@bpmn-sdk/feel

Complete FEEL (Friendly Enough Expression Language) implementation — parser, evaluator, and highlighter

Documentation · GitHub · Changelog

Overview

@bpmn-sdk/feel is a complete implementation of the FEEL expression language used in DMN decision tables and BPMN condition expressions. It includes a tokenizer, recursive-descent parser, AST evaluator, formatter, and syntax highlighter.

Features

• Full FEEL grammar — arithmetic, comparisons, logic, function calls, paths, filters
• Temporal types — date, time, datetime, duration (ISO 8601, full spec compliance)
• Unary tests — DMN input expression syntax (> 5, "gold", "silver", [1..10])
• Built-in functions — 50+ standard FEEL functions (string, list, numeric, date, context)
• Range expressions — [1..10], (0..1), [today..end]
• Context literals — { key: value, nested: { x: 1 } }
• Syntax highlighting — semantic token classification for editors
• Zero dependencies

Installation

npm install @bpmn-sdk/feel

Quick Start

Evaluate an expression

import { parseExpression, evaluate } from "@bpmn-sdk/feel"

const parsed = parseExpression("amount * 1.2 + fee")
if (!parsed.errors.length) {
  const result = evaluate(parsed.ast!, { amount: 100, fee: 5 })
  console.log(result) // 125
}

Evaluate unary tests (DMN input expressions)

import { parseUnaryTests, evaluateUnaryTests } from "@bpmn-sdk/feel"

// Does input value match any listed condition?
const parsed = parseUnaryTests('"gold","silver"')
const matches = evaluateUnaryTests(parsed.ast!, "gold", { /* context */ })
console.log(matches) // true

Syntax highlighting

import { highlightFeel } from "@bpmn-sdk/feel"

const tokens = highlightFeel('if x > 10 then "high" else "low"')
for (const token of tokens) {
  console.log(token.type, token.value) // keyword, number, string, ...
}

API Reference

ParseResult

interface ParseResult {
  ast: FeelNode | null
  errors: ParseError[]   // { message, position }
}

Related Packages

License

MIT © bpmn-sdk