davidbludlow/rfdc

Really Fast Deep Clone

rfdc

Really Fast Deep Clone

Usage

const clone = require('rfdc')()
clone({a: 1, b: {c: 2}}) // => {a: 1, b: {c: 2}}

API

require('rfdc')(opts = { proto: false, circles: false }) => clone(obj) => obj2

proto option

Copy prototype properties as well as own properties into the new object.

It's marginally faster to allow enumerable properties on the prototype to be copied into the cloned object (not onto it's prototype, directly onto the object).

To explain by way of code:

require('rfdc')({ proto: false })(Object.create({a: 1})) // => {}
require('rfdc')({ proto: true })(Object.create({a: 1})) // => {a: 1}

Setting proto to true will provide an additional 2% performance boost.

circles option

Keeping track of circular references will slow down performance with an additional 25% overhead. Even if an object doesn't have any circular references, the tracking overhead is the cost. By default if an object with a circular reference is passed to rfdc, it will throw (similar to how JSON.stringify
would throw).

Use the circles option to detect and preserve circular references in the object. If performance is important, try removing the circular reference from the object (set to undefined) and then add it back manually after cloning instead of using this option.

default import

It is also possible to directly import the clone function with all options set to their default:

const clone = require("rfdc/default")
clone({a: 1, b: {c: 2}}) // => {a: 1, b: {c: 2}}

Types

rfdc clones all JSON types:

• Object
• Array
• Number
• String
• null

With additional support for:

• Date (copied)
• undefined (copied)
• Buffer (copied)
• TypedArray (copied)
• Map (copied)
• Set (copied)
• Function (referenced)
• AsyncFunction (referenced)
• GeneratorFunction (referenced)
• arguments (copied to a normal object)

All other types have output values that match the output of JSON.parse(JSON.stringify(o)).

For instance:

const rfdc = require('rfdc')()
const err = Error()
err.code = 1
JSON.parse(JSON.stringify(e)) // {code: 1}
rfdc(e) // {code: 1}

JSON.parse(JSON.stringify({rx: /foo/})) // {rx: {}}
rfdc({rx: /foo/}) // {rx: {}}

Benchmarks

npm run bench

benchDeepCopy*100: 550.406ms
benchLodashCloneDeep*100: 1.479s
benchCloneDeep*100: 761.46ms
benchFastCopy*100: 800.38ms
benchFastestJsonCopy*100: 369.979ms // See note below
benchJsondiffpatchClone*100: 285.914ms // See note below
benchPlainObjectClone*100: 544ms
benchNanoCopy*100: 640.636ms
benchRamdaClone*100: 2.936s
benchJsonParseJsonStringify*100: 2.239s // JSON.parse(JSON.stringify(obj))
benchRfdc*100: 403.151ms
benchRfdcProto*100: 416.896ms
benchRfdcCircles*100: 453.706ms
benchRfdcCirclesProto*100: 404.127ms

It is true that jsondiffpatch.clone() from jsondiffpatch is the fastest in that benchmark run, but it cannot handle as many situations as rfdc can. Also, this and this benchmark result suggest that rfdc is faster in Firefox and Chrome at this different benchmark.

It is true that fastest-json-copy is faster than rfdc in this particular benchmark, but we have seen that, with different input, the opposite is true. Also, fastest-json-copy has such huge limitations that it is rarely useful. For example, it treats things like Date and Map instances the same as empty {}. It can't handle circular references.

plain-object-clone is also really limited in capability.

Tests

npm test

169 passing (342.514ms)

Coverage

npm run cov

----------|----------|----------|----------|----------|-------------------|
File      |  % Stmts | % Branch |  % Funcs |  % Lines | Uncovered Line #s |
----------|----------|----------|----------|----------|-------------------|
All files |      100 |      100 |      100 |      100 |                   |
 index.js |      100 |      100 |      100 |      100 |                   |
----------|----------|----------|----------|----------|-------------------|

__proto__ own property copying

rfdc works the same way as Object.assign when it comes to copying ['__proto__'] (e.g. when an object has an own property key called 'proto'). It results in the target object prototype object being set per the value of the ['__proto__'] own property.

For detailed write-up on how a way to handle this security-wise see https://www.fastify.io/docs/latest/Guides/Prototype-Poisoning/.

License

MIT