Skip to content

peakchen90/decimal-eval

Repository files navigation

decimal-eval

A tiny, safe, fast JavaScript library for decimal arithmetic expressions.

Build Status Codecov npm BUNDLE SIZE GitHub

English | 简体中文

Features

  • ✌️ Automatically deal with the JavaScript decimal precision problem by bignumber.js, and supports big number
  • 🚀 Fast and tiny, only 28.7 KB minified and 11.2 KB gzipped (pure only 10.5 KB minified and 3.3 KB gzipped without bignumber.js)
  • ✍️ Easy to extend custom operator
  • 🖖 Supports expression scope variables

Get Started

Installation

# use npm
npm i -S decimal-eval

# or use yarn
yarn add decimal-eval

Usage

Supports the four arithmetic operations (+, -, *, /), and automatically deal with JavaScript decimal precision by bignumber.js, and supports big number.

import {evaluate} from 'decimal-eval';

evaluate('0.1 + 0.2') // '0.3'
evaluate('100 * (0.08 - 0.01)'); // '7'
evaluate('9007199254740992 + 1'); // '9007199254740993'
evaluate('1 + abc', { abc: 2 }); // '3'

In addition to the above operators, it also supports custom operator expansion, and supports unary operators and binary operators. See also reference document for the precedence of built-in operators.

import {evaluate, Parser} from 'decimal-eval';

// create binary operator `add`, the precedence is 13
const addOp = Parser.createBinaryOperator('add', 13, (left, right) => {
  return String(Number(left) + Number(right));
});

// create unary operator `sin`, the precedence is 16
const sinOp = Parser.createUnaryOperator('sin', 16, (value) => {
  return String(Math.sin(value));
});

// install custom operators
Parser.useOperator(addOp);
Parser.useOperator(sinOp);

// same as: `1 + Math.sin(-2)`
evaluate('1 add sin -2') // '0.09070257317431829'

API

evaluate(expression: string, scope?: Record<string, number>): number

Parse the arithmetic expression and then calculate the result. The name of scope has the following restrictions:

  • muse be start with a-z or A-Z
  • only includes a-zA-Z0-9 and _
import {evaluate} from 'decimal-eval';

evaluate('1 + 2'); // '3'
evaluate('1 + abc', { abc: 2 }); // '3'

Parser

new Parser(expression: string).parse(): Node | null

Parse arithmetic expressions.

import {Parser} from 'decimal-eval';

const node = new Parser('1 + 2').parse();

new Parser(expression: string).compile(): (scope: Record<string, number | string>) => string

Compile and cache the arithmetic expression.

import {Parser} from 'decimal-eval';

const evaluate = new Parser('1 + abc').compile();
evaluate({ abc: 2 }); // '3'
evaluate({ abc: 9 }); // '10'
evaluate({ def: 1 }); // throw error, the scope name `abc` is not initialized

Parser.createBinaryOperator(value: string, precedence: number, calc: (left: string, right: string) => string): Operator

Create a binary operator.

Parser.createUnaryOperator(value: string, precedence: number, calc: (value: string) => string): Operator

Create a unary operator.

Parser.useOperator(operator: Operator): void

Install an operator which created by the Parser.createBinaryOperator() or Parser.createUnaryOperator() method.

Parser.useAdapter(adapter)

Set custom calculation adapter methods for four arithmetic (+, -, *, /). bignumber.js is used by default.

Parser.useAdapter({
  '+': (left, right) => String(Number(left) + Number(right)),
  '-': (left, right) => String(Number(left) - Number(right)),
  '*': (left, right) => String(Number(left) * Number(right)),
  '/': (left, right) => String(Number(left) / Number(right))
})

Advanced

Use pure package (no dependencies)

When using a custom method to deal with the decimal precision problem, you can use a pure package, which can reduce the size by about 60%. It doesn't include bignumber.js.

import {evaluate, Parser} from 'decimal-eval/dist/pure';

// set custom calculation adapter
// Parser.useAdapter(adapter);

// Does not deal with precision by default
evaluate('0.1 + 0.2'); // '0.30000000000000004'

// not supports big number by default
evaluate('9007199254740992 + 1'); // '9007199254740992'

Re-export bignumber.js

Useful for deal JavaScript decimal precision problem without having to install bignumber.js again.

import {BigNumber} from 'decimal-eval';
const val = new BigNumber(0.1).plus(0.2);
console.log(String(val)); // '0.3'

Precedence of built-in operators

operator precedence
... + ... 13
... - ... 13
... * ... 14
... / ... 14
+ ... 16
- ... 16