A cache module for nodejs that allows easy wrapping of functions in cache, tiered caches, and a consistent interface.
- Easy way to wrap any function in cache.
- Tiered caches -- data gets stored in each cache and fetched from the highest priority cache(s) first.
- Use any cache you want, as long as it has the same API.
- 100% test coverage via mocha, istanbul, and sinon.
See the Express.js cache-manager example app to see how to use
node-cache-manager in your applications.
npm install cache-manager
First, it includes a wrap function that lets you wrap any function in cache.
(Note, this was inspired by node-caching.)
This is probably the feature you're looking for. As an example, where you might have to do this:
function getCachedUser(id, cb) {
memoryCache.get(id, function (err, result) {
if (err) {
return cb(err);
}
if (result) {
return cb(null, result);
}
getUser(id, function (err, result) {
if (err) {
return cb(err);
}
memoryCache.set(id, result);
cb(null, result);
});
});
}... you can instead use the wrap function:
function getCachedUser(id, cb) {
memoryCache.wrap(
id,
function (cacheCallback) {
getUser(id, cacheCallback);
},
{ ttl: ttl },
cb,
);
}Second, node-cache-manager features a built-in memory cache (using node-lru-cache), with the standard functions you'd expect in most caches:
set(key, val, {ttl: ttl}, cb) // * see note below
get(key, cb)
del(key, cb)
mset(key1, val1, key2, val2, {ttl: ttl}, cb) // set several keys at once
mget(key1, key2, key3, cb) // get several keys at once
// * Note that depending on the underlying store, you may be able to pass the
// ttl as the third param, like this:
set(key, val, ttl, cb)
// ... or pass no ttl at all:
set(key, val, cb)
Third, node-cache-manager lets you set up a tiered cache strategy. This may be of limited use in most cases, but imagine a scenario where you expect tons of traffic, and don't want to hit your primary cache (like Redis) for every request. You decide to store the most commonly-requested data in an in-memory cache, perhaps with a very short timeout and/or a small data size limit. But you still want to store the data in Redis for backup, and for the requests that aren't as common as the ones you want to store in memory. This is something node-cache-manager handles easily and transparently.
Fourth, it allows you to get and set multiple keys at once for caching store that support it. This means that when getting muliple keys it will go through the different caches starting from the highest priority one (see multi store below) and merge the values it finds at each level.
See examples below and in the examples directory. See examples/redis_example for an example of how to implement a
Redis cache store with connection pooling.
var cacheManager = require('cache-manager');
var memoryCache = cacheManager.caching({
store: 'memory',
max: 100,
ttl: 10 /*seconds*/,
});
var ttl = 5;
// Note: callback is optional in set() and del().
// Note: memory cache clones values before setting them unless
// shouldCloneBeforeSet is set to false
memoryCache.set('foo', 'bar', { ttl: ttl }, function (err) {
if (err) {
throw err;
}
memoryCache.get('foo', function (err, result) {
console.log(result);
// >> 'bar'
memoryCache.del('foo', function (err) {});
});
});
function getUser(id, cb) {
setTimeout(function () {
console.log('Returning user from slow database.');
cb(null, { id: id, name: 'Bob' });
}, 100);
}
var userId = 123;
var key = 'user_' + userId;
// Note: ttl is optional in wrap()
memoryCache.wrap(
key,
function (cb) {
getUser(userId, cb);
},
{ ttl: ttl },
function (err, user) {
console.log(user);
// Second time fetches user from memoryCache
memoryCache.wrap(
key,
function (cb) {
getUser(userId, cb);
},
function (err, user) {
console.log(user);
},
);
},
);
// Outputs:
// Returning user from slow database.
// { id: 123, name: 'Bob' }
// { id: 123, name: 'Bob' }The ttl can also be computed dynamically by passing in a function. E.g.,
var opts = {
ttl: function(user) {
if (user.id === 1) {
return 0.1;
} else {
return 0.5;
}
}
};
memoryCache.wrap(key, function(cb) {
getUser(userId, cb);
}, opts, function(err, user) {
console.log(user);
}You can get several keys at once. Note that this will return whatever records it
finds in the cache and it is up to the user to check the results against the
supplied keys and make any calls to the underlying data store to fill in
missing records. In practice, this should not be much of a concern if you are
only using the wrap function to set these records in cache.
Side note: Ideally the wrap function would get what it can from the cache and fill in
the missing records from the data store, but I can't think of a way to do this
that is generic to all situations. Another option is to only return the data
from the cache if all records are found, but this woul break multi-caching.
See unit tests in caching.unit.js for more information.
Example:
var key1 = 'user_1';
var key2 = 'user_1';
memoryCache.wrap(
key1,
key2,
function (cb) {
getManyUser([key1, key2], cb);
},
function (err, users) {
console.log(users[0]);
console.log(users[1]);
},
);memoryCache.mset('foo', 'bar', 'foo2', 'bar2', { ttl: ttl }, function (err) {
if (err) {
throw err;
}
memoryCache.mget('foo', 'foo2', function (err, result) {
console.log(result);
// >> ['bar', 'bar2']
// Delete keys with del() passing arguments...
memoryCache.del('foo', 'foo2', function (err) {});
// ...passing an Array of keys
memoryCache.del(['foo', 'foo2'], function (err) {});
});
});memoryCache
.wrap(key, function () {
return getUserPromise(userId);
})
.then(function (user) {
console.log('User:', user);
});If you are using a Node version that does not include native promises, you can specify your promise dependency in the options passed to the cache module. E.g.,
var Promise = require('es6-promise').Promise;
cache = caching({ store: store, promiseDependency: Promise });try {
let user = await memoryCache.wrap(key, function () {
return getUserPromise(userId);
});
} catch (err) {
// error handling
}Hint: should wrap await call with try - catch to handle promise error.
