zod

安装依赖

npm

yarn

pnpm

bun

npm i drizzle-orm@rc zod

yarn add drizzle-orm@rc zod

pnpm add drizzle-orm@rc zod

bun add drizzle-orm@rc zod

选择模式

定义从数据库查询的数据形状 - 可用于验证 API 响应。

import { pgTable, text, integer } from 'drizzle-orm/pg-core';
import { createSelectSchema } from 'drizzle-orm/zod';

const users = pgTable('users', {
  id: integer().generatedAlwaysAsIdentity().primaryKey(),
  name: text().notNull(),
  age: integer().notNull()
});

const userSelectSchema = createSelectSchema(users);

const rows = await db.select({ id: users.id, name: users.name }).from(users).limit(1);
const parsed: { id: number; name: string; age: number } = userSelectSchema.parse(rows[0]); // 错误：上面的查询未返回 `age`

const rows = await db.select().from(users).limit(1);
const parsed: { id: number; name: string; age: number } = userSelectSchema.parse(rows[0]); // 将成功解析

也支持视图和枚举。

import { pgEnum, pgView } from 'drizzle-orm/pg-core';
import { gt } from 'drizzle-orm';
import { createSelectSchema } from 'drizzle-orm/zod';

const roles = pgEnum('roles', ['admin', 'basic']);
const rolesSchema = createSelectSchema(roles);
const parsed: 'admin' | 'basic' = rolesSchema.parse(...);

const usersView = pgView('users_view').as((qb) => qb.select().from(users).where(gt(users.age, 18)));
const usersViewSchema = createSelectSchema(usersView);
const parsed: { id: number; name: string; age: number } = usersViewSchema.parse(...);

插入模式

定义要插入到数据库中的数据形状 - 可用于验证 API 请求。

import { pgTable, text, integer } from 'drizzle-orm/pg-core';
import { createInsertSchema } from 'drizzle-orm/zod';

const users = pgTable('users', {
  id: integer().generatedAlwaysAsIdentity().primaryKey(),
  name: text().notNull(),
  age: integer().notNull()
});

const userInsertSchema = createInsertSchema(users);

const user = { name: 'John' };
const parsed: { name: string, age: number } = userInsertSchema.parse(user); // 错误：未定义 `age`

const user = { name: 'Jane', age: 30 };
const parsed: { name: string, age: number } = userInsertSchema.parse(user); // 将成功解析
await db.insert(users).values(parsed);

更新模式

定义要在数据库中更新的数据形状 - 可用于验证 API 请求。

import { pgTable, text, integer } from 'drizzle-orm/pg-core';
import { createUpdateSchema } from 'drizzle-orm/zod';
import { eq } from "drizzle-orm";

const users = pgTable('users', {
  id: integer().generatedAlwaysAsIdentity().primaryKey(),
  name: text().notNull(),
  age: integer().notNull()
});

const userUpdateSchema = createUpdateSchema(users);

const user = { age: 35 };
const parsed: { name?: string | undefined, age?: number | undefined } = userUpdateSchema.parse(user); // 将成功解析
await db.update(users).set(parsed).where(eq(users.name, 'Jane'));

精炼

每个 create schema 函数都接受一个额外的可选参数，你可以用它来扩展、修改或完全覆盖某个字段的 schema。提供回调函数会进行扩展或修改，而提供 Zod schema 则会覆盖它。

import { pgTable, text, integer, json } from 'drizzle-orm/pg-core';
import { createSelectSchema } from 'drizzle-orm/zod';
import { z } from 'zod/v4';

const users = pgTable('users', {
  id: integer().primaryKey(),
  name: text().notNull(),
  bio: text(),
  preferences: json()
});

const userSelectSchema = createSelectSchema(users, {
  name: (schema) => schema.max(20), // 扩展 schema
  bio: (schema) => schema.max(1000), // 在变为 nullable/optional 之前扩展 schema
  preferences: z.object({ theme: z.string() }) // 覆盖该字段，包括其可空性
});

const parsed: {
    id: number;
    name: string;
    bio: string | null;
    preferences: {
        theme: string;
    };
} = userSelectSchema.parse(...);

