mirror of
https://github.com/MBombeck/HealthLog.git
synced 2026-11-08 04:32:42 +00:00
Marc-André Bombeck
3a99de312e
- evaluate the fence in the wrapper above findCached, so a stale request neither replays a body, learns a cell exists, nor inserts a claim - return the refusal rather than throwing: one route composes the wrapper outside apiHandler, where a throw escapes the envelope - read the claim and the session's own context on one session lookup
55 lines
2.5 KiB
TypeScript
55 lines
2.5 KiB
TypeScript
/**
|
|
* Shared `next/headers` mock state for the integration suite.
|
|
*
|
|
* Background: vitest runs the integration suite with `isolate: false`
|
|
* (one worker, one container). The factory passed to
|
|
* `vi.mock("next/headers", ...)` is resolved ONCE per worker —
|
|
* whichever test file loads first wins. If two files each declare a
|
|
* top-level `const cookieJar = new Map()` and reference it from their
|
|
* own `vi.mock("next/headers", ...)` factory, only the first file's
|
|
* Map is ever read; the second file's writes silently disappear and
|
|
* its tests flake based on import order.
|
|
*
|
|
* Fix: every integration file imports `cookieJar` + `headerJar` from
|
|
* THIS module (a singleton) and clears them in `beforeEach`. The
|
|
* `vi.mock` factory inside each test file resolves the Maps via
|
|
* `await import()` so it sees the same Map instances regardless of
|
|
* which file's mock factory ran first.
|
|
*/
|
|
export const cookieJar = new Map<string, string>();
|
|
export const headerJar = new Map<string, string>();
|
|
|
|
/**
|
|
* Session snapshots for invoking concurrent route handlers as different users.
|
|
* A request consumes one entry when its mocked cookie store is created.
|
|
*/
|
|
export const queuedSessionIds: string[] = [];
|
|
|
|
/**
|
|
* v1.37.0 — attach the record-session assertion the shipped browser client
|
|
* attaches to every same-origin request.
|
|
*
|
|
* The record-session fence refuses a cookie request whose asserted context does
|
|
* not match its session row, and it refuses a request that asserts nothing at
|
|
* all once the session has been inside somebody's record. A test that moves
|
|
* `acting_as_user_id` directly and then calls a route is therefore exercising
|
|
* the fence's pre-fence-bundle arm, not whatever the test was written for — so
|
|
* every such fixture calls this immediately after the switch.
|
|
*
|
|
* Truthful by construction: the values are read back off the row rather than
|
|
* computed, so this can only ever say what the session actually holds. The
|
|
* cases that want a STALE or ABSENT assertion set the headers themselves, and
|
|
* they live in `tests/integration/record-session-fence*.test.ts`.
|
|
*
|
|
* Takes the epoch and scope rather than reading the database, so this module
|
|
* keeps its single responsibility (shared mock state) and no test file has to
|
|
* import a Prisma client it does not otherwise use.
|
|
*/
|
|
export function assertRecordContext(
|
|
recordEpoch: number,
|
|
actingAsUserId: string | null,
|
|
): void {
|
|
headerJar.set("x-healthlog-record-epoch", String(recordEpoch));
|
|
headerJar.set("x-healthlog-record-scope", actingAsUserId ?? "self");
|
|
}
|