# emmentaler_kerning.py # # This file is part of LilyPond, the GNU music typesetter. # # Copyright (C) 2022--2023 Werner Lemberg # # LilyPond is free software: you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation, either version 3 of the License, or # (at your option) any later version. # # LilyPond is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with LilyPond. If not, see . """ OpenType kerning for LilyPond's Emmentaler Font """ import fontforge # It is no longer trivial to import kerning data from # `feta-alphabetXXX.tfm` because we have kerning values for glyphs # that don't get assigned to input character code points by default. # Additionally, we use OpenType font features. For these reasons we # add kerning to the Emmentaler fonts while post-processing the # Metafont subfonts. # # * `feta-numbers.mf` originally contained kerning data for time # signature digits. # # ``` # [number_design_size = design_size / 2] # # space# := number_design_size / 2; # # ligtable "3": # "3" kern 0.1 space#, # "0" kern 0.1 space#; # # ligtable "2": # "7" kern 0.15 space#; # ``` # # * `feta-dynamics.mf` originally contained kerning data for letters # used in dynamics. # # ``` # [pt# = 1] # [dynamic_design_size = 14 design_size / 20] # # ex# := (dynamic_design_size / 2.4) * pt#; # horizontal_space# := .66 ex#; # # fet_beginchar ("space", "space"); # set_char_box (0, horizontal_space#, 0, ex#); # fet_endchar; # # ligtable "m": # "p" kern 0.2 ex#, # "f" kern -0.1 ex#; # # ligtable "n": # "p" kern 0.2 ex#, # "f" kern -0.1 ex#; # # ligtable "f": # "f" kern -0.13 ex#; # # ligtable "r": # "f" kern 0.1 ex#; # ``` # # Based on this, we use the width of the 'space' character (which can # be found in the intermediate PFB file and is thus accessible by # FontForge) to derive kern values. # # We change the names of `horizontal_space#`, `space#`, and `ex#` to # the more meaningful names `SW` (space width), `NKU` (numbers kerning # unit), and `DKU` (dynamics kerning unit), respectively. We express # `NKU` and `DKU` as multiples of `SW`. # # ``` # ex# = (dynamic_design_size / 2.4) * pt# # = (14 design_size / 20) / 2.4 # = design_size * 14/20 * 10/24 # = design_size * 7/2 * 1/12 # = 7/24 design_size # # SW = .66 ex# = 33/50 DKU # DKU = 50/33 SW # # design_size = 24/7 * ex# # = 24/7 * 50/33 SW # = 8/7 * 50/11 SW # = 400/77 SW # # space# = number_design_size / 2 # = design_size / 4 # NKU = 100/77 SW # ``` # The kerning should always be applied regardless of script and # language; we thus use the 'DFLT' script and 'dflt' (dummy) language # tags. def add_feature_kern(font): # TODO: Check whether kerning should also depend on design size. SW = font["space"].width DKU = 50.0/33.0 * SW NKU = 100.0/77.0 * SW # It is easier to keep track of kerning by our own than to use # the generic FontForge API to get data from the 'GPOS' table. kerning = {} font.addLookup("kern", "gpos_pair", 0, (("kern", (("DFLT", ("dflt")), )), )) font.addLookupSubtable("kern", "kern") # The left side bearing factors are taken from `feta-numbers.mf`. lsb_factors = { "zero": 1.0, "one": 1.2, "two": 1.0, "three": 1.0, "four": 1.0, "four.alt": 1.0, # same value as 'four' "five": 0.7, "six": 1.0, "seven": 1.2, "seven.alt": 1.2, # same value as 'seven' "eight": 1.0, "nine": 1.0, # A zero lsb factor indicates zero lsb and rsb values. We # also use this to distinguish between digits and non-digits # in this dictionary. "hyphen": 0.0, "plus": 0.0, "period": 0.0, "comma": 0.0 } lsb_factors_keys = sorted(lsb_factors.keys()) def