def modify_fonts(font_names,
presentation='emoji',
output=None,
suffix=None,
dst_dir=None,
vs_added=None):
assert dst_dir
if output:
assert len(font_names) == 1
for font_name in font_names:
font = ttLib.TTFont(font_name)
if font_data.get_variation_sequence_cmap(font):
# process no font if any already has a var selector cmap
raise ValueError('font %s already has a format 14 cmap' %
font_name)
if not path.exists(dst_dir):
os.makedirs(dst_dir)
emoji_variants = unicode_data.get_unicode_emoji_variants()
if vs_added:
emoji_variants = emoji_variants | vs_added
for font_name in font_names:
font = ttLib.TTFont(font_name)
modify_font(font_name, font, presentation, emoji_variants)
if output:
new_name = output
else:
new_name = path.basename(font_name)
if suffix:
name, ext = path.splitext(new_name)
new_name = name + suffix + ext
font.save(path.join(dst_dir, new_name))
def modify_fonts(font_names, presentation='emoji', output=None, suffix=None,
dst_dir=None, vs_added=None):
assert dst_dir
if output:
assert len(font_names) == 1
for font_name in font_names:
font = ttLib.TTFont(font_name)
if font_data.get_variation_sequence_cmap(font):
# process no font if any already has a var selector cmap
raise ValueError('font %s already has a format 14 cmap' % font_name)
if not path.exists(dst_dir):
os.makedirs(dst_dir)
emoji_variants = unicode_data.get_unicode_emoji_variants()
if vs_added:
emoji_variants = emoji_variants | vs_added
for font_name in font_names:
font = ttLib.TTFont(font_name)
modify_font(font_name, font, presentation, emoji_variants)
if output:
new_name = output
else:
new_name = path.basename(font_name)
if suffix:
name, ext = path.splitext(new_name)
new_name = name + suffix + ext
font.save(path.join(dst_dir, new_name))
def _create_flag_sets(data_dir):
"""Returns map from flag name to pairs of cp_set, boolean.
These get added to a codepoint name if the the boolean matches
the result of 'cp in cp_set'."""
# These are hardcoded for now, should be able to specify on
# command line... (TODO)
# I propose supporting some emoji in Noto even if they don't have text
# variation sequences proposed, we can remove those for Android if they
# disagree.
emoji_only = (unicode_data.get_emoji() -
unicode_data.get_unicode_emoji_variants('proposed_extra'))
current_sym2_path = path.join(data_dir, 'NotoSansSymbols2-Regular.ttf')
current_sym2 = CodeList.fromfontcmap(current_sym2_path).codeset()
sym2_path = path.join(data_dir, 'notosanssymbols2_cmap.txt')
with open(sym2_path, 'r') as f:
sym2_cmap = f.read()
expect_sym2 = tool_utils.parse_int_ranges(sym2_cmap)
add_sym2 = expect_sym2 - current_sym2
# True means set flag if cp in set, False means set if not in set
flag_sets = {
'ref only': (expect_sym2, False),
'emoji only': (emoji_only, True),
'add': (add_sym2, True),
}
return flag_sets
def _create_codeset_from_expr(expr_list, flag_sets, data_dir, codelist_map):
"""Processes expr_list in order, building a codeset.
See _read_flag_data_from_file for information on expr_list.
This can modify flag_sets and codelist_map."""
result = ()
for op, exp in expr_list:
if exp not in flag_sets:
# its a codelist
codes = _load_codelist(exp, data_dir, codelist_map).codeset()
else:
codes_or_spec = flag_sets[exp]
if isinstance(codes_or_spec, (set, frozenset)):
codes = codes_or_spec
else:
# replace the spec with the actual codes
if codes_or_spec == None:
# we only know about '_emoji_' and '_math_'
if exp == '_emoji_':
codes = (
unicode_data.get_emoji() -
unicode_data.get_unicode_emoji_variants('proposed_extra'))
elif exp == '_math_':
codes = unicode_data.chars_with_property('Math')
else:
raise Exception('unknown special codeset "%s"' % exp)
else:
codes = _load_codelist(
codes_or_spec, data_dir, codelist_map).codeset()
flag_sets[exp] = codes
if op == '|':
if not result:
# it appers that python 'optimizes' |= by replacing the lhs by rhs if
# lhs is an empty set, but this changes the type of lhs to frozenset...
result = set(codes)
else:
result |= codes
elif op == '&':
result &= codes
elif op == '-':
result -= codes
else:
raise Exception('unknown op "%s"' % op)
return result
def subset_symbols(srcdir, dstdir):
"""Subset Noto Sans Symbols in a curated way.
Noto Sans Symbols is now subsetted in a curated way. Changes include:
* Currency symbols now included in Roboto are removed.
* All combining marks for symbols (except for combining keycap) are
added, to combine with other symbols if needed.
* Characters in symbol blocks that are also covered by Noto CJK fonts
are added, for better harmony with the rest of the fonts in non-CJK
settings. The dentistry characters at U+23BE..23CC are not added,
since they appear to be Japan-only and full-width.
