问题
I have a UITableView that I am attempting to expand and collapse using its beginUpdates
and endUpdates
methods and have a drop shadow displayed as that's happening. In my custom UITableViewCell, I have a layer which I create a shadow for in layoutSubviews
:
self.shadowLayer.shadowColor = self.shadowColor.CGColor;
self.shadowLayer.shadowOffset = CGSizeMake(self.shadowOffsetWidth, self.shadowOffsetHeight);
self.shadowLayer.shadowOpacity = self.shadowOpacity;
self.shadowLayer.masksToBounds = NO;
self.shadowLayer.frame = self.layer.bounds;
// this is extremely important for performance when drawing shadows
UIBezierPath *shadowPath = [UIBezierPath bezierPathWithRoundedRect:self.shadowLayer.frame cornerRadius:self.cornerRadius];
self.shadowLayer.shadowPath = shadowPath.CGPath;
I add this layer to the UITableViewCell in viewDidLoad
:
self.shadowLayer = [CALayer layer];
self.shadowLayer.backgroundColor = [UIColor whiteColor].CGColor;
[self.layer insertSublayer:self.shadowLayer below:self.contentView.layer];
As I understand it, when I call beginUpdates
, an implicit CALayerTransaction
is made for the current run loop if none exists. Additionally, layoutSubviews
also gets called. The problem here is that the resulting shadow is drawn immediately based on the new size of the UITableViewCell. I really need to shadow to continue to cast in the expected way as the actual layer is animating.
Since my created layer is not a backing CALayer it animates without explicitly specifying a CATransaction, which is expected. But, as I understand it, I really need some way to grab hold of beginUpdates
/endUpdates
CATransaction and perform the animation in that. How do I do that, if at all?
回答1:
So I guess you have something like this:
(I turned on “Debug > Slow Animations” in the simulator.) And you don't like the way the shadow jumps to its new size. You want this instead:
You can find my test project in this github repository.
It is tricky but not impossible to pick up the animation parameters and add an animation in the table view's animation block. The trickiest part is that you need to update the shadowPath
in the layoutSubviews
method of the shadowed view itself, or of the shadowed view's immediate superview. In my demo video, that means that the shadowPath
needs to be updated by the layoutSubviews
method of the green box view or the green box's immediate superview.
I chose to create a ShadowingView
class whose only job is to draw and animate the shadow of one of its subviews. Here's the interface:
@interface ShadowingView : UIView
@property (nonatomic, strong) IBOutlet UIView *shadowedView;
@end
To use ShadowingView
, I added it to my cell view in my storyboard. Actually it's nested inside a stack view inside the cell. Then I added the green box as a subview of the ShadowingView
and connected the shadowedView
outlet to the green box.
The ShadowingView
implementation has three parts. One is its layoutSubviews
method, which sets up the layer shadow properties on its own layer to draw a shadow around its shadowedView
subview:
@implementation ShadowingView
- (void)layoutSubviews {
[super layoutSubviews];
CALayer *layer = self.layer;
layer.backgroundColor = nil;
CALayer *shadowedLayer = self.shadowedView.layer;
if (shadowedLayer == nil) {
layer.shadowColor = nil;
return;
}
NSAssert(shadowedLayer.superlayer == layer, @"shadowedView must be my direct subview");
layer.shadowColor = UIColor.blackColor.CGColor;
layer.shadowOffset = CGSizeMake(0, 1);
layer.shadowOpacity = 0.5;
layer.shadowRadius = 3;
layer.masksToBounds = NO;
CGFloat radius = shadowedLayer.cornerRadius;
layer.shadowPath = CGPathCreateWithRoundedRect(shadowedLayer.frame, radius, radius, nil);
}
When this method is run inside an animation block (as is the case when the table view animates a change in the size of a cell), and the method sets shadowPath
, Core Animation looks for an “action” to run after updating shadowPath
. One of the ways it looks is by sending actionForLayer:forKey:
to the layer's delegate, and the delegate is the ShadowingView
. So we override actionForLayer:forKey:
to provide an action if possible and appropriate. If we can't, we just call super
.
It is important to understand that Core Animation asks for the action from inside the shadowPath
setter, before actually changing the value of shadowPath
.
To provide the action, we make sure the key is @"shadowPath"
, that there is an existing value for shadowPath
, and that there is already an animation on the layer for bounds.size
. Why do we look for an existing bounds.size
animation? Because that existing animation has the duration and timing function we should use to animate shadowPath
. If everything is in order, we grab the existing shadowPath
, make a copy of the animation, store them in an action, and return the action:
- (id<CAAction>)actionForLayer:(CALayer *)layer forKey:(NSString *)event {
if (![event isEqualToString:@"shadowPath"]) { return [super actionForLayer:layer forKey:event]; }
CGPathRef priorPath = layer.shadowPath;
if (priorPath == NULL) { return [super actionForLayer:layer forKey:event]; }
CAAnimation *sizeAnimation = [layer animationForKey:@"bounds.size"];
if (![sizeAnimation isKindOfClass:[CABasicAnimation class]]) { return [super actionForLayer:layer forKey:event]; }
CABasicAnimation *animation = [sizeAnimation copy];
animation.keyPath = @"shadowPath";
ShadowingViewAction *action = [[ShadowingViewAction alloc] init];
action.priorPath = priorPath;
action.pendingAnimation = animation;
return action;
}
@end
What does the action look like? Here's the interface:
@interface ShadowingViewAction : NSObject <CAAction>
@property (nonatomic, strong) CABasicAnimation *pendingAnimation;
@property (nonatomic) CGPathRef priorPath;
@end
The implementation requires a runActionForKey:object:arguments:
method. In this method, we update the animation that we created in actionForLayer:forKey:
using the saved-away old shadowPath
and the new shadowPath
, and then we add the animation to the layer.
