starred/windmill
1
0
Fork 0
mirror of https://github.com/windmill-labs/windmill.git synced 2026-10-03 08:02:19 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor(ai-evals): drop two unreachable job routes from the fetch stub (#11251)

Browse Source 
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
AlexRV12
2026-09-21 14:46:50 +00:00
committed by GitHub
co-authored by Claude Opus 5
parent 49e0974c07
commit dbac7cce99
2 changed files with 12 additions and 107 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ai_evals/adapters/frontend/mockBackend.ts
-53
View File
@@ -1178,27 +1178,6 @@ const BENCHMARK_WORKERS = [
}
]
const BENCHMARK_JOB_GET_PATH = /^\/api\/w\/([^/]+)\/jobs_u\/get\/([^/]+)$/
const BENCHMARK_RUN_BY_PATH = /^\/api\/w\/([^/]+)\/jobs\/run\/(p|f)\/([^/]+)$/
/** An intercepted request carries its args as a JSON string; anything else means none
* were supplied. */
function parseBenchmarkRequestBody(
body: BodyInit | null | undefined
): Record<string, unknown> | undefined {
if (typeof body !== 'string') {
return undefined
}
try {
const parsed = JSON.parse(body)
return typeof parsed === 'object' && parsed !== null
? (parsed as Record<string, unknown>)
: undefined
} catch {
return undefined
}
}
/** True when `handleBenchmarkApiFetch` has an answer for this `/api/...` url.
* Any other relative fetch must keep its normal (non-benchmark) behavior —
* intercepting it with a synthetic 404 sends the model into retry loops. */
@@ -1210,8 +1189,6 @@ export function hasBenchmarkApiHandler(url: string): boolean {
const path = url.split('?')[0]
return (
path === '/api/workers/list' ||
BENCHMARK_JOB_GET_PATH.test(path) ||
BENCHMARK_RUN_BY_PATH.test(path) ||
path === '/api/embeddings/query_hub_scripts' ||
path.startsWith('/api/scripts/hub/get_full/') ||
BENCHMARK_AI_MODELS_PATH.test(path)
@@ -1235,36 +1212,6 @@ export function handleBenchmarkApiFetch(url: string, init?: RequestInit): Respon
?.aiProviders?.find((entry) => entry.path === resourcePath)
return Response.json({ data: (seed?.models ?? []).map((id) => ({ id })) })
}
const jobGet = BENCHMARK_JOB_GET_PATH.exec(path)
if (jobGet) {
const id = decodeURIComponent(jobGet[2])
const job = getBenchmarkCompletedJob(decodeURIComponent(jobGet[1]), id)
if (!job) {
return Response.json({ error: `Job not found for "${id}"` }, { status: 404 })
}
// The real endpoint lets a caller drop the bulky fields. Ignoring that here would
// size the model's context off a payload it explicitly asked to shrink.
const query = new URLSearchParams(url.split('?')[1] ?? '')
if (query.get('no_logs') === 'true') {
delete job.logs
}
if (query.get('no_code') === 'true') {
delete job.raw_code
}
return Response.json(job)
}
const runByPath = BENCHMARK_RUN_BY_PATH.exec(path)
if (runByPath) {
const workspace = decodeURIComponent(runByPath[1])
const runnablePath = decodeURIComponent(runByPath[3])
const args = parseBenchmarkRequestBody(init?.body)
// The real endpoint answers with the bare job id as text, not JSON.
return new Response(
runByPath[2] === 'f'
? runBenchmarkFlowByPath({ workspace, path: runnablePath, args })
: runBenchmarkScriptByPath({ workspace, path: runnablePath, args })
)
}
if (path === '/api/embeddings/query_hub_scripts') {
const text = new URLSearchParams(url.split('?')[1] ?? '').get('text') ?? ''
return Response.json(searchBenchmarkHubScripts(text))
ai_evals/adapters/frontend/mockBackendApi.test.ts
+12 -54
View File
@@ -1,60 +1,18 @@
import { afterEach, beforeEach, describe, expect, it } from 'bun:test'
import {
createBenchmarkCompletedJob,
getBenchmarkCompletedJob,
handleBenchmarkApiFetch,
hasBenchmarkApiHandler,
resetBenchmarkMockBackend,
registerBenchmarkWorkspaceRunnables
} from './mockBackend'
const WORKSPACE = 'benchmark-api-ws'
import { describe, expect, it } from 'bun:test'
import { handleBenchmarkApiFetch, hasBenchmarkApiHandler } from './mockBackend'
// `list_workers`'s client call is not faked, so its request really does leave as a relative
// `/api/...` url that node's fetch cannot resolve. Without this route the tool throws instead
// of answering.
describe('benchmark API fetch handlers', () => {
beforeEach(() => resetBenchmarkMockBackend())
afterEach(() => resetBenchmarkMockBackend())
it('lists workers, the way list_workers reaches them', async () => {
const url = '/api/workers/list?per_page=100'
it('runs a deployed script by path', async () => {
registerBenchmarkWorkspaceRunnables(WORKSPACE, {
scripts: [
{
path: 'f/evals/greet',
summary: 'Greet',
language: 'bun',
content: 'export async function main() {}'
}
]
})
expect(hasBenchmarkApiHandler(url)).toBe(true)
const body = await handleBenchmarkApiFetch(url).json()
const res = handleBenchmarkApiFetch(
`/api/w/${WORKSPACE}/jobs/run/p/${encodeURIComponent('f/evals/greet')}`,
{ method: 'POST', body: JSON.stringify({ name: 'ada' }) }
)
expect(res.status).toBe(200)
const job = getBenchmarkCompletedJob(WORKSPACE, (await res.text()).trim())
expect(job).toMatchObject({ success: true, args: { name: 'ada' } })
})
it('serves a recorded job so a model can check the run it just started', async () => {
const id = createBenchmarkCompletedJob({
workspace: WORKSPACE,
jobKind: 'preview',
result: 'Hello, World!'
})
const res = handleBenchmarkApiFetch(`/api/w/${WORKSPACE}/jobs_u/get/${id}`)
expect(res.status).toBe(200)
expect(await res.json()).toMatchObject({
id,
success: true,
result: 'Hello, World!'
})
})
it('404s an unknown job id instead of letting the fetch fall through', () => {
expect(hasBenchmarkApiHandler(`/api/w/${WORKSPACE}/jobs_u/get/missing`)).toBe(true)
expect(handleBenchmarkApiFetch(`/api/w/${WORKSPACE}/jobs_u/get/missing`).status).toBe(404)
// The tool counts the list and maps over it, so it needs a bare array, not an envelope.
expect(Array.isArray(body)).toBe(true)
expect(body[0]).toMatchObject({ worker: expect.any(String), worker_group: expect.any(String) })
})
})