joesdu / objectcloner

Insanely fast and capable Deep Clone implementation for .NET based on Expression Trees

ObjectCloner

ObjectCloner aims to be the fastest, yet most capable deep (or shallow) clone solution for .NET objects. It utilizes compiled Expression Trees under the hood, making it just as fast as (or faster than) a hand-written DeepClone method.

Features

• Native speed using code generation
• Works on any type
• Does not require a parameterless constructor or custom attribute
• Clones properties
• Clones read-only fields
• Preserves reference equality in the tree¹
• Handles circular dependencies
• Handles polymorphism²
• MIT license
• 100% test coverage

¹ For example if you put the same object into an array 3 times, you will get back a different list which also contains the same single clone object 3 times.

² Interface-typed fields (IEnumerable, etc.) will be cloned using their actual runtime types.

Installation

The package is available from NuGet as ObjectCloner and maybe ObjectCloner.Extensions (see Usage). Install via the package manager console: PM> Install-Package ObjectCloner the dotnet command line: dotnet add package ObjectCloner or using your IDE's visual NuGet tool.

Usage

The library exposes a static class with two generic methods, making the usage very straightforward.

var clone = ObjectCloner.DeepClone(original);

or

var clone = ObjectCloner.ShallowClone(original);

If you would rather use an extension method format and don't mind polluting your object type install the package ObjectCloner.Extensions as well and use it like so:

var clone = original.DeepClone();

Performance

In a benchmark cloning a custom class hierarchy ObjectCloner beats every traditional solution except custom written cloning code.

Method	Mean	Error	StdDev	Ratio	RatioSD
Custom Code	88.71 ns	4.242 ns	2.826 ns	1.00	0.00
ObjectCloner	520.09 ns	54.632 ns	36.136 ns	5.88	0.56
Reflection	2,648.47 ns	172.854 ns	112.711 ns	29.88	1.51
Newtonsoft.Json	9,315.16 ns	661.012 ns	437.218 ns	105.10	6.05
BinaryFormatter	22,743.60 ns	1,369.405 ns	1,121.254 ns	256.52	10.89

You can find the benchmarks in the samples folder of the solution and run them yourself.