Files
gstack/test/lib-eval-tier.test.ts
Garry Tan 9bc6c9416f feat: add eval format validation, tier selection, cost tracking
- lib/eval-format.ts: StandardEvalResult interfaces, validateEvalResult(),
  normalizeFromLegacy/normalizeToLegacy round-trip converters
- lib/eval-tier.ts: EvalTier type, resolveTier/resolveJudgeTier from env,
  tierToModel mapping, TIER_ALIASES (haiku→fast, sonnet→standard, opus→full)
- lib/eval-cost.ts: MODEL_PRICING (last verified 2025-05-01), computeCosts(),
  formatCostDashboard(), aggregateCosts(), fallback for unknown models
- 42 tests across 3 test files

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-15 09:39:18 -05:00

95 lines
2.8 KiB
TypeScript

/**
* Tests for lib/eval-tier.ts — model tier selection.
*/
import { describe, test, expect, beforeEach, afterEach } from 'bun:test';
import { resolveTier, resolveJudgeTier, tierToModel, TIER_ALIASES } from '../lib/eval-tier';
describe('lib/eval-tier', () => {
const origEvalTier = process.env.EVAL_TIER;
const origJudgeTier = process.env.EVAL_JUDGE_TIER;
afterEach(() => {
if (origEvalTier === undefined) delete process.env.EVAL_TIER;
else process.env.EVAL_TIER = origEvalTier;
if (origJudgeTier === undefined) delete process.env.EVAL_JUDGE_TIER;
else process.env.EVAL_JUDGE_TIER = origJudgeTier;
});
describe('resolveTier', () => {
test('defaults to standard when unset', () => {
delete process.env.EVAL_TIER;
expect(resolveTier()).toBe('standard');
});
test('resolves tier names directly', () => {
process.env.EVAL_TIER = 'fast';
expect(resolveTier()).toBe('fast');
process.env.EVAL_TIER = 'full';
expect(resolveTier()).toBe('full');
});
test('resolves model aliases', () => {
process.env.EVAL_TIER = 'haiku';
expect(resolveTier()).toBe('fast');
process.env.EVAL_TIER = 'sonnet';
expect(resolveTier()).toBe('standard');
process.env.EVAL_TIER = 'opus';
expect(resolveTier()).toBe('full');
});
test('is case-insensitive', () => {
process.env.EVAL_TIER = 'HAIKU';
expect(resolveTier()).toBe('fast');
process.env.EVAL_TIER = 'Full';
expect(resolveTier()).toBe('full');
});
test('defaults to standard for unknown value', () => {
process.env.EVAL_TIER = 'gpt-4';
expect(resolveTier()).toBe('standard');
});
});
describe('resolveJudgeTier', () => {
test('falls back to EVAL_TIER when EVAL_JUDGE_TIER unset', () => {
delete process.env.EVAL_JUDGE_TIER;
process.env.EVAL_TIER = 'fast';
expect(resolveJudgeTier()).toBe('fast');
});
test('uses EVAL_JUDGE_TIER when set', () => {
process.env.EVAL_TIER = 'fast';
process.env.EVAL_JUDGE_TIER = 'full';
expect(resolveJudgeTier()).toBe('full');
});
test('resolves aliases for judge tier', () => {
process.env.EVAL_JUDGE_TIER = 'opus';
expect(resolveJudgeTier()).toBe('full');
});
});
describe('tierToModel', () => {
test('maps fast to haiku', () => {
expect(tierToModel('fast')).toBe('claude-haiku-4-5');
});
test('maps standard to sonnet', () => {
expect(tierToModel('standard')).toBe('claude-sonnet-4-6');
});
test('maps full to opus', () => {
expect(tierToModel('full')).toBe('claude-opus-4-6');
});
});
describe('TIER_ALIASES', () => {
test('contains expected aliases', () => {
expect(TIER_ALIASES.haiku).toBe('fast');
expect(TIER_ALIASES.sonnet).toBe('standard');
expect(TIER_ALIASES.opus).toBe('full');
});
});
});