StyledTextKit is a declarative attributed string library for fast rendering and easy string building.

didTap and didLongPress are not working in the current version (f4927d2). StyledTextView doesn't receive touch events because its layer has no content. The content has been moved to contentLayer at the commit 1271289. I added hitTest to forward hit testing to contentLayer to fix that.

Implemented a way to always have the highlight view always present around a range of text, instead of just having to tap on the text to get it to appear.

First, thanks for this amazing library.

This is a question though, not an issue.

Doesn't CATextLayer provide a similar functionality? I specifically mean the async drawing bit, great work!

CATextLayer has an inherited drawsAsynchronously property. The docs says:

When this property is set to true, the graphics context used to draw the layer’s contents queues drawing commands and executes them on a background thread rather than executing them synchronously. Performing these commands asynchronously can improve performance in some apps. However, you should always measure the actual performance benefits before enabling this capability.

Aside from caching, what advantage does asynchronously outputting to a CGImage have over what CATextLayer provides?

Thanks!

Hi @rnystrom, thanks for this library and let me build attributed string much easier than before. 🎉

I want to add an example project for it, can I make a PR?

And also ask a question here, is it possible to let .link used other styles? Thanks again! ❤️

Hi, Thank you for this amazing library.

When I use StyledTextView in UICollectionViewCell, I found a reuse problem when reloading renderer, and found that the animatable of layer’s contents caused the view content to follow the effect of an animation transition.

https://github.com/GitHawkApp/StyledTextKit/blob/fe3232c980c374d59e520b086b04869e3c920f86/Source/StyledTextView.swift#L190

Suggested to change to:

   CATransaction.begin()
   CATransaction.setValue(kCFBooleanTrue, forKey: kCATransactionDisableActions)
   layer.contents = result.0
   CATransaction.commit()

I believe that the doubly-linked list in the LRUCache impl has a retain cycle in its nodes:

https://github.com/ocrickard/StyledTextKit/blob/master/Source/LRUCache.swift#L24

cc @BasThomas https://developer.apple.com/documentation/foundation/nsattributedstringkey

These accessibility keys are kind of incredible. It'd be really amazing to bake these into the framework somehow.

Can probably get away w/ enumerating all attributes on render() and injecting mapped accessibility attributes? Most of them are straightforward:

• Font
• Foreground color
• Background color
• Strikethrough
• Underline
• Link

I saw your comment about the article I wrote in #52 and figured I had to take a quick stab at fixing that for you. 😄 Let me know if you'd like to see more tests, you can come up with a nicer name than scalingTextStyle, or if there is any documentation I should update.

On iOS 11 and up, use UIFontMetrics to scale font sizes. Falls back to a calculation using the current content size category / large as a multiplier.

Also adds the scalingTextStyle attribute to TextStyle with a default value of body so that users may specify how they'd prefer to scale the specific font.

This should not be a breaking change for anyone since it's an added field with a default value. All tests still pass without adding this new value anywhere.

i'm working on a Japanese project , it has a feature to show/hide Hurigana(Ruby) text. i've some knowledge that it uses kCTRubyAnnotation but it is harsh on main thread..

so, is this library supports RubyText?

This is my first pass at a threadsafe async drawing system.

Problems to resolve before merge:

1. If the bitmap is cached, there's no purpose to this, it should short-circuit instead of trampolining.
2. The logic inside this function feels a little icky. We should clean it up a little bit.

Known problems:

1. Testing, as with all async modes, is nearly impossible.

I'm getting a reproducible but isolated cache miss in GitHawk. Did some digging, nothing obvious. But added this test just as a safety.

GitHawk key's are unequal b/c the UIFont instances are different. Pretty weird.

I learned about the use of unfair locks from this library. Thanks!

Anyway, today I stumbled upon this StackOverflow answer that mentioned that the direct use of unfair locks in Swift can fail at runtime. The fix is really simple; a pointer wrapper that provides a stable memory address (the assumption the C API makes). Any thoughts about this?

I'm trying to create a string that has a mix of two font sizes, but I can't seem to make it work. Is this supported by StyledTextKit? (if not, would this be addressed by issue #61 ?)

Here's a sample of what I tried. Adding calls to save() / restore() don't affect the behavior, nor does changing the contentSizeCategory

let attrStr = StyledTextBuilder(text: "")
        .add(text: "Foo", attributes: [.font: UIFont.systemFont(ofSize: 32)])
        .add(text: "Bar", attributes: [.font: UIFont.systemFont(ofSize: 14)])
        .build()
        .render(contentSizeCategory: .unspecified)

Love this blog post! Removing all color/font from building would be amazing. Instead:

• Use "style builder" pattern that registers styles
  • Need to still be able to "push" and "pop" styles
  • Should all props be in one style (.font, .foregroundColor, etc), or should there be a style per prop? Latter seems harder to maintain
• Remove UIContentSizeCategory from all params until render

Necessary to make progress towards GitHawkApp/GitHawk#198 so you can have themes/styles and re-render (missing all caches) w/out having to rebuild everything (which you'd have to do today).