lru-cache

A cache object that deletes the least-recently-used items.

Specify a max number of the most recently used items that you want to keep, and this cache will keep that many of the most recently accessed items.

This is not primarily a TTL cache, and does not make strong TTL guarantees. There is no preemptive pruning of expired items by default, but you may set a TTL on the cache or on a single set. If you do so, it will treat expired items as missing, and delete them when fetched. If you are more interested in TTL caching than LRU caching, check out @isaacs/ttlcache.

As of version 7, this is one of the most performant LRU implementations available in JavaScript, and supports a wide diversity of use cases. However, note that using some of the features will necessarily impact performance, by causing the cache to have to do more work. See the "Performance" section below.

Installation

npm install lru-cache --save

Usage

const LRU = require('lru-cache')

// At least one of 'max', 'ttl', or 'maxSize' is required, to prevent
// unsafe unbounded storage.
//
// In most cases, it's best to specify a max for performance, so all
// the required memory allocation is done up-front.
//
// All the other options are optional, see the sections below for
// documentation on what each one does.  Most of them can be
// overridden for specific items in get()/set()
const options = {
  max: 500,

  // for use with tracking overall storage size
  maxSize: 5000,
  sizeCalculation: (value, key) => {
    return 1
  },

  // for use when you need to clean up something when objects
  // are evicted from the cache
  dispose: (value, key) => {
    freeFromMemoryOrWhatever(value)
  },

  // how long to live in ms
  ttl: 1000 * 60 * 5,

  // return stale items before removing from cache?
  allowStale: false,

  updateAgeOnGet: false,
  updateAgeOnHas: false,

  // async method to use for cache.fetch(), for
  // stale-while-revalidate type of behavior
  fetch: async (key, staleValue, { options, signal }) => {}
}

const cache = new LRU(options)

cache.set("key", "value")
cache.get("key") // "value"

// non-string keys ARE fully supported
// but note that it must be THE SAME object, not
// just a JSON-equivalent object.
var someObject = { a: 1 }
cache.set(someObject, 'a value')
// Object keys are not toString()-ed
cache.set('[object Object]', 'a different value')
assert.equal(cache.get(someObject), 'a value')
// A similar object with same keys/values won't work,
// because it's a different object identity
assert.equal(cache.get({ a: 1 }), undefined)

cache.clear()    // empty the cache

If you put more stuff in it, then items will fall out.

Options

max

The maximum number (or size) of items that remain in the cache (assuming no TTL pruning or explicit deletions). Note that fewer items may be stored if size calculation is used, and maxSize is exceeded. This must be a positive finite intger.

At least one of max, maxSize, or TTL is required. This must be a positive integer if set.

It is strongly recommended to set a max to prevent unbounded growth of the cache. See "Storage Bounds Safety" below.

maxSize

Set to a positive integer to track the sizes of items added to the cache, and automatically evict items in order to stay below this size. Note that this may result in fewer than max items being stored.

Optional, must be a positive integer if provided. Required if other size tracking features are used.

At least one of max, maxSize, or TTL is required. This must be a positive integer if set.

Even if size tracking is enabled, it is strongly recommended to set a max to prevent unbounded growth of the cache. See "Storage Bounds Safety" below.

sizeCalculation

Function used to calculate the size of stored items. If you're storing strings or buffers, then you probably want to do something like n => n.length. The item is passed as the first argument, and the key is passed as the second argument.

This may be overridden by passing an options object to cache.set().

Requires maxSize to be set.

Deprecated alias: length

fetchMethod

Function that is used to make background asynchronous fetches. Called with fetchMethod(key, staleValue, { signal, options }). May return a Promise.

If fetchMethod is not provided, then cache.fetch(key) is equivalent to Promise.resolve(cache.get(key)).

The signal object is an AbortSignal if that's available in the global object, otherwise it's a pretty close polyfill.

If at any time, signal.aborted is set to true, or if the signal.onabort method is called, or if it emits an 'abort' event which you can listen to with addEventListener, then that means that the fetch should be abandoned. This may be passed along to async functions aware of AbortController/AbortSignal behavior.

