@ipld/schema

JavaScript IPLD Schema parser, validator, and code generator.

What are IPLD Schemas?

IPLD Schemas define a type system for content-addressed data. Think of them like TypeScript or Protocol Buffers, but designed for immutable, linked data structures. They help you:

• 🛡️ Validate that your data matches an expected structure
• 🔄 Transform between compact binary formats and developer-friendly JSON
• 🏗️ Generate type-safe code in multiple languages (Go, Rust, TypeScript)
• 🔗 Link between data structures using content identifiers (CIDs)

Learn more at https://ipld.io/docs/schemas/

Features

• Parse and validate IPLD Schema DSL (Domain Specific Language)
• Runtime validation with automatic format conversion
• Code generation for Go, Rust, and TypeScript
• Advanced type features: optional fields, defaults, renames, type aliases
• Multiple data representations for space-efficient storage
• CLI tools for validation and code generation

Installation

npm install @ipld/schema

For CLI usage:

npm install -g @ipld/schema

Quick Start

Basic Schema Definition

IPLD Schemas use a simple, readable syntax:

import { fromDSL } from '@ipld/schema/from-dsl.js'
import { create } from '@ipld/schema/typed.js'

// Define your schema
const schema = fromDSL(`
  # A simple user profile schema
  type UserProfile struct {
    username String
    email String
    age Int
    isActive Bool
  }
`)

// Create a validator
const validator = create(schema, 'UserProfile')

// Validate some data
const userData = {
  username: 'alice',
  email: '[email protected]',
  age: 25,
  isActive: true
}

const validatedData = validator.toTyped(userData)
if (validatedData === undefined) {
  console.error('Invalid data!')
} else {
  console.log('Valid user profile:', validatedData)
}

Schema Language Basics

Basic Types

type MyString String     # UTF-8 string
type MyInt Int           # Signed integer
type MyFloat Float       # Floating point
type MyBool Bool         # Boolean
type MyBytes Bytes       # Binary data
type MyLink &Any         # IPLD Link (CID)

Structs (Objects)

type Person struct {
  name String
  age Int
  email String optional          # Optional field
  nickname String (implicit "")  # Default value
}

Lists and Maps

# List of strings
type Names [String]

# Map from string to integers
type Scores {String: Int}

# Nested structures
type Team struct {
  name String
  members [Person]
  metadata {String: String}
}

Enums

type Status enum {
  | Active
  | Inactive
  | Pending
}

# With custom string values
type Color enum {
  | Red ("red")
  | Green ("green")
  | Blue ("blue")
} representation string

Type Aliases

type UserID = String
type Timestamp = Int
type EmailAddress = String

Data Representations

IPLD Schemas separate the logical structure (what developers work with) from the storage format (how it's encoded). This allows for space-efficient storage while maintaining developer-friendly APIs.

Example: Tuple Representation

import { fromDSL } from '@ipld/schema/from-dsl.js'
import { create } from '@ipld/schema/typed.js'

const schema = fromDSL(`
  # Store as array instead of object to save space
  type Point struct {
    x Float
    y Float
  } representation tuple
`)

const validator = create(schema, 'Point')

// Work with nice objects in your code
const point = { x: 10.5, y: 20.3 }

// But it's stored as a compact array
const stored = validator.toRepresentation(point)
console.log(stored) // [10.5, 20.3]

// And automatically converted back
const restored = validator.toTyped(stored)
console.log(restored) // { x: 10.5, y: 20.3 }

Code Generation

Generate type-safe code from your schemas:

Go

import { fromDSL } from '@ipld/schema/from-dsl.js'
import { generateGo } from '@ipld/schema/gen/go.js'

const schema = fromDSL(`
  type Person struct {
    name String
    age Int optional
  }
`)

const goCode = generateGo(schema, { packageName: 'person' })
// Generates Go structs with proper JSON tags and pointer types for optionals

Rust

import { generateRust } from '@ipld/schema/gen/rust.js'

const rustCode = generateRust(schema)
// Generates Rust structs with serde derives and Option<T> for optionals

TypeScript

import { generateTypeScript } from '@ipld/schema/gen/typescript.js'

const tsCode = generateTypeScript(schema)
// Generates TypeScript interfaces and runtime validators

Advanced Features

Field Renames

Control JSON field names separately from your schema field names:

type ServerConfig struct {
  serverPort Int (rename "server_port")
  debugMode Bool (rename "debug_mode")
  apiKey String (rename "api_key")
}

Annotations

Add language-specific type information:

# Use big integers in Go
# @gotype(big.Int)
type Balance Int

type Transaction struct {
  # Custom serialization in Rust
  # @rustserde(with = "chrono::serde::ts_seconds")
  timestamp Int

  # Multiple annotations
  # @gotag(`json:"tx_id" db:"transaction_id"`)
  id String
}

Custom Transforms

Handle special encoding requirements:

const customTransforms = {
  Base64String: {
    // Decode base64 strings to bytes
    toTyped: (str) => {
      try {
        return Uint8Array.from(atob(str), c => c.charCodeAt(0))
      } catch {
        return undefined
      }
    },
    // Encode bytes to base64 strings
    toRepresentation: (bytes) => {
      return btoa(String.fromCharCode(...bytes))
    }
  }
}

const validator = create(schema, 'MyType', { customTransforms })

Command Line Interface

The ipld-schema command provides tools for working with schemas:

Validation

# Validate schema files
ipld-schema validate schema.ipldsch

# Extract and validate schemas from markdown
ipld-schema validate README.md

Conversion

# Convert to JSON format
ipld-schema to-json schema.ipldsch

# Pretty print as canonical schema
ipld-schema to-schema schema.ipldsch

Code Generation

# Generate JavaScript validators
ipld-schema to-js schema.ipldsch

# Generate TypeScript definitions
ipld-schema to-tsdefs schema.ipldsch

API Reference

Parsing Schemas

• fromDSL(dsl: string) - Parse schema DSL into an AST
• toDSL(schema: Schema) - Convert AST back to DSL

Validation

• create(schema: Schema, type: string, options?) - Create a validator
  • Returns { toTyped, toRepresentation }
  • toTyped(data) - Validate and convert from storage format
  • toRepresentation(data) - Validate and convert to storage format

Code Generation

• generateGo(schema, options) - Generate Go code
• generateRust(schema, options) - Generate Rust code
• generateTypeScript(schema, options) - Generate TypeScript code

License & Copyright

Copyright 2019-2025 Rod Vagg

Licensed under either of

• Apache 2.0, (LICENSE-APACHE / http://www.apache.org/licenses/LICENSE-2.0)
• MIT (LICENSE-MIT / http://opensource.org/licenses/MIT)

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.