go-godash

An experimental generic functional library for transforming collections of data, inspired by packages such as Lodash.

This package aims to implement many of the methods from Lodash over Go arrays and Maps. All functions in this module are intended to work with native Go types (i.e. do not define their own containers/iterators), and should be type safe at compile-time.

The functions defined here are only loosely inspired by Lodash. Sometimes they deviate because of impedance between the two languages, or sometimes because I didn't bother to read the documentation properly.

Some categories of Lodash functions are completely omitted. e.g. there are plenty of libraries out there providing String manipulation tools, there's no need to create a new set in this library.

Caveat: I have very little experience with Go generics, so the API will evolve heavily. It may never hit version 1, depending on how collections evolve in Go once generics lands.

Example

Here's a contrived example of implementing and incrementAndSum function. The examples themselves are defined for a concrete type, but the underlying Sum and Map functions are generic across Numeric and all types respectively.

// A non-generic increment function for integers
func increment(x int) int {
	return x + 1
}

// A non-generic increment and sum function for arrays of integers
func incrementAndSum(arr []int) int {
	// Utilises generic Map and Sum functions
	result := Sum(Map(arr, increment))
}

// Returns 9
incrementAndSum([]int{1, 2, 3})

Implemented functions

Arrays:

• Map
• Reduce
• Filter
• Take
• TakeWhile
• Drop
• DropWhile
• Partition
• First (Head)
• Last
• Reversed
• Find
• FindPredicate
• Intersection
• Difference
• Union
• Chunk
• Flatten
• Without
• Any/Some
• All/Every
• FlatMap

Numeric functions:

• Sum
• Mean
• Mode (defined over all comparables, not just Numerics)

Not implemented

Some functions are unimplemented because they do not translate in a type-safe manner.

• FlattenDeep/FlattenDepth - It's not immediately obvious to me how to achieve this in a type safe way, so these functions are omitted
• FromPairs - Assumes an input array of untyped data which cannot be achieved at compile time

Limitations

Complex types are not supported on operations for Numeric types (Sum, Mean etc.).
I was able to find solutions to zero-initialise complex types for Sum, but this technique does not translate to Mean (and probably other generic Numeric operations) where we need to divide an arbitrary complex type (complex64 or 128) by the length of the array. There is no way to infer whether it should be divided by a float32 or float64 type.

See this section and this section of the generics proposal for more information.

gotip

This library uses generics which is not yet available in go. You'll need to use gotip to test this out.