private ILogicalExpression makeCondition(
List<Mutable<ILogicalExpression>> predList, IOptimizationContext context) {
if (predList.size() > 1) {
IFunctionInfo finfo =
context.getMetadataProvider().lookupFunction(AlgebricksBuiltinFunctions.AND);
return new ScalarFunctionCallExpression(finfo, predList);
} else {
return predList.get(0).getValue();
}
}
private int removeFromAssigns(
AbstractLogicalOperator op, Set<LogicalVariable> toRemove, IOptimizationContext context)
throws AlgebricksException {
switch (op.getOperatorTag()) {
case ASSIGN:
{
AssignOperator assign = (AssignOperator) op;
if (removeUnusedVarsAndExprs(toRemove, assign.getVariables(), assign.getExpressions())) {
context.computeAndSetTypeEnvironmentForOperator(assign);
}
return assign.getVariables().size();
}
case AGGREGATE:
{
AggregateOperator agg = (AggregateOperator) op;
if (removeUnusedVarsAndExprs(toRemove, agg.getVariables(), agg.getExpressions())) {
context.computeAndSetTypeEnvironmentForOperator(agg);
}
return agg.getVariables().size();
}
case UNNEST:
{
UnnestOperator uOp = (UnnestOperator) op;
LogicalVariable pVar = uOp.getPositionalVariable();
if (pVar != null && toRemove.contains(pVar)) {
uOp.setPositionalVariable(null);
}
break;
}
case UNIONALL:
{
UnionAllOperator unionOp = (UnionAllOperator) op;
if (removeUnusedVarsFromUnionAll(unionOp, toRemove)) {
context.computeAndSetTypeEnvironmentForOperator(unionOp);
}
return unionOp.getVariableMappings().size();
}
}
return -1;
}
@Override
public boolean rewritePre(Mutable<ILogicalOperator> opRef, IOptimizationContext context)
throws AlgebricksException {
if (context.checkIfInDontApplySet(this, opRef.getValue())) {
return false;
}
Set<LogicalVariable> toRemove = new HashSet<LogicalVariable>();
collectUnusedAssignedVars((AbstractLogicalOperator) opRef.getValue(), toRemove, true, context);
boolean smthToRemove = !toRemove.isEmpty();
if (smthToRemove) {
removeUnusedAssigns(opRef, toRemove, context);
}
return !toRemove.isEmpty();
}
@Override
public boolean rewritePost(Mutable<ILogicalOperator> opRef, IOptimizationContext context) {
AbstractLogicalOperator op = (AbstractLogicalOperator) opRef.getValue();
if (op.getOperatorTag() != LogicalOperatorTag.GROUP) {
return false;
}
if (context.checkIfInDontApplySet(this, op)) {
return false;
}
vars.clear();
boolean modified = false;
GroupByOperator groupOp = (GroupByOperator) op;
Iterator<Pair<LogicalVariable, Mutable<ILogicalExpression>>> iter =
groupOp.getDecorList().iterator();
while (iter.hasNext()) {
Pair<LogicalVariable, Mutable<ILogicalExpression>> decor = iter.next();
if (decor.first != null
|| decor.second.getValue().getExpressionTag() != LogicalExpressionTag.VARIABLE) {
continue;
}
VariableReferenceExpression varRefExpr =
(VariableReferenceExpression) decor.second.getValue();
LogicalVariable var = varRefExpr.getVariableReference();
if (vars.contains(var)) {
iter.remove();
modified = true;
} else {
vars.add(var);
}
}
if (modified) {
context.addToDontApplySet(this, op);
}
return modified;
}
@Override
public boolean rewritePost(Mutable<ILogicalOperator> opRef, IOptimizationContext context)
throws AlgebricksException {
AbstractLogicalOperator op = (AbstractLogicalOperator) opRef.getValue();
if (op.getOperatorTag() != LogicalOperatorTag.INNERJOIN) {
return false;
}
