Improve compute to handle arrays

[millsp]

🐞 Bug Report

Describe the bug

@millsp I have a question regarding the A.Compute<> type provided. It seems to be incorrectly computing arrays. (e.g. it's pulling them apart as if it's an object instead of keeping it as an array.)

Reproduce the bug

Expected behavior

Possible Solution

import { A, M } from 'ts-toolbelt';
import { Depth } from 'Object/_Internal';
type ComputeFlat<A extends any> = A extends M.BuiltInObject
  ? A
  : A extends Array<any>
  ? A extends Array<Record<string | number | symbol, any>>
    ? Array<
        {
          [K in keyof A[number]]: A[number][K];
        } & {}
      >
    : A
  : A extends ReadonlyArray<any>
  ? A extends ReadonlyArray<Record<string | number | symbol, any>>
    ? ReadonlyArray<
        {
          [K in keyof A[number]]: A[number][K];
        } & {}
      >
    : A
  : {
      [K in keyof A]: A[K];
    } & {};

type ComputeDeep<A extends any> = A extends M.BuiltInObject
  ? A
  : A extends Array<any>
  ? A extends Array<Record<string | number | symbol, any>>
    ? Array<
        {
          [K in keyof A[number]]: ComputeDeep<A[number][K]>;
        } & {}
      >
    : A
  : A extends ReadonlyArray<any>
  ? A extends ReadonlyArray<Record<string | number | symbol, any>>
    ? ReadonlyArray<
        {
          [K in keyof A[number]]: ComputeDeep<A[number][K]>;
        } & {}
      >
    : A
  : {
      [K in keyof A]: ComputeDeep<A[K]>;
    } & {};

export type Compute<A extends any, depth extends Depth = 'deep'> = {
  flat: ComputeFlat<A>;
  deep: ComputeDeep<A>;
}[depth];

type test = {
  a: ReadonlyArray<{
    a: number;
    c: ReadonlyArray<{
      a: number;
    }>;
  }>;
  b: { a: number };
} & {
  a: ReadonlyArray<{
    d: Array<{
      b: number;
    }>;
    c: ReadonlyArray<{
      b: number;
    }>;
  }>;
  c: { c: number };
};

Screenshots

Additional context

@wesselvdv

[millsp]

I found that the performance is quite OK. So I'll be adding these changes in the next release of ts-toolbelt. Thanks for your contribution @wesselvdv

[wesselvdv]

This can be closed I believe @millsp ?

[millsp]

In a few minutes, yes :)