feat: phase 1 of AI pipeline hardening — single LLM client + tier-aware models

Implements phase 1 of AI_PIPELINE_HARDENING_PLAN.md. Every Anthropic call
now goes through one module that owns retry, timeout, abort, structured-
output parsing, schema validation, and best-effort call telemetry.

* src/lib/llm.js — single callLLM entry point. Resolves model per tier
  (fast / standard / reasoning) with admin:model legacy fallback for the
  standard tier; 60s default timeout via AbortController; balanced-brace
  JSON extraction; LLMHttpError, LLMTruncatedError, LLMOutputError, and
  LLMValidationError surface clearly distinct failure modes.
* src/lib/llmRetry.js — exponential backoff with full jitter, retries
  only on transient HTTP statuses, honours Retry-After up to 60s, never
  retries on AbortError.
* src/lib/llmSchemas.js — Zod schemas for every structured task plus
  normalizeHandbookResult (collapses legacy "executes" relations into
  the canonical "executed_by" vocabulary).
* src/lib/api.js — thin shim over callLLM so existing callers (extraction
  pipeline, learning, quiz, R42, knowledge graph) keep working unchanged.
* src/lib/__tests__/ — 32 Vitest cases covering parse paths, error
  surfaces, simulation mode, model resolution, and schema validation.
* src/pages/Admin/index.jsx — three model inputs (fast / standard /
  reasoning) replacing the single legacy field; legacy value falls back
  for the standard tier so existing overrides survive.

Adds Zod and Vitest, plus an "npm run test" script.

Also cleans up the pre-existing repo-wide ESLint failures so phase 1's
"npm run lint passes" acceptance criterion can be checked: drops unused
React imports across the JSX tree (React 19 JSX runtime auto-imports),
attaches cause to rethrown errors in the service modules, ignores
pb_migrations in the ESLint config (PocketBase JSVM globals), and
removes one dead handleCreateCustom function in Leren.jsx. A real
behaviour bug surfaced in Testen.jsx — the quiz timer captured a stale
finishQuiz via setInterval closure; now updated via finishQuizRef so the
timer always invokes the latest callback.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
RaymondVerhoef
2026-05-20 13:50:09 +02:00
parent db5bb854c3
commit 4a8dbee7df
36 changed files with 1612 additions and 233 deletions

96
src/lib/llmRetry.js Normal file
View File

@@ -0,0 +1,96 @@
/**
* Retry policy for LLM calls.
*
* Exponential backoff with full jitter, base 1000ms, cap 16000ms. Only
* retries on transient HTTP statuses (408, 425, 429, 5xx, 529). Honours a
* `Retry-After` hint up to 60 seconds; longer waits fail fast. Never retries
* on AbortError.
*/
const RETRYABLE_STATUSES = new Set([408, 425, 429, 500, 502, 503, 504, 529]);
const BASE_DELAY_MS = 1000;
const CAP_DELAY_MS = 16000;
const MAX_RETRY_AFTER_MS = 60 * 1000;
export class RetryableError extends Error {
constructor(status, retryAfterMs = null, message) {
super(message || `Retryable HTTP ${status}`);
this.name = 'RetryableError';
this.status = status;
this.retryAfterMs = retryAfterMs;
}
}
export function isRetryableStatus(status) {
return RETRYABLE_STATUSES.has(status);
}
function backoffWithJitter(attempt) {
const exp = Math.min(CAP_DELAY_MS, BASE_DELAY_MS * 2 ** attempt);
return Math.floor(Math.random() * exp);
}
/**
* Parse a `Retry-After` header. Returns null when absent or unusable, or
* a millisecond delay otherwise. Supports both seconds and HTTP-date forms.
*/
export function parseRetryAfter(value, now = Date.now()) {
if (value == null) return null;
const s = String(value).trim();
if (!s) return null;
if (/^\d+$/.test(s)) return Number(s) * 1000;
const dateMs = Date.parse(s);
if (Number.isFinite(dateMs)) {
const delta = dateMs - now;
return delta > 0 ? delta : 0;
}
return null;
}
function sleep(ms, signal) {
return new Promise((resolve, reject) => {
if (signal?.aborted) return reject(signal.reason ?? new DOMException('Aborted', 'AbortError'));
const t = setTimeout(() => {
signal?.removeEventListener('abort', onAbort);
resolve();
}, ms);
const onAbort = () => {
clearTimeout(t);
reject(signal.reason ?? new DOMException('Aborted', 'AbortError'));
};
signal?.addEventListener('abort', onAbort, { once: true });
});
}
/**
* Run `fn(attempt)` with retry. `fn` may throw a `RetryableError` to request
* a retry, or any other error to fail immediately.
*
* @template T
* @param {(attempt:number) => Promise<T>} fn
* @param {{ maxRetries?: number, signal?: AbortSignal }} [opts]
* @returns {Promise<T>}
*/
export async function withRetry(fn, { maxRetries = 4, signal } = {}) {
let attempt = 0;
for (;;) {
if (signal?.aborted) throw signal.reason ?? new DOMException('Aborted', 'AbortError');
try {
return await fn(attempt);
} catch (err) {
if (err?.name === 'AbortError') throw err;
const retryable = err instanceof RetryableError;
if (!retryable || attempt >= maxRetries) throw err;
let delayMs;
if (err.retryAfterMs != null) {
if (err.retryAfterMs > MAX_RETRY_AFTER_MS) throw err;
delayMs = err.retryAfterMs;
} else {
delayMs = backoffWithJitter(attempt);
}
await sleep(delayMs, signal);
attempt++;
}
}
}