** Let function overloading possible in rust **
With a single attribute on top of the function, you can overload the function with different parameters. Current implementation still has some flaws and todo items, so use at your own risk.
This library is based on some unstable features. To use this library, please put the following lines in crate root and the beginning of test files:
#![feature(fn_traits, unboxed_closures)]There are some features that cannot be achieved until now:
- unsafe function overloading
- const function overloading
- different privacy setting on function overloading (will pickup the privacy setting in first function and apply to all)
- function overloading inside traits (for limited cases)
simple one:
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
#[overload]
pub fn xdd() -> i32 {
5_i32
}
#[overload]
pub fn xdd(number: i32) -> i32 {
number * 3
}
#[overload]
pub fn xdd(number: u8) {
println!("{}", number);
}
#[overload]
pub unsafe fn xdd(number: &u64) -> u64 {
let n = number as *const u64;
*n * 4
}
assert_eq!(xdd(3_i32), 9_i32);
let c: &u64 = &6_u64;
assert_eq!(xdd(c), 24_u64); // unsafe function is not supported.
assert_eq!(xdd(), 5_i32);with generic and custom type:
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
use std::ops::MulAssign;
use std::fmt::Debug;
#[overload]
pub fn xdd<T: Copy + Debug + MulAssign<i32>>(mut number: T) -> T
where
T: PartialEq,
{
println!("number {:?}", number);
number *= 3_i32;
number
}
struct ABC;
impl Debug for ABC {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
write!(f, "xdddd")
}
}
#[overload]
pub fn xdd(number: ABC) -> () {
println!("number {:?}", number);
}
let aa = 123;
assert_eq!(xdd(aa), 369);
assert_eq!(xdd(ABC), ());async / await:
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
#[overload]
pub async fn xdd(number: i32) -> i32 {
number + 3
}
#[overload]
pub async fn xdd(number: i64) -> i64 {
number + 4
}
assert_eq!(futures::executor::block_on(xdd(3_i32)), 6);
assert_eq!(futures::executor::block_on(xdd(3_i64)), 7);type conflict might happen if generic overlaps with the definition of implemented types:
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
use std::ops::Mul;
use std::fmt::Debug;
#[overload]
pub fn xdd(number: i32) -> i32 {
number * 2
}
#[overload]
pub fn xdd<T: Copy + Debug + Mul<i32>>(number: T) -> T {
number * 3_i32
}for trait methods (notice that trait for overload must inherit Sized):
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
#[overload]
trait Xdd: Sized {
fn new(input: i32) -> Self;
fn new(input :u32) -> Self;
}
struct Haha {
a: u32,
b: i32,
}
#[overload]
impl Xdd for Haha {
fn new(b: i32) -> Self {
Self {
a: 1,
b,
}
}
fn new(a: u32) -> Self {
Self {
a,
b: 2,
}
}
}
let haha = Haha::new(12_i32);
assert_eq!(haha.a, 1_u32);
assert_eq!(haha.b, 12_i32);
let haha = Haha::new(9_u32);
assert_eq!(haha.a, 9_u32);
assert_eq!(haha.b, 2_i32);dynamic trait object implementation:
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
trait Xdd: 'static {}
#[overload]
impl dyn Xdd {
fn abc(&self, c: i32) -> i32 {
c
}
fn abc(&self, c: u32) -> u32 {
c
}
}
impl<T: 'static + ?Sized> Xdd for T {}
let obj = (&3_i32 as &(dyn Xdd));
// here we lost the syntax sugar of calling obj.abc(c) directly
assert_eq!(Xdd::abc(obj, 3_i32), 3_i32);
assert_eq!(Xdd::abc(obj, 3_u32), 3_u32);trait with generics:
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
#[overload]
trait Xdd<T: Sized>: Sized {
type J: Into<i32>;
fn new(input: i32) -> T where T: Debug;
// trait function with default implementation
fn new(input: u32) -> String {
"please use new(input1, input2) instead".to_string()
}
fn new<I: Into<u32>>(input1: I, input2: Self::J) -> T where T: Debug;
}
#[derive(Debug)]
struct Haha {
a: u32,
b: i32,
}
#[overload]
impl Xdd<Haha> for Haha {
type J = i32;
fn new(b: i32) -> Haha {
Self {
a: 1,
b,
}
}
fn new<I: Into<u32>>(a: I, b: Self::J) -> Haha {
Self {
a: a.into(),
b,
}
}
}
let haha = Haha::new(12_i32);
assert_eq!(haha.a, 1_u32);
assert_eq!(haha.b, 12_i32);
let haha = Haha::new(9_u32, 2_i32);
assert_eq!(haha.a, 9_u32);
assert_eq!(haha.b, 2_i32);
assert_eq!(Haha::new(3_u32), "please use new(input1, input2) instead".to_string());non-trait impl:
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
#[derive(Debug)]
pub struct Haha {
a: u32,
b: i32,
}
#[overload]
impl Haha {
pub fn new(b: i32) -> Self {
Self {
a: 1,
b,
}
}
pub fn new(a: u32) -> Self {
Self {
a,
b: 2,
}
}
// self in function parameter will be converted to associated type
// that means no more syntax sugar like haha.normal()
pub fn normal(&self) -> String {
format!("{:?}", self)
}
pub fn normal(&self, prefix: &str) -> String {
format!("{} {:?}", prefix, self)
}
// function without overloading is not influenced
pub fn display(&self) -> String {
format!("{:?}", self)
}
}
let haha = Haha::new(12_i32);
assert_eq!(haha.a, 1_u32);
assert_eq!(haha.b, 12_i32);
let haha = Haha::new(9_u32);
assert_eq!(haha.a, 9_u32);
assert_eq!(haha.b, 2_i32);
assert_eq!(Haha::normal(&haha), "Haha { a: 9, b: 2 }");
assert_eq!(Haha::normal(&haha, "abc"), "abc Haha { a: 9, b: 2 }");
assert_eq!(haha.display(), "Haha { a: 9, b: 2 }");default attribute:
in functions(not yet for traits), now you could decorate parameters with default values.
the reason why we don't support parameter = value syntax directly is that in derived
TokenStream, values inside will go through compiler first for syntax check.
The default value syntax will be treated as a compile error and forbid us from parsing
and generating valid Tokens.
example:
#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
#[overload]
fn xdd(#[default(= 5_i32)] a: i32, #[default(= 32_u64)] b: u64) -> u64 { b - (a as u64) }
assert_eq!(xdd(3_i32), 29_u64);
assert_eq!(xdd(4_i32, 4_u64), 0);
assert_eq!(xdd(), 27_u64);
// compile error: expect i32 found u64
// assert_eq!(xdd(6_u64), 1_u64);#![feature(fn_traits, unboxed_closures)]
use overloadf::*;
#[overload]
fn xdd(#[default(= 5_i32)] a: i32, b: u8, #[default(= 32_u64)] c: u64) -> u64 { c + (b as u64) - (a as u64) }
assert_eq!(xdd(4_i32, 7_u8), 35_u64);
assert_eq!(xdd(3_u8), 30_u64);Licensed under
- MIT license (LICENSE-MIT or https://opensource.org/licences/MIT)
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, shall be licensed as above, without any additional terms or conditions.