blacklist function that may output multivalued

Author: jan-elastic

x-pack/plugin/esql/src/main/java/org/elasticsearch/xpack/esql/approximate/Approximate.java

@@ -30,6 +30,7 @@
 import org.elasticsearch.xpack.esql.expression.function.aggregate.StdDev;
 import org.elasticsearch.xpack.esql.expression.function.aggregate.Sum;
 import org.elasticsearch.xpack.esql.expression.function.aggregate.WeightedAvg;
+import org.elasticsearch.xpack.esql.expression.function.scalar.EsqlScalarFunction;
 import org.elasticsearch.xpack.esql.expression.function.scalar.convert.ToLong;
 import org.elasticsearch.xpack.esql.expression.function.scalar.multivalue.ConfidenceInterval;
 import org.elasticsearch.xpack.esql.expression.function.scalar.multivalue.MvAppend;
@@ -142,6 +143,10 @@ public interface LogicalPlanRunner {
         org.elasticsearch.xpack.esql.expression.function.aggregate.Sample.class
     );
 
+    private static final Set<Class<? extends EsqlScalarFunction>> MULTI_VALUED_OUTPUT_FUNCTIONS = Set.of(
+        MvAppend.class
+    );
+
     // TODO: find a good default value, or alternative ways of setting it
     private static final int SAMPLE_ROW_COUNT = 100000;
 
@@ -402,7 +407,7 @@ private LogicalPlan approximatePlan(double sampleProbability) {
                     case Eval eval:
                         List<Alias> newFields = new ArrayList<>(eval.fields());
                         for (Alias field : eval.fields()) {
-                            if (field.dataType().isNumeric() == false || field.child().anyMatch(expr -> expr instanceof MvAppend)) {
+                            if (field.dataType().isNumeric() == false || field.child().anyMatch(expr -> MULTI_VALUED_OUTPUT_FUNCTIONS.contains(expr.getClass()))) {
                                 continue;
                             }
                             if (field.child().anyMatch(expr -> expr instanceof NamedExpression named && variablesWithConfidenceInterval.containsKey(named.id()))) {

x-pack/plugin/esql/src/main/java/org/elasticsearch/xpack/esql/expression/function/scalar/multivalue/ConfidenceInterval.java

@@ -131,7 +131,9 @@ public boolean equals(Object obj) {
 
     @Evaluator(extraName = "Double")
     static void process(DoubleBlock.Builder builder, @Position int position, DoubleBlock bestEstimateBlock, DoubleBlock estimatesBlock) {
-        assert bestEstimateBlock.getValueCount(position) == 1 : "bestEstimate: expected 1 element, got " + bestEstimateBlock.getValueCount(position);
+        if (bestEstimateBlock.getValueCount(position) != 1) {
+            builder.appendNull();
+        }
         Number bestEstimate = bestEstimateBlock.getDouble(bestEstimateBlock.getFirstValueIndex(position));
 
         Number[] estimates = new Number[estimatesBlock.getValueCount(position)];
@@ -145,13 +147,14 @@ static void process(DoubleBlock.Builder builder, @Position int position, DoubleB
             builder.appendDouble(v.doubleValue());
         }
         builder.endPositionEntry();
-
-        System.out.println("@@@ bestEstimate = " + bestEstimate + ", estimates = " + Arrays.toString(estimates) + " --> confidenceInterval = " + Arrays.toString(confidenceInterval));
     }
 
     @Evaluator(extraName = "Int")
     static void process(IntBlock.Builder builder, @Position int position, IntBlock bestEstimateBlock, IntBlock estimatesBlock) {
-        assert bestEstimateBlock.getValueCount(position) == 1 : "bestEstimate: expected 1 element, got " + bestEstimateBlock.getValueCount(position);
+        if (bestEstimateBlock.getValueCount(position) != 1) {
+            builder.appendNull();
+            return;
+        }
         Number bestEstimate = bestEstimateBlock.getInt(bestEstimateBlock.getFirstValueIndex(position));
 
         Number[] estimates = new Number[estimatesBlock.getValueCount(position)];
@@ -169,7 +172,10 @@ static void process(IntBlock.Builder builder, @Position int position, IntBlock b
 
     @Evaluator(extraName = "Long")
     static void process(LongBlock.Builder builder, @Position int position, LongBlock bestEstimateBlock, LongBlock estimatesBlock) {
-        assert bestEstimateBlock.getValueCount(position) == 1 : "bestEstimate: expected 1 element, got " + bestEstimateBlock.getValueCount(position);
+        if (bestEstimateBlock.getValueCount(position) != 1) {
+            builder.appendNull();
+            return;
+        }
         Number bestEstimate = bestEstimateBlock.getLong(bestEstimateBlock.getFirstValueIndex(position));
 
         Number[] estimates = new Number[estimatesBlock.getValueCount(position)];