x

/** * Loads lib/agent-web-fetch.js in a VM (same bundle shape as /bin/agent preamble). */ import test from 'brittle' import { readFileSync } from 'node:fs' import vm from 'node:vm' const CODE = readFileSync( new URL('../lib/agent-web-fetch.js', import.meta.url), 'utf8' ) /** @returns {Record} */ function loadSandbox(extra = {}) { const sandbox = { URL, TextDecoder, TextEncoder, Uint8Array, AbortController, AbortSignal, ReadableStream, Response, Request, setTimeout: globalThis.setTimeout.bind(globalThis), clearTimeout: globalThis.clearTimeout.bind(globalThis), console, ...extra } vm.createContext(sandbox) vm.runInContext(CODE, sandbox, { filename: 'agent-web-fetch.js' }) return sandbox } test('bareWebExtractHtmlText strips scripts and tags', async (t) => { const s = loadSandbox() const fn = /** @type {(h: string) => string} */ (s.bareWebExtractHtmlText) const html = 'x

Hello & world

' const out = fn(html) t.ok(out.includes('Hello')) t.ok(out.includes('world')) t.absent(out.includes('script')) t.ok(out.includes('&') || out.includes('world')) }) test('bareWebExtractLinks resolves relative hrefs', async (t) => { const s = loadSandbox() const fn = /** @type {(h: string, b: string) => { links: string[] }} */ ( s.bareWebExtractLinks ) const r = fn( 'x y', 'https://example.com/pages/' ) t.ok(r.links.includes('https://example.com/doc')) t.ok(r.links.includes('https://ex.org/a')) }) test('bareWebExtractMeta title and og', async (t) => { const s = loadSandbox() const fn = /** @type {(h: string) => Record} */ ( s.bareWebExtractMeta ) const html = `T < Test ` const m = fn(html) t.is(m.title, 'T < Test') t.ok(String(m.description).includes('Desc')) t.is(m.og_title, 'OG Title') t.is(m.og_description, 'OG Desc') }) test('bareWebMaybeParseJson', async (t) => { const s = loadSandbox() const fn = /** @type {(x: string) => { ok: boolean, value?: unknown }} */ ( s.bareWebMaybeParseJson ) t.ok(fn('{"a":1}').ok) t.is(/** @type {{a:number}} */ (fn('{"a":1}').value).a, 1) t.absent(fn('{').ok) }) test('bareWebRunTool follows redirect then reads body', async (t) => { let calls = 0 const fetchFn = async (url, init) => { calls++ if (calls === 1) { t.is(String(url), 'https://a.test/start') return new Response('', { status: 302, headers: { Location: '/final' } }) } t.is(String(url), 'https://a.test/final') t.is(init.method, 'GET') return new Response('

OK

', { status: 200, headers: { 'Content-Type': 'text/html; charset=utf-8' } }) } const s = loadSandbox({ fetch: fetchFn }) const run = /** @type {typeof bareWebRunTool} */ (s.bareWebRunTool) const ctx = { httpFetch: fetchFn } const out = await run({ ctx, url: 'https://a.test/start', format: 'markdownish' }) t.ok(out.ok) t.is(out.status, 200) t.ok(String(out.url_final).includes('final')) t.ok(out.extract && typeof out.extract === 'object') const ex = /** @type {{ text?: string }} */ (out.extract) t.ok(ex.text && ex.text.includes('OK')) }) test('bareWebRunTool truncates large body stream', async (t) => { const chunk = new Uint8Array(60000).fill(65) const fetchFn = async () => new Response( new ReadableStream({ start(controller) { controller.enqueue(chunk) controller.enqueue(chunk) controller.close() } }), { status: 200, headers: { 'Content-Type': 'text/plain' } } ) const s = loadSandbox({ fetch: fetchFn }) const run = /** @type {typeof bareWebRunTool} */ (s.bareWebRunTool) const out = await run({ ctx: { httpFetch: fetchFn }, url: 'https://big.test/x', format: 'raw', max_response_bytes: 50000 }) t.ok(out.ok) t.ok(out.truncated) }) test('bareWebFmtErr null/undefined and node-like errors', async (t) => { const s = loadSandbox() const fmt = /** @type {(e: unknown) => string} */ (s.bareWebFmtErr) t.ok(fmt(null).includes('rejected_with_null')) t.ok(fmt(undefined).includes('rejected_with_undefined')) t.absent(fmt({ message: undefined }) === 'undefined') t.ok(fmt({ name: 'TypeError', code: 'ETIMEDOUT' }).includes('ETIMEDOUT')) const withCause = new Error('outer') withCause.cause = new Error('inner') t.ok(fmt(withCause).includes('inner')) }) test('bareWebRunTool without timeout_ms uses finite default (not NaN)', async (t) => { const fetchFn = async () => new Response('ok', { status: 200, headers: { 'Content-Type': 'text/html; charset=utf-8' } }) const s = loadSandbox({ fetch: fetchFn }) const run = /** @type {typeof bareWebRunTool} */ (s.bareWebRunTool) const out = await run({ ctx: { httpFetch: fetchFn }, url: 'https://x.test/', format: 'meta' }) t.ok(out.ok) t.absent(JSON.stringify(out).includes('NaN')) }) test('bareWebRunTool timeout passes abort reason (matches bare-fetch contract)', async (t) => { /** Simulates bare-fetch: rejects with `signal.reason` when aborted. */ const fetchFn = async (_url, init) => { const sig = init && /** @type {{ signal?: AbortSignal }} */ (init).signal if (!sig) return new Promise(() => { /* hang */ }) return new Promise((_res, rej) => { if (sig.aborted) { rej(/** @type {AbortSignal} */ (sig).reason) return } sig.addEventListener( 'abort', () => rej(/** @type {AbortSignal} */ (sig).reason), { once: true } ) }) } const s = loadSandbox({ fetch: fetchFn }) const run = /** @type {typeof bareWebRunTool} */ (s.bareWebRunTool) const out = await run({ ctx: { httpFetch: fetchFn }, url: 'https://slow.example/hang', format: 'meta', timeout_ms: 25 }) t.absent(out.ok) const err = String(/** @type {{ error?: string }} */ (out).error || '') t.ok(err.includes('web_fetch: request failed')) t.ok(err.includes('timeout') || err.includes('TimeoutError') || err.includes('exceeded')) t.absent(err.includes('promise_rejected_with_undefined')) })