The options object is a union of the options that may be provided to set() and get(). If they are modified, then that will result in modifying the settings to cache.set() when the value is resolved. For example, a DNS cache may update the TTL based on the value returned from a remote DNS server by changing options.ttl in the fetchMethod.

noDeleteOnFetchRejection

If a fetchMethod throws an error or returns a rejected promise, then by default, any existing stale value will be removed from the cache.

If noDeleteOnFetchRejection is set to true, then this behavior is suppressed, and the stale value remains in the cache in the case of a rejected fetchMethod.

This is important in cases where a fetchMethod is only called as a background update while the stale value is returned, when allowStale is used.

This may be set in calls to fetch(), or defaulted on the constructor.

dispose

Function that is called on items when they are dropped from the cache, as this.dispose(value, key, reason).

This can be handy if you want to close file descriptors or do other cleanup tasks when items are no longer stored in the cache.

NOTE: It is called before the item has been fully removed from the cache, so if you want to put it right back in, you need to wait until the next tick. If you try to add it back in during the dispose() function call, it will break things in subtle and weird ways.

Unlike several other options, this may not be overridden by passing an option to set(), for performance reasons. If disposal functions may vary between cache entries, then the entire list must be scanned on every cache swap, even if no disposal function is in use.

The reason will be one of the following strings, corresponding to the reason for the item's deletion:

  • evict Item was evicted to make space for a new addition
  • set Item was overwritten by a new value
  • delete Item was removed by explicit cache.delete(key) or by calling cache.clear(), which deletes everything.

The dispose() method is not called for canceled calls to fetchMethod(). If you wish to handle evictions, overwrites, and deletes of in-flight asynchronous fetches, you must use the AbortSignal provided.

Optional, must be a function.

disposeAfter

The same as dispose, but called after the entry is completely removed and the cache is once again in a clean state.

It is safe to add an item right back into the cache at this point. However, note that it is very easy to inadvertently create infinite recursion in this way.

The disposeAfter() method is not called for canceled calls to fetchMethod(). If you wish to handle evictions, overwrites, and deletes of in-flight asynchronous fetches, you must use the AbortSignal provided.

noDisposeOnSet

Set to true to suppress calling the dispose() function if the entry key is still accessible within the cache.

This may be overridden by passing an options object to cache.set().

Boolean, default false. Only relevant if dispose or disposeAfter options are set.

ttl

Max time to live for items before they are considered stale. Note that stale items are NOT preemptively removed by default, and MAY live in the cache, contributing to its LRU max, long after they have expired.

Also, as this cache is optimized for LRU/MRU operations, some of the staleness/TTL checks will reduce performance, as they will incur overhead by deleting from Map objects rather than simply throwing old Map objects away.

This is not primarily a TTL cache, and does not make strong TTL guarantees. There is no pre-emptive pruning of expired items, but you may set a TTL on the cache, and it will treat expired items as missing when they are fetched, and delete them.

Optional, but must be a positive integer in ms if specified.

This may be overridden by passing an options object to cache.set().

At least one of max, maxSize, or TTL is required. This must be a positive integer if set.

Even if ttl tracking is enabled, it is strongly recommended to set a max to prevent unbounded growth of the cache. See "Storage Bounds Safety" below.

If ttl tracking is enabled, and max and maxSize are not set, and ttlAutopurge is not set, then a warning will be emitted cautioning about the potential for unbounded memory consumption.

Deprecated alias: maxAge

noUpdateTTL

Boolean flag to tell the cache to not update the TTL when setting a new value for an existing key (ie, when updating a value rather than inserting a new value). Note that the TTL value is always set (if provided) when adding a new entry into the cache.

This may be passed as an option to cache.set().

Boolean, default false.

ttlResolution

Minimum amount of time in ms in which to check for staleness. Defaults to 1, which means that the current time is checked at most once per millisecond.

Set to 0 to check the current time every time staleness is tested.

Note that setting this to a higher value will improve performance somewhat while using ttl tracking, albeit at the expense of keeping stale items around a bit longer than intended.

ttlAutopurge

Preemptively remove stale items from the cache.

Note that this may significantly degrade performance, especially if the cache is storing a large number of items. It is almost always best to just leave the stale items in the cache, and let them fall out as new items are added.

