rtlopez/decimal

An object oriented immutable arbitrary-precision arithmetic library for PHP

1.0.1 2014-12-16 20:30 UTC

README

An object oriented arbitrary precision arithmetic class for PHP (>5.3.3).

Installation

You can use Composer

To add this package as a local, per-project dependency to your project, simply add a dependency on rtlopez/decimal to your project's composer.json file. Here is a minimal example of a composer.json file that just defines a dependency on Decimal:

{
  "require": {
    "rtlopez/decimal": "dev-master"
  }
}

Features

The Decimal class can be used for an arbitrary precision calculation. Main features are:

  • Immutable result of each operation,
  • The same result precision,
  • Safe for finance calculations,
  • Fluent interface,
  • Many unit test (more than 1100 tests and 3700 assertions),
  • Easy to use and install,

Why should I use it?

The Decimal solve several inconveniences known in other solutions.

  • Float is only an aproximation and have limited precision,
  • Fixed-point have limited size (64-bits on 64-bit system),
  • Pure bcmath cannot correctly round, only truncate,
  • GMP is not object-oriented

Usage

Creating objects with simple factory

use RtLopez\Decimal;

$n1 = Decimal::create(123, 2);
$n2 = Decimal::create(123.45, 2); // not recommended
$n3 = Decimal::create('123.45', 2);
$n4 = Decimal::create($n1, 2);

Printing and formatting numbers

$number = Decimal::create('12345.671', 2);

echo (string)$number;
// 12345.67

echo $number->format(1, ',', ' ');
// 12 345,7

echo $number->truncate()->format(null, ',', ' ', false);
// 12 345,00

Arithmetics

$n1 = Decimal::create('12345.671', 3);
$n2 = Decimal::create('11111.111', 3);
$n3 = $n1->add($n2)->mul(-1);
echo $n3; // -23456.782

// fluid interface
$n4 = $n3->pow(2)->sqrt()->mul(-2)->div(2)->abs();

Immutable

$n1 = Decimal::create('12345.671', 3);
$n1->add(2);
...
echo $n1;
// 12345.671

Comparisions

$n1 = Decimal::create('12345.671', 3);
$n2 = Decimal::create('11111.111', 3);
$n1->gt($n2);        // true
$n1->eq(0);          // false
$n2->le(100);        // false

Rounding

$n = Decimal::create('12345.671', 3);
echo $n->round(2);   // 12345.67
echo $n->round(1);   // 12345.7
echo $n->truncate(); // 12345
echo $n->ceil(1);    // 12345.6
echo $n->floor();    // 12346

Implementations

There are exists three internal implementations of this library.

  • RtLopez\DecimalBCMath: (default)based on bcmath library
  • RtLopez\DecimalFixed: based on scaled integers
  • RtLopez\DecimalFloat: based on floats

Only BCMath implementation works correctly in wide spectrum of numbers. All others can fail in many border cases. They was implemented only as proof of concept and you can using it only for your own risk.

The default implementation or precision can be changed by functions:

Decimal::setDefaultImplementation('RtLopez\DecimalBCMath');
Decimal::setDefaultPrecision(6);

Licence

This library is distributed under MIT Licence.

Issue reporting or feature request

Coming soon.

TODO

  • GMP implementation

Donation

BTC: 1Lopez7yPtbyjfLGe892JfheDFJMMt43tW LTC: LV3G3sJxz9AYpDMYUp8e1LCmerFYxVY3ak