zod_gen

Generate Zod schemas and TypeScript types from Rust types with zero runtime overhead and full type safety.

🚀 Features

• Zero-cost abstractions - No runtime overhead, pure compile-time code generation
• Full type safety - End-to-end type safety from Rust to TypeScript
• Serde rename support - Automatic handling of #[serde(rename = "...")] attributes
• Derive macro support - #[derive(ZodSchema)] for automatic schema generation
• Primitive type support - Built-in support for all common Rust types
• Generic types - Automatic handling of Option<T>, Vec<T>, and more
• Custom schemas - Manual implementation for complex types
• Batch generation - Generate multiple schemas in a single TypeScript file

📦 Installation

Add both crates to your Cargo.toml:

[dependencies]
zod_gen = "1.2.0"
zod_gen_derive = "1.2.0"

🔧 Quick Start

Using the derive macro (recommended)

use zod_gen::ZodSchema;
use zod_gen_derive::ZodSchema;

#[derive(ZodSchema)]
struct User {
    id: u64,
    name: String,
    email: String,
    is_admin: bool,
    tags: Vec<String>,
    profile: Option<UserProfile>,
}

#[derive(ZodSchema)]
struct UserProfile {
    bio: String,
    avatar_url: Option<String>,
}

#[derive(ZodSchema)]
enum UserStatus {
    #[serde(rename = "active")]
    Active,
    #[serde(rename = "inactive")]
    Inactive,
    #[serde(rename = "suspended")]
    Suspended,
}

fn main() {
    // Generate Zod schema
    println!("{}", User::zod_schema());
    // Output: z.object({
    //   id: z.number(),
    //   name: z.string(),
    //   email: z.string(),
    //   is_admin: z.boolean(),
    //   tags: z.array(z.string()),
    //   profile: z.object({ bio: z.string(), avatar_url: z.string().optional() }).optional()
    // })
}

Manual implementation

use zod_gen::{ZodSchema, zod_object, zod_string, zod_number, zod_boolean};

struct User {
    id: u64,
    name: String,
    is_admin: bool,
}

impl ZodSchema for User {
    fn zod_schema() -> String {
        zod_object(&[
            ("id", zod_number()),
            ("name", zod_string()),
            ("is_admin", zod_boolean()),
        ])
    }
}

Single file generation with ZodGenerator

use zod_gen::ZodGenerator;
use std::fs;

fn generate_types() {
    let mut generator = ZodGenerator::new();

    // Add all your types with meaningful names
    generator.add_schema::<User>("User");
    generator.add_schema::<UserProfile>("UserProfile");
    generator.add_schema::<UserStatus>("UserStatus");

    // Generate a single TypeScript file with all schemas
    let content = generator.generate();
    fs::write("types/schemas.ts", content).unwrap();
}

📚 Generated TypeScript

The generated TypeScript provides both Zod schemas and inferred types in a single file:

// Generated schemas.ts
import * as z from 'zod';

export const UserSchema = z.object({
  id: z.number(),
  name: z.string(),
  email: z.string(),
  is_admin: z.boolean(),
  tags: z.array(z.string()),
  profile: z.object({
    bio: z.string(),
    avatar_url: z.string().optional()
  }).optional()
});
export type User = z.infer<typeof UserSchema>;

export const UserProfileSchema = z.object({
  bio: z.string(),
  avatar_url: z.string().optional()
});
export type UserProfile = z.infer<typeof UserProfileSchema>;

export const UserStatusSchema = z.union([z.literal('active'), z.literal('inactive'), z.literal('suspended')]);
export type UserStatus = z.infer<typeof UserStatusSchema>;

Use it in your TypeScript code:

import { UserSchema, type User } from './types/schemas';

// Runtime validation
const validateUser = (data: unknown): User => {
  return UserSchema.parse(data);
};

// Type-safe API calls
const createUser = async (user: User): Promise<User> => {
  const response = await fetch('/api/users', {
    method: 'POST',
    body: JSON.stringify(user),
  });
  return UserSchema.parse(await response.json());
};

