Known Names

FEEL allows variable names to contain spaces: Monthly Salary, Order Line Item, Full Name. The DMN spec calls these Name tokens assembled from multiple words.

The problem

At the lexer level, Monthly Salary is two separate Name tokens. Without additional context the parser cannot know whether they form a single identifier or two consecutive expressions.

The solution

Pass a Set<string> of known multi-word names as the third argument to parse():

import { parse } from '@veridtools/feel-parser';

const knownNames = new Set([
  'Monthly Salary',
  'Full Name',
  'Order Line Item',
]);

const ast = parse('Monthly Salary * 12', 'expression', knownNames);
// BinaryOp {
//   op: '*',
//   left: Identifier { name: 'Monthly Salary' },  ← assembled
//   right: NumberLiteral { value: '12' }
// }

Without the hint:

parse('Monthly Salary * 12')
// BinaryOp {
//   op: '*',
//   left: Identifier { name: 'Monthly' },
//   right: ...                                ← Salary is orphaned
// }

How it works

The parser performs a greedy longest-match when it sees consecutive Name tokens. It checks whether the concatenated phrase (with spaces) exists in knownNames, extending the match as far as possible before falling back to the shorter prefix.

In the playground

The Known Names input field accepts a comma-separated list:

Monthly Salary, Full Name, Order Line Item

Each entry is trimmed and added to the Set passed to parse().