uzyn / bigdenary Goto Github PK

View Code? Open in Web Editor NEW

11.0 3.0 2.0 77 KB

🦕🧮 Arbitrary-length decimal library, implemented with ES2020's native BigInt.

Home Page: https://deno.land/x/bigdenary

License: MIT License

TypeScript 100.00%

bignum bignumber bigint decimal decimals bigdecimals javascript typescript deno es6

bigdenary's Introduction

BigDenary

Arbitrary-length decimal implementation using JavaScript's native BigInt with no dependencies.

Supported on Node >= 10.4 and Deno.
Check caniuse.com for browser support.

Features

Deno module first. Soon to be available as ES Module (ESM) and CommonJS (Node) module.
Compute methods are largely implemented through native BigInt, without much string manipulation required.
Standalone & lightweight. Zero dependencies.
Intuitive data structure – base amount and decimal places, similar to that of cryptocurrency esp. Bitcoin.
API is similar with the popular BigNumber libraries such as bignumber.js, big.js, decimal.js. Not all methods are supported, yet.

Usage

import { BigDenary } from "https://deno.land/x/bigdenary/mod.ts";

const bd = new BigDenary("123.4512");
const sum = bd.add(56.1e2);

console.log(sum.toString()); // 5733.4512
console.log(sum); // BigDenary { base: 57334512n, _decimals: 4 }

API is largely inspired by and attempts to be compatible with decimal.js-light.

Available API

Core

constructor(): supports type BDNumberInput = number | string | bigint | BigDenary | BigDenaryRaw.
toString(): Returns string representation
valueOf(): Returns number approximation
toFixed(digits?): Returns string representation to the number of digits to appear decimal point.

Operations

plus() or add()
minus() or sub()
multipliedBy() or mul()
dividedBy() or div()
negated() or neg()
absoluteValue() or abs()

Comparisons

comparedTo() or cmp()
equals() or eq()
greaterThan() or gt()
greaterThanOrEqualTo() or gte()
lessThan() or lt()
lessThanOrEqualTo() or lte()

Develop and running of tests

Install Deno
Run unit tests

deno test

Notes

JavaScript native decimal support is currently being proposed (Stage 1) to ECMA.

License

MIT · U-Zyn Chua

Contributions are welcomed.

bigdenary's People

Contributors

Stargazers

Watchers

Forkers

77it

bigdenary's Issues

Rounding issues and unsafe operations

I think this line in the implementation is not safe it could cause rounding issues:

https://github.com/uzyn/bigdenary/blob/master/mod.ts#L31C7-L31C60

For example the following number will fail: 0.0023537

You can see the problem with this if you try the following calculation:

// 23536.999999999996 instead of 23537
0.0023537 * Math.pow(10, 7)

The response will be a float instead of an integer so it will fail with BigInt.

This module name has been reserved for a repository

Please take a look at https://deno.land/x/bigdenary

(100 / 3) * 3 != 100 and (100 / 3) + (100 / 3) + (100 / 3) != 100

Hi,
I found an issue I want to highlight, curious about the position of the maintainer of this library on it.

I found that for bigdenary (100 / 3) * 3 != 100 and (100 / 3) + (100 / 3) + (100 / 3) != 100

the failing test are the following

Deno.test("BigDenary: (100 / 3) + (100 / 3) + (100 / 3) == 100", () => {
	const b_d100 = new BigDenary(100);
	const b_d33 = b_d100.div(3);
	const b_d100sum = new BigDenary(0).add(b_d33).add(b_d33).add(b_d33);
	assert(b_d100.eq(b_d100sum));
});

Deno.test("BigDenary: (100 / 3) * 3 == 100", () => {
	const b_d100 = new BigDenary(100);
	const b_d100mul = new BigDenary(100).div(3).mul(3);
	assert(b_d100.eq(b_d100mul));
});

The tests succeed when a BigDenary is converted to number with valueOf(), because valueOf() internally convert to string and then uses parseFloat() for the conversion to number, and because the converted string has more than 14 decimals, the returned number is rounded.

I decided then to test also decimal.js and number object, and I found that: number process them correctly (failing on others) while decimal.js fails at them as this library.

Test file attached.

For completeness, I also tested C #, and using the decimal type the problem is not present:

	decimal d100 = 100;
	decimal d33 = d100/ 3;
	decimal d100sum = d33 + d33 + d33;
	decimal d100mul = d33 * 3;
	Console.WriteLine(d33);
	Console.WriteLine(d100 == d100sum);
	Console.WriteLine(d100sum);
	Console.WriteLine(d100 == d100mul);
	Console.WriteLine(d100mul);

test.txt