🏗️ Architecture

This repository contains two crates:

zod_gen - Core Library

• ZodSchema trait for defining schemas
• Helper functions for building Zod expressions
• ZodGenerator for batch file generation
• Built-in implementations for primitive types

zod_gen_derive - Derive Macro

• #[derive(ZodSchema)] procedural macro
• Supports structs with named fields
• Supports enums with unit variants
• Automatic dependency resolution

🎯 Supported Types

Primitives

• String, &str → z.string() (TypeScript: string)
• i32, i64, u32, u64, f32, f64 → z.number() (TypeScript: number)
• bool → z.boolean() (TypeScript: boolean)

Generics

• Option<T> → T.optional() (TypeScript: Optional<T>)
• Vec<T> → z.array(T) (TypeScript: Array<T>)
• HashMap<String, T> → z.record(z.string(), T) (TypeScript: Record<string, T>)
• Custom collections via manual implementation

Structs

• Named fields → z.object({ ... })
• Nested structs supported

Enums

• Unit variants → z.union([z.literal('A'), z.literal('B')])
• Serde rename support → #[serde(rename = "custom_name")] → z.literal('custom_name')

🎯 Serde Rename Support

zod_gen automatically handles #[serde(rename = "...")] attributes, ensuring your TypeScript types match your serialized JSON exactly:

use serde::{Serialize, Deserialize};
use zod_gen_derive::ZodSchema;

#[derive(ZodSchema, Serialize, Deserialize)]
enum ApiStatus {
    #[serde(rename = "success")]
    Success,
    #[serde(rename = "error")]
    Error,
    #[serde(rename = "pending")]
    Pending,
}

#[derive(ZodSchema, Serialize, Deserialize)]
enum UserRole {
    #[serde(rename = "admin")]
    Administrator,
    #[serde(rename = "user")]
    RegularUser,
    #[serde(rename = "guest")]
    GuestUser,
}

Generated TypeScript:

export const ApiStatusSchema = z.union([z.literal('success'), z.literal('error'), z.literal('pending')]);
export type ApiStatus = z.infer<typeof ApiStatusSchema>;

export const UserRoleSchema = z.union([z.literal('admin'), z.literal('user'), z.literal('guest')]);
export type UserRole = z.infer<typeof UserRoleSchema>;

Type Safety Benefits:

// ✅ TypeScript accepts the serde-renamed values
const status: ApiStatus = 'success';
const role: UserRole = 'admin';

// ❌ TypeScript catches typos with the Rust variant names
const badStatus: ApiStatus = 'Success'; // Error: Type '"Success"' is not assignable to type 'ApiStatus'
const badRole: UserRole = 'Administrator'; // Error: Type '"Administrator"' is not assignable to type 'UserRole'

This ensures perfect alignment between your Rust API and TypeScript frontend, catching serialization mismatches at compile time.

🔧 Advanced Usage

Schema Generation Strategy

By default, zod_gen generates inline schemas for nested objects. This means that complex types are expanded directly into their Zod representation:

#[derive(ZodSchema)]
struct User {
    id: u64,
    name: String,
    profile: Option<UserProfile>,
}

#[derive(ZodSchema)]
struct UserProfile {
    bio: String,
    avatar_url: Option<String>,
}

// Generates inline schema:
// z.object({
//   id: z.number(),
//   name: z.string(),
//   profile: z.object({
//     bio: z.string(),
//     avatar_url: z.string().nullable()
//   }).nullable()
// })

This approach works consistently across all generic types:

use std::collections::HashMap;

// HashMap<String, User> generates:
// z.record(z.string(), z.object({
//   id: z.number(),
//   name: z.string(),
//   profile: z.object({
//     bio: z.string(),
//     avatar_url: z.string().nullable()
//   }).nullable()
// }))

Benefits of inline schemas:

• ✅ Self-contained - no external dependencies
• ✅ Works with any nesting level
• ✅ Consistent behavior across all types
• ✅ No need to manage schema imports