We also need to manage the retain count of the saved path, because ARC doesn't manage CGPath
objects.
@implementation ShadowingViewAction
- (void)runActionForKey:(NSString *)event object:(id)anObject arguments:(NSDictionary *)dict {
if (![anObject isKindOfClass:[CALayer class]] || _pendingAnimation == nil) { return; }
CALayer *layer = anObject;
_pendingAnimation.fromValue = (__bridge id)_priorPath;
_pendingAnimation.toValue = (__bridge id)layer.shadowPath;
[layer addAnimation:_pendingAnimation forKey:@"shadowPath"];
}
- (void)setPriorPath:(CGPathRef)priorPath {
CGPathRetain(priorPath);
CGPathRelease(_priorPath);
_priorPath = priorPath;
}
- (void)dealloc {
CGPathRelease(_priorPath);
}
@end
回答2:
This is Rob Mayoff's answer written in Swift. Could save someone some time.
Please don't upvote this. Upvote Rob Mayoff's solution. It is awesome, and correct.
import UIKit
class AnimatingShadowView: UIView {
struct DropShadowParameters {
var shadowOpacity: Float = 0
var shadowColor: UIColor? = .black
var shadowRadius: CGFloat = 0
var shadowOffset: CGSize = .zero
static let defaultParameters = DropShadowParameters(shadowOpacity: 0.15,
shadowColor: .black,
shadowRadius: 5,
shadowOffset: CGSize(width: 0, height: 1))
}
@IBOutlet weak var contentView: UIView! // no sense in have a shadowView without content!
var shadowParameters: DropShadowParameters = DropShadowParameters.defaultParameters
private func apply(dropShadow: DropShadowParameters) {
let layer = self.layer
layer.shadowColor = dropShadow.shadowColor?.cgColor
layer.shadowOffset = dropShadow.shadowOffset
layer.shadowOpacity = dropShadow.shadowOpacity
layer.shadowRadius = dropShadow.shadowRadius
layer.masksToBounds = false
}
override func layoutSubviews() {
super.layoutSubviews()
let layer = self.layer
layer.backgroundColor = nil
let contentLayer = self.contentView.layer
assert(contentLayer.superlayer == layer, "contentView must be a direct subview of AnimatingShadowView!")
self.apply(dropShadow: self.shadowParameters)
let radius = contentLayer.cornerRadius
layer.shadowPath = UIBezierPath(roundedRect: contentLayer.frame, cornerRadius: radius).cgPath
}
override func action(for layer: CALayer, forKey event: String) -> CAAction? {
guard event == "shadowPath" else {
return super.action(for: layer, forKey: event)
}
guard let priorPath = layer.shadowPath else {
return super.action(for: layer, forKey: event)
}
guard let sizeAnimation = layer.animation(forKey: "bounds.size") as? CABasicAnimation else {
return super.action(for: layer, forKey: event)
}
let animation = sizeAnimation.copy() as! CABasicAnimation
animation.keyPath = "shadowPath"
let action = ShadowingViewAction()
action.priorPath = priorPath
action.pendingAnimation = animation
return action
}
}
private class ShadowingViewAction: NSObject, CAAction {
var pendingAnimation: CABasicAnimation? = nil
var priorPath: CGPath? = nil
// CAAction Protocol
func run(forKey event: String, object anObject: Any, arguments dict: [AnyHashable : Any]?) {
guard let layer = anObject as? CALayer, let animation = self.pendingAnimation else {
return
}
animation.fromValue = self.priorPath
animation.toValue = layer.shadowPath
layer.add(animation, forKey: "shadowPath")
}
}
回答3:
Assuming that you're manually setting your shadowPath
, here's a solution inspired by the others here that accomplishes the same thing using less code.
Note that I'm intentionally constructing my own CABasicAnimation
rather than copying the bounds.size
animation exactly, as in my own tests I found that toggling the copied animation while it was still in progress could cause the animation to snap to it's toValue
, rather than transitioning smoothly from its current value.
class ViewWithAutosizedShadowPath: UIView {
override func layoutSubviews() {
super.layoutSubviews()
let oldShadowPath = layer.shadowPath
let newShadowPath = CGPath(rect: bounds, transform: nil)
if let boundsAnimation = layer.animation(forKey: "bounds.size") as? CABasicAnimation {
let shadowPathAnimation = CABasicAnimation(keyPath: "shadowPath")
shadowPathAnimation.duration = boundsAnimation.duration
shadowPathAnimation.timingFunction = boundsAnimation.timingFunction
shadowPathAnimation.fromValue = oldShadowPath
shadowPathAnimation.toValue = newShadowPath
layer.add(shadowPathAnimation, forKey: "shadowPath")
}
layer.shadowPath = newShadowPath
}
}
回答4:
UITableView
is likely not creating a CATransaction
, or if it is, it's waiting until after you end the updates. My understanding is that table views just coalesce all changes between those functions and then creates the animations as necessary. You don't have a way to get a handle on the actual animation parameters it's committing, because we don't know when that actually happens. The same thing happens when you animate a content offset change in UIScrollView
: the system provides no context about the animation itself, which is frustrating. There is also no way to query the system for current CATransaction
s.
Probably the best you can do is inspect the animation that UITableView
is creating and just mimic the same timing parameters in your own animation. Swizzling add(_:forKey:)
on CALayer
can allow you to inspect all animations being added. You certainly don't want to actually ship with this, but I often use this technique in debugging to figure out what animations are being added and what their properties are.
I suspect that you're going to have to commit your own shadow layer animations in tableView(_:willDisplayCell:for:row:)
for the appropriate cells.
来源:https://stackoverflow.com/questions/43944020/animating-calayer-shadow-simultaneously-as-uitableviewcell-height-animates