Package org.elasticsearch.xpack.esql.optimizer.rules.logical.preoptimizer

Class FoldInferenceFunctions

java.lang.Object

org.elasticsearch.xpack.esql.optimizer.rules.logical.preoptimizer.FoldInferenceFunctions

All Implemented Interfaces:

LogicalPlanPreOptimizerRule

public class FoldInferenceFunctions
extends Object
implements LogicalPlanPreOptimizerRule

Pre-optimizer rule that performs constant folding for inference functions in logical plans.

This rule identifies inference functions with constant parameters and evaluates them at optimization time, replacing them with their computed results.

The folding process is recursive and handles nested inference functions by processing them in multiple passes until no more foldable functions remain.

Example transformation: TEXT_EMBEDDING("hello world", "model1") → [0.1, 0.2, 0.3, ...]

Constructor Summary

Constructors

Modifier	Constructor	Description
protected	FoldInferenceFunctions(InferenceFunctionEvaluator inferenceFunctionEvaluator)
	FoldInferenceFunctions(LogicalPreOptimizerContext preOptimizerContext)

Method Summary

Modifier and Type	Method	Description
void	apply(LogicalPlan plan, ActionListener<LogicalPlan> listener)	Apply the rule to the logical plan.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

FoldInferenceFunctions

public FoldInferenceFunctions(LogicalPreOptimizerContext preOptimizerContext)

FoldInferenceFunctions

protected FoldInferenceFunctions(InferenceFunctionEvaluator inferenceFunctionEvaluator)

Method Details

apply

public void apply(LogicalPlan plan,
 ActionListener<LogicalPlan> listener)

Description copied from interface: LogicalPlanPreOptimizerRule

Apply the rule to the logical plan.

Specified by:

apply in interface LogicalPlanPreOptimizerRule

Parameters:

plan - the analyzed logical plan to pre-optimize

listener - the listener returning the pre-optimized plan when pre-optimization rule is applied