kern(left, right, val): v = int(round(val)) kerning[(left, right)] = v font[left].addPosSub("kern", right, 0, 0, v, 0, 0, 0, 0, 0) # First, the standard kerning pairs for digits and punctuation. kern("zero", "zero", 0.15 * NKU) kern("zero", "one", 0.1 * NKU) kern("zero", "five", 0.05 * NKU) kern("zero", "six", 0.1 * NKU) kern("zero", "seven", 0.1 * NKU) kern("zero", "eight", 0.05 * NKU) kern("zero", "nine", 0.1 * NKU) kern("zero", "hyphen", 0.1 * NKU) kern("zero", "plus", 0.1 * NKU) kern("one", "zero", -0.1 * NKU) kern("one", "four", -0.1 * NKU) kern("one", "six", -0.1 * NKU) kern("one", "period", 0.1 * NKU) kern("one", "comma", 0.1 * NKU) kern("two", "one", 0.1 * NKU) kern("two", "three", 0.1 * NKU) kern("two", "six", 0.1 * NKU) kern("two", "seven", 0.15 * NKU) kern("two", "eight", 0.05 * NKU) kern("two", "nine", 0.1 * NKU) kern("two", "period", 0.05 * NKU) kern("two", "comma", 0.05 * NKU) kern("three", "one", 0.1 * NKU) kern("three", "zero", 0.1 * NKU) kern("three", "two", 0.05 * NKU) kern("three", "three", 0.1 * NKU) kern("three", "five", 0.1 * NKU) kern("three", "six", 0.1 * NKU) kern("three", "seven", 0.1 * NKU) kern("three", "eight", 0.1 * NKU) kern("three", "nine", 0.1 * NKU) kern("three", "hyphen", 0.1 * NKU) kern("three", "period", 0.1 * NKU) kern("three", "comma", 0.1 * NKU) kern("four", "hyphen", 0.1 * NKU) kern("four", "period", 0.1 * NKU) kern("four", "comma", 0.1 * NKU) kern("five", "zero", 0.15 * NKU) kern("five", "one", 0.15 * NKU) kern("five", "two", 0.1 * NKU) kern("five", "three", 0.15 * NKU) kern("five", "four", 0.1 * NKU) kern("five", "five", 0.15 * NKU) kern("five", "six", 0.15 * NKU) kern("five", "seven", 0.15 * NKU) kern("five", "eight", 0.15 * NKU) kern("five", "nine", 0.15 * NKU) kern("five", "plus", 0.1 * NKU) kern("five", "hyphen", 0.1 * NKU) kern("five", "period", 0.1 * NKU) kern("five", "comma", 0.1 * NKU) kern("six", "zero", 0.1 * NKU) kern("six", "one", 0.05 * NKU) kern("six", "two", 0.15 * NKU) kern("six", "three", 0.1 * NKU) kern("six", "five", 0.1 * NKU) kern("six", "six", 0.1 * NKU) kern("six", "seven", 0.1 * NKU) kern("six", "eight", 0.1 * NKU) kern("six", "nine", 0.1 * NKU) kern("six", "hyphen", 0.1 * NKU) kern("six", "period", 0.1 * NKU) kern("six", "comma", 0.1 * NKU) kern("seven", "four", -0.2 * NKU) kern("seven", "five", 0.1 * NKU) kern("seven", "seven", 0.15 * NKU) kern("seven", "nine", 0.1 * NKU) kern("seven", "hyphen", -0.05 * NKU) kern("seven", "period", -0.1 * NKU) kern("seven", "comma", -0.1 * NKU) kern("eight", "zero", 0.1 * NKU) kern("eight", "one", 0.1 * NKU) kern("eight", "two", 0.05 * NKU) kern("eight", "three", 0.1 * NKU) kern("eight", "four", 0.05 * NKU) kern("eight", "five", 0.1 * NKU) kern("eight", "six", 0.1 * NKU) kern("eight", "seven", 0.1 * NKU) kern("eight", "eight", 0.1 * NKU) kern("eight", "nine", 0.1 * NKU) kern("eight", "plus", 0.1 * NKU) kern("eight", "hyphen", 0.1 * NKU) kern("eight", "period", 0.1 * NKU) kern("eight", "comma", 0.1 * NKU) kern("nine", "zero", 0.15 * NKU) kern("nine", "one", 0.15 * NKU) kern("nine", "two", 0.05 * NKU) kern("nine", "three", 0.1 * NKU) kern("nine", "six", 0.1 * NKU) kern("nine", "seven", 0.15 * NKU) kern("nine", "eight", 0.05 * NKU) kern("nine", "nine", 0.1 * NKU) kern("nine", "plus", 0.1 * NKU) kern("nine", "hyphen", 0.1 * NKU) kern("hyphen", "zero", 0.1 * NKU) kern("hyphen", "one", 0.1 * NKU) kern("hyphen", "three", 0.1 * NKU) kern("hyphen", "four", 0.1 * NKU) kern("hyphen", "five", 0.1 * NKU) kern("hyphen", "six", 0.1 * NKU) kern("hyphen", "seven", 