AbstractBinaryJoinOperator join = (AbstractBinaryJoinOperator) op;
ILogicalExpression expr = join.getCondition().getValue();
if (expr.getExpressionTag() != LogicalExpressionTag.FUNCTION_CALL) {
return false;
}
AbstractFunctionCallExpression fexp = (AbstractFunctionCallExpression) expr;
FunctionIdentifier fi = fexp.getFunctionIdentifier();
if (!fi.equals(AlgebricksBuiltinFunctions.AND)) {
return false;
}
List<Mutable<ILogicalExpression>> eqVarVarComps = new ArrayList<Mutable<ILogicalExpression>>();
List<Mutable<ILogicalExpression>> otherPredicates =
new ArrayList<Mutable<ILogicalExpression>>();
for (Mutable<ILogicalExpression> arg : fexp.getArguments()) {
if (isEqVarVar(arg.getValue())) {
eqVarVarComps.add(arg);
} else {
otherPredicates.add(arg);
}
}
if (eqVarVarComps.isEmpty() || otherPredicates.isEmpty()) {
return false;
}
// pull up
ILogicalExpression pulledCond = makeCondition(otherPredicates, context);
SelectOperator select =
new SelectOperator(new MutableObject<ILogicalExpression>(pulledCond), false, null);
ILogicalExpression newJoinCond = makeCondition(eqVarVarComps, context);
join.getCondition().setValue(newJoinCond);
select.getInputs().add(new MutableObject<ILogicalOperator>(join));
opRef.setValue(select);
context.computeAndSetTypeEnvironmentForOperator(select);
return true;
}
private void collectUnusedAssignedVars(
AbstractLogicalOperator op,
Set<LogicalVariable> toRemove,
boolean first,
IOptimizationContext context)
throws AlgebricksException {
if (!first) {
context.addToDontApplySet(this, op);
}
for (Mutable<ILogicalOperator> c : op.getInputs()) {
collectUnusedAssignedVars((AbstractLogicalOperator) c.getValue(), toRemove, false, context);
}
if (op.hasNestedPlans()) {
AbstractOperatorWithNestedPlans opWithNested = (AbstractOperatorWithNestedPlans) op;
for (ILogicalPlan plan : opWithNested.getNestedPlans()) {
for (Mutable<ILogicalOperator> r : plan.getRoots()) {
collectUnusedAssignedVars(
(AbstractLogicalOperator) r.getValue(), toRemove, false, context);
}
}
}
boolean removeUsedVars = true;
switch (op.getOperatorTag()) {
case ASSIGN:
{
AssignOperator assign = (AssignOperator) op;
toRemove.addAll(assign.getVariables());
break;
}
case AGGREGATE:
{
AggregateOperator agg = (AggregateOperator) op;
toRemove.addAll(agg.getVariables());
break;
}
case UNNEST:
{
UnnestOperator uOp = (UnnestOperator) op;
LogicalVariable pVar = uOp.getPositionalVariable();
if (pVar != null) {
toRemove.add(pVar);
}
break;
}
case UNIONALL:
{
UnionAllOperator unionOp = (UnionAllOperator) op;
for (Triple<LogicalVariable, LogicalVariable, LogicalVariable> varMapping :
unionOp.getVariableMappings()) {
toRemove.add(varMapping.third);
}
removeUsedVars = false;
break;
}
}
if (removeUsedVars) {
List<LogicalVariable> used = new LinkedList<LogicalVariable>();
VariableUtilities.getUsedVariables(op, used);
toRemove.removeAll(used);
}
}
@Override
public boolean rewritePost(Mutable<ILogicalOperator> opRef, IOptimizationContext context)
throws AlgebricksException {
AbstractLogicalOperator unnest = (AbstractLogicalOperator) opRef.getValue();
if (unnest.getOperatorTag() != LogicalOperatorTag.UNNEST) {
return false;
}
UnnestOperator unnestOpRef = (UnnestOperator) opRef.getValue();
Mutable<ILogicalOperator> unionOp = unnest.getInputs().get(0);
AbstractLogicalOperator unionAbstractOp = (AbstractLogicalOperator) unionOp.getValue();
