path: root/scripts/subset-fonts.pe

#!/usr/bin/env fontforge

# Prerequisites:
#
# - fontforge: https://fontforge.github.io/en-US/
#
# Run using:
#
# $ ./subset-fonts.pe <source directory> <output directory>
#
# Example:
#
# $ ./subset-fonts.pe source-fonts ../data/fonts
#
# NOTE: Ignore "GID out of range" warnings; fontforge has to kick the tires a bit to flush out these dead references.
#
# Use with Noto Serif fonts from https://code.google.com/p/noto/source/browse/#svn%2Ftrunk%2Ffonts%2Findividual%2Funhinted
# Use with M+ fonts from http://mplus-fonts.sourceforge.jp/mplus-outline-fonts/download/index-en.html
#
# See Unicode Blocks: http://jrgraphix.net/research/unicode_blocks.php
#
# IMPORTANT: Must generate Apple format (0x10) or include old-fashioned 'kern' table for kerning to work in Prawn.
# Prawn misses some of the kern pairs when using the Apply format, so the old-fashioned 'kern' table is best (0x90).
# Keep in mind, though, that the old-fashioned 'kern' table only supported a limited number of pairs.
#
# generate flags (additive):
# * 0x00 - OpenType
# * 0x10 - Apple
# * 0x80 - OpenType and Apple
# * 0x90 - Neither OpenType or Apple (implicitly generates an old-style 'kern' table)
# * 0x800 - Generate old-style 'kern' table
# * 0x08 - exclude TrueType instructions
# * 0x04 - exclude PS glyph names; short format only (PDF readers use these names to guess characters when copying text)
# * 0x4000000 - drop glyphs with Unicode value of -1
#
# NOTE best choice for Prawn only is 0x90 + 0x08
# NOTE best choice for web only is 0x00 + 0x08
# NOTE best choice for Prawn & web is 0x00 + 0x800 + 0x08

genflags = 0x90 + 0x08
copy_fonts = ["NotoSerif-Regular.ttf", "NotoSerif-Bold.ttf", "NotoSerif-Italic.ttf", "NotoSerif-BoldItalic.ttf", "mplus-1p-regular.ttf"]
# FIXME reenable copy_scripts to be customized at top level; disabled due to support for mplus1p-regular-fallback
# copy_scripts = ["subset"]
code_fonts = ["mplus-1mn-light.ttf", "mplus-1mn-regular.ttf", "mplus-1mn-medium.ttf", "mplus-1mn-bold.ttf"]
code_scripts = ["ascii"]

if ($argc >= 2)
  source_dir = $argv[1]
else
  source_dir = "."
endif
if ($argc >= 3)
  output_dir = $argv[2]
else
  output_dir = "generated"
endif

fa_src_filepath = StrJoin([source_dir, "fontawesome-webfont.ttf"], "/")
mplus1p_src_filepath = StrJoin([source_dir, "mplus-1p-regular.ttf"], "/")

