nakprok - Python with Enforced Static Typing

Same Python syntax. Zero dynamic typing.

nakprok (นาคปรก) is a subset of Python that enforces static type annotations everywhere. It's not a new language, but a stricter runtime for the Python you already know.

Why the name? Nakprok is a depiction of the Buddha sheltered by the multi-headed serpent Nāga in Thai Buddhist art. Just as the Naga protected the Buddha, this project protects your code from untyped or mistyped logic.

🚀 Key Concepts

1. Explicit is Better than Implicit: If a value's type can't be seen in the code, it shouldn't be there.
2. Zero Runtime Overhead: Validation happens at the AST level before execution. Once validated, it runs as standard CPython.
3. Ecosystem Compatibility: Works with all existing Python libraries and tools.
4. No New Syntax: Uses standard Python 3.10+ type hinting syntax.

---

🆚 How is this different from mypy?

nakprok is not a replacement for mypy. They tackle type safety from different angles and work well together:

	mypy	nakprok
When	Static analysis (separate step)	AST validation before execution
What	Deep type inference & checking	Enforces explicit type declarations
Scope	All valid Python	Strict subset (blocks lambda, untyped unpacking, etc.)
Result	Advisory report (can be ignored)	Hard block — code won't run

Think of it this way:

• nakprok ensures every value is explicitly typed at the source level.
• mypy then verifies those types are semantically correct (e.g., protocol compliance, generic constraints, int where float is expected).

You can (and should) run both:

nakprok check .     # Enforces explicit typing
mypy .              # Validates type correctness

---

🛠️ Implemented Rules

nakprok currently enforces the following rules inside functions (and classes):

1. Function Signatures

• All parameters must be typed: def add(a: int, b: int) is required.
• Return types are mandatory: -> None or -> int must be specified.
• Exceptions: self and cls parameters in methods are exempt.
• Variadic arguments: *args: str and **kwargs: Any must also have annotations.

2. Variable Declarations

• Strict Local Typing: Every local variable must be declared with a type.

  • x: int = 10 (Annotated assignment)
  • x: int followed by x = 10 (Declaration then assignment)

• Re-assignment: Once a variable is typed in the current scope, it can be re-assigned without repeating the annotation.

  x: int = 10
  x = 20  # Valid
  

3. Control Flow

• For Loops: Loop variables must be previously declared.

  i: int
  for i in range(10): ...
  

• With Statements: Context manager variables must be previously declared.

  f: TextIO
  with open("file.txt") as f: ...
  

• Match Case: Variable captures via as in match patterns must be pre-declared with a type annotation:

  val: int
  match data:
      case int() as val:
          print(val)
  

  Star patterns ([*rest]) follow the same rule:

  rest: list[int]
  match items:
      case [*rest]:
          print(rest)
  

  Type-only matches without binding (case int():) are always valid.

4. Global Constants

• Immutability: UPPERCASE variables are treated as constants and are immutable everywhere. Once assigned a value, any attempt to re-assign or shadow them will result in an error.
• Exemptions: UPPERCASE constants at the module level are exempt from strict typing by convention (e.g., MAX_SIZE = 100).

5. Prohibited Features

• Lambdas: Blocked because Python provides no syntax for annotating their parameters or return types.
• Untyped Unpacking: x, y = (1, 2) is blocked unless x and y were previously declared.

---

🔮 Future Roadmap

The project aims to become a complete "strict mode" for Python. Future plans include:

1. Module-Level Enforcement: Extend strict typing to all module-level variables (not just UPPERCASE constants).
2. Type-Safe Imports: Validate that imported names are used consistently with their types (integration with mypy or pyright stubs).
3. Strict Class Attributes: Enforce that all class attributes are declared in the class body with types.
4. Decorator Validation: Ensure decorators preserve type information and are themselves strictly typed.
5. Linter Integration: A dedicated VS Code / PyCharm extension to show nakprok errors in real-time.

---

📦 Installation & Usage

Installation

pip install -e .

Usage

# Check and run (default)
nakprok file.py

# Check types only (file or directory)
nakprok check file.py
nakprok check src/

# Explicit run
nakprok run file.py

Example: factorial.py

def factorial(n: int) -> int:
    if n <= 1:
        return 1
    return n * factorial(n - 1)

def main() -> None:
    result: int = factorial(5)
    print(f"5! = {result}")

if __name__ == "__main__":
    main()

$ nakprok factorial.py
5! = 120

---

🧪 Testing

We maintain a comprehensive test suite to ensure all rules are correctly enforced.

pytest tests/ -v