if (unionAbstractOp.getOperatorTag() != LogicalOperatorTag.UNIONALL) {
return false;
}
LogicalVariable unnestVar1 = context.newVar();
UnnestOperator unnest1 =
new UnnestOperator(
unnestVar1,
new MutableObject<ILogicalExpression>(
unnestOpRef.getExpressionRef().getValue().cloneExpression()));
LogicalVariable unnestVar2 = context.newVar();
UnnestOperator unnest2 =
new UnnestOperator(
unnestVar2,
new MutableObject<ILogicalExpression>(
unnestOpRef.getExpressionRef().getValue().cloneExpression()));
// Getting the two topmost branched and adding them as an input to the unnests:
Mutable<ILogicalOperator> branch1 = unionAbstractOp.getInputs().get(0);
ILogicalOperator agg1 = branch1.getValue();
List<LogicalVariable> agg1_var = new ArrayList<LogicalVariable>();
VariableUtilities.getLiveVariables(agg1, agg1_var);
Mutable<ILogicalOperator> branch2 = unionAbstractOp.getInputs().get(1);
ILogicalOperator agg2 = branch2.getValue();
List<LogicalVariable> agg2_var = new ArrayList<LogicalVariable>();
VariableUtilities.getLiveVariables(agg2, agg2_var);
// Modifying the unnest so it has the right variable
List<LogicalVariable> var_unnest_1 = new ArrayList<LogicalVariable>();
unnest1.getExpressionRef().getValue().getUsedVariables(var_unnest_1);
unnest1.getExpressionRef().getValue().substituteVar(var_unnest_1.get(0), agg1_var.get(0));
List<LogicalVariable> var_unnest2 = new ArrayList<LogicalVariable>();
unnest2.getExpressionRef().getValue().getUsedVariables(var_unnest2);
unnest2.getExpressionRef().getValue().substituteVar(var_unnest2.get(0), agg2_var.get(0));
unnest1.getInputs().add(branch1);
unnest2.getInputs().add(branch2);
context.computeAndSetTypeEnvironmentForOperator(unnest1);
context.computeAndSetTypeEnvironmentForOperator(unnest2);
// creating a new union operator with the updated logical variables
List<Triple<LogicalVariable, LogicalVariable, LogicalVariable>> varMap =
new ArrayList<Triple<LogicalVariable, LogicalVariable, LogicalVariable>>(1);
Triple<LogicalVariable, LogicalVariable, LogicalVariable> union_triple_vars =
new Triple<LogicalVariable, LogicalVariable, LogicalVariable>(
unnestVar1, unnestVar2, unnestOpRef.getVariables().get(0));
varMap.add(union_triple_vars);
UnionAllOperator unionOpFinal = new UnionAllOperator(varMap);
unionOpFinal.getInputs().add(new MutableObject<ILogicalOperator>(unnest1));
unionOpFinal.getInputs().add(new MutableObject<ILogicalOperator>(unnest2));
context.computeAndSetTypeEnvironmentForOperator(unionOpFinal);
opRef.setValue(unionOpFinal);
return true;
}
@Override
public boolean rewritePost(Mutable<ILogicalOperator> opRef, IOptimizationContext context)
throws AlgebricksException {
AbstractLogicalOperator op = (AbstractLogicalOperator) opRef.getValue();
// current opperator is join
if (op.getOperatorTag() != LogicalOperatorTag.INNERJOIN
&& op.getOperatorTag() != LogicalOperatorTag.LEFTOUTERJOIN) {
return false;
}
// Find GET_ITEM function.
AbstractBinaryJoinOperator joinOp = (AbstractBinaryJoinOperator) op;
Mutable<ILogicalExpression> expRef = joinOp.getCondition();
Mutable<ILogicalExpression> getItemExprRef = getSimilarityExpression(expRef);
if (getItemExprRef == null) {
return false;
}
// Check if the GET_ITEM function is on one of the supported similarity-check functions.
