Add code for validating OpenType GLYF table entries

[qwerasd205]

This code was motivated by the need for the glyph protocol handler (#12352) to be able to validate the provided glyf payload, without having to link freetype or anything (because libghostty-vt needs to be static). As such it's written specifically to meet those needs, but in such a way that it can be expanded if we find a need for more in-depth inspection of glyfs in the future.

[mitchellh]

From the spec

If a glyph has zero contours, no additional glyph data beyond the header is required. A glyph with zero contours may have additional data, however; in particular, it may have instructions that operate on phantom points.

This test should pass:

test "glyf: zero-contour glyph can be header-only" {
    const testing = std.testing;

    const header: Glyf.Entry.Header = .{
        .numberOfContours = 0,
        .xMin = 0,
        .yMin = 0,
        .xMax = 0,
        .yMax = 0,
    };
    const glyph = try Glyf.Entry.init(std.mem.asBytes(&header));
    try testing.expectEqual(@sizeOf(Glyf.Entry.Header), try glyph.size());
}

[qwerasd205]

Haha yeah I actually realized I had written that bug while I was going to sleep last night and going back over the spec in my head- adding a test (and if need be, a fix) for it was on my todo list for today.

[mitchellh]

I realize for the purpose of the glyph protocol its a silly glyph cause its... empty... but it is valid. 🤷

[qwerasd205]

But actually, no that test is incorrect-- the additional data after the header is a few bytes still since it has to encode the instructions length. At least 2 extra bytes for the 16-bit instructionLength field, and then however many bytes for the instructions array.

The spec is worded in a confusing way that seems to imply a bare header should be allowed but I have no clue how that would function in an actual font file, since there's nothing in the header that encodes whether or not it potentially contains instructions... oh, oh no actually the length of any entry is determined by the difference to the subsequence loca offset, so I guess it does need to allow truncation after the header if the contour count is 0.

[mitchellh]

It is confusing, and semantically for our use case it probably doesn't matter. I read this:

however; in particular, it may have instructions that operate on phantom points.

As meaning the instructions part was optional.

[qwerasd205]

Yep, see my edits on my last reply. I'll make a fix for it in an hour or so.