Fix behavior of null in IN and NOT IN operators with empty sets (#2310)

Co-authored-by: copilot-swe-agent[bot] <[email protected]>
Co-authored-by: mathiasrw <[email protected]>
Co-authored-by: Mathias Wulff <[email protected]>

Author: Copilot

src/50expression.js

@@ -294,6 +294,7 @@
 			var s;
 			let refs = [];
 			let op = this.op;
+			let skipNullCheck = false; // Flag to skip null checking for deterministic empty set operations
 			let _this = this;
 			let ref = function (expr) {
 				if (expr.toJS) {
@@ -375,7 +376,17 @@
 					const uncachedLookup = `(alasql.utils.flatArray(this.queriesfn[${this.queriesidx}](params, null, ${context})).indexOf(alasql.utils.getValueOf(${leftJS()})) > -1)`;
 					s = `(${checkCorrelated} ? ${uncachedLookup} : ${cachedLookup})`;
 				} else if (Array.isArray(this.right)) {
-					if (!alasql.options.cache || this.right.some(value => value instanceof yy.ParamValue)) {
+					// Empty array: nothing is IN an empty set, always false
+					if (this.right.length === 0) {
+						// Must call leftJS() to populate the refs array for the declareRefs statement,
+						// even though the result is not used in the final expression
+						leftJS();
+						s = 'false';
+						skipNullCheck = true; // Result is deterministic even with null operands
+					} else if (
+						!alasql.options.cache ||
+						this.right.some(value => value instanceof yy.ParamValue)
+					) {
 						// Leverage JS Set for faster lookups than arrays
 						s = `(new Set([${this.right.map(ref).join(',')}]).has(alasql.utils.getValueOf(${leftJS()})))`;
 					} else {
@@ -399,7 +410,17 @@
 					const uncachedLookup = `(alasql.utils.flatArray(this.queriesfn[${this.queriesidx}](params, null, ${context})).indexOf(alasql.utils.getValueOf(${leftJS()})) < 0)`;
 					s = `(${checkCorrelated} ? ${uncachedLookup} : ${cachedLookup})`;
 				} else if (Array.isArray(this.right)) {
-					if (!alasql.options.cache || this.right.some(value => value instanceof yy.ParamValue)) {
+					// Empty array: everything is NOT IN an empty set, always true
+					if (this.right.length === 0) {
+						// Must call leftJS() to populate the refs array for the declareRefs statement,
+						// even though the result is not used in the final expression
+						leftJS();
+						s = 'true';
+						skipNullCheck = true; // Result is deterministic even with null operands
+					} else if (
+						!alasql.options.cache ||
+						this.right.some(value => value instanceof yy.ParamValue)
+					) {
 						// Leverage JS Set for faster lookups than arrays
 						s = `(!(new Set([${this.right.map(ref).join(',')}]).has(alasql.utils.getValueOf(${leftJS()}))))`;
 					} else {
@@ -475,7 +496,14 @@
 			var expr = s || '(' + leftJS() + op + rightJS() + ')';
 
 			var declareRefs = 'y=[(' + refs.join('), (') + ')]';
-			if (op === '&&' || op === '||' || op === 'IS' || op === 'IS NULL' || op === 'IS NOT NULL') {
+			if (
+				skipNullCheck ||
+				op === '&&' ||
+				op === '||' ||
+				op === 'IS' ||
+				op === 'IS NULL' ||
+				op === 'IS NOT NULL'
+			) {
 				return '(' + declareRefs + ', ' + expr + ')';
 			}
 

test/test1529.js

@@ -0,0 +1,28 @@
+if (typeof exports === 'object') {
+	var assert = require('assert');
+	var alasql = require('..');
+}
+
+describe('Test 1529 - SELECT null [not] IN ()', function () {
+	it('1. null IN () should return false', function (done) {
+		var res = alasql('SELECT VALUE null IN ()');
+		assert.strictEqual(res, false, 'null IN () should return false, got ' + res);
+		done();
+	});
+
+	it('2. null NOT IN () should return true', function (done) {
+		var res = alasql('SELECT VALUE null NOT IN ()');
+		assert.strictEqual(res, true, 'null NOT IN () should return true, got ' + res);
+		done();
+	});
+
+	it('3. Verify non-null values still work correctly', function (done) {
+		var res1 = alasql('SELECT VALUE 1 IN ()');
+		assert.strictEqual(res1, false, '1 IN () should return false');
+
+		var res2 = alasql('SELECT VALUE 1 NOT IN ()');
+		assert.strictEqual(res2, true, '1 NOT IN () should return true');
+
+		done();
+	});
+});