AbstractFunctionCallExpression getItemFuncExpr =
(AbstractFunctionCallExpression) getItemExprRef.getValue();
Mutable<ILogicalExpression> argRef = getItemFuncExpr.getArguments().get(0);
AbstractFunctionCallExpression simFuncExpr = (AbstractFunctionCallExpression) argRef.getValue();
if (!simFuncs.contains(simFuncExpr.getFunctionIdentifier())) {
return false;
}
// Skip this rule based on annotations.
if (simFuncExpr.getAnnotations().containsKey(IndexedNLJoinExpressionAnnotation.INSTANCE)) {
return false;
}
List<Mutable<ILogicalOperator>> inputOps = joinOp.getInputs();
ILogicalOperator leftInputOp = inputOps.get(0).getValue();
ILogicalOperator rightInputOp = inputOps.get(1).getValue();
List<Mutable<ILogicalExpression>> inputExps = simFuncExpr.getArguments();
ILogicalExpression inputExp0 = inputExps.get(0).getValue();
ILogicalExpression inputExp1 = inputExps.get(1).getValue();
// left and right expressions are variables
if (inputExp0.getExpressionTag() != LogicalExpressionTag.VARIABLE
|| inputExp1.getExpressionTag() != LogicalExpressionTag.VARIABLE) {
return false;
}
LogicalVariable inputVar0 = ((VariableReferenceExpression) inputExp0).getVariableReference();
LogicalVariable inputVar1 = ((VariableReferenceExpression) inputExp1).getVariableReference();
LogicalVariable leftInputVar;
LogicalVariable rightInputVar;
liveVars.clear();
VariableUtilities.getLiveVariables(leftInputOp, liveVars);
if (liveVars.contains(inputVar0)) {
leftInputVar = inputVar0;
rightInputVar = inputVar1;
} else {
leftInputVar = inputVar1;
rightInputVar = inputVar0;
}
List<LogicalVariable> leftInputPKs = context.findPrimaryKey(leftInputVar);
List<LogicalVariable> rightInputPKs = context.findPrimaryKey(rightInputVar);
// Bail if primary keys could not be inferred.
if (leftInputPKs == null || rightInputPKs == null) {
return false;
}
// primary key has only one variable
if (leftInputPKs.size() != 1 || rightInputPKs.size() != 1) {
return false;
}
IAType leftType =
(IAType) context.getOutputTypeEnvironment(leftInputOp).getVarType(leftInputVar);
IAType rightType =
(IAType) context.getOutputTypeEnvironment(rightInputOp).getVarType(rightInputVar);
// left-hand side and right-hand side of "~=" has the same type
IAType left2 = TypeHelper.getNonOptionalType(leftType);
IAType right2 = TypeHelper.getNonOptionalType(rightType);
if (!left2.deepEqual(right2)) {
return false;
}
//
// -- - FIRE - --
//
AqlMetadataProvider metadataProvider = ((AqlMetadataProvider) context.getMetadataProvider());
FunctionIdentifier funcId = FuzzyUtils.getTokenizer(leftType.getTypeTag());
String tokenizer;
if (funcId == null) {
tokenizer = "";
} else {
tokenizer = funcId.getName();
}
float simThreshold = FuzzyUtils.getSimThreshold(metadataProvider);
String simFunction = FuzzyUtils.getSimFunction(metadataProvider);
// finalize AQL+ query
String prepareJoin;
switch (joinOp.getJoinKind()) {
case INNER:
{
prepareJoin = "join" + AQLPLUS;
break;
}
case LEFT_OUTER:
{
// TODO To make it work for Left Outer Joins, we should permute
// the #LEFT and #RIGHT at the top of the AQL+ query. But, when
// doing this, the
// fuzzyjoin/user-vis-int-vis-user-lot-aqlplus_1.aql (the one
// doing 3-way fuzzy joins) gives a different result. But even
// if we don't change the FuzzyJoinRule, permuting the for
// clauses in fuzzyjoin/user-vis-int-vis-user-lot-aqlplus_1.aql
// leads to different results, which suggests there is some
// other sort of bug.
return false;
// prepareJoin = "loj" + AQLPLUS;
// break;
}
default:
{
throw new IllegalStateException();
}
}
String aqlPlus =
String.format(
Locale.US,
prepareJoin,
tokenizer,
tokenizer,
simFunction,
simThreshold,
tokenizer,
tokenizer,
simFunction,
simThreshold,
simFunction,
simThreshold,
simThreshold);
LogicalVariable leftPKVar = leftInputPKs.get(0);
LogicalVariable rightPKVar = rightInputPKs.get(0);
Counter counter = new Counter(context.getVarCounter());
AQLPlusParser parser = new AQLPlusParser(new StringReader(aqlPlus));
parser.initScope();
parser.setVarCounter(counter);
List<Clause> clauses;
try {
clauses = parser.Clauses();
} catch (ParseException e) {
throw new AlgebricksException(e);
}
// The translator will compile metadata internally. Run this compilation
// under the same transaction id as the "outer" compilation.
