config.ts · Source

config.tsapps/worker/src/config.tsraw ← src6,427 B← Back to the app
1/**
2 * apps/worker — environment configuration (zod-validated, dev-safe defaults).
3 *
4 * `loadEnv()` reads `process.env` once and returns a frozen, fully-typed
5 * config object. Every processor takes this object (dependency injection) so
6 * tests can override hosts/keys without touching the environment.
7 */
8import { z } from "zod";
9 
10const intFrom = (def: number) =>
11  z.coerce
12    .number()
13    .int()
14    .positive()
15    .default(def);
16 
17const boolFrom = (def: boolean) =>
18  z
19    .enum(["true", "false"])
20    .default(def ? "true" : "false")
21    .transform((v) => v === "true");
22 
23/**
24 * Environment contract. All values are strings by default (env is strings);
25 * coercion happens here, nowhere else.
26 */
27export const envSchema = z.object({
28  NODE_ENV: z.string().default("development"),
29  REDIS_URL: z.string().url().default("redis://localhost:6379"),
30  DATABASE_URL: z
31    .string()
32    .default("postgresql://factory:factory@localhost:5432/factory"),
33 
34  S3_ENDPOINT: z.string().url().default("http://localhost:9000"),
35  S3_REGION: z.string().default("us-east-1"),
36  /** TASK-0101: the two private buckets (files + thumbs). `S3_BUCKET` is the
37   *  legacy alias for the files bucket (kept for older .env files). */
38  S3_BUCKET_FILES: z.string().default("factory-files"),
39  S3_BUCKET_THUMBS: z.string().default("factory-thumbs"),
40  S3_BUCKET: z.string().default("factory-files"),
41  S3_ACCESS_KEY: z.string().default("minioadmin"),
42  S3_SECRET_KEY: z.string().default("minioadmin"),
43  S3_PUBLIC_BASE: z.string().default("http://localhost:9001"),
44 
45  SEARCH_URL: z.string().url().default("http://localhost:9200"),
46  SEARCH_USERNAME: z.string().default("admin"),
47  SEARCH_PASSWORD: z.string().default("changeme"),
48  // The content index's dense_vector dimension (search §0: a parameter, not a
49  // constant — `dims` is immutable per index; a change rolls a new index via
50  // the alias, TASK-0099 `scripts/opensearch-reindex.ts`). 1536 =
51  // text-embedding-3-small. HNSW construction tuning (search §3).
52  SEARCH_VECTOR_DIMENSION: intFrom(1536),
53  SEARCH_KNN_M: intFrom(16),
54  SEARCH_KNN_EF_CONSTRUCTION: intFrom(100),
55 
56  MASTER_ENCRYPTION_KEY: z.string().default(""),
57 
58  /** TASK-0130: allow webhook delivery URLs targeting private/loopback/
59   *  link-local addresses (dev/testing override; production must stay false).
60   *  When false, the shared SSRF guard (assertPublicUrl*) is enforced. */
61  WEBHOOK_ALLOW_PRIVATE: boolFrom(false),
62 
63  /** TASK-0130 (C-13.8d): webhook retry backoff — the task default is
64   *  30 s base × 2 per attempt, capped at 15 min, with ±20 % jitter
65   *  (packages/shared `webhookBackoffMs`). Overridable per deployment; the
66   *  integration/E2E suites set short values so retries stay visible fast. */
67  WEBHOOK_BACKOFF_BASE_MS: intFrom(30_000),
68  WEBHOOK_BACKOFF_MAX_MS: intFrom(900_000),
69 
70  AI_API_KEY: z.string().default(""),
71  AI_PROVIDER: z.string().default("openai"),
72  // Empty = default OpenAI endpoint (see ai.processor fallback).
73  AI_BASE_URL: z.string().default(""),
74  AI_MODEL: z.string().default("gpt-4o-mini"),
75  // Embedding provider for the `embed:chunks` job (TASK-0099 §4, Phase-12
76  // stub). Empty (the default) ⇒ no real provider is wired: the job logs
77  // "embedding disabled (Phase 12)" and skips (no AiEmbedding row, no index
78  // write, the chunk docs' `vector` stays null). Set to a provider name to
79  // record placeholder AiEmbedding cache rows (Phase 12 turns this into a
80  // real provider call). AI_EMBEDDING_MODEL names the embedding model.
