C FFI & Python Bridge

C Foreign Function Interface

Thagore provides direct access to C functions through extern func declarations. Any function available in a C shared library or the system’s C standard library can be called.

Basic C FFI

# Declare C standard library functions
extern func strlen(s: String) -> i32
extern func puts(text: String) -> i32
extern func printf(fmt: String, value: i32) -> i32

func main() -> i32:
    let msg = "Hello from FFI!"
    puts(msg)
    printf("String length: %d\n", strlen(msg))
    return 0

Math Library FFI

Call C math functions directly:

# Float math functions
extern func sqrtf(x: f32) -> f32
extern func powf(base: f32, exp: f32) -> f32
extern func sinf(x: f32) -> f32
extern func cosf(x: f32) -> f32
extern func fabsf(x: f32) -> f32

# Double math functions
extern func sqrt(x: f64) -> f64
extern func pow(base: f64, exp: f64) -> f64

func calc_hypotenuse(a: f32, b: f32) -> f32:
    return sqrtf(a * a + b * b)

func main() -> i32:
    let h = calc_hypotenuse(3.0, 4.0)
    print(h)    # Output: 5.0
    return 0

Memory Management FFI

extern func malloc(size: i32) -> ptr
extern func free(p: ptr) -> void
extern func memcpy(dst: ptr, src: ptr, size: i32) -> ptr

func allocate_buffer(size: i32) -> ptr:
    let buffer = malloc(size)
    return buffer

func main() -> i32:
    let buf = allocate_buffer(1024)
    # ... use buffer ...
    free(buf)
    return 0

FFI Type Mapping Reference

Thagore Type	C Type	Notes
`i32`	`int` / `int32_t`	32-bit signed integer
`f32`	`float`	IEEE 754 single precision
`f64`	`double`	IEEE 754 double precision
`String`	`const char*`	Null-terminated C string
`ptr`	`void*`	Generic pointer
`void`	`void`	No return value

Python Bridge

Thagore can interface with Python through the runtime’s Python bridge. This enables calling Python libraries (including NumPy, PyTorch, etc.) from Thagore code.

Python Bridge Functions

Function	Purpose
`__thg_py_initialize()`	Initialize the Python interpreter
`__thg_py_finalize()`	Shutdown the Python interpreter
`__thg_py_import(name)`	Import a Python module
`__thg_py_getattr(obj, name)`	Get an attribute from a Python object
`__thg_py_call0(callable)`	Call a Python callable with zero arguments
`__thg_py_call_2(func, a1, a2)`	Call with two arguments
`__thg_py_from_i32(val)`	Convert Thagore i32 to Python int
`__thg_py_print_obj(obj)`	Print a Python object to stdout

Basic Python Usage

extern func __thg_py_initialize() -> void
extern func __thg_py_finalize() -> void
extern func __thg_py_import(name: String) -> ptr
extern func __thg_py_print_obj(obj: ptr) -> void

func main() -> i32:
    print("Importing Python math module...")
    __thg_py_initialize()

    let math_mod = __thg_py_import("math")
    __thg_py_print_obj(math_mod)

    __thg_py_finalize()
    return 0

PyTorch Integration

Call PyTorch from Thagore for tensor operations:

extern func __thg_py_initialize() -> void
extern func __thg_py_import(name: String) -> ptr
extern func __thg_py_getattr(obj: ptr, name: String) -> ptr
extern func __thg_py_print_obj(obj: ptr) -> void
extern func __thg_py_from_i32(val: i32) -> ptr
extern func __thg_py_call_2(func_obj: ptr, a1: ptr, a2: ptr) -> ptr

func main() -> i32:
    __thg_py_initialize()

    print("Loading PyTorch...")
    let torch = __thg_py_import("torch")

    # Get torch.rand function
    let rand_func = __thg_py_getattr(torch, "rand")

    # Create dimensions
    let dim1 = __thg_py_from_i32(3)
    let dim2 = __thg_py_from_i32(3)

    # Generate a 3x3 random tensor
    print("Generating Tensor(3, 3)...")
    let tensor = __thg_py_call_2(rand_func, dim1, dim2)

    __thg_py_print_obj(tensor)
    return 0

Requirements

To use the Python bridge:

Python 3.x must be installed and on PATH
Link against the Python shared library
Required Python packages must be installed (pip install torch for PyTorch)