AqlPlusExpressionToPlanTranslator translator =
new AqlPlusExpressionToPlanTranslator(
metadataProvider.getJobId(), metadataProvider, counter, null, null);
context.setVarCounter(counter.get());
LogicalOperatorDeepCopyWithNewVariablesVisitor deepCopyVisitor =
new LogicalOperatorDeepCopyWithNewVariablesVisitor(context);
translator.addOperatorToMetaScope(new Identifier("#LEFT"), leftInputOp);
translator.addVariableToMetaScope(new Identifier("$$LEFT"), leftInputVar);
translator.addVariableToMetaScope(new Identifier("$$LEFTPK"), leftPKVar);
translator.addOperatorToMetaScope(new Identifier("#RIGHT"), rightInputOp);
translator.addVariableToMetaScope(new Identifier("$$RIGHT"), rightInputVar);
translator.addVariableToMetaScope(new Identifier("$$RIGHTPK"), rightPKVar);
translator.addOperatorToMetaScope(
new Identifier("#LEFT_1"), deepCopyVisitor.deepCopy(leftInputOp, null));
translator.addVariableToMetaScope(
new Identifier("$$LEFT_1"), deepCopyVisitor.varCopy(leftInputVar));
translator.addVariableToMetaScope(
new Identifier("$$LEFTPK_1"), deepCopyVisitor.varCopy(leftPKVar));
deepCopyVisitor.updatePrimaryKeys(context);
deepCopyVisitor.reset();
// translator.addOperatorToMetaScope(new Identifier("#LEFT_2"),
// deepCopyVisitor.deepCopy(leftInputOp, null));
// translator.addVariableToMetaScope(new Identifier("$$LEFT_2"),
// deepCopyVisitor.varCopy(leftInputVar));
// translator.addVariableToMetaScope(new Identifier("$$LEFTPK_2"),
// deepCopyVisitor.varCopy(leftPKVar));
// deepCopyVisitor.updatePrimaryKeys(context);
// deepCopyVisitor.reset();
//
// translator.addOperatorToMetaScope(new Identifier("#LEFT_3"),
// deepCopyVisitor.deepCopy(leftInputOp, null));
// translator.addVariableToMetaScope(new Identifier("$$LEFT_3"),
// deepCopyVisitor.varCopy(leftInputVar));
// translator.addVariableToMetaScope(new Identifier("$$LEFTPK_3"),
// deepCopyVisitor.varCopy(leftPKVar));
// deepCopyVisitor.updatePrimaryKeys(context);
// deepCopyVisitor.reset();
translator.addOperatorToMetaScope(
new Identifier("#RIGHT_1"), deepCopyVisitor.deepCopy(rightInputOp, null));
translator.addVariableToMetaScope(
new Identifier("$$RIGHT_1"), deepCopyVisitor.varCopy(rightInputVar));
translator.addVariableToMetaScope(
new Identifier("$$RIGHTPK_1"), deepCopyVisitor.varCopy(rightPKVar));
deepCopyVisitor.updatePrimaryKeys(context);
deepCopyVisitor.reset();
// TODO pick side to run Stage 1, currently always picks RIGHT side
translator.addOperatorToMetaScope(
new Identifier("#RIGHT_2"), deepCopyVisitor.deepCopy(rightInputOp, null));
translator.addVariableToMetaScope(
new Identifier("$$RIGHT_2"), deepCopyVisitor.varCopy(rightInputVar));
translator.addVariableToMetaScope(
new Identifier("$$RIGHTPK_2"), deepCopyVisitor.varCopy(rightPKVar));
deepCopyVisitor.updatePrimaryKeys(context);
deepCopyVisitor.reset();
translator.addOperatorToMetaScope(
new Identifier("#RIGHT_3"), deepCopyVisitor.deepCopy(rightInputOp, null));
translator.addVariableToMetaScope(
new Identifier("$$RIGHT_3"), deepCopyVisitor.varCopy(rightInputVar));
translator.addVariableToMetaScope(
new Identifier("$$RIGHTPK_3"), deepCopyVisitor.varCopy(rightPKVar));
deepCopyVisitor.updatePrimaryKeys(context);
deepCopyVisitor.reset();
ILogicalPlan plan;
try {
plan = translator.translate(clauses);