fi = 0
num_copy_fonts = SizeOf(copy_fonts)
while (fi < num_copy_fonts)
  copy_font = copy_fonts[fi]
  src_filepath = StrJoin([source_dir, copy_font], "/")
  new_basename = ToLower(copy_font:r)
  if (Strstr(new_basename, "bolditalic") >= 0)
    new_basename = StrJoin(StrSplit(new_basename, "bolditalic"), "bold_italic")
  endif
  is_mplus = Strstr(new_basename, "mplus") >= 0
  is_noto = Strstr(new_basename, "noto") >= 0
  copy_scripts = ["subset"]
  # remove hyphen from mplus-1 file basename
  if (is_mplus)
    new_basename = "mplus1" + StrJoin(StrSplit(new_basename, "mplus-1"), "")
    copy_scripts = ["fallback"]
  endif
  si = 0
  num_copy_scripts = SizeOf(copy_scripts)
  while (si < num_copy_scripts)
    script = copy_scripts[si]
    new_filename = new_basename + "-" + script + ".ttf"
    new_filepath = output_dir + "/" + new_filename
    if (is_noto)
      Print("Stealing glyphs from other fonts...")
      if (new_basename == "notoserif-regular")
        Print("Stealing ballot boxes from FontAwesome...")
        # Grab ballot boxes from FontAwesome
        Open(fa_src_filepath)
        # relocate 0uf046 -> 0u2611
        Select(0uf046); Copy(); Select(0u2611); Paste()
        # relocate 0uf096 -> 0u2610
        Select(0uf096); Copy(); Select(0u2610); Paste()
        # select and copy ballot boxes
        Select(0u2610,0u2611)
        Copy()
        Close()
        Open(src_filepath)
        Select(0u2610,0u2611)
        Paste()
        SetWidth(1664)
        SelectNone()
        # NOTE it shouldn't be necessary to write the file here, but for some reason it doesn't work otherwise in this script
        Generate(new_filepath, "", genflags)
        src_filepath = new_filepath
        Close()
      endif
      Print("Stealing double arrows from M+ 1p...")
      Open(mplus1p_src_filepath)
      from_em = $em
      SelectSingletons(0u21d0,0u21d2)
      Copy()
      Close()
      Open(src_filepath)
      SelectSingletons(0u21d0,0u21d2)
      Paste()
      # TODO to be strictly accurate, multiply $em by 1.0
      scale_factor = ($em / from_em) * 100
      Scale(scale_factor, scale_factor)
      SetWidth(2048)
      CenterInWidth()
      # Move single and double arrows up to align with middle of X
      SelectMoreSingletons(0u2190,0u2192)
      Move(0, 380)
      SelectNone()
      Print("Done stealing glyphs from other fonts")
    else
      Open(src_filepath)
    endif
    SelectAll()
    # Remove TrueType instructions (i.e., hinting)
    ClearInstrs()
    SelectNone()
    # Basic Latin (e.g., English)
    SelectMore(0u0020,0u007e)
    # Latin-1 Supplement (covers core Western European languages)
    SelectMore(0u00a1,0u00fd)
    # Latin Extended-A (covers Czech, Dutch, Polish & Turkish, esp. names)
    SelectMore(0u0100,0u017f)
    # General Punctuation (most of it) (e.g., dashes, curved quotes, bullet, ellipsis)
    SelectMore(0u2000,0u203a)
    # More picky general punctuation
    ## Spaces
    #SelectMore(0u2000,0u200b)
    ## Dashes
    #SelectMore(0u2012,0u2015)
    ## Curved quotes
    #SelectMore(0u2018,0u2019)
    #SelectMore(0u201c,0u201d)
    #SelectMore(0u2039,0u203a)
    ## Daggars
    #SelectMore(0u2020,0u2021)
    ## Bullet
    #SelectMore(0u2022)
    ## Ellipsis
    #SelectMore(0u2026)
    # Additional Currency Symbols
    #SelectMore(0u20a0,0u20d0)
    # ...or just the Euro sign
    SelectMore(0u20ac)
    # Trademark sign (selected from Letterlike Symbols set)
    SelectMore(0u2122)
    # Mathematical Operators (e.g., infinity, sum, partial differential)
    SelectMore(0u2200,0u22ff)
    # Geometric Shapes (e.g., list bullets)
    SelectMore(0u25a0,0u25ff)
    # Greek (frequently used for math and bullets)
    SelectMore(0u0370,0u03ff)
    if (is_noto)
      # Single arrows (present in Noto Serif, but misaligned) and double arrows (imported from M+ 1p)
      SelectMoreSingletons(0u2190,0u2192,0u21d0,0u21d2)
