Very cool project but I fear that this approach will only work for relatively simple glyphs that can be broken down into 3 sections (main, :before and :after) whereas Chinese characters, for example, may take up a few dozen strokes.