} catch (AsterixException e) {
throw new AlgebricksException(e);
}
context.setVarCounter(counter.get());
ILogicalOperator outputOp = plan.getRoots().get(0).getValue();
SelectOperator extraSelect = null;
if (getItemExprRef != expRef) {
// more than one join condition
getItemExprRef.setValue(ConstantExpression.TRUE);
switch (joinOp.getJoinKind()) {
case INNER:
{
extraSelect = new SelectOperator(expRef, false, null);
extraSelect.getInputs().add(new MutableObject<ILogicalOperator>(outputOp));
outputOp = extraSelect;
break;
}
case LEFT_OUTER:
{
if (((AbstractLogicalOperator) outputOp).getOperatorTag()
!= LogicalOperatorTag.LEFTOUTERJOIN) {
throw new IllegalStateException();
}
LeftOuterJoinOperator topJoin = (LeftOuterJoinOperator) outputOp;
topJoin.getCondition().setValue(expRef.getValue());
break;
}
default:
{
throw new IllegalStateException();
}
}
}
opRef.setValue(outputOp);
OperatorPropertiesUtil.typeOpRec(opRef, context);
return true;
}
@Override
public boolean rewritePost(Mutable<ILogicalOperator> opRef, IOptimizationContext context)
throws AlgebricksException {
if (!opRef.getValue().getOperatorTag().equals(LogicalOperatorTag.ASSIGN)) {
return false;
}
AssignOperator assignUnion = (AssignOperator) opRef.getValue();
if (assignUnion.getExpressions().get(0).getValue().getExpressionTag()
!= LogicalExpressionTag.FUNCTION_CALL) return false;
AbstractFunctionCallExpression u =
(AbstractFunctionCallExpression) assignUnion.getExpressions().get(0).getValue();
if (!AsterixBuiltinFunctions.UNION.equals(u.getFunctionIdentifier())) {
return false;
}
// Retrieving the logical variables for the union from the two aggregates which are inputs to
// the join
Mutable<ILogicalOperator> join = assignUnion.getInputs().get(0);
LogicalOperatorTag tag1 = join.getValue().getOperatorTag();
if (tag1 != LogicalOperatorTag.INNERJOIN && tag1 != LogicalOperatorTag.LEFTOUTERJOIN) {
return false;
}
AbstractBinaryJoinOperator join1 = (AbstractBinaryJoinOperator) join.getValue();
ILogicalExpression cond1 = join1.getCondition().getValue();
// don't try to push a product down
if (!OperatorPropertiesUtil.isAlwaysTrueCond(cond1)) {
return false;
}
List<Mutable<ILogicalOperator>> joinInputs = join.getValue().getInputs();
Mutable<ILogicalOperator> left_branch = joinInputs.get(0);
Mutable<ILogicalOperator> right_branch = joinInputs.get(1);
List<LogicalVariable> input1Var = new ArrayList<LogicalVariable>();
VariableUtilities.getProducedVariables(left_branch.getValue(), input1Var);
List<LogicalVariable> input2Var = new ArrayList<LogicalVariable>();
VariableUtilities.getProducedVariables(right_branch.getValue(), input2Var);
List<Triple<LogicalVariable, LogicalVariable, LogicalVariable>> varMap =
new ArrayList<Triple<LogicalVariable, LogicalVariable, LogicalVariable>>(1);
Triple<LogicalVariable, LogicalVariable, LogicalVariable> triple =
new Triple<LogicalVariable, LogicalVariable, LogicalVariable>(
input1Var.get(0), input2Var.get(0), assignUnion.getVariables().get(0));
varMap.add(triple);
UnionAllOperator unionOp = new UnionAllOperator(varMap);
unionOp.getInputs().add(left_branch);
unionOp.getInputs().add(right_branch);
context.computeAndSetTypeEnvironmentForOperator(unionOp);
opRef.setValue(unionOp);
return true;
}