(Also see the Express.js cache-manager example app).
function respond(res, err, data) {
if (err) {
res.json(500, err);
} else {
res.json(200, data);
}
}
app.get('/foo/bar', function (req, res) {
var cacheKey = 'foo-bar:' + JSON.stringify(req.query);
var ttl = 10;
memoryCache.wrap(
cacheKey,
function (cacheCallback) {
DB.find(req.query, cacheCallback);
},
{ ttl: ttl },
function (err, result) {
respond(res, err, result);
},
);
});You can use your own custom store by creating one with the same API as the built-in memory stores (such as a redis or memcached store). To use your own store just pass in an instance of it.
E.g.,
var myStore = require('your-homemade-store');
var cache = cacheManager.caching({ store: myStore });var multiCache = cacheManager.multiCaching([memoryCache, someOtherCache]);
userId2 = 456;
key2 = 'user_' + userId;
ttl = 5;
// Sets in all caches.
// The "ttl" option can also be a function (see example below)
multiCache.set('foo2', 'bar2', { ttl: ttl }, function (err) {
if (err) {
throw err;
}
// Fetches from highest priority cache that has the key.
multiCache.get('foo2', function (err, result) {
console.log(result);
// >> 'bar2'
// Delete from all caches
multiCache.del('foo2');
});
});
// Set the ttl value by context depending on the store.
function getTTL(data, store) {
if (store === 'redis') {
return 6000;
}
return 3000;
}
// Sets multiple keys in all caches.
// You can pass as many key,value pair as you want
multiCache.mset(
'key',
'value',
'key2',
'value2',
{ ttl: getTTL },
function (err) {
if (err) {
throw err;
}
// mget() fetches from highest priority cache.
// If the first cache does not return all the keys,
// the next cache is fetched with the keys that were not found.
// This is done recursively until either:
// - all have been found
// - all caches has been fetched
multiCache.mget('key', 'key2', function (err, result) {
console.log(result[0]);
console.log(result[1]);
// >> 'bar2'
// >> 'bar3'
// Delete from all caches
multiCache.del('key', 'key2');
// ...or with an Array
multiCache.del(['key', 'key2']);
});
},
);
// Note: options with ttl are optional in wrap()
multiCache.wrap(
key2,
function (cb) {
getUser(userId2, cb);
},
{ ttl: ttl },
function (err, user) {
console.log(user);
// Second time fetches user from memoryCache, since it's highest priority.
// If the data expires in the memory cache, the next fetch would pull it from
// the 'someOtherCache', and set the data in memory again.
multiCache.wrap(
key2,
function (cb) {
getUser(userId2, cb);
},
function (err, user) {
console.log(user);
},
);
},
);
// Multiple keys
multiCache.wrap(
'key1',
'key2',
function (cb) {
getManyUser(['key1', 'key2'], cb);
},
{ ttl: ttl },
function (err, users) {
console.log(users[0]);
console.log(users[1]);
},
);Both the caching and multicaching modules allow you to pass in a callback function named
isCacheableValue which is called by the wrap function with every value returned from cache or from the wrapped function.
This lets you specify which values should and should not be cached by wrap. If the function returns true, it will be
stored in cache. By default the caches cache everything except undefined.
NOTE: The set functions in caching and multicaching do not use isCacheableValue.
For example, if you don't want to cache false and null, you can pass in a function like this:
var isCacheableValue = function (value) {
return value !== null && value !== false && value !== undefined;
};Then pass it to caching like this:
var memoryCache = cacheManager.caching({
store: 'memory',
isCacheableValue: isCacheableValue,
});And pass it to multicaching like this:
var multiCache = cacheManager.multiCaching([memoryCache, someOtherCache], {
isCacheableValue: isCacheableValue,
});Both the caching and multicaching modules support a mechanism to refresh expiring cache keys in background when using wrap function.
This is done by adding a refreshThreshold attribute while creating the caching store.
If refreshThreshold is set and if the ttl method is available for the used store, after retrieving a value from cache TTL will be checked.
If the remaining TTL is less than refreshThreshold, the system will spawn a background worker to update the value, following same rules as standard fetching. In the meantime, the system will return the old value until expiration.
In case of multicaching, the store that will be used for refresh is the one where the key will be found first (highest priority). The value will then be set in all the stores.
NOTES:
- In case of multicaching, the store that will be checked for refresh is the one where the key will be found first (highest priority).
- If the threshold is low and the worker function is slow, the key may expire and you may encounter a racing condition with updating values.
- The background refresh mechanism currently does not support providing multiple keys to
wrapfunction. - The caching store needs to provide the
ttlmethod.
For example, pass the refreshThreshold to caching like this:
var redisStore = require('cache-manager-ioredis');
var redisCache = cacheManager.caching({
store: redisStore,
refreshThreshold: 3,
isCacheableValue: isCacheableValue,
});When a value will be retrieved from Redis with a remaining TTL < 3sec, the value will be updated in background.
You may disable real caching but still get all the callback functionality working by setting none store.
To generate JSDOC 3 documentation:
make docs
To run tests, first run:
npm install -d
Run the tests and JShint:
make
If you would like to contribute to the project, please fork it and send us a pull request. Please add tests
for any new features or bug fixes. Also run make before submitting the pull request.
node-cache-manager is licensed under the MIT license.