// traversing origin, find ExprNodeDesc in sources and replaces it with ExprNodeDesc
// in targets having same index.
// return null if failed to find
public static ExprNodeDesc replace(
ExprNodeDesc origin, List<ExprNodeDesc> sources, List<ExprNodeDesc> targets) {
int index = indexOf(origin, sources);
if (index >= 0) {
return targets.get(index);
}
// encountered column or field which cannot be found in sources
if (origin instanceof ExprNodeColumnDesc || origin instanceof ExprNodeFieldDesc) {
return null;
}
// for ExprNodeGenericFuncDesc, it should be deterministic and stateless
if (origin instanceof ExprNodeGenericFuncDesc) {
ExprNodeGenericFuncDesc func = (ExprNodeGenericFuncDesc) origin;
if (!FunctionRegistry.isDeterministic(func.getGenericUDF())
|| FunctionRegistry.isStateful(func.getGenericUDF())) {
return null;
}
List<ExprNodeDesc> children = new ArrayList<ExprNodeDesc>();
for (int i = 0; i < origin.getChildren().size(); i++) {
ExprNodeDesc child = replace(origin.getChildren().get(i), sources, targets);
if (child == null) {
return null;
}
children.add(child);
}
// duplicate function with possibly replaced children
ExprNodeGenericFuncDesc clone = (ExprNodeGenericFuncDesc) func.clone();
clone.setChildren(children);
return clone;
}
// constant or null, just return it
return origin;
}
/** return true if predicate is already included in source */
public static boolean containsPredicate(ExprNodeDesc source, ExprNodeDesc predicate) {
if (source.isSame(predicate)) {
return true;
}
if (FunctionRegistry.isOpAnd(source)) {
if (containsPredicate(source.getChildren().get(0), predicate)
|| containsPredicate(source.getChildren().get(1), predicate)) {
return true;
}
}
return false;
}
/** Return false if the expression has any non deterministic function */
public static boolean isDeterministic(ExprNodeDesc desc) {
if (desc instanceof ExprNodeGenericFuncDesc) {
if (!FunctionRegistry.isDeterministic(((ExprNodeGenericFuncDesc) desc).getGenericUDF())) {
return false;
}
}
if (desc.getChildren() != null) {
for (ExprNodeDesc child : desc.getChildren()) {
if (!isDeterministic(child)) {
return false;
}
}
}
return true;
}
/** bind two predicates by AND op */
public static ExprNodeGenericFuncDesc mergePredicates(ExprNodeDesc prev, ExprNodeDesc next) {
final List<ExprNodeDesc> children = new ArrayList<ExprNodeDesc>(2);
if (FunctionRegistry.isOpAnd(prev)) {
children.addAll(prev.getChildren());
} else {
children.add(prev);
}
if (FunctionRegistry.isOpAnd(next)) {
children.addAll(next.getChildren());
} else {
children.add(next);
}
return new ExprNodeGenericFuncDesc(
TypeInfoFactory.booleanTypeInfo, FunctionRegistry.getGenericUDFForAnd(), children);
}
// Check if ExprNodeColumnDesc is wrapped in expr.
// If so, peel off. Otherwise return itself.
private ExprNodeDesc getColumnExpr(ExprNodeDesc expr) {
if (expr instanceof ExprNodeColumnDesc) {
return expr;
}
ExprNodeGenericFuncDesc funcDesc = null;
if (expr instanceof ExprNodeGenericFuncDesc) {
funcDesc = (ExprNodeGenericFuncDesc) expr;
}
if (null == funcDesc) {
return expr;
}
GenericUDF udf = funcDesc.getGenericUDF();
// check if its a simple cast expression.
if ((udf instanceof GenericUDFBridge
|| udf instanceof GenericUDFToBinary
|| udf instanceof GenericUDFToChar
|| udf instanceof GenericUDFToVarchar
|| udf instanceof GenericUDFToDecimal
|| udf instanceof GenericUDFToDate
|| udf instanceof GenericUDFToUnixTimeStamp
|| udf instanceof GenericUDFToUtcTimestamp)
&& funcDesc.getChildren().size() == 1
&& funcDesc.getChildren().get(0) instanceof ExprNodeColumnDesc) {
return expr.getChildren().get(0);
}
return expr;
}
/**
* Get Map of ExprNodeColumnDesc HashCode to ExprNodeColumnDesc.
*
* @param exprDesc
* @param hashCodeToColumnDescMap Assumption: If two ExprNodeColumnDesc have same hash code then
* they are logically referring to same projection
*/
public static void getExprNodeColumnDesc(
ExprNodeDesc exprDesc, Map<Integer, ExprNodeDesc> hashCodeToColumnDescMap) {
if (exprDesc instanceof ExprNodeColumnDesc) {
hashCodeToColumnDescMap.put(exprDesc.hashCode(), exprDesc);
} else if (exprDesc instanceof ExprNodeColumnListDesc) {
for (ExprNodeDesc child : exprDesc.getChildren()) {
getExprNodeColumnDesc(child, hashCodeToColumnDescMap);
}
} else if (exprDesc instanceof ExprNodeGenericFuncDesc) {
for (ExprNodeDesc child : exprDesc.getChildren()) {
getExprNodeColumnDesc(child, hashCodeToColumnDescMap);
}
} else if (exprDesc instanceof ExprNodeFieldDesc) {
getExprNodeColumnDesc(((ExprNodeFieldDesc) exprDesc).getDesc(), hashCodeToColumnDescMap);
}
}
void dumpFilterExpr(ExprNodeDesc expr) {
if (expr == null) return;
List<ExprNodeDesc> children = expr.getChildren();
if (children != null && children.size() > 0) {
for (ExprNodeDesc e : children) {
dumpFilterExpr(e);
}
}
}
public static String extractColName(ExprNodeDesc root) {
if (root instanceof ExprNodeColumnDesc) {
return ((ExprNodeColumnDesc) root).getColumn();
} else {
if (root.getChildren() == null) {
return null;
}
String column = null;
for (ExprNodeDesc d : root.getChildren()) {
String candidate = extractColName(d);
if (column != null && candidate != null) {
return null;
} else if (candidate != null) {
column = candidate;
}
}
return column;
}
}
/** Recommend name for the expression */
public static String recommendInputName(ExprNodeDesc desc) {
if (desc instanceof ExprNodeColumnDesc) {
return ((ExprNodeColumnDesc) desc).getColumn();
}
List<ExprNodeDesc> children = desc.getChildren();
if (FunctionRegistry.isOpPreserveInputName(desc)
&& !children.isEmpty()
&& children.get(0) instanceof ExprNodeColumnDesc) {
return ((ExprNodeColumnDesc) children.get(0)).getColumn();
}
return null;
}
/** split predicates by AND op */
public static List<ExprNodeDesc> split(ExprNodeDesc current, List<ExprNodeDesc> splitted) {
if (FunctionRegistry.isOpAnd(current)) {
for (ExprNodeDesc child : current.getChildren()) {
split(child, splitted);
}
return splitted;
}
if (indexOf(current, splitted) < 0) {
splitted.add(current);
}
return splitted;
}
public static ExprNodeColumnDesc getColumnExpr(ExprNodeDesc expr) {
while (FunctionRegistry.isOpCast(expr)) {
expr = expr.getChildren().get(0);
}
return (expr instanceof ExprNodeColumnDesc) ? (ExprNodeColumnDesc) expr : null;
}