DataLoader is a generic utility to be used as part of your application's data fetching layer to provide a simplified and consistent API over various remote data sources such as databases or web services via batching and caching.
This is a Swift version of the Facebook DataLoader.
Include this repo in your Package.swift
file.
.Package(url: "https://github.com/GraphQLSwift/DataLoader.git", .upToNextMajor(from: "1.1.0"))
To get started, create a DataLoader. Each DataLoader instance represents a unique cache. Typically instances are created per request when used within a web-server if different users can see different things.
Batching is not an advanced feature, it's DataLoader's primary feature.
Create a DataLoader by providing a batch loading function:
let userLoader = Dataloader<Int, User>(batchLoadFunction: { keys in
try User.query(on: req).filter(\User.id ~~ keys).all().map { users in
keys.map { key in
DataLoaderFutureValue.success(users.filter{ $0.id == key })
}
}
})
The order of the returned DataLoaderFutureValues must match the order of the keys.
let future1 = try userLoader.load(key: 1, on: eventLoopGroup)
let future2 = try userLoader.load(key: 2, on: eventLoopGroup)
let future3 = try userLoader.load(key: 1, on: eventLoopGroup)
The example above will only fetch two users, because the user with key 1
is present twice in the list.
There is also a method to load multiple keys at once
try userLoader.loadMany(keys: [1, 2, 3], on: eventLoopGroup)
By default, a DataLoader will wait for a short time from the moment load
is called to collect keys prior
to running the batchLoadFunction
and completing the load
futures. This is to let keys accumulate and
batch into a smaller number of total requests. This amount of time is configurable using the executionPeriod
option:
let myLoader = DataLoader<String, String>(
options: DataLoaderOptions(executionPeriod: .milliseconds(50)),
batchLoadFunction: { keys in
self.someBatchLoader(keys: keys).map { DataLoaderFutureValue.success($0) }
}
)
Longer execution periods reduce the number of total data requests, but also reduce the responsiveness of the
load
futures.
If desired, you can manually execute the batchLoadFunction
and complete the futures at any time, using the
.execute()
method.
Scheduled execution can be disabled by setting executionPeriod
to nil
, but be careful - you must call .execute()
manually in this case. Otherwise, the futures will never complete!
It is possible to disable batching by setting the batchingEnabled
option to false
.
In this case, the batchLoadFunction
will be invoked immediately when a key is loaded.
DataLoader provides a memoization cache. After .load()
is called with a key, the resulting value is cached
for the lifetime of the DataLoader object. This eliminates redundant loads.
In addition to relieving pressure on your data storage, caching results also creates fewer objects which may relieve memory pressure on your application:
let userLoader = DataLoader<Int, Int>(...)
let future1 = userLoader.load(key: 1, on: eventLoopGroup)
let future2 = userLoader.load(key: 1, on: eventLoopGroup)
print(future1 == future2) // true
DataLoader caching does not replace Redis, Memcache, or any other shared
application-level cache. DataLoader is first and foremost a data loading mechanism,
and its cache only serves the purpose of not repeatedly loading the same data in
the context of a single request to your Application. To do this, it maintains a
simple in-memory memoization cache (more accurately: .load()
is a memoized function).
Avoid multiple requests from different users using the DataLoader instance, which could result in cached data incorrectly appearing in each request. Typically, DataLoader instances are created when a Request begins, and are not used once the Request ends.
In certain uncommon cases, clearing the request cache may be necessary.
The most common example when clearing the loader's cache is necessary is after a mutation or update within the same request, when a cached value could be out of date and future loads should not use any possibly cached value.
Here's a simple example using SQL UPDATE to illustrate.
// Request begins...
let userLoader = DataLoader<Int, Int>(...)
// And a value happens to be loaded (and cached).
userLoader.load(key: 4, on: eventLoopGroup)
// A mutation occurs, invalidating what might be in cache.
sqlRun('UPDATE users WHERE id=4 SET username="zuck"').whenComplete { userLoader.clear(key: 4) }
// Later the value load is loaded again so the mutated data appears.
userLoader.load(key: 4, on: eventLoopGroup)
// Request completes.
If a batch load fails (that is, a batch function throws or returns a DataLoaderFutureValue.failure(Error)),
then the requested values will not be cached. However if a batch
function returns an Error
instance for an individual value, that Error
will
be cached to avoid frequently loading the same Error
.
