<ILogicalExpression>> p : g.getDecorList()) {
ILogicalExpression e = p.second.getValue();
if (e.getExpressionTag() == LogicalExpressionTag.VARIABLE) {
LogicalVariable decorVar = GroupByOperator.getDecorVariable(p);
if (inter.contains(decorVar)) {
inter.remove(decorVar);
LogicalVariable v1 = ((VariableReferenceExpression) e).getVariableReference();
varMappings.add(new Pair<>(decorVar, v1));
}
}
}
if (inter.isEmpty()) {
boolean changed = false;
for (Pair<LogicalVariable, LogicalVariable> m : varMappings) {
LogicalVariable v2 = context.newVar();
LogicalVariable oldVar = access.getVariables().get(0);
g.getDecorList().add(new Pair<LogicalVariable, Mutable<ILogicalExpression>>(oldVar, new MutableObject<ILogicalExpression>(new VariableReferenceExpression(v2))));
changed = true;
access.getVariables().set(0, v2);
VariableUtilities.substituteVariables(access, m.first, m.second, context);
}
if (changed) {
context.computeAndSetTypeEnvironmentForOperator(g);
}
usedInAccess.clear();
VariableUtilities.getUsedVariables(access, usedInAccess);
pushItDown = true;
}
}
if (pushItDown) {
if (op2.getOperatorTag() == LogicalOperatorTag.NESTEDTUPLESOURCE) {
Mutable<ILogicalOperator> childOfSubplan = ((NestedTupleSourceOperator) op2).getDataSourceReference().getValue().getInputs().get(0);
pushAccessDown(opRef, op2, childOfSubplan, context, finalAnnot);
return true;
}
if (op2.getInputs().size() == 1 && !op2.hasNestedPlans()) {
pushAccessDown(opRef, op2, op2.getInputs().get(0), context, finalAnnot);
return true;
} else {
for (Mutable<ILogicalOperator> inp : op2.getInputs()) {
HashSet<LogicalVariable> v2 = new HashSet<>();
VariableUtilities.getLiveVariables(inp.getValue(), v2);
if (v2.containsAll(usedInAccess)) {
pushAccessDown(opRef, op2, inp, context, finalAnnot);
return true;
}
}
}
if (op2.hasNestedPlans()) {
AbstractOperatorWithNestedPlans nestedOp = (AbstractOperatorWithNestedPlans) op2;
for (ILogicalPlan plan : nestedOp.getNestedPlans()) {
for (Mutable<ILogicalOperator> root : plan.getRoots()) {
HashSet<LogicalVariable> v2 = new HashSet<>();
VariableUtilities.getLiveVariables(root.getValue(), v2);
if (v2.containsAll(usedInAccess)) {
pushAccessDown(opRef, op2, root, context, finalAnnot);
return true;
}
}
}
}
throw new AlgebricksException("Field access " + access.getExpressions().get(0).getValue() + " does not correspond to any input of operator " + op2);
} else {
// fields. If yes, we can equate the two variables.