      if (new_basename == "notoserif-regular")
        # Ballot boxes (imported from FontAwesome)
        SelectMore(0u2610,0u2611)
      endif
      # Check mark (missing from Noto Serif)
      #SelectMore(0u2713)
    endif
    if (is_mplus)
      # Single arrows
      SelectMore(0u2190,0u2195)
      # Double arrows
      SelectMore(0u21d0,0u21d5)
      # Ballot boxes
      SelectMore(0u2610,0u2611)
      # Check mark
      SelectMore(0u2713)
    endif
    if (script == "latin-ext" || script == "latin-ext-cyrillic" || script == "fallback")
      # Latin Extended-B
      SelectMore(0u0180,0u024f)
      # IPA Extensions (i.e., core phonetics)
      #SelectMore(0u0250,0u02af)
      # Upside-down e (from IPA Extensions)
      SelectMore(0u0259)
      # Spacing Modifier Letters (i.e., IPA tone marks, and modifiers for aspiration and palatalization) (missing from Noto Serif)
      SelectMore(0u02b0,0u02ff)
      # Latin Ligatures (e.g., fi) (Noto Serif doesn't auto-detect them, so leave them off)
      #SelectMore(0ufb00,0ufb06)
    endif
    if (script == "latin-cyrillic" || script == "latin-ext-cyrillic" || script == "subset" || script == "fallback")
      # Cyrillic
      SelectMore(0u0400,0u04ff)
    endif
    if (script == "subset" || script == "fallback")
      # Non-optimal selection for Vietnamese
      # Latin Extended-A, Latin Extended Additional
      #SelectMore(0u0100,0u017f)
      #SelectMore(0u1e00,0u1eff)
      # Optimal selection for Vietnamese (see http://blog.int3ractive.com/2010/06/optimal-unicode-range-for-vietnamese.html)
      # NOTE Latin Extended-A may already included at this point, so 0u0102-0u0169 may be redundant
      SelectMore(0u0102,0u0103)
      SelectMore(0u0110,0u0111)
      SelectMore(0u0128,0u0129)
      SelectMore(0u0168,0u0169)
      SelectMore(0u01a0,0u01b0)
      SelectMore(0u1ea0,0u1ef9)
      if (is_mplus)
        # CJK Symbols and Punctuation (not present in mainstream Noto Serif fonts)
        SelectMore(0u3000,0u303f)
        # Hiragana
        SelectMore(0u3040,0u309f)
        # Katakana
        SelectMore(0u30a0,0u30ff)
        # Full-width roman characters and half-width katakana
        SelectMore(0uff00,0uffef)
        # CJK Unified Ideographs (for Japanese, aka kanji) (not present in mainstream Noto Serif fonts)
        SelectMore(0u4e00,0u9faf)
      endif
    endif
    #if (script == "fallback")
    #  # Cyrillic Supplement
    #  #SelectMore(0u0500,0u052f)
    #  # Greek Extended (i.e., Polytonic)
    #  #SelectMore(0u1f00,0u1fff)
    #  # Or just select all them symbols...
    #  SelectAll()
    #endif
    # BOM (zero-width no-break space) and no-break space
    SelectMoreSingletons(0ufeff,0u00a0)
    # Drop all glyphs that weren't selected
    SelectInvert()
    Clear()
    SelectNone()
    if (is_mplus)
      # Generate BOM (zero-width no-break space), zero-width space, and word joiner from no-break space (for M+ fonts)
      Select(0u00a0)
      Copy()
      SelectNone()
      SelectSingletons(0ufeff,0u200b,0u2060)
      Paste()
      SetWidth(0)
      SelectNone()
      # Generate narrow no-break space from thin space (for M+ fonts)
      Select(0u2009)
      Copy()
      SelectNone()
      Select(0u202f)
      Paste()
      SetWidth(226)
      # Generate hair space from thin space (for M+ fonts)
      Select(0u2009)
      Copy()
      SelectNone()
      Select(0u200a)
      Paste()
      SetWidth(94)
      SelectNone()
    else
      # Generate word joiner from BOM (zero-width no-break space)
      Select(0ufeff)
      Copy()
      Select(0u2060)
      Paste()
      SelectNone()
      # Generate no-break hyphen from hyphen (for Noto Serif fonts)
      Select(0u002d)
      Copy()
      SelectNone()
      Select(0u2011)
      Paste()
      SelectNone()
      # Fix width of .null character
      #Select(0u0000)
      #SetWidth(0)
      #SelectNone()
    endif

