Open martin-df opened 5 years ago
I need the actual encoded text to explain / investigate this further. In general HarfBuzz doesn't apply the BiDi algorithm and also just checks the first character of the text for Unicode properties. If the text starts with a LTR character the whole buffer is shaped with this direction.
I am using SkiaSharp with SkiaSharp.Harfbuzz in a Xamarin App and it works fine with RTL languages like arabic.
I am currently facing a problem displaying an arabic date. It works fine if it starts with arabic characters, but if it starts with a number it does not shape the text correctly.
The following date shows up correctly in canvas: 8 yluJ
In this example yluJ is July in arabic.
This date is not shaped correctly: yluJ 8
I figured out it is a problem if the text starts with an LTR-Text following by an RTL text.
Is there any way to force SkiaSharp.Hafbuzz to use arabic shaping? Or is there any other way around this?