Support woff2

[stephanemagnenat]

Hello,

Is there planned support for the woff2 format? For example, allsorts supports it. This format is convenient to embed fonts, for example in games.

kind regards

[RazrFalcon]

Plans - yes, but no timeline. It will be done when it will be done.

[inferiorhumanorgans]

I wonder if it would be more appropriate to put WOFF support in a separate crate (or at least outside of Face/RawFace). I'm just spitballing here but…

WOFF Considerations

• Dependencies (brotli, zlib). There are pure rust implementations of both, but this is still a big first step.
• Heap allocation. It's certainly possible to inflate stuff in a no-std environment and/or without heap allocations, but it's that much more archaic.
• WOFF1 compresses each table individually, so it's probably not too intrusive to track whether a table should be inflated and use something like a ZlibStream struct.
• WOFF2 encodes everything as one brotli stream after optionally applying some per table transformations. This is different enough there's a risk of creating spaghetti code.
• A separate struct could simply handle the table lookup and inflation and pass the "raw" inflated data back to the appropriate ttf-parser structs. At that point it's probably (maybe?) worth extracting most of Face into a trait for ease if interoperability.

[RazrFalcon]

My approach would be to have a separate WOFF to TTF converter crate. I don't see much benefits of having WOFF support directly in ttf-parser. But I haven't looked deep enough into it.

It definitely feels like something that should not belong in ttf-parser.