0.1 * NKU) kern("hyphen", "eight", 0.1 * NKU) kern("hyphen", "nine", 0.1 * NKU) kern("plus", "zero", 0.1 * NKU) kern("plus", "one", 0.05 * NKU) kern("plus", "six", 0.05 * NKU) kern("plus", "seven", 0.1 * NKU) kern("period", "one", 0.1 * NKU) kern("period", "two", 0.15 * NKU) kern("period", "three", 0.1 * NKU) kern("period", "five", 0.1 * NKU) kern("period", "seven", 0.05 * NKU) kern("period", "eight", 0.1 * NKU) kern("period", "nine", 0.1 * NKU) kern("period", "period", 0.15 * NKU) kern("period", "comma", 0.15 * NKU) kern("comma", "one", 0.1 * NKU) kern("comma", "two", 0.15 * NKU) kern("comma", "three", 0.1 * NKU) kern("comma", "five", 0.1 * NKU) kern("comma", "seven", 0.05 * NKU) kern("comma", "eight", 0.1 * NKU) kern("comma", "nine", 0.1 * NKU) kern("comma", "period", 0.15 * NKU) kern("comma", "comma", 0.15 * NKU) # Kerning for alternative digits forms; they are identical to the # standard glyphs. for left in lsb_factors_keys: for right in lsb_factors_keys: if (not (lsb_factors[left] or lsb_factors[right])): continue if (not (left == "four" or left == "seven" or right == "four" or right == "seven")): continue if (left, right) in kerning: kern_val = kerning[(left, right)] left_suffix = ".alt" \ if (left == "four" or left == "seven") else "" right_suffix = ".alt" \ if (right == "four" or right == "seven") else "" kern(left + left_suffix, right + right_suffix, kern_val) # Compute kern values for 'fattened' digits. They are the same # as for normal digits. for left in lsb_factors_keys: for right in lsb_factors_keys: if (not (lsb_factors[left] or lsb_factors[right])): continue if (left, right) in kerning: kern_val = kerning[(left, right)] left_prefix = "fattened." if lsb_factors[left] else "" right_prefix = "fattened." if lsb_factors[right] else "" kern(left_prefix + left, right_prefix + right, kern_val) # For fixed-width digits (both normal and fattened) the goal is to # make all such digit pairs have the same horizontal distance as # non-fixed-width digits if the 'kern' feature is active. # # +---------+ # | | # | glyph | # | | # +---------+ # # |-------|---------|-------| # lsb digit rsb # width # |-------------------------| # fixed-width digit width fixed_width_digit_width = font["four"].width def lsb(glyph): width = font[glyph].width lsb_factor = lsb_factors[glyph] if (lsb_factor == 0.0): return 0 return lsb_factor * (fixed_width_digit_width - width) / 2.0 def rsb(glyph): width = font[glyph].width lsb_factor = lsb_factors[glyph] if (lsb_factor == 0.0): return 0 return fixed_width_digit_width - width - lsb(glyph) for left in lsb_factors_keys: for right in lsb_factors_keys: if (not (lsb_factors[left] or lsb_factors[right])): continue if (left, right) in kerning: kern_val = kerning[(left, right)] else: kern_val = 0 correction = rsb(left) + lsb(right) figbass_kern = kern_val - correction if (figbass_kern == 0.0): continue left_prefix = "fixedwidth." if lsb_factors[left] else "" right_prefix = "fixedwidth." if lsb_factors[right] else "" kern(left_prefix + left, right_prefix + right, figbass_kern) left_prefix = "fattened.fixedwidth." if lsb_factors[left] else "" right_prefix = "fattened.fixedwidth." if lsb_factors[right] else "" kern(left_prefix + left, right_prefix + right, figbass_kern) # Finally, the kerning for the few letters. kern("f", "f", -0.13 * DKU) kern("m", "p", 0.2 * DKU) kern("m", "f", -0.1 * DKU) kern("n", "p", 0.2 * DKU) kern("n", "f", -0.1 * DKU) kern("r", "f", 0.1 * DKU) kern("s", "p", 0.3 * DKU) kern("z", "p", 0.2 * DKU) # eof