* Characters that UTR #51 defines as default text are added, although
they may also exist in the color emoji font, to make sure they get
a default text style.
* Characters that UTR #51 defines as default emoji are removed, to
make sure they don't block the fallback to the color emoji font.
* A few math symbols that are currently included in Roboto are added,
to prepare for potentially removing them from Roboto when they are
lower-quality in Roboto.
Based on subset_noto_sans_symbols.py from AOSP external/noto-fonts."""
# TODO see if we need to change this subset based on Noto Serif coverage
# (so the serif fallback chain would support them)
target_coverage = set()
# Add all characters in BLOCKS_TO_INCLUDE
for first, last, _ in unicode_data._parse_code_ranges(BLOCKS_TO_INCLUDE):
target_coverage.update(range(first, last + 1))
# Add one-off characters
target_coverage |= ONE_OFF_ADDITIONS
# Remove characters preferably coming from Roboto
target_coverage -= LETTERLIKE_CHARS_IN_ROBOTO
# Remove default emoji presentation (including ones Android prefers default)
target_coverage -= EMOJI
# Remove COMBINING ENCLOSING KEYCAP. It's needed for Android's color emoji
# mechanism to work properly
target_coverage.remove(0x20E3)
# Remove dentistry symbols, as their main use appears to be for CJK:
# http://www.unicode.org/L2/L2000/00098-n2195.pdf
target_coverage -= set(range(0x23BE, 0x23CC + 1))
for font_file in glob.glob(path.join(srcdir, 'NotoSansSymbols-*.ttf')):
print 'main subset', font_file
out_file = path.join(dstdir,
path.basename(font_file)[:-4] + '-Subsetted.ttf')
subset.subset_font(font_file, out_file, include=target_coverage)
# The second subset will be a fallback after the color emoji, for
# explicit text presentation sequences.
target_coverage = EMOJI | unicode_data.get_unicode_emoji_variants()
for font_file in glob.glob(path.join(srcdir, 'NotoSansSymbols-*.ttf')):
print 'secondary subset', font_file
out_file = path.join(dstdir,
path.basename(font_file)[:-4] + '-Subsetted2.ttf')
subset.subset_font(font_file, out_file, include=target_coverage)
def subset_symbols(srcdir, dstdir):
"""Subset Noto Sans Symbols in a curated way.
Noto Sans Symbols is now subsetted in a curated way. Changes include:
* Currency symbols now included in Roboto are removed.
* All combining marks for symbols (except for combining keycap) are
added, to combine with other symbols if needed.
* Characters in symbol blocks that are also covered by Noto CJK fonts
are added, for better harmony with the rest of the fonts in non-CJK
settings. The dentistry characters at U+23BE..23CC are not added,
since they appear to be Japan-only and full-width.
* Characters that UTR #51 defines as default text are added, although
they may also exist in the color emoji font, to make sure they get
a default text style.
* Characters that UTR #51 defines as default emoji are removed, to
make sure they don't block the fallback to the color emoji font.
* A few math symbols that are currently included in Roboto are added,
to prepare for potentially removing them from Roboto when they are
lower-quality in Roboto.
Based on subset_noto_sans_symbols.py from AOSP external/noto-fonts."""
# TODO see if we need to change this subset based on Noto Serif coverage
# (so the serif fallback chain would support them)
target_coverage = set()
# Add all characters in BLOCKS_TO_INCLUDE
for first, last, _ in unicode_data._parse_code_ranges(BLOCKS_TO_INCLUDE):
target_coverage.update(range(first, last+1))
# Add one-off characters
target_coverage |= ONE_OFF_ADDITIONS
# Remove characters preferably coming from Roboto
target_coverage -= LETTERLIKE_CHARS_IN_ROBOTO
# Remove default emoji presentation (including ones Android prefers default)
target_coverage -= EMOJI
# Remove COMBINING ENCLOSING KEYCAP. It's needed for Android's color emoji
# mechanism to work properly
target_coverage.remove(0x20E3)
# Remove dentistry symbols, as their main use appears to be for CJK:
# http://www.unicode.org/L2/L2000/00098-n2195.pdf
target_coverage -= set(range(0x23BE, 0x23CC+1))
for font_file in glob.glob(path.join(srcdir, 'NotoSansSymbols-*.ttf')):
print 'main subset', font_file
out_file = path.join(
dstdir, path.basename(font_file)[:-4] + '-Subsetted.ttf')
subset.subset_font(font_file, out_file, include=target_coverage)
# The second subset will be a fallback after the color emoji, for
# explicit text presentation sequences.
target_coverage = EMOJI | unicode_data.get_unicode_emoji_variants()
for font_file in glob.glob(path.join(srcdir, 'NotoSansSymbols-*.ttf')):
print 'secondary subset', font_file
out_file = path.join(
dstdir, path.basename(font_file)[:-4] + '-Subsetted2.ttf')
subset.subset_font(font_file, out_file, include=target_coverage)