工厂函数

对于更高级的用例，你可以使用 createSchemaFactory 函数。

用例：使用扩展的 Zod 实例

import { pgTable, text, integer } from 'drizzle-orm/pg-core';
import { createSchemaFactory } from 'drizzle-orm/zod';
import { z } from '@hono/zod-openapi'; // 扩展的 Zod 实例

const users = pgTable('users', {
  id: integer().generatedAlwaysAsIdentity().primaryKey(),
  name: text().notNull(),
  age: integer().notNull()
});

const { createInsertSchema } = createSchemaFactory({ zodInstance: z });

const userInsertSchema = createInsertSchema(users, {
  // 现在我们可以使用扩展实例
  name: (schema) => schema.openapi({ example: 'John' })
});

用例：类型强制转换

import { pgTable, timestamp } from 'drizzle-orm/pg-core';
import { createSchemaFactory } from 'drizzle-orm/zod';
import { z } from 'zod/v4';

const users = pgTable('users', {
  ...,
  createdAt: timestamp().notNull()
});

const { createInsertSchema } = createSchemaFactory({
  // 此配置只会强制转换日期。将 `coerce` 设为 `true` 可强制转换所有数据类型，或者指定其他类型
  coerce: {
    date: true
  }
});

const userInsertSchema = createInsertSchema(users);
// 上面这段与以下内容相同：
const userInsertSchema = z.object({
  ...,
  createdAt: z.coerce.date()
});

数据类型参考

pg.boolean();

// Schema
z.boolean();

pg.pgEnum('name', ['val1', 'val2']);

// Schema
z.enum(['val1', 'val2']);

pg.date({ mode: 'date' });
pg.timestamp({ mode: 'date' });

// Schema
z.date();

pg.date({ mode: 'string' });
pg.timestamp({ mode: 'string' });
pg.cidr();
pg.inet();
pg.interval();
pg.macaddr();
pg.macaddr8();
pg.numeric();
pg.text();
pg.sparsevec();
pg.time();

// Schema
z.string();

pg.bit({ dimensions: ... });

// Schema
z.string().regex(/^[01]+$/).max(dimensions);

pg.uuid();

// Schema
z.string().uuid();

pg.char({ length: ... });

// Schema
z.string().length(length);

pg.varchar({ length: ... });

// Schema
z.string().max(length);

pg.text({ enum: ... });
pg.char({ enum: ... });
pg.varchar({ enum: ... });

// Schema
z.enum(enum);

pg.smallint();
pg.smallserial();

// Schema
z.number().min(-32_768).max(32_767).int(); // 16 位整数下限和上限

pg.real();

// Schema
z.number().min(-8_388_608).max(8_388_607); // 24 位整数下限和上限

pg.integer();
pg.serial();

// Schema
z.number().min(-2_147_483_648).max(2_147_483_647).int(); // 32 位整数下限和上限

pg.doublePrecision();

// Schema
z.number().min(-140_737_488_355_328).max(140_737_488_355_327); // 48 位整数下限和上限

pg.bigint({ mode: 'number' });
pg.bigserial({ mode: 'number' });

// Schema
z.number().min(-9_007_199_254_740_991).max(9_007_199_254_740_991).int(); // JavaScript 的最小和最大安全整数

pg.bigint({ mode: 'bigint' });
pg.bigserial({ mode: 'bigint' });

// Schema
z.bigint().min(-9_223_372_036_854_775_808n).max(9_223_372_036_854_775_807n); // 64 位整数下限和上限

pg.geometry({ type: 'point', mode: 'tuple' });
pg.point({ mode: 'tuple' });

// Schema
z.tuple([z.number(), z.number()]);

pg.geometry({ type: 'point', mode: 'xy' });
pg.point({ mode: 'xy' });

// Schema
z.object({ x: z.number(), y: z.number() });

pg.halfvec({ dimensions: ... });
pg.vector({ dimensions: ... });

// Schema
z.array(z.number()).length(dimensions);

pg.line({ mode: 'abc' });

// Schema
z.object({ a: z.number(), b: z.number(), c: z.number() });

pg.line({ mode: 'tuple' });

// Schema
z.tuple([z.number(), z.number(), z.number()]);

pg.json();
pg.jsonb();

// Schema
z.union([z.union([z.string(), z.number(), z.boolean(), z.null()]), z.record(z.any()), z.array(z.any())]);

pg.dataType().array(...);

// Schema
z.array(baseDataTypeSchema).length(size);

pg.bytea();

// Schema
zod.custom<Buffer>((v) => v instanceof Buffer)