Note that this means that allowStale is a bit pointless, as stale items will be deleted almost as soon as they expire.

Use with caution!

Boolean, default false

allowStale

By default, if you set ttl, it'll only delete stale items from the cache when you get(key). That is, it's not preemptively pruning items.

If you set allowStale:true, it'll return the stale value as well as deleting it. If you don't set this, then it'll return undefined when you try to get a stale entry.

Note that when a stale entry is fetched, even if it is returned due to allowStale being set, it is removed from the cache immediately. You can immediately put it back in the cache if you wish, thus resetting the TTL.

This may be overridden by passing an options object to cache.get(). The cache.has() method will always return false for stale items.

Boolean, default false, only relevant if ttl is set.

Deprecated alias: stale

updateAgeOnGet

When using time-expiring entries with ttl, setting this to true will make each item's age reset to 0 whenever it is retrieved from cache with get(), causing it to not expire. (It can still fall out of cache based on recency of use, of course.)

This may be overridden by passing an options object to cache.get().

Boolean, default false, only relevant if ttl is set.

updateAgeOnHas

When using time-expiring entries with ttl, setting this to true will make each item's age reset to 0 whenever its presence in the cache is checked with has(), causing it to not expire. (It can still fall out of cache based on recency of use, of course.)

This may be overridden by passing an options object to cache.has().

Boolean, default false, only relevant if ttl is set.

API

new LRUCache(options)

Create a new LRUCache. All options are documented above, and are on the cache as public members.

cache.max, cache.maxSize, cache.allowStale,

cache.noDisposeOnSet, cache.sizeCalculation, cache.dispose, cache.maxSize, cache.ttl, cache.updateAgeOnGet, cache.updateAgeOnHas

All option names are exposed as public members on the cache object.

These are intended for read access only. Changing them during program operation can cause undefined behavior.

cache.size

The total number of items held in the cache at the current moment.

cache.calculatedSize

The total size of items in cache when using size tracking.

`set(key, value, [{ size, sizeCalculation, ttl,

noDisposeOnSet }])`

Add a value to the cache.

Optional options object may contain ttl and sizeCalculation as described above, which default to the settings on the cache object.

Options object my also include size, which will prevent calling the sizeCalculation function and just use the specified number if it is a positive integer, and noDisposeOnSet which will prevent calling a dispose function in the case of overwrites.

Will update the recency of the entry.

Returns the cache object.

get(key, { updateAgeOnGet, allowStale } = {}) => value

Return a value from the cache.

Will update the recency of the cache entry found.

If the key is not found, get() will return undefined. This can be confusing when setting values specifically to undefined, as in cache.set(key, undefined). Use cache.has() to determine whether a key is present in the cache at all.

`async fetch(key, { updateAgeOnGet, allowStale, size,

sizeCalculation, ttl, noDisposeOnSet } = {}) => Promise`

If the value is in the cache and not stale, then the returned Promise resolves to the value.

If not in the cache, or beyond its TTL staleness, then fetchMethod(key, staleValue, options) is called, and the value returned will be added to the cache once resolved.

If called with allowStale, and an asynchronous fetch is currently in progress to reload a stale value, then the former stale value will be returned.

Multiple fetches for the same key will only call fetchMethod a single time, and all will be resolved when the value is resolved, even if different options are used.

If fetchMethod is not specified, then this is effectively an alias for Promise.resolve(cache.get(key)).

When the fetch method resolves to a value, if the fetch has not been aborted due to deletion, eviction, or being overwritten, then it is added to the cache using the options provided.

peek(key, { allowStale } = {}) => value

Like get() but doesn't update recency or delete stale items.

Returns undefined if the item is stale, unless allowStale is set either on the cache or in the options object.

has(key, { updateAgeOnHas } = {}) => Boolean

Check if a key is in the cache, without updating the recency of use. Age is updated if updateAgeOnHas is set to true in either the options or the constructor.

Will return false if the item is stale, even though it is technically in the cache.

delete(key)

Deletes a key out of the cache.

Returns true if the key was deleted, false otherwise.

clear()

Clear the cache entirely, throwing away all values.

