Financial Ops

This crate provides a set of operations for working with financial data, more specifically, avoiding the usage of floating point types.

Usage

use financial_ops::CheckedDecimalOperations;

fn test_add_decimals() {
    let a: u64 = 1_0000;
    let a_decimals = 4;
    let b: u64 = 2_00;
    let b_decimals = 2;

    let (result, decimals) = a.add_decimals_checked(b, a_decimals, b_decimals)?;
    assert_eq!(result, 3_0000);
    assert_eq!(decimals, 4);

    let a: u32 = 123_45;
    let a_decimals = 2;
    let b: u32 = 0_45;
    let b_decimals = 2;

    let (result, decimals) = a.add_decimals_checked(b, a_decimals, b_decimals)?;
    assert_eq!(result, 123_90);
    assert_eq!(decimals, 2);
}

Very useful when dealing with money or blockchain transactions.

Supported operations

Checked

This set of operations will return an Result with the result and the number of decimals, if the operation is successful. If the operation is not successful, it will return a DecimalOperationError.

use financial_ops::CheckedDecimalOperations;

• add_decimals_checked
• sub_decimals_checked
• mul_decimals_checked
• div_decimals_checked
• rem_decimals_checked

Each of these has an _or variant that lets you choose the error type (any value — your own enum, or even a &str — no external crate required), mirroring the @ syntax of the checked! macro:

use financial_ops::CheckedDecimalOperations;

#[derive(Debug, PartialEq)]
enum MyError { Math }

let result = u32::MAX.add_decimals_checked_or(1, 0, 0, MyError::Math);
assert_eq!(result, Err(MyError::Math));

// The error can also just be a string:
let r: Result<(u64, u32), &str> = 10u64.divide_decimals_checked_or(0, 2, 2, "division by zero");
assert_eq!(r, Err("division by zero"));

• add_decimals_checked_or
• sub_decimals_checked_or
• multiply_decimals_checked_or
• divide_decimals_checked_or
• rem_decimals_checked_or

Unchecked

This set of operations will return the result and the number of decimals, without any checks, carrying the underlying operation way of handling overflows and underflows.

use financial_ops::DecimalOperations;

• add_decimals
• sub_decimals
• mul_decimals
• div_decimals
• rem_decimals

The checked! macro

The checked! macro rewrites a normal arithmetic expression into a chain of checked operations, recursively, while respecting operator precedence and parentheses. Without an error it evaluates to an Option; with a trailing @ <error> it evaluates to a Result.

use financial_ops::checked;

// Respects precedence: this is `a + (b * c)`, fully checked.
let value: Option<u64> = checked! { 2u64 + 3 * 4 };
assert_eq!(value, Some(14));

// Overflow short-circuits to `None`.
assert_eq!(checked! { u8::MAX + 1u8 }, None);

// With `@ <error>` you get a `Result<T, E>`, where `E` is simply the type of
// the expression you pass. A string literal makes `E = &str` — no `anyhow`,
// no external crate, no trait bounds.
let ok: Result<u64, &str> = checked! { 2u64 + 2 @ "overflow" };
assert_eq!(ok, Ok(4));

let err: Result<u8, &str> = checked! { u8::MAX + 1u8 @ "overflow" };
assert_eq!(err, Err("overflow"));

Because @ <error> expands to .ok_or(<error>), the error can be anything: a &str, your own enum variant, etc. To use ? on the result, the surrounding function's error type just needs to be the same type (or From it).

Supported operators: +, -, *, /, % (mapped to checked_add, checked_sub, checked_mul, checked_div, checked_rem). Each operand is evaluated exactly once, in left-to-right order.