if (op2.getOperatorTag() == LogicalOperatorTag.DATASOURCESCAN) {
DataSourceScanOperator scan = (DataSourceScanOperator) op2;
int n = scan.getVariables().size();
LogicalVariable scanRecordVar = scan.getVariables().get(n - 1);
AbstractFunctionCallExpression accessFun = (AbstractFunctionCallExpression) access.getExpressions().get(0).getValue();
ILogicalExpression e0 = accessFun.getArguments().get(0).getValue();
LogicalExpressionTag tag = e0.getExpressionTag();
if (tag == LogicalExpressionTag.VARIABLE) {
VariableReferenceExpression varRef = (VariableReferenceExpression) e0;
if (varRef.getVariableReference() == scanRecordVar) {
ILogicalExpression e1 = accessFun.getArguments().get(1).getValue();
if (e1.getExpressionTag() == LogicalExpressionTag.CONSTANT) {
IDataSource<DataSourceId> dataSource = (IDataSource<DataSourceId>) scan.getDataSource();
byte dsType = ((DataSource) dataSource).getDatasourceType();
if (dsType == DataSource.Type.FEED || dsType == DataSource.Type.LOADABLE) {
return false;
}
DataSourceId asid = dataSource.getId();
MetadataProvider mp = (MetadataProvider) context.getMetadataProvider();
Dataset dataset = mp.findDataset(asid.getDataverseName(), asid.getDatasourceName());
if (dataset == null) {
throw new AlgebricksException("Dataset " + asid.getDatasourceName() + " not found.");
}
if (dataset.getDatasetType() != DatasetType.INTERNAL) {
setAsFinal(access, context, finalAnnot);
return false;
}
ConstantExpression ce = (ConstantExpression) e1;
IAObject obj = ((AsterixConstantValue) ce.getValue()).getObject();
String fldName;
if (obj.getType().getTypeTag() == ATypeTag.STRING) {
fldName = ((AString) obj).getStringValue();
} else {
int pos = ((AInt32) obj).getIntegerValue();
String tName = dataset.getItemTypeName();
IAType t = mp.findType(dataset.getItemTypeDataverseName(), tName);
if (t.getTypeTag() != ATypeTag.OBJECT) {
return false;
}
ARecordType rt = (ARecordType) t;
if (pos >= rt.getFieldNames().length) {
setAsFinal(access, context, finalAnnot);
return false;
}
fldName = rt.getFieldNames()[pos];
}
int p = DatasetUtil.getPositionOfPartitioningKeyField(dataset, fldName);
if (p < 0) {
// not one of the partitioning fields
setAsFinal(access, context, finalAnnot);
return false;
}
LogicalVariable keyVar = scan.getVariables().get(p);
access.getExpressions().get(0).setValue(new VariableReferenceExpression(keyVar));
return true;
}
}
}
}
setAsFinal(access, context, finalAnnot);
return false;
}
}
use of org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression in project asterixdb by apache.
the class AccessMethodJobGenParams method writeVarList.
protected void writeVarList(List<LogicalVariable> varList, List<Mutable<ILogicalExpression>> funcArgs) {
Mutable<ILogicalExpression> numKeysRef = new MutableObject<>(new ConstantExpression(new AsterixConstantValue(new AInt32(varList.size()))));
funcArgs.add(numKeysRef);
for (LogicalVariable keyVar : varList) {
Mutable<ILogicalExpression> keyVarRef = new MutableObject<>(new VariableReferenceExpression(keyVar));
funcArgs.add(keyVarRef);
}
}
use of org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression in project asterixdb by apache.
the class AccessMethodJobGenParams method readVarList.
protected int readVarList(List<Mutable<ILogicalExpression>> funcArgs, int index, List<LogicalVariable> varList) {
int numLowKeys = AccessMethodUtils.getInt32Constant(funcArgs.get(index));
if (numLowKeys > 0) {
for (int i = 0; i < numLowKeys; i++) {
LogicalVariable var = ((VariableReferenceExpression) funcArgs.get(index + 1 + i).getValue()).getVariableReference();
varList.add(var);
}
}
return index + numLowKeys + 1;
}
use of org.apache.hyracks.algebricks.core.algebra.expressions.VariableReferenceExpression in project asterixdb by apache.
the class AccessMethodUtils method createPrimaryIndexUnnestMap.
public static AbstractUnnestMapOperator createPrimaryIndexUnnestMap(AbstractDataSourceOperator dataSourceOp, Dataset dataset, ARecordType recordType, ARecordType metaRecordType, ILogicalOperator inputOp, IOptimizationContext context, boolean sortPrimaryKeys, boolean retainInput, boolean retainNull, boolean requiresBroadcast) throws AlgebricksException {
List<LogicalVariable> primaryKeyVars = AccessMethodUtils.getPrimaryKeyVarsFromSecondaryUnnestMap(dataset, inputOp);
// Optionally add a sort on the primary-index keys before searching the primary index.
