A performant implementation of the SIEVE caching algorithm that implements Map properties and methods designed for Node.js 20+.
npm install @neophi/sieve-cache --save
import { SieveCache } from "@neophi/sieve-cache";
const sieveCache = new SieveCache<string, string>(2);
sieveCache.set("key", "value");
// "value"
sieveCache.get("key");
sieveCache.set("key2", "value2");
// triggers eviction via SIEVE algorithm
sieveCache.set("key3", "value3");
// false
sieveCache.has("key2");
The API exposes the JS Map properties and methods.
constructor(maxSize: number, options: Partial<SieveCacheOptions<K, V>> = {})
The maxSize
of the cache must be at least 1 and less than 4,294,967,295.
The existingSetBehavior
option changes the behavior of the SIEVE visited
state when updating a key that already exists in the Map. The default is ExistingSetBehavior.VISITED_SKIP
which means the existing visited state is not modified.
The evictHook
option is a function (key: K, value: V) => void
that can be configured to be called with the details of the evicted item after the cache has been updated.
import { SieveCache, ExistingSetBehavior } from "@neophi/sieve-cache";
const sieveCache = new SieveCache<string, string>(2, {
existingSetBehavior: ExistingSetBehavior.VISITED_TRUE,
evictHook: (key, value) => { console.log(`Evicted ${key} => ${value}`); }
});
Always benchmark your specific use case since many runtime, key type, and object type differences can impact the performance of an implementation. With that said this implementation compares favorably to established libraries such as lru-cache and mnemonist. For an extensive benchmark see cache-playground.
This implementation was inspired by Implementing an efficient LRU cache in JavaScript.
Everyone interacting with this project is expected to follow the code of conduct.
Released under the MIT License.