Deprecated alias: reset()

keys()

Return a generator yielding the keys in the cache, in order from most recently used to least recently used.

rkeys()

Return a generator yielding the keys in the cache, in order from least recently used to most recently used.

values()

Return a generator yielding the values in the cache, in order from most recently used to least recently used.

rvalues()

Return a generator yielding the values in the cache, in order from least recently used to most recently used.

entries()

Return a generator yielding [key, value] pairs, in order from most recently used to least recently used.

rentries()

Return a generator yielding [key, value] pairs, in order from least recently used to most recently used.

find(fn, [getOptions])

Find a value for which the supplied fn method returns a truthy value, similar to Array.find().

fn is called as fn(value, key, cache).

The optional getOptions are applied to the resulting get() of the item found.

dump()

Return an array of [key, entry] objects which can be passed to cache.load()

Note: this returns an actual array, not a generator, so it can be more easily passed around.

load(entries)

Reset the cache and load in the items in entries in the order listed. Note that the shape of the resulting cache may be different if the same options are not used in both caches.

purgeStale()

Delete any stale entries. Returns true if anything was removed, false otherwise.

Deprecated alias: prune

getRemainingTTL(key)

Return the number of ms left in the item's TTL. If item is not in cache, returns 0. Returns Infinity if item is in cache without a defined TTL.

forEach(fn, [thisp])

Call the fn function with each set of fn(value, key, cache) in the LRU cache, from most recent to least recently used.

Does not affect recency of use.

If thisp is provided, function will be called in the this-context of the provided object.

rforEach(fn, [thisp])

Same as cache.forEach(fn, thisp), but in order from least recently used to most recently used.

pop()

Evict the least recently used item, returning its value.

Returns undefined if cache is empty.

Internal Methods and Properties

In order to optimize performance as much as possible, "private" members and methods are exposed on the object as normal properties, rather than being accessed via Symbols, private members, or closure variables.

Do not use or rely on these. They will change or be removed without notice. They will cause undefined behavior if used inappropriately. There is no need or reason to ever call them directly.

This documentation is here so that it is especially clear that this not "undocumented" because someone forgot; it is documented, and the documentation is telling you not to do it.

Do not report bugs that stem from using these properties. They will be ignored.

  • initializeTTLTracking() Set up the cache for tracking TTLs
  • updateItemAge(index) Called when an item age is updated, by internal ID
  • setItemTTL(index) Called when an item ttl is updated, by internal ID
  • isStale(index) Called to check an item's staleness, by internal ID
  • initializeSizeTracking() Set up the cache for tracking item size. Called automatically when a size is specified.
  • removeItemSize(index) Updates the internal size calculation when an item is removed or modified, by internal ID
  • addItemSize(index) Updates the internal size calculation when an item is added or modified, by internal ID
  • indexes() An iterator over the non-stale internal IDs, from most recently to least recently used.
  • rindexes() An iterator over the non-stale internal IDs, from least recently to most recently used.
  • newIndex() Create a new internal ID, either reusing a deleted ID, evicting the least recently used ID, or walking to the end of the allotted space.
  • evict() Evict the least recently used internal ID, returning its ID. Does not do any bounds checking.
  • connect(p, n) Connect the p and n internal IDs in the linked list.
  • moveToTail(index) Move the specified internal ID to the most recently used position.
  • keyMap Map of keys to internal IDs
  • keyList List of keys by internal ID
  • valList List of values by internal ID
  • sizes List of calculated sizes by internal ID
  • ttls List of TTL values by internal ID
  • starts List of start time values by internal ID
  • next Array of "next" pointers by internal ID
  • prev Array of "previous" pointers by internal ID
  • head Internal ID of least recently used item
  • tail Internal ID of most recently used item
  • free Stack of deleted internal IDs

Storage Bounds Safety

This implementation aims to be as flexible as possible, within the limits of safe memory consumption and optimal performance.

At initial object creation, storage is allocated for max items. If max is set to zero, then some performance is lost, and item count is unbounded. Either maxSize or ttl must be set if max is not specified.