In some circumstances you may wish to clear the cache for these individual Errors:
userLoader.load(key: 1, on: eventLoopGroup).whenFailure { error in
if (/* determine if should clear error */) {
userLoader.clear(key: 1);
}
throw error
}
In certain uncommon cases, a DataLoader which does not cache may be desirable.
Calling DataLoader(options: DataLoaderOptions(cachingEnabled: false), batchLoadFunction: batchLoadFunction)
will ensure that every
call to .load()
will produce a new Future, and previously requested keys will not be
saved in memory.
However, when the memoization cache is disabled, your batch function will
receive an array of keys which may contain duplicates! Each key will be
associated with each call to .load()
. Your batch loader should provide a value
for each instance of the requested key.
For example:
let myLoader = DataLoader<String, String>(
options: DataLoaderOptions(cachingEnabled: false),
batchLoadFunction: { keys in
self.someBatchLoader(keys: keys).map { DataLoaderFutureValue.success($0) }
}
)
myLoader.load(key: "A", on: eventLoopGroup)
myLoader.load(key: "B", on: eventLoopGroup)
myLoader.load(key: "A", on: eventLoopGroup)
// > [ "A", "B", "A" ]
More complex cache behavior can be achieved by calling .clear()
or .clearAll()
rather than disabling the cache completely. For example, this DataLoader will
provide unique keys to a batch function due to the memoization cache being
enabled, but will immediately clear its cache when the batch function is called
so later requests will load new values.
let myLoader = DataLoader<String, String>(batchLoadFunction: { keys in
identityLoader.clearAll()
return someBatchLoad(keys: keys)
})
DataLoader pairs nicely well with GraphQL and Graphiti. GraphQL fields are designed to be stand-alone functions. Without a caching or batching mechanism, it's easy for a naive GraphQL server to issue new database requests each time a field is resolved.
Consider the following GraphQL request:
{
me {
name
bestFriend {
name
}
friends(first: 5) {
name
bestFriend {
name
}
}
}
}
Naively, if me
, bestFriend
and friends
each need to request the backend,
there could be at most 12 database requests!
By using DataLoader, we could batch our requests to a User
type, and
only require at most 4 database requests, and possibly fewer if there are cache hits.
Here's a full example using Graphiti:
struct User : Codable {
let id: Int
let name: String
let bestFriendID: Int
let friendIDs: [Int]
func getBestFriend(context: UserContext, arguments: NoArguments, group: EventLoopGroup) throws -> EventLoopFuture<User> {
return try context.userLoader.load(key: user.bestFriendID, on: group)
}
struct FriendArguments {
first: Int
}
func getFriends(context: UserContext, arguments: FriendArguments, group: EventLoopGroup) throws -> EventLoopFuture<[User]> {
return try context.userLoader.loadMany(keys: user.friendIDs[0..<arguments.first], on: group)
}
}
struct UserResolver {
public func me(context: UserContext, arguments: NoArguments) -> User {
...
}
}
class UserContext {
let database = ...
let userLoader = DataLoader<Int, User>() { [weak self] keys in
guard let self = self else { throw ContextError }
return User.query(on: self.database).filter(\.$id ~~ keys).all().map { users in
keys.map { key in
users.first { $0.id == key }!
}
}
}
}
struct UserAPI : API {
let resolver = UserResolver()
let schema = Schema<UserResolver, UserContext> {
Type(User.self) {
Field("name", at: \.content)
Field("bestFriend", at: \.getBestFriend, as: TypeReference<User>.self)
Field("friends", at: \.getFriends, as: [TypeReference<User>]?.self) {
Argument("first", at: .\first)
}
}
Query {
Field("me", at: UserResolver.hero, as: User.self)
}
}
}
All your feedback and help to improve this project is very welcome. Please create issues for your bugs, ideas and enhancement requests, or better yet, contribute directly by creating a PR. π
When reporting an issue, please add a detailed example, and if possible a code snippet or test to reproduce your problem. π₯
When creating a pull request, please adhere to the current coding style where possible, and create tests with your code so it keeps providing an awesome test coverage level πͺ
This repo uses SwiftFormat, and includes lint checks to enforce these formatting standards.
To format your code, install swiftformat
and run:
swiftformat .
This library is entirely a Swift version of Facebooks DataLoader. Developed by Lee Byron and Nicholas Schrock from Facebook.