OrderOperator order = null;
if (sortPrimaryKeys) {
order = new OrderOperator();
for (LogicalVariable pkVar : primaryKeyVars) {
Mutable<ILogicalExpression> vRef = new MutableObject<>(new VariableReferenceExpression(pkVar));
order.getOrderExpressions().add(new Pair<>(OrderOperator.ASC_ORDER, vRef));
}
// The secondary-index search feeds into the sort.
order.getInputs().add(new MutableObject<>(inputOp));
order.setExecutionMode(ExecutionMode.LOCAL);
context.computeAndSetTypeEnvironmentForOperator(order);
}
// The job gen parameters are transferred to the actual job gen via the UnnestMapOperator's function arguments.
List<Mutable<ILogicalExpression>> primaryIndexFuncArgs = new ArrayList<>();
BTreeJobGenParams jobGenParams = new BTreeJobGenParams(dataset.getDatasetName(), IndexType.BTREE, dataset.getDataverseName(), dataset.getDatasetName(), retainInput, requiresBroadcast);
// Set low/high inclusive to true for a point lookup.
jobGenParams.setLowKeyInclusive(true);
jobGenParams.setHighKeyInclusive(true);
jobGenParams.setLowKeyVarList(primaryKeyVars, 0, primaryKeyVars.size());
jobGenParams.setHighKeyVarList(primaryKeyVars, 0, primaryKeyVars.size());
jobGenParams.setIsEqCondition(true);
jobGenParams.writeToFuncArgs(primaryIndexFuncArgs);
// Variables and types coming out of the primary-index search.
List<LogicalVariable> primaryIndexUnnestVars = new ArrayList<>();
List<Object> primaryIndexOutputTypes = new ArrayList<>();
// Append output variables/types generated by the primary-index search (not forwarded from input).
primaryIndexUnnestVars.addAll(dataSourceOp.getVariables());
appendPrimaryIndexTypes(dataset, recordType, metaRecordType, primaryIndexOutputTypes);
// An index search is expressed as an unnest over an index-search function.
IFunctionInfo primaryIndexSearch = FunctionUtil.getFunctionInfo(BuiltinFunctions.INDEX_SEARCH);
AbstractFunctionCallExpression primaryIndexSearchFunc = new ScalarFunctionCallExpression(primaryIndexSearch, primaryIndexFuncArgs);
// This is the operator that jobgen will be looking for. It contains an unnest function that has all necessary arguments to determine
// which index to use, which variables contain the index-search keys, what is the original dataset, etc.
AbstractUnnestMapOperator primaryIndexUnnestOp = null;
if (retainNull) {
if (retainInput) {
primaryIndexUnnestOp = new LeftOuterUnnestMapOperator(primaryIndexUnnestVars, new MutableObject<ILogicalExpression>(primaryIndexSearchFunc), primaryIndexOutputTypes, retainInput);
} else {
// Left-outer-join without retainNull and retainInput doesn't make sense.
throw new AlgebricksException("Left-outer-join should propagate all inputs from the outer branch.");
}
} else {
primaryIndexUnnestOp = new UnnestMapOperator(primaryIndexUnnestVars, new MutableObject<ILogicalExpression>(primaryIndexSearchFunc), primaryIndexOutputTypes, retainInput);
}
// Fed by the order operator or the secondaryIndexUnnestOp.
if (sortPrimaryKeys) {
primaryIndexUnnestOp.getInputs().add(new MutableObject<ILogicalOperator>(order));
} else {
primaryIndexUnnestOp.getInputs().add(new MutableObject<>(inputOp));
}
context.computeAndSetTypeEnvironmentForOperator(primaryIndexUnnestOp);
primaryIndexUnnestOp.setExecutionMode(ExecutionMode.PARTITIONED);
return primaryIndexUnnestOp;
}
Aggregations