downloading and caching images from url asynchronously - swift

Question

I'm trying to download images from my firebase database and load them into collectionviewcells. The images download, however I am having trouble having them all download and load asynchronously.
Currently when I run my code the last image downloaded loads. However, if I update my database the collection view updates and the new last user profile image also loads in but the remainder are missing.
I'd prefer to not use a 3rd party library so any resources or suggestions would be greatly appreciated.
Here's the code that handles the downloading:
func loadImageUsingCacheWithUrlString(_ urlString: String) {
self.image = nil
// checks cache
if let cachedImage = imageCache.object(forKey: urlString as NSString) as? UIImage {
self.image = cachedImage
return
}
//download
let url = URL(string: urlString)
URLSession.shared.dataTask(with: url!, completionHandler: { (data, response, error) in
//error handling
if let error = error {
print(error)
return
}
DispatchQueue.main.async(execute: {
if let downloadedImage = UIImage(data: data!) {
imageCache.setObject(downloadedImage, forKey: urlString as NSString)
self.image = downloadedImage
}
})
}).resume()
}
I believe the solution lies somewhere in reloading the collectionview I just don't know where exactly to do it.
Any suggestions?
EDIT:
Here is where the function is being called; my cellForItem at indexpath
override func collectionView(_ collectionView: UICollectionView, cellForItemAt indexPath: IndexPath) -> UICollectionViewCell {
let cell = collectionView.dequeueReusableCell(withReuseIdentifier: userResultCellId, for: indexPath) as! FriendCell
let user = users[indexPath.row]
cell.nameLabel.text = user.name
if let profileImageUrl = user.profileImageUrl {
cell.profileImage.loadImageUsingCacheWithUrlString(profileImageUrl)
}
return cell
}
The only other thing that I believe could possibly affect the images loading is this function I use to download the user data, which is called in viewDidLoad, however all the other data downloads correctly.
func fetchUser(){
Database.database().reference().child("users").observe(.childAdded, with: {(snapshot) in
if let dictionary = snapshot.value as? [String: AnyObject] {
let user = User()
user.setValuesForKeys(dictionary)
self.users.append(user)
print(self.users.count)
DispatchQueue.main.async(execute: {
self.collectionView?.reloadData()
})
}
}, withCancel: nil)
}
Current Behavior:
As for the current behavior the last cell is the only cell that displays the downloaded profile image; if there are 5 cells, the 5th is the only one that displays a profile image. Also when I update the database, ie register a new user into it, the collectionview updates and displays the newly registered user correctly with their profile image in addition to the old last cell that downloaded it's image properly. The rest however, remain without profile images.

Answer 1

I know you found your problem and it was unrelated to the above code, yet I still have an observation. Specifically, your asynchronous requests will carry on, even if the cell (and therefore the image view) have been subsequently reused for another index path. This results in two problems:
If you quickly scroll to the 100th row, you are going to have to wait for the images for the first 99 rows to be retrieved before you see the images for the visible cells. This can result in really long delays before images start popping in.
If that cell for the 100th row was reused several times (e.g. for row 0, for row 9, for row 18, etc.), you may see the image appear to flicker from one image to the next until you get to the image retrieval for the 100th row.
Now, you might not immediately notice either of these are problems because they will only manifest themselves when the image retrieval has a hard time keeping up with the user's scrolling (the combination of slow network and fast scrolling). As an aside, you should always test your app using the network link conditioner, which can simulate poor connections, which makes it easier to manifest these bugs.
Anyway, the solution is to keep track of (a) the current URLSessionTask associated with the last request; and (b) the current URL being requested. You can then (a) when starting a new request, make sure to cancel any prior request; and (b) when updating the image view, make sure the URL associated with the image matches what the current URL is.
The trick, though, is when writing an extension, you cannot just add new stored properties. So you have to use the associated object API to associate these two new stored values with the UIImageView object. I personally wrap this associated value API with a computed property, so that the code for retrieving the images does not get too buried with this sort of stuff. Anyway, that yields:
extension UIImageView {
private static var taskKey = 0
private static var urlKey = 0
private var currentTask: URLSessionTask? {
get { objc_getAssociatedObject(self, &UIImageView.taskKey) as? URLSessionTask }
set { objc_setAssociatedObject(self, &UIImageView.taskKey, newValue, .OBJC_ASSOCIATION_RETAIN_NONATOMIC) }
}
private var currentURL: URL? {
get { objc_getAssociatedObject(self, &UIImageView.urlKey) as? URL }
set { objc_setAssociatedObject(self, &UIImageView.urlKey, newValue, .OBJC_ASSOCIATION_RETAIN_NONATOMIC) }
}
func loadImageAsync(with urlString: String?, placeholder: UIImage? = nil) {
// cancel prior task, if any
weak var oldTask = currentTask
currentTask = nil
oldTask?.cancel()
// reset image view’s image
self.image = placeholder
// allow supplying of `nil` to remove old image and then return immediately
guard let urlString = urlString else { return }
// check cache
if let cachedImage = ImageCache.shared.image(forKey: urlString) {
self.image = cachedImage
return
}
// download
let url = URL(string: urlString)!
currentURL = url
let task = URLSession.shared.dataTask(with: url) { [weak self] data, response, error in
self?.currentTask = nil
// error handling
if let error = error {
// don't bother reporting cancelation errors
if (error as? URLError)?.code == .cancelled {
return
}
print(error)
return
}
guard let data = data, let downloadedImage = UIImage(data: data) else {
print("unable to extract image")
return
}
ImageCache.shared.save(image: downloadedImage, forKey: urlString)
if url == self?.currentURL {
DispatchQueue.main.async {
self?.image = downloadedImage
}
}
}
// save and start new task
currentTask = task
task.resume()
}
}
Also, note that you were referencing some imageCache variable (a global?). I would suggest an image cache singleton, which, in addition to offering the basic caching mechanism, also observes memory warnings and purges itself in memory pressure situations:
class ImageCache {
private let cache = NSCache<NSString, UIImage>()
private var observer: NSObjectProtocol?
static let shared = ImageCache()
private init() {
// make sure to purge cache on memory pressure
observer = NotificationCenter.default.addObserver(
forName: UIApplication.didReceiveMemoryWarningNotification,
object: nil,
queue: nil
) { [weak self] notification in
self?.cache.removeAllObjects()
}
}
deinit {
NotificationCenter.default.removeObserver(observer!)
}
func image(forKey key: String) -> UIImage? {
return cache.object(forKey: key as NSString)
}
func save(image: UIImage, forKey key: String) {
cache.setObject(image, forKey: key as NSString)
}
}
A bigger, more architectural, observation: One really should decouple the image retrieval from the image view. Imagine you have a table where you have a dozen cells using the same image. Do you really want to retrieve the same image a dozen times just because the second image view scrolled into view before the first one finished its retrieval? No.
Also, what if you wanted to retrieve the image outside of the context of an image view? Perhaps a button? Or perhaps for some other reason, such as to download images to store in the user’s photos library. There are tons of possible image interactions above and beyond image views.
Bottom line, fetching images is not a method of an image view, but rather a generalized mechanism of which an image view would like to avail itself. An asynchronous image retrieval/caching mechanism should generally be incorporated in a separate “image manager” object. It can then detect redundant requests and be used from contexts other than an image view.
As you can see, the asynchronous retrieval and caching is starting to get a little more complicated, and this is why we generally advise considering established asynchronous image retrieval mechanisms like AlamofireImage or Kingfisher or SDWebImage. These guys have spent a lot of time tackling the above issues, and others, and are reasonably robust. But if you are going to “roll your own,” I would suggest something like the above at a bare minimum.