Allow table names starting with numbers to fix #1185 (#2316)

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

Author: Copilot

src/alasqlparser.jison

@@ -270,10 +270,9 @@ SETS                                        	return 'SET'
 'WITH'                                          return 'WITH'
 'WORK'                                          return 'TRANSACTION'  /* Is this keyword required? */
 
-/* Issue #1173: Reject invalid identifiers like 6minAvgOpac (number followed by letters without space) */
-/* This rule must precede NUMBER to catch invalid patterns before they're tokenized as NUMBER + LITERAL */
-\d+[a-zA-Z_][a-zA-Z_0-9]*						return 'INVALID'
-(\d+\.?\d*|\.\d+)([eE][+-]?\d+)?				return 'NUMBER'
+/* LITERAL must come before NUMBER to match identifiers like 50k, 100x as single tokens */
+[0-9]*[a-zA-Z_]+[a-zA-Z_0-9]* 					return 'LITERAL'
+(\d+\.?\d*|\.\d+)([eE][+-]?\d+)?(?![a-zA-Z_0-9])				return 'NUMBER'
 '->'											return 'ARROW'
 '#'												return 'SHARP'
 '+'												return 'PLUS'
@@ -318,7 +317,6 @@ SETS                                        	return 'SET'
 
 '~'												return 'TILDA'
 
-[0-9]*[a-zA-Z_]+[a-zA-Z_0-9]* 					return 'LITERAL'
 '@'												return 'AT'
 <<EOF>>               							return 'EOF'
 .												return 'INVALID'

src/alasqlparser.js

@@ -3,7 +3,7 @@ if (typeof exports === 'object') {
 	var alasql = require('..');
 }
 
-describe('Test 1173 - Parser should error on column names starting with numbers', function () {
+describe('Test 1173 - Parser behavior for identifiers with special patterns', function () {
 	const test = '1173';
 
 	before(function () {
@@ -31,30 +31,21 @@ describe('Test 1173 - Parser should error on column names starting with numbers'
 		assert.equal(ast.statements[0].columns[1].columnid, '6minAvgOpac');
 	});
 
-	it('C) Should throw error for unquoted column name starting with number', function () {
+	it('C) Should now parse unquoted column name starting with number (behavior changed in #1185)', function () {
+		// After fix for #1185, identifiers starting with numbers are now allowed
 		var sql = 'SELECT ID, 6minAvgOpac FROM test';
-		assert.throws(
-			function () {
-				alasql.parse(sql);
-			},
-			function (err) {
-				return err instanceof SyntaxError || err instanceof Error;
-			},
-			'Should throw a SyntaxError for invalid column name starting with number'
-		);
+		var ast = alasql.parse(sql);
+		assert.equal(ast.statements[0].columns.length, 2);
+		assert.equal(ast.statements[0].columns[0].columnid, 'ID');
+		assert.equal(ast.statements[0].columns[1].columnid, '6minAvgOpac');
 	});
 
-	it('D) Should throw error for similar invalid identifier pattern', function () {
+	it('D) Should now parse identifiers like 123abc (behavior changed in #1185)', function () {
+		// After fix for #1185, identifiers starting with numbers are now allowed
 		var sql = 'SELECT 123abc FROM test';
-		assert.throws(
-			function () {
-				alasql.parse(sql);
-			},
-			function (err) {
-				return err instanceof SyntaxError || err instanceof Error;
-			},
-			'Should throw a SyntaxError for invalid identifier 123abc'
-		);
+		var ast = alasql.parse(sql);
+		assert.equal(ast.statements[0].columns.length, 1);
+		assert.equal(ast.statements[0].columns[0].columnid, '123abc');
 	});
 
 	it('E) Valid number literals followed by proper aliases should still work', function () {

test/test1173.js

@@ -0,0 +1,47 @@
+if (typeof exports === 'object') {
+	var assert = require('assert');
+	var alasql = require('..');
+}
+
+describe('Test 1185 - Table names starting with numbers', function () {
+	const test = '1185'; // insert test file number
+
+	before(function () {
+		alasql('create database test' + test);
+		alasql('use test' + test);
+	});
+
+	after(function () {
+		alasql('drop database test' + test);
+	});
+
+	it('A) Table name starting with number (e.g., 50ks)', function () {
+		alasql('CREATE TABLE `50ks` (id INT, name STRING)');
+		alasql('INSERT INTO `50ks` VALUES (1, "test")');
+		var res = alasql('SELECT * FROM `50ks`');
+		assert.deepEqual(res, [{id: 1, name: 'test'}]);
+	});
+
+	it('B) Table name starting with number without backticks', function () {
+		alasql('CREATE TABLE `50k` (id INT, name STRING)');
+		alasql('INSERT INTO `50k` VALUES (2, "test2")');
+		// This should work without backticks after the fix
+		var res = alasql('SELECT * FROM 50k');
+		assert.deepEqual(res, [{id: 2, name: 'test2'}]);
+	});
+
+	it('C) Table name with schema prefix starting with number', function () {
+		// This is the specific case from the issue: csv.50k where csv is schema, 50k is table
+		// Create a separate database called 'csv'
+		alasql('CREATE DATABASE csv');
+		alasql('USE csv');
+		alasql('CREATE TABLE `50k` (id INT, name STRING)');
+		alasql('INSERT INTO `50k` VALUES (3, "test3")');
+		// Query with schema.table notation where table starts with number
+		var res = alasql('SELECT * FROM csv.50k');
+		assert.deepEqual(res, [{id: 3, name: 'test3'}]);
+		// Cleanup
+		alasql('USE test1185');
+		alasql('DROP DATABASE csv');
+	});
+});