Considerations:

• Schema duplication if the same type is used in multiple places
• Larger generated schemas for deeply nested structures

User-Controlled Naming

You provide the TypeScript type names when adding schemas to the generator:

let mut gen = ZodGenerator::new();

// Use meaningful names for your TypeScript types
gen.add_schema::<HashMap<String, User>>("UserMap");
gen.add_schema::<Vec<User>>("UserList");
gen.add_schema::<Option<UserProfile>>("OptionalProfile");

This generates clean TypeScript with your chosen names:

export const UserMapSchema = z.record(z.string(), UserSchema);
export type UserMap = z.infer<typeof UserMapSchema>;

export const UserListSchema = z.array(UserSchema);
export type UserList = z.infer<typeof UserListSchema>;

Benefits:

• Full control over TypeScript type names
• Meaningful names instead of auto-generated ones
• No magic - you decide what gets exported
• TypeScript types are inferred from Zod schemas using z.infer<>

Single File Output

The ZodGenerator creates a single TypeScript file containing all your schemas. This approach:

• Simplifies file management - No need to track multiple files
• Reduces complexity - One file, one import
• Improves maintainability - All schemas in one place
• Enables tree-shaking - Import only what you need

If you need multiple files, simply create multiple generators:

// Generate API types
let mut api_gen = ZodGenerator::new();
api_gen.add_schema::<User>("User");
api_gen.add_schema::<Post>("Post");
std::fs::write("types/api.ts", api_gen.generate()).unwrap();

// Generate config types
let mut config_gen = ZodGenerator::new();
config_gen.add_schema::<AppConfig>("AppConfig");
std::fs::write("types/config.ts", config_gen.generate()).unwrap();

Custom Schema Implementation

use zod_gen::ZodSchema;
use chrono::{DateTime, Utc};

impl ZodSchema for DateTime<Utc> {
    fn zod_schema() -> String {
        "z.string().datetime()".to_string()
    }
}

Integration with Build Scripts

Create a build.rs file:

use zod_gen::ZodGenerator;

fn main() {
    let mut generator = ZodGenerator::new();
    generator.add_schema::<MyType>("MyType");

    // Generate during build
    let content = generator.generate();
    std::fs::write("frontend/types/schemas.ts", content).unwrap();
}

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

🔄 CI/CD Pipeline

This project uses GitHub Actions for comprehensive automation:

• 🧪 Continuous Integration: Tests run on every PR and push

  • Multi-version Rust testing (stable, beta, nightly)
  • Code formatting and linting with clippy
  • Documentation building and security audits
  • Example validation and code coverage

• 🚀 Automated Releases: Tag-triggered releases

  • Automatic publishing to crates.io
  • GitHub release creation with changelog
  • Full validation before publishing

• 🔒 Security & Maintenance:

  • Weekly dependency updates via automated PRs
  • Security vulnerability scanning
  • Breaking change detection on PRs

• 📚 Documentation: Auto-deployed to GitHub Pages

📋 PR Guidelines

• Follow Conventional Commits format
• Update CHANGELOG.md for non-documentation changes
• Ensure all tests pass and examples work
• Code must be formatted (cargo fmt) and pass clippy lints

Development Setup

git clone https://github.com/cimatic/zod_gen.git
cd zod_gen

# The rust-toolchain.toml ensures you use the same Rust version as CI
cargo test --workspace

# Run clippy with CI-equivalent flags
./scripts/clippy-ci.sh

See DEVELOPMENT.md for detailed development guidelines and CI consistency tips.

Running Examples

# Manual ZodSchema implementation
cargo run --example basic_usage

# Using the derive macro for automatic schema generation
cargo run --example derive_example

# Using ZodGenerator for multiple schemas with custom naming
cargo run --example generator_example

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Zod - Runtime type validation for TypeScript
• serde - Inspiration for the derive macro pattern
• ts-rs - Alternative approach to Rust→TypeScript codegen

---

Built with ❤️ by the Cimatic team