Creative challenges now have auto-generated MC choices with correct answers. Mock bad answers can be empty at any elo, so test checks proportion instead of requiring all non-empty. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
153 lines
5.2 KiB
TypeScript
153 lines
5.2 KiB
TypeScript
import { describe, it, expect } from 'vitest'
|
|
import { mockResponse } from './mock.js'
|
|
import { pickChallenge, getAllChallengeTypes } from './challenges.js'
|
|
|
|
describe('mockResponse', () => {
|
|
it('returns a valid response structure', () => {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
const resp = mockResponse(challenge, 'confident', 1500)
|
|
expect(resp).toHaveProperty('answer')
|
|
expect(resp).toHaveProperty('trashTalk')
|
|
expect(resp).toHaveProperty('timeMs')
|
|
expect(resp).toHaveProperty('timedOut')
|
|
expect(resp).toHaveProperty('error')
|
|
expect(typeof resp.timeMs).toBe('number')
|
|
expect(typeof resp.timedOut).toBe('boolean')
|
|
expect(typeof resp.error).toBe('boolean')
|
|
})
|
|
|
|
it('high ELO bots answer correctly more often (factual)', () => {
|
|
let highCorrect = 0
|
|
let lowCorrect = 0
|
|
const runs = 200
|
|
|
|
for (let i = 0; i < runs; i++) {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
if (challenge.scoring !== 'factual' || !challenge.answers?.length) continue
|
|
|
|
const highResp = mockResponse(challenge, 'confident', 1900)
|
|
const lowResp = mockResponse(challenge, 'clueless', 900)
|
|
|
|
if (!highResp.timedOut && !highResp.error && challenge.answers.includes(highResp.answer)) highCorrect++
|
|
if (!lowResp.timedOut && !lowResp.error && challenge.answers.includes(lowResp.answer)) lowCorrect++
|
|
}
|
|
|
|
// High ELO should answer correctly more often
|
|
expect(highCorrect).toBeGreaterThan(lowCorrect)
|
|
})
|
|
|
|
it('low ELO bots time out / error more often', () => {
|
|
let highFails = 0
|
|
let lowFails = 0
|
|
const runs = 500
|
|
|
|
for (let i = 0; i < runs; i++) {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
const highResp = mockResponse(challenge, 'confident', 1900)
|
|
const lowResp = mockResponse(challenge, 'clueless', 900)
|
|
|
|
if (highResp.timedOut || highResp.error) highFails++
|
|
if (lowResp.timedOut || lowResp.error) lowFails++
|
|
}
|
|
|
|
expect(lowFails).toBeGreaterThan(highFails)
|
|
})
|
|
|
|
it('high ELO bots respond faster on average', () => {
|
|
let highTotal = 0
|
|
let lowTotal = 0
|
|
let highCount = 0
|
|
let lowCount = 0
|
|
const runs = 200
|
|
|
|
for (let i = 0; i < runs; i++) {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
const highResp = mockResponse(challenge, 'confident', 1900)
|
|
const lowResp = mockResponse(challenge, 'clueless', 900)
|
|
|
|
if (!highResp.timedOut && !highResp.error) { highTotal += highResp.timeMs; highCount++ }
|
|
if (!lowResp.timedOut && !lowResp.error) { lowTotal += lowResp.timeMs; lowCount++ }
|
|
}
|
|
|
|
const highAvg = highTotal / Math.max(highCount, 1)
|
|
const lowAvg = lowTotal / Math.max(lowCount, 1)
|
|
expect(highAvg).toBeLessThan(lowAvg)
|
|
})
|
|
|
|
it('creative challenges return answers (most non-empty at high elo)', () => {
|
|
let nonEmpty = 0
|
|
let total = 0
|
|
for (let i = 0; i < 50; i++) {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
if (challenge.scoring !== 'creative') continue
|
|
|
|
const resp = mockResponse(challenge, 'witty', 1800)
|
|
if (!resp.timedOut && !resp.error) {
|
|
total++
|
|
if (resp.answer.length > 0) nonEmpty++
|
|
}
|
|
}
|
|
// At elo 1800, bad answer chance is very low; most should be non-empty
|
|
expect(nonEmpty).toBeGreaterThan(total * 0.7)
|
|
})
|
|
|
|
it('trash talk is always a string', () => {
|
|
for (let i = 0; i < 20; i++) {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
const resp = mockResponse(challenge, 'confident', 1900)
|
|
expect(typeof resp.trashTalk).toBe('string')
|
|
}
|
|
})
|
|
|
|
it('timeMs is always positive', () => {
|
|
for (let i = 0; i < 50; i++) {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
const resp = mockResponse(challenge, 'confident', 1500)
|
|
expect(resp.timeMs).toBeGreaterThan(0)
|
|
}
|
|
})
|
|
|
|
it('produces valid responses for all 16 challenge types', () => {
|
|
const types = getAllChallengeTypes()
|
|
expect(types.length).toBe(16)
|
|
|
|
for (const type of types) {
|
|
// Force pick a challenge of this type by using all other types
|
|
const otherTypes = new Set(types.filter(t => t !== type))
|
|
const challenge = pickChallenge(otherTypes, null)
|
|
// May not get exact type if creative/factual split causes fallback, but should still work
|
|
const resp = mockResponse(challenge, 'confident', 1500)
|
|
expect(resp).toHaveProperty('answer')
|
|
expect(resp).toHaveProperty('timeMs')
|
|
expect(resp.timeMs).toBeGreaterThan(0)
|
|
}
|
|
})
|
|
|
|
it('answer is empty string on timeout', () => {
|
|
// Run many times with very low elo to trigger timeouts
|
|
let foundTimeout = false
|
|
for (let i = 0; i < 200; i++) {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
const resp = mockResponse(challenge, 'clueless', 500)
|
|
if (resp.timedOut) {
|
|
expect(resp.answer).toBe('')
|
|
foundTimeout = true
|
|
}
|
|
}
|
|
expect(foundTimeout).toBe(true)
|
|
})
|
|
|
|
it('answer is empty string on error', () => {
|
|
let foundError = false
|
|
for (let i = 0; i < 200; i++) {
|
|
const challenge = pickChallenge(new Set(), null)
|
|
const resp = mockResponse(challenge, 'clueless', 500)
|
|
if (resp.error) {
|
|
expect(resp.answer).toBe('')
|
|
foundError = true
|
|
}
|
|
}
|
|
expect(foundError).toBe(true)
|
|
})
|
|
})
|