81  AI_EMBEDDING_PROVIDER: z.string().default(""),
82  AI_EMBEDDING_MODEL: z.string().default("text-embedding-3-small"),
83 
84  SMTP_HOST: z.string().default(""),
85  SMTP_PORT: intFrom(587),
86  SMTP_USER: z.string().default(""),
87  SMTP_PASS: z.string().default(""),
88  SMTP_FROM: z.string().default("noreply@factory.local"),
89  SMTP_STARTTLS: boolFrom(true),
90 
91  CLAMAV_HOST: z.string().default(""),
92  CLAMAV_PORT: intFrom(3310),
93  // TASK-0102 §3: UNIX ClamAV socket (e.g. /var/run/clamav/clamd.ctl). Empty
94  // = TCP (CLAMAV_HOST/CLAMAV_PORT) or, if both are empty, dev mode
95  // (scan skipped + WARN). REQUIRED in production (boot fails — fail closed).
96  CLAMD_SOCKET: z.string().default(""),
97 
98  // TASK-0102 §7: files pipeline tuning.
99  FILES_WORKER_CONCURRENCY: intFrom(4),
100  // ClamAV scan timeout (ms) — large files take a while; a hung scan is a
101  // RETRYABLE failure (BullMQ backoff), never a silent pass.
102  CLAMAV_TIMEOUT_MS: intFrom(120_000),
103  FILES_TEMP_DIR: z.string().default(""), // empty = os.tmpdir()/factory-files
104  FILES_TEMP_MAX_AGE_MS: intFrom(600_000), // cleanup: delete older than 10 min
105  FILES_TEMP_FREE_MIN_BYTES: intFrom(512 * 1024 * 1024), // disk guard
106 
107  // TASK-0094 R6: base URL for ABSOLUTE links in notification/digest emails
108  // (e.g. https://factory.example.com). The API calls the same value
109  // `WEB_ORIGIN` (recorded naming divergence); both default to the local web
110  // origin so a fresh setup needs no configuration.
111  PUBLIC_URL: z.string().url().default("http://localhost:3000"),
112 
113  WORKER_CONCURRENCY: intFrom(10),
114  // TASK-0113 §1: automation run pipeline concurrency (run jobs + schedule
115  // fires share this pool; separate from the general WORKER_CONCURRENCY).
116  AUTOMATIONS_WORKER_CONCURRENCY: intFrom(4),
117  // TASK-0113 §2: run-level retry cap (retryable errors). Dev-only override
118  // — production stays at 8.
119  AUTO_RUN_MAX_RETRIES: intFrom(8),
120  WORKER_ID: z.string().default("worker-1"),
121  WORKER_RETENTION_EXECUTE: boolFrom(false),
122  // HTTP health port (TASK-0009 item 4): /health + /health/ready on 4003.
123  HEALTH_PORT: intFrom(4003)
124});
125 
126export type WorkerEnv = z.infer<typeof envSchema>;
127 
128let cached: WorkerEnv | null = null;
129 
130/**
131 * Parse the environment once (cached). Pass an explicit object in tests.
132 * Throws on invalid config — a worker must never start half-configured.
133 */
134export function loadEnv(
135  source: Record<string, string | undefined> = process.env
136): WorkerEnv {
137  if (cached) return cached;
138  const parsed = envSchema.safeParse(source);
139  if (!parsed.success) {
140    throw new Error(
141      `Invalid worker environment: ${parsed.error.issues
142        .map((i) => `${i.path.join(".")}: ${i.message}`)
143        .join("; ")}`
144    );
145  }
146  cached = Object.freeze(parsed.data);
147  return cached;
148}
149 
150/** Test helper: reset the cached env (loadEnv re-parses next call). */
151export function resetEnvCache(): void {
152  cached = null;
153}
154