Response caching

Use case

Cache GET responses in memory to avoid redundant network calls within a time window.

Simple LRU cache recipe

import { createClient } from '@parcely/core'
import type { HttpResponse } from '@parcely/core'

const http = createClient({ baseURL: 'https://api.example.com' })

interface CacheEntry {
  response: HttpResponse<unknown>
  expiry: number
}

const cache = new Map<string, CacheEntry>()
const TTL = 30_000 // 30 seconds

http.interceptors.request.use((config) => {
  if ((config.method ?? 'GET').toUpperCase() !== 'GET') return config

  const key = `GET:${config.url}`
  const entry = cache.get(key)
  if (entry && entry.expiry > Date.now()) {
    // We will handle this in the response interceptor
    return { ...config, headers: { ...config.headers, 'x-cache': 'HIT' } }
  }
  cache.delete(key)
  return config
})

http.interceptors.response.use((response) => {
  if ((response.config.method ?? 'GET').toUpperCase() === 'GET') {
    const key = `GET:${response.config.url}`
    cache.set(key, { response, expiry: Date.now() + TTL })
  }
  return response
})

Notes

• This is a simple in-memory cache. For production, consider libraries like lru-cache.
• Cache invalidation is the hard part. Clear the cache on mutations or use TTLs.
• Consider the Map size growing unbounded. Implement eviction or use a bounded cache.
• For server-side caching, respect Cache-Control headers from the API.