Re: [Fontforge-users] how to subset a complex font without inadvertently breaking it?

[James C. <cl...@jh...>]

>>>>> "AT" == Aaron Turner <aar...@ma...> writes:

AT> I have a large number of fonts (all downloaded from "copyleft" sources 
AT> such as Google fonts, SIL, etc) and I wish to use Fontforge/Python to 
AT> generate specific Unicode subsets of these fonts, for WOFF / @font-face 
AT> purposes.

I'd take a look at cairo's code for subsetting.  Perhaps also pdftex
and/or luatex.

Of course, there are some table those projects do not need to emit which
you will need to, so they can only be a starting point.

The problem boils down to creating a graph of how each entry of each
table interacts, weeding out what you do not want, and saving the rest.

After forming the graph, mark each needed node starting with the char-
acters you want, and recusively drilling down a la find(1) to every node
which those nodes depend on (glyphs, GPOS, GSUB, et al).

As the tables shrink, the IDs and offsets of their contents change; you
have to keep track of those changes.  You might find some ideas for this
step from tree-based filesystems like btrfs and tux3.  Even if you use
fontforge (or libfontforge) to do everything to ff's sfd representation,
I suspect you'll still need to keep track of renumbering once you know
what you want to keep and elide.

-JimC
-- 
James Cloos <cl...@jh...>         OpenPGP: 1024D/ED7DAEA6