refactoring-inline-types

1. Analyze Current Types

Read the target file specified by the user.

Identify all inline type definitions:

• Object type literals in function parameters
• Union types without names
• Intersection types without names
• Complex type expressions
• Mapped types without names
• Conditional types without names

Find duplicated type patterns:

• Same object shapes in multiple places
• Repeated union combinations
• Similar generic patterns
• Common property subsets

Locate complex inline types:

• Nested object types (3+ levels deep)
• Large union types (4+ members)
• Complex mapped types
• Utility type compositions

Detect types that could be reused:

• Function signatures used multiple times
• Data structures shared across functions
• API response/request shapes
• Event handler types

2. Design Type Structure

Determine appropriate type organization:

Interfaces for object shapes (extensible):

interface User {
  id: string
  name: string
  email: string
}

interface AdminUser extends User {
  permissions: string[]
}

Type aliases for unions/primitives/functions:

type Status = "pending" | "active" | "inactive"
type ID = string | number
type Handler = (event: Event) => void

Generics for reusable patterns:

type Result<T, E = Error> =
  | { success: true; data: T }
  | { success: false; error: E }

type Nullable<T> = T | null
type Optional<T> = T | undefined

Utility types for transformations:

type PartialUser = Partial<User>
type UserCredentials = Pick<User, "email" | "password">
type PublicUser = Omit<User, "password">
type ReadonlyUser = Readonly<User>

Plan type hierarchy:

• Base types first
• Derived types after
• Utility types at end
• Group related types together

3. Extract and Refactor

Create types module if needed:

• types.ts for single domain
• types/index.ts for multiple domains
• types/{domain}.ts for large projects

Extract inline types to named definitions:

Before:

function createUser(data: {
  name: string
  email: string
  age: number
}): { id: string; name: string; email: string; age: number } {
  return { id: generateId(), ...data }
}

After:

interface UserInput {
  name: string
  email: string
  age: number
}

interface User extends UserInput {
  id: string
}

function createUser(data: UserInput): User {
  return { id: generateId(), ...data }
}

Give descriptive names:

• Follow project conventions
• Use domain language
• Be specific and clear
• Avoid generic names (Data, Info, Params)

Add JSDoc for complex types:

/**
 * Represents the result of an async operation.
 * Success case includes data, failure case includes error.
 */
type AsyncResult<T, E = Error> =
  | { success: true; data: T }
  | { success: false; error: E }

Use generic constraints:

interface Repository<T extends { id: string }> {
  findById(id: string): Promise<T | null>
  save(item: T): Promise<T>
}

Organize types logically:

// Base types
export interface User {
  id: string
  name: string
}

// Derived types
export interface AdminUser extends User {
  permissions: string[]
}

// Input/output types
export type CreateUserInput = Omit<User, "id">
export type UserResponse = Readonly<User>

// Utility types
export type PartialUser = Partial<User>

Export public types:

export interface User { }
export type Status = "active" | "inactive"

Keep internal types private:

interface InternalCache { }
type ValidationState = "valid" | "invalid"

Update original file:

import { User, CreateUserInput, UserResponse } from "./types"

4. Optimization

Replace duplicate types:

• Single source of truth
• Import instead of redefine
• Use extends for variations

Simplify complex inline types:

// Before
function process(
  data: { id: string } & ({ type: "user"; name: string } | { type: "admin"; permissions: string[] })
): void { }

// After
interface BaseData {
  id: string
}

type UserData = BaseData & {
  type: "user"
  name: string
}

type AdminData = BaseData & {
  type: "admin"
  permissions: string[]
}

type ProcessData = UserData | AdminData

function process(data: ProcessData): void { }

Add type aliases for readability:

type UserID = string
type Timestamp = number
type EmailAddress = string

Use discriminated unions:

type Shape =
  | { kind: "circle"; radius: number }
  | { kind: "square"; size: number }
  | { kind: "rectangle"; width: number; height: number }

function area(shape: Shape): number {
  switch (shape.kind) {
    case "circle":
      return Math.PI * shape.radius ** 2
    case "square":
      return shape.size ** 2
    case "rectangle":
      return shape.width * shape.height
  }
}

Apply utility types:

type UserInput = Omit<User, "id" | "createdAt">
type PartialUpdate = Partial<UserInput>
type ReadonlyConfig = Readonly<Config>
type RequiredFields = Required<OptionalConfig>

5. Validation

Run type check on the target file:

pnpm type-check 2>&1 | grep "target-file"

Replace target-file with the actual file path.

Verify:

• No type errors introduced
• Type definitions properly exported/imported
• Refactored code maintains same type safety
• All references updated correctly

If errors found:

• Fix import/export issues
• Correct type definitions
• Update all references
• Re-run validation