refactor: switch to local custom LRUCache

[zonemeen]

No description provided.

[zhangyx1998]

Will it help if LRUCache works like a function wrapper? Just like python's @cache decorator provided by functools.

e.g.

export function cached<K extends any[], V>(
  fn: (...args: K) => V,
  {
    serialize = JSON.stringify,
    max
  }: {
    serialize: (args: K) => string
    max?: number
  }
): (...args: K) => V {
  const cache = new LRUCache<string, V>(max)
  return Object.assign(
    (...args: K) => {
      const k = serialize(args)
      if (!cache.has(k)) cache.set(k, fn(...args))
      return cache.get(k)!
    },
    { cache } // fn.cache?.clear()
  )
}

It should provide an non-intrusive way of implementing cache:

import { cached } from '...'
import { createFn } from '...'

fn = config.cache ? cached(createFn(...)) : createFn(...)

fn.cache?.clear()