fix(ai-chat): целостность delta-spec (молчаливый скип) + 2 hardening (#491)

Внутреннее ре-ревью: DB-backed delta-spec молча СКИПАЛСЯ — нулевое покрытие
инварианта DB-clock курсора.

- Импорт `import postgres from 'postgres'` (default) при tsconfig
  module:commonjs без esModuleInterop компилился в `postgres_1.default(...)`,
  а CJS-`postgres` не имеет `.default` → TypeError в beforeAll → пустой catch →
  reachable=false → все 6 тестов уходили в console.warn('SKIP') и return БЕЗ
  ассертов (suite оставался бы зелёным даже при регрессии на new Date()).
  Фикс: `import * as postgres from 'postgres'` (как в рабочем int-харнессе).
- Хардненг харнесса: реальная ошибка программирования в beforeAll больше НЕ
  маскируется под «DB unreachable» — скип легитимен только для сетевого отказа
  (ECONNREFUSED и т.п.), иначе rethrow → suite падает громко.
- Два DB-clock теста использовали jest.useFakeTimers() целиком, что замораживало
  внутренние таймеры postgres.js → awaited DB round-trip зависал на 5s-капе (и
  вешал afterAll). Фейкаем ТОЛЬКО Date (doNotFake всех таймеров) — запрос
  резолвится, а инвариант «стамп от часов БД, не app-clock» по-прежнему доказан
  (скос процесс-часов в 2099 → стамп остаётся на времени БД). Теперь все 6
  тестов РЕАЛЬНО исполняются и зелёные против живого Postgres.

