我正在开发一个使用Swift 3的iOS应用程序,使用ReactiveSwift 1.1.1,MVVM + Flow Coordinator模式和Firebase作为后端。我最近才开始适应FRP,我仍在努力弄清楚如何将新功能集成到我现有的代码库中。
例如,我的模型使用Firebase中的异步方法从Web下载缩略图,我想提供SignalProducer<Content, NoError>
来订阅我的ViewModel类并观察是否已下载缩略图,然后更新UI。
// field to be used from the view-models to observe
public let thumbnailContentSignalProducer = SignalProducer<Content, NoError> { (observer, disposable) in
// TODO: send next content via completion below
}
// thumbnail download method
public func findThumbnail(bucketId: String, contentId: String) {
guard let userId = userService.getCurrentUserId() else {
debugPring("Error id")
return
}
let ref = self.storageThumbnail.reference()
let contentRef = ref
.child(userId)
.child(bucketId)
.child(FirebaseConstants.pathImages)
.child("\(contentId).jpg")
contentRef.data(withMaxSize: 1 * 1024 * 1024, completion: { (data, error) in
guard let data = data else {
debugPrint("Error download")
return
}
let content = Image(data: data)
content.id = contentId
content.userId = userId
content.bucketId = bucketId
// TODO: emit signal with content
// How to send the content via the SignalProducer above?
})
}
我也尝试过与Signal<Content, NoError>
类似的东西,而我使用Signal<Content, NoError>.pipe()
方法接收(observer, disposable)
元组,并将观察者保存为私有全局字段以从中访问它Firebase回调。
问题:
这是正确的做法还是我错过了什么?
如何在完成时发出内容对象?
更新
经过几个小时的痛苦,我发现了如何设计SingalProducer以发出信号并从ViewModel订阅。
也许以下代码段也可以帮助其他代码:
// model protocol
import ReactiveSwift
import enum Result.NoError
public protocol ContentService {
func findThumbnail(bucketId: String, contentId: String)
var thumbnailContentProducer: SignalProducer<Content, NoError> { get }
}
// model implementation using firebase
import Firebase
import FirebaseStorage
import ReactiveSwift
public class FirebaseContentService: ContentService {
// other fields, etc.
// ...
private var thumbnailContentObserver: Observer<Content, NoError>?
private var thumbnailContentSignalProducer: SignalProducer<Content, NoError>?
var thumbnailContentProducer: SignalProducer<Content, NoError> {
return thumbnailContentSignalProducer!
}
init() {
thumbnailContentSignalProducer = SignalProducer<Content, NoError> { (observer, disposable) in
self.thumbnailContentObserver = observer
}
}
func findThumbnail(bucketId: String, contentId: String) {
guard let userId = userService.getCurrentUserId() else {
// TODO handle error
return
}
let ref = self.storageThumbnail.reference()
let contentRef = ref
.child(userId)
.child(bucketId)
.child(FirebaseConstants.pathImages)
.child("\(contentId).jpg")
contentRef.data(withMaxSize: 1 * 1024 * 1024, completion: { (data, error) in
guard let data = data else {
// TODO handle error
return
}
let content = Image(data: data)
content.id = contentId
content.userId = userId
content.bucketId = bucketId
// emit signal
self.thumbnailContentObserver?.send(value: content)
})
}
}
// usage from a ViewModel
contentService.thumbnailContentProducer
.startWithValues { content in
self.contents.append(content)
}
也许有人可以验证上面的代码,并说这是正确的方法。
答案 0 :(得分:1)
当您考虑将Signal
与pipe
一起使用时,我认为您走的是正确的道路。关键是您需要为每个缩略图请求创建一个新的SignalProducer
,并且您需要一种方法将所有这些请求组合成一个结果信号。我正在考虑这样的事情(注意这是未经测试的代码,但它应该得到这个想法):
class FirebaseContentService {
// userService and storageThumbnail defined here
}
extension FirebaseContentService: ReactiveExtensionsProvider { }
extension Reactive where Base: FirebaseContentService {
private func getThumbnailContentSignalProducer(bucketId: String, contentId: String) -> SignalProducer<Content, ContentError> {
return SignalProducer<Content, ContentError> { (observer, disposable) in
guard let userId = self.base.userService.getCurrentUserId() else {
observer.send(error: ContentError.invalidUserLogin)
return
}
let ref = self.base.storageThumbnail.reference()
let contentRef = ref
.child(userId)
.child(bucketId)
.child(FirebaseConstants.pathImages)
.child("\(contentId).jpg")
contentRef.data(withMaxSize: 1 * 1024 * 1024, completion: { (data, error) in
guard let data = data else {
observer.send(error: ContentError.contentNotFound)
return
}
let content = Image(data: data)
content.id = contentId
content.userId = userId
content.bucketId = bucketId
observer.send(value: content)
observer.sendCompleted()
})
}
}
}
class ThumbnailProvider {
public let thumbnailSignal: Signal<Content, NoError>
private let input: Observer<(bucketId: String, contentId: String), NoError>
init(contentService: FirebaseContentService) {
let (signal, observer) = Signal<(bucketId: String, contentId: String), NoError>.pipe()
self.input = observer
self.thumbnailSignal = signal
.flatMap(.merge) { param in
return contentService.reactive.getThumbnailContentSignalProducer(bucketId: param.bucketId, contentId: param.contentId)
.flatMapError { error in
debugPrint("Error download")
return SignalProducer.empty
}
}
}
public func findThumbnail(bucketId: String, contentId: String) {
input.send(value: (bucketId: bucketId, contentId: contentId))
}
}
像这样使用ReactiveExtensionsProvider
是通过reactive
属性将反应API添加到现有功能的惯用方法。
实际的请求代码仅限于getThumbnailContentSignalProducer
,它为每个请求创建一个SignalProducer。请注意,此处会传递错误,稍后会处理和转换为NoError
。
findThumbnails
只需bucketId
和contentId
,然后通过输入可观察对象发送。
thumbnailSignal
中init
的构造是神奇发生的地方。每个输入(包含bucketId
和contentId
的元组)都会通过flatMap
转换为请求。请注意,.merge
策略意味着以请求完成的任何顺序尽快发送缩略图。如果您想确保缩略图按照请求的顺序返回,您可以使用.concat
。
flatMapError
是处理潜在错误的地方。在这种情况下,它只是打印&#34;错误下载&#34;什么都不做。