If maxSize is set, then this creates a safe limit on the maximum storage consumed, but without the performance benefits of pre-allocation. When maxSize is set, every item must provide a size, either via the sizeCalculation method provided to the constructor, or via a size or sizeCalculation option provided to cache.set(). The size of every item must be a positive integer.

If neither max nor maxSize are set, then ttl tracking must be enabled. Note that, even when tracking item ttl, items are not preemptively deleted when they become stale, unless ttlAutopurge is enabled. Instead, they are only purged the next time the key is requested. Thus, if ttlAutopurge, max, and maxSize are all not set, then the cache will potentially grow unbounded.

In this case, a warning is printed to standard error. Future versions may require the use of ttlAutopurge if max and maxSize are not specified.

If you truly wish to use a cache that is bound only by TTL expiration, consider using a Map object, and calling setTimeout to delete entries when they expire. It will perform much better than an LRU cache.

Here is an implementation you may use, under the same license as this package:

// a storage-unbounded ttl cache that is not an lru-cache
const cache = {
  data: new Map(),
  timers: new Map(),
  set: (k, v, ttl) => {
    if (cache.timers.has(k)) {
      clearTimeout(cache.timers.get(k))
    }
    cache.timers.set(k, setTimeout(() => cache.del(k), ttl))
    cache.data.set(k, v)
  },
  get: k => cache.data.get(k),
  has: k => cache.data.has(k),
  delete: k => {
    if (cache.timers.has(k)) {
      clearTimeout(cache.timers.get(k))
    }
    cache.timers.delete(k)
    return cache.data.delete(k)
  },
  clear: () => {
    cache.data.clear()
    for (const v of cache.timers.values()) {
      clearTimeout(v)
    }
    cache.timers.clear()
  }
}

If that isn't to your liking, check out @isaacs/ttlcache.

Performance

As of January 2022, version 7 of this library is one of the most performant LRU cache implementations in JavaScript.

Benchmarks can be extremely difficult to get right. In particular, the performance of set/get/delete operations on objects will vary wildly depending on the type of key used. V8 is highly optimized for objects with keys that are short strings, especially integer numeric strings. Thus any benchmark which tests solely using numbers as keys will tend to find that an object-based approach performs the best.

Note that coercing anything to strings to use as object keys is unsafe, unless you can be 100% certain that no other type of value will be used. For example:

const myCache = {}
const set = (k, v) => myCache[k] = v
const get = (k) => myCache[k]

set({}, 'please hang onto this for me')
set('[object Object]', 'oopsie')

Also beware of "Just So" stories regarding performance. Garbage collection of large (especially: deep) object graphs can be incredibly costly, with several "tipping points" where it increases exponentially. As a result, putting that off until later can make it much worse, and less predictable. If a library performs well, but only in a scenario where the object graph is kept shallow, then that won't help you if you are using large objects as keys.

In general, when attempting to use a library to improve performance (such as a cache like this one), it's best to choose an option that will perform well in the sorts of scenarios where you'll actually use it.

This library is optimized for repeated gets and minimizing eviction time, since that is the expected need of a LRU. Set operations are somewhat slower on average than a few other options, in part because of that optimization. It is assumed that you'll be caching some costly operation, ideally as rarely as possible, so optimizing set over get would be unwise.

If performance matters to you:

  1. If it's at all possible to use small integer values as keys, and you can guarantee that no other types of values will be used as keys, then do that, and use a cache such as lru-fast, or mnemonist's LRUCache which uses an Object as its data store.
  2. Failing that, if at all possible, use short non-numeric strings (ie, less than 256 characters) as your keys, and use mnemonist's LRUCache.
  3. If the types of your keys will be long strings, strings that look like floats, null, objects, or some mix of types, or if you aren't sure, then this library will work well for you.
  4. Do not use a dispose function, size tracking, or especially ttl behavior, unless absolutely needed. These features are convenient, and necessary in some use cases, and every attempt has been made to make the performance impact minimal, but it isn't nothing.

Breaking Changes in Version 7

This library changed to a different algorithm and internal data structure in version 7, yielding significantly better performance, albeit with some subtle changes as a result.

If you were relying on the internals of LRUCache in version 6 or before, it probably will not work in version 7 and above.

For more info, see the change log.