2nd/10_Wiki/Topic_Programming/Programming & Language/Zod 런타임 유효성 검사 통합.md

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

id	title	category	status	canonical_id	aliases	duplicate_of	source_trust_level	confidence_score	verification_status	tags	raw_sources	last_reinforced	github_commit	tech_stack
wiki-2026-0508-zod-런타임-유효성-검사-통합	Zod 런타임 유효성 검사 통합	10_Wiki/Topics	verified	self	Zod TypeScript runtime validation Zod schema	none	A	0.9	applied	typescript validation zod runtime-types schema		2026-05-10	pending	language framework TypeScript Zod 3.x

Zod 런타임 유효성 검사 통합

매 한 줄

"매 type schema = single source of truth". Zod는 매 TypeScript type을 런타임 schema 로 정의하고, parse 시점에 매 validation + type narrowing 을 동시 제공한다. API boundary, env vars, form input 매 unsafe input 의 매 첫 검증선.

매 핵심

매 동기 (Why Zod over alternatives)

• TypeScript types are erased: 매 컴파일 후 interface User는 매 사라짐 → API response 의 data as User 매 lying.
• Zod = schema → type: z.infer<typeof Schema> 로 매 schema 가 source of truth.
• Composability: 매 .merge, .partial, .extend, .transform 으로 매 schema 합성.
• Error-rich: parse failure 시 매 path, code, message tree 반환.

매 경쟁 라이브러리

• Yup: 매 older, schema → type 약함, 매 Zod 가 대체.
• io-ts: 매 더 functional (fp-ts), 매 learning curve 높음.
• Valibot: 매 tree-shakable, 매 bundle size 우선이면 고려 (~10x smaller).
• ArkType: 매 string-based syntax, 매 빠르지만 ecosystem 작음.
• Zod: 매 default choice in 2026 — DX, ecosystem (tRPC, React Hook Form), maturity.

매 응용

1. API boundary: fetch response 매 parse 후 typed 으로 사용.
2. Form validation: React Hook Form + zodResolver.
3. env vars: process.env 의 매 schema parse, missing key 시 즉시 fail.
4. DB row → domain: ORM 결과 매 Schema.parse(row).
5. LLM structured output: Claude/GPT JSON response 매 schema 로 검증.

💻 패턴

Pattern 1: 기본 schema + type inference

import { z } from "zod";

export const UserSchema = z.object({
  id: z.string().uuid(),
  email: z.string().email(),
  age: z.number().int().min(0).max(150),
  role: z.enum(["admin", "user", "guest"]),
  createdAt: z.coerce.date(),
});

export type User = z.infer<typeof UserSchema>;

// usage
const raw: unknown = await fetchUser();
const user = UserSchema.parse(raw); // throws ZodError if invalid
//    ^? User

Pattern 2: safeParse for non-throw

const result = UserSchema.safeParse(raw);
if (!result.success) {
  console.error(result.error.flatten());
  return;
}
const user = result.data; // typed

Pattern 3: env validation (fail fast at boot)

const EnvSchema = z.object({
  DATABASE_URL: z.string().url(),
  PORT: z.coerce.number().int().positive().default(3000),
  NODE_ENV: z.enum(["development", "production", "test"]),
  ANTHROPIC_API_KEY: z.string().startsWith("sk-ant-"),
});

export const env = EnvSchema.parse(process.env);
// process exits at startup if invalid — better than runtime surprise

Pattern 4: discriminated union

const ResultSchema = z.discriminatedUnion("status", [
  z.object({ status: z.literal("ok"), data: z.string() }),
  z.object({ status: z.literal("error"), code: z.number() }),
]);

type Result = z.infer<typeof ResultSchema>;
// narrowing on .status works correctly

Pattern 5: transform for parse-not-validate

const DateSchema = z.string().transform((s, ctx) => {
  const d = new Date(s);
  if (isNaN(d.getTime())) {
    ctx.addIssue({ code: "custom", message: "Invalid date" });
    return z.NEVER;
  }
  return d;
});

const out = DateSchema.parse("2026-05-10"); // Date instance

Pattern 6: API client with Zod

async function fetchUser(id: string): Promise<User> {
  const res = await fetch(`/api/users/${id}`);
  const json = await res.json();
  return UserSchema.parse(json); // unknown → User
}

Pattern 7: tRPC integration

import { router, procedure } from "./trpc";

export const userRouter = router({
  create: procedure
    .input(UserSchema.omit({ id: true, createdAt: true }))
    .mutation(async ({ input }) => {
      // input is fully typed + validated
      return db.user.create({ data: input });
    }),
});

Pattern 8: React Hook Form + Zod

import { useForm } from "react-hook-form";
import { zodResolver } from "@hookform/resolvers/zod";

const FormSchema = z.object({
  email: z.string().email(),
  password: z.string().min(8),
});

const { register, handleSubmit, formState } = useForm({
  resolver: zodResolver(FormSchema),
});

Pattern 9: schema composition

const Base = z.object({ id: z.string(), createdAt: z.date() });
const Post = Base.extend({ title: z.string(), body: z.string() });
const PostUpdate = Post.partial().required({ id: true });

Pattern 10: refinement (custom rules)

const PasswordSchema = z
  .object({ password: z.string(), confirm: z.string() })
  .refine((d) => d.password === d.confirm, {
    message: "Passwords do not match",
    path: ["confirm"],
  });

매 결정 기준

상황	Approach
API boundary, untrusted input	Zod parse
Internal pure-TS code	Type only, no Zod
Bundle size critical (mobile, edge)	Valibot
Functional ergonomics	io-ts
LLM structured output (Claude/GPT)	Zod + tool schema
Performance hot path (>10k parses/sec)	Compile to TypeBox/AJV

기본값: Zod 3.x — 매 modern TS app 의 default validation layer.

🔗 Graph

• 부모: TypeScript · Runtime_Validation
• 변형: Valibot · Yup · ArkType
• 응용: React Hook Form
• Adjacent: JSON Schema · 과잉 속성 체크 (Excess Property Checking)

🤖 LLM 활용

언제: untrusted boundary (API, form, env, LLM output) 매 parse. tool/function calling 의 매 input schema. 언제 X: internal pure-TS code 매 over-validation 불필요. hot loop 의 매 매 parse 호출.

❌ 안티패턴

• Anti1: parse everywhere: 매 internal function 매 Zod parse — 매 overhead 누적, 매 type only 충분.
• Anti2: as cast after parse: Schema.parse(x) as MyType — 매 redundant, parse 가 이미 typed return.
• Anti3: schema duplication: type + schema 따로 정의 — 매 z.infer 사용.
• Anti4: nested transforms with side effects: transform 안에서 fetch/IO — 매 pure 하게 유지.

🧪 검증 / 중복

• Verified (Zod docs colinhacks.com/zod, 2026 ecosystem).
• 신뢰도 A.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — Zod runtime validation patterns + 2026 ecosystem context