Два дешёвых hardening из ревью:
- registry coverageFloor: пустая ветка возвращает max(currentStamp,
  persistedFloor) — инвариант «клиент с n=persistedFloor всегда покрыт»
  структурный, а не тайминг-зависимый.
- GetChatDeltaDto.cursor: @IsString → @IsISO8601 — битый курсор отсекается 400
  на уровне DTO, а не 500 на `::timestamptz`.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-07-11 19:16:59 +03:00
parent daeeb1f3f2
commit daca5ce8d6
3 changed files with 71 additions and 9 deletions
@@ -440,7 +440,13 @@ export class AiChatStreamRegistryService implements OnModuleDestroy {
* An attach at frontier `n` is covered ⟺ coverageFloor <= n.
*/
private coverageFloor(entry: Entry): number {
if (entry.frames.length === 0) return entry.currentStamp;
// Empty ring: only the live tail is coming. The floor is the current step,
// but never below persistedFloor — a confirmed persist can rotate the ring
// empty while currentStamp still lags a beat behind on another connection, so
// max() keeps the invariant STRUCTURAL (a client with n = persistedFloor is
// always covered) rather than timing-dependent.
if (entry.frames.length === 0)
return Math.max(entry.currentStamp, entry.persistedFloor);
const min = entry.stamps[0];
return entry.overflowThroughStamp >= min ? min + 1 : min;
}
@@ -1,4 +1,10 @@
import { IsOptional, IsString, MaxLength, MinLength } from 'class-validator';
import {
IsISO8601,
IsOptional,
IsString,
MaxLength,
MinLength,
} from 'class-validator';
/** Identify a chat by id (workspace-scoped on the server). */
export class ChatIdDto {
@@ -47,8 +53,11 @@ export class GetChatDeltaDto {
@IsString()
chatId: string;
// ISO-8601 timestamp echoed from the previous poll's response. Validated as
// ISO-8601 (not a bare string): a malformed cursor would otherwise reach the
// `::timestamptz` cast in findByChatUpdatedAfter and 500 instead of a clean 400.
@IsOptional()
@IsString()
@IsISO8601()
cursor?: string;
}
@@ -1,7 +1,12 @@
import { randomUUID } from 'crypto';
import { CamelCasePlugin, Kysely, sql } from 'kysely';
import { PostgresJSDialect } from 'kysely-postgres-js';
import postgres from 'postgres';
// NOT a default import: the project tsconfig is `module: commonjs` with NO
// esModuleInterop, so `import postgres from 'postgres'` compiles to
// `postgres_1.default(...)` and the CJS `postgres` export has no `.default` —
// it threw in beforeAll, was swallowed as "DB unreachable", and SILENTLY voided
// all six tests. Mirror the working integration harness (test/integration/db.ts).
import * as postgres from 'postgres';
import { AiChatMessageRepo } from './ai-chat-message.repo';
import { AiChatRunRepo } from './ai-chat-run.repo';
@@ -50,8 +55,21 @@ beforeAll(async () => {
await sql`set session_replication_role = replica`.execute(db);
await sql`select 1`.execute(db);
reachable = true;
} catch {
} catch (err) {
reachable = false;
// A genuine connection failure (ECONNREFUSED etc.) is a legitimate skip on a
// DB-less CI. A PROGRAMMING error (bad import, typo, driver misuse) must NOT
// masquerade as "DB unreachable" and silently void the whole suite (that is
// exactly the bug that hid this spec's zero coverage) — rethrow it so the
// suite fails LOUDLY.
const msg = String((err as Error)?.message ?? err);
if (
!/ECONNREFUSED|ENOTFOUND|ETIMEDOUT|EHOSTUNREACH|connect|terminating|password|authentication|role .* does not exist|database .* does not exist/i.test(
msg,
)
) {
throw err;
}
}
msgRepo = new AiChatMessageRepo(db as never);
runRepo = new AiChatRunRepo(db as never);
@@ -96,6 +114,34 @@ async function dbNow(): Promise<string> {
return r.rows[0].now.toISOString();
}
// Fake ONLY the Date object (so in-process `new Date()`/`Date.now()` jump), while
// leaving every TIMER function real. Faking timers wholesale freezes postgres.js's
// internal connection/query timers, so the awaited DB round-trip would hang the
// test (and the afterAll cleanup) at the jest 5s cap. With Date-only faking the
// query resolves normally, and we still prove the stamp is the DB clock (not the
// skewed process clock).
function fakeDateOnly(iso: string): void {
jest.useFakeTimers({
doNotFake: [
'hrtime',
'nextTick',
'performance',
'queueMicrotask',
'requestAnimationFrame',
'cancelAnimationFrame',
'requestIdleCallback',
'cancelIdleCallback',
'setImmediate',
'clearImmediate',
'setInterval',
'clearInterval',
'setTimeout',
'clearTimeout',
],
now: new Date(iso),
});
}
const maybe = (name: string, fn: () => Promise<void>) =>
it(name, async () => {
if (!reachable) {
@@ -188,9 +234,10 @@ describe('AiChatMessageRepo.findByChatUpdatedAfter (#491 delta poll)', () => {
'update() stamps updatedAt from the DB clock, not the app clock',
async () => {
const m = await seedMessage();
// Fake the PROCESS clock ~73 years into the future. If the stamp came from
// `new Date()` the row would read year 2099; sql now() keeps it on DB time.
jest.useFakeTimers().setSystemTime(new Date('2099-01-01T00:00:00Z'));
// Skew the PROCESS clock ~73 years into the future (Date only). If the stamp
// came from `new Date()` the row would read year 2099; sql now() keeps it on
// DB time.
fakeDateOnly('2099-01-01T00:00:00Z');
const updated = await msgRepo.update(m.id, workspaceId, {
content: 'y',
});
@@ -211,7 +258,7 @@ describe('AiChatMessageRepo.findByChatUpdatedAfter (#491 delta poll)', () => {
status: 'running',
stepCount: 0,
} as never);
jest.useFakeTimers().setSystemTime(new Date('2099-01-01T00:00:00Z'));
fakeDateOnly('2099-01-01T00:00:00Z');
const updated = await runRepo.update(run.id, workspaceId, {
stepCount: 1,
});