    # Generate line feed from no-break space (works around error "cmap format 14 is not supported" in ttfunk)
    # FIXME another option here is to select all the characters referenced by the cmap format 14 table
    Select(0u00a0)
    Copy()
    SelectNone()
    Select(0u000a)
    Paste()
    SetWidth(0)
    SelectNone()

    Print("Generating " + new_filename + "...")
    Generate(new_filepath, "", genflags)
    Close()
    if (is_mplus)
      # Regenerate font to drop invalid cmap format 14 table (ignore warnings)
      Open(new_filepath)
      Generate(new_filepath, "", genflags)
      Close()
    endif
    si = si + 1
  endloop
  fi = fi + 1
endloop

fi = 0
num_code_fonts = SizeOf(code_fonts)
while (fi < num_code_fonts)
  code_font = code_fonts[fi]
  src_filepath = StrJoin([source_dir, code_font], "/")
  new_basename_base = code_font:r
  # remove hyphen from mplus-1 file basename
  new_basename_base = "mplus1" + StrJoin(StrSplit(new_basename_base, "mplus-1"), "")
  si = 0
  num_code_scripts = SizeOf(code_scripts)
  while (si < num_code_scripts)
    new_basename = new_basename_base
    script = code_scripts[si]
    new_suffix = "-" + script + ".ttf"
    Open(src_filepath)
    SelectAll()
    # NOTE: M+ fonts don't have hinting, so technically this is redundant
    ClearInstrs()
    SelectNone()
    # Basic Latin (e.g., English)
    SelectMore(0u0020,0u007e)
    # No-break space
    SelectMore(0u00a0)
    # Box drawing symbols (for unix `tree` output)
    SelectMore(0u2500,0u257f)
    if (new_basename == "mplus1mn-regular")
      # Enclosed numbers (1-20 circled and filled)
      SelectMore(0u2460,0u2473)
      SelectMore(0u2776,0u277f)
      SelectMore(0u24eb,0u24f4)
      if (script == "ascii")
        new_suffix = "-" + script + "-conums.ttf"
      endif
    endif
    SelectInvert()
    Clear()
    SelectNone()
    SetFontNames(new_basename, "M+ 1mn")
    # repurpose light as italic
    if (new_basename == "mplus1mn-light")
      SetFontNames("mplus1mn-italic", "M+ 1mn", "M+ 1mn Italic")
      SetOS2Value("Weight", 400)
      SetPanose(2, 5)
      SetTTFName(0x409, 2, "Italic")
      SetTTFName(0x409, 16, "")
      SetTTFName(0x409, 17, "")
      SetTTFName(0x411, 16, "")
      SetTTFName(0x411, 17, "")
      new_basename = "mplus1mn-italic"
    # repurpose medium as bold
    elseif (new_basename == "mplus1mn-medium")
      SetFontNames("mplus1mn-bold", "M+ 1mn", "M+ 1mn Bold")
      SetOS2Value("Weight", 700)
      SetPanose(2, 8)
      SetTTFName(0x409, 2, "Bold")
      SetTTFName(0x409, 16, "")
      SetTTFName(0x409, 17, "")
      SetTTFName(0x411, 16, "")
      SetTTFName(0x411, 17, "")
      new_basename = "mplus1mn-bold"
    # repurpose bold as bold italic
    elseif (new_basename == "mplus1mn-bold")
      SetFontNames("mplus1mn-bold_italic", "M+ 1mn", "M+ 1mn Bold Italic")
      SetOS2Value("Weight", 700)
      SetPanose(2, 8)
      SetTTFName(0x409, 2, "Bold Italic")
      SetTTFName(0x409, 16, "")
      SetTTFName(0x409, 17, "")
      SetTTFName(0x411, 16, "")
      SetTTFName(0x411, 17, "")
      new_basename = "mplus1mn-bold_italic"
    endif

    ## Adjust width of box drawing symbols (not working)
    #Select(0u2500,0u257f)
    #SetWidth(50, 2)
    #SelectNone()

    # Generate BOM (zero-width no-break space) from no-break space
    Select(0u00a0)
    Copy()
    SelectNone()
    Select(0ufeff)
    Paste()
    SetWidth(0)
    SelectNone()

    # Generate line feed from no-break space
    Select(0u00a0)
    Copy()
    SelectNone()
    Select(0u000a)
    Paste()
    SetWidth(0)
    SelectNone()

    new_filename = new_basename + new_suffix
    new_filepath = output_dir + "/" + new_filename
    Print("Generating " + new_filename + "...")
    Generate(new_filepath, "", genflags)
    Close()

    # Regenerate font to drop invalid cmap format 14 table (ignore warnings)
    Open(new_filepath)
    Generate(new_filepath, "", genflags)
    Close()
    si = si + 1
  endloop

  fi = fi + 1
endloop