Refine Blocks

Refine blocks allow conditional field overrides based on other field values. They're useful for creating variant-specific constraints without duplicating schemas.

Basic Syntax

Use refine { } after a schema to define conditional overrides:

schema Player {
  position: "GK" | "DEF" | "MID" | "FWD",
  goals: int in 0..30,
  clean_sheets: int in 0..20
} refine {
  if position == "GK" {
    goals: int in 0..2
  },
  if position == "FWD" {
    clean_sheets: int in 0..3
  }
}

After initial generation:

• If position == "GK", goals is regenerated with range 0..2
• If position == "FWD", clean_sheets is regenerated with range 0..3

How It Works

1. All fields are generated normally
2. Each if condition is evaluated
3. If true, the specified fields are regenerated with new constraints
4. Multiple conditions can match and apply sequentially

Multiple Conditions

schema Employee {
  department: "engineering" | "sales" | "marketing",
  level: "junior" | "senior" | "lead",
  salary: int in 30000..200000
} refine {
  if department == "engineering" {
    salary: int in 50000..200000
  },
  if department == "sales" {
    salary: int in 40000..150000
  },
  if level == "junior" {
    salary: int in 30000..60000
  },
  if level == "lead" {
    salary: int in 100000..200000
  }
}

Note: If both conditions match (e.g., junior engineer), both refinements apply in order.

Refine vs Constraints

Feature	assume Constraints	refine Blocks
When applied	During generation	After generation
Failure mode	Retry (rejection sampling)	Override value
Use case	Hard constraints	Variant-specific ranges

// Using assume: might need many retries
schema Player {
  position: "GK" | "FWD",
  goals: int in 0..30,
  assume if position == "GK" { goals <= 2 }
}

// Using refine: guaranteed to satisfy
schema Player {
  position: "GK" | "FWD",
  goals: int in 0..30
} refine {
  if position == "GK" { goals: int in 0..2 }
}

Practical Examples

Product Pricing Tiers

schema Product {
  category: "electronics" | "clothing" | "food",
  tier: "budget" | "standard" | "premium",
  price: decimal in 1..1000
} refine {
  if category == "electronics" {
    price: decimal in 50..2000
  },
  if category == "food" {
    price: decimal in 1..50
  },
  if tier == "budget" {
    price: decimal in 1..30
  },
  if tier == "premium" {
    price: decimal in 200..2000
  }
}

Vehicle Specifications

schema Vehicle {
  type: "sedan" | "suv" | "truck" | "motorcycle",
  doors: int in 2..5,
  seats: int in 1..8,
  cargo_capacity: int in 100..2000
} refine {
  if type == "motorcycle" {
    doors: int in 0..0,
    seats: int in 1..2,
    cargo_capacity: int in 0..50
  },
  if type == "truck" {
    doors: int in 2..4,
    cargo_capacity: int in 500..2000
  },
  if type == "sedan" {
    doors: int in 4..4,
    seats: int in 4..5
  }
}

User Permissions

schema User {
  role: "guest" | "member" | "admin",
  can_read: boolean,
  can_write: boolean,
  can_delete: boolean,
  can_admin: boolean
} refine {
  if role == "guest" {
    can_read: true,
    can_write: false,
    can_delete: false,
    can_admin: false
  },
  if role == "member" {
    can_read: true,
    can_write: true,
    can_delete: false,
    can_admin: false
  },
  if role == "admin" {
    can_read: true,
    can_write: true,
    can_delete: true,
    can_admin: true
  }
}

Combining with Other Features

With Computed Fields

schema Order {
  type: "retail" | "wholesale",
  quantity: int in 1..100,
  unit_price: decimal in 10..100
} refine {
  if type == "wholesale" {
    quantity: int in 50..500,
    unit_price: decimal in 5..50
  }
}

// Computed fields use refined values
schema OrderWithTotal {
  type: "retail" | "wholesale",
  quantity: int in 1..100,
  unit_price: decimal in 10..100,
  total: quantity * unit_price
} refine {
  if type == "wholesale" {
    quantity: int in 50..500,
    unit_price: decimal in 5..50
  }
}

Note: Computed fields are recalculated after refinement.