fix(hocuspocus): unblock typecheck on link-metadata tests + html scrape

Four CI typecheck errors after the link-metadata module landed:

- htmlScrape.ts: Bun's TextDecoder ctor types the encoding as a closed
  union, but the value is parsed from arbitrary HTTP / HTML headers
  and cannot be narrowed statically. Cast to never (matches the
  receiver's "trust nothing" stance) — invalid encodings throw and
  the existing catch falls back to utf-8.
- http.test.ts: Response.json() returns unknown under bun:test. Added
  a local WireBody type + readBody helper so every assertion reads
  through a typed boundary instead of `as any`. Kept WireBody
  independent of the production discriminated union — intersecting
  MetadataResponse & ErrorResponse collapses to never on success.
- htmlScrape.test.ts: dropped the as HeadersInit cast (the type is
  not in the bun-types lib set, and Headers ctor accepts
  RequestInit['headers'] directly).
- oembed.test.ts: typed the mockImplementation params as
  (_url: unknown, init?: RequestInit) so the spy honors the fetch
  signature instead of falling back to implicit any.

Made-with: Cursor

Author: HMarzban

packages/hocuspocus.server/src/modules/link-metadata/__tests__/integration/http.test.ts

@@ -4,6 +4,23 @@ import pino from 'pino'
 
 import { init } from '../../module'
 
+// Wire shape that tests assert against. Kept independent of the production
+// types because the discriminated union (success: true vs false) collapses
+// to `never` when intersected, and toBe() narrows expected values from the
+// receiver's type. `success: boolean` here is intentional.
+interface WireBody {
+  success?: boolean
+  message?: string
+  code?: string
+  title?: string
+  publisher?: { name?: string }
+  cached?: boolean
+  fetched_at?: string
+  oembed?: { html?: string }
+}
+const readBody = async (response: Response): Promise<WireBody> =>
+  (await response.json()) as WireBody
+
 const silentLogger = pino({ level: 'silent' })
 
 const makeMockRedis = () => {
@@ -40,7 +57,7 @@ describe('GET /api/metadata (integration)', () => {
     const app = buildApp()
     const response = await app.request('http://localhost/api/metadata')
     expect(response.status).toBe(400)
-    const body = await response.json()
+    const body = await readBody(response)
     expect(body.success).toBe(false)
     expect(body.code).toBe('INVALID_URL')
     expect(typeof body.message).toBe('string')
@@ -50,7 +67,7 @@ describe('GET /api/metadata (integration)', () => {
     const app = buildApp()
     const response = await app.request('http://localhost/api/metadata?url=not-a-url')
     expect(response.status).toBe(400)
-    const body = await response.json()
+    const body = await readBody(response)
     expect(body.success).toBe(false)
     expect(body.code).toBe('INVALID_URL')
   })
@@ -73,7 +90,7 @@ describe('GET /api/metadata (integration)', () => {
       'http://localhost/api/metadata?url=' +
         encodeURIComponent('https://www.youtube.com/watch?v=xss')
     )
-    const body = await response.json()
+    const body = (await response.json()) as { oembed?: { html?: string } }
     expect(body.oembed?.html).toBeUndefined()
   })
 
@@ -83,7 +100,7 @@ describe('GET /api/metadata (integration)', () => {
       'http://localhost/api/metadata?url=' + encodeURIComponent('http://192.168.1.1')
     )
     expect(response.status).toBe(400)
-    const body = await response.json()
+    const body = await readBody(response)
     expect(body.code).toBe('BLOCKED_URL')
   })
 
@@ -109,10 +126,10 @@ describe('GET /api/metadata (integration)', () => {
     expect(response.status).toBe(200)
     expect(response.headers.get('cache-control')).toContain('max-age=3600')
     expect(response.headers.get('vary')).toBe('Accept-Language')
-    const body = await response.json()
+    const body = await readBody(response)
     expect(body.success).toBe(true)
     expect(body.title).toBe('Test')
-    expect(body.publisher.name).toBe('YouTube')
+    expect(body.publisher?.name).toBe('YouTube')
     expect(body.cached).toBe(false)
     expect(body.fetched_at).toBeDefined()
   })
@@ -127,7 +144,7 @@ describe('GET /api/metadata (integration)', () => {
 
     expect(response.status).toBe(200)
     expect(response.headers.get('cache-control')).toContain('max-age=600')
-    const body = await response.json()
+    const body = await readBody(response)
     expect(body.success).toBe(true)
     expect(body.title).toBe('random-blog.example.com')
   })
@@ -146,10 +163,10 @@ describe('GET /api/metadata (integration)', () => {
       encodeURIComponent('https://www.youtube.com/watch?v=cached')
 
     const first = await app.request(url)
-    expect((await first.json()).cached).toBe(false)
+    expect((await readBody(first)).cached).toBe(false)
 
     const second = await app.request(url)
-    const body = await second.json()
+    const body = await readBody(second)
     expect(body.cached).toBe(true)
   })
 })

packages/hocuspocus.server/src/modules/link-metadata/__tests__/unit/htmlScrape.test.ts

@@ -30,7 +30,7 @@ describe('runHtmlScrape', () => {
     await runHtmlScrape('https://example.com', noopScraper, 'fr-FR')
 
     const init = fetchSpy.mock.calls[0]![1] as RequestInit
-    const headers = new Headers(init.headers as HeadersInit)
+    const headers = new Headers(init.headers)
     expect(headers.get('accept-language')).toBe('fr-FR')
     expect(headers.get('user-agent')).toContain('DocsPlusBot')
     expect(headers.get('user-agent')).toContain('facebookexternalhit')

packages/hocuspocus.server/src/modules/link-metadata/__tests__/unit/oembed.test.ts

@@ -71,10 +71,10 @@ describe('runOembed', () => {
 
   test('respects per-call timeout via AbortSignal', async () => {
     let receivedSignal: AbortSignal | undefined
-    fetchSpy.mockImplementation(async (_url, init) => {
-      receivedSignal = (init as RequestInit).signal as AbortSignal
+    fetchSpy.mockImplementation((async (_url: unknown, init?: RequestInit) => {
+      receivedSignal = init?.signal as AbortSignal
       return new Response('{}', { status: 200, headers: { 'content-type': 'application/json' } })
-    })
+    }) as typeof fetch)
     await runOembed('https://www.youtube.com/watch?v=abc')
     expect(receivedSignal).toBeDefined()
   })

packages/hocuspocus.server/src/modules/link-metadata/domain/stages/htmlScrape.ts

@@ -41,7 +41,11 @@ const decodeBody = (bytes: Uint8Array, contentType: string | null): string => {
     charset = parseCharsetFromMeta(firstKb)
   }
   try {
-    return new TextDecoder(charset || 'utf-8', { fatal: false }).decode(bytes)
+    // Bun's TextDecoder ctor types the encoding as a closed union, but the
+    // value here is parsed from arbitrary HTTP / HTML and cannot be narrowed
+    // statically. Invalid encodings throw RangeError, which the catch below
+    // turns into a utf-8 fallback.
+    return new TextDecoder((charset || 'utf-8') as never, { fatal: false }).decode(bytes)
   } catch {
     return new TextDecoder('utf-8', { fatal: false }).decode(bytes)
   }