def __init__(self, raw_or_get_table):
self.tables = {}
if isinstance(raw_or_get_table, bytes):
raw = raw_or_get_table
self.sfnt_version = raw[:4]
if self.sfnt_version not in {
b'\x00\x01\x00\x00', b'OTTO', b'true', b'type1'
}:
raise UnsupportedFont('Font has unknown sfnt version: %r' %
self.sfnt_version)
for table_tag, table, table_index, table_offset, table_checksum in get_tables(
raw):
self.tables[table_tag] = self.TABLE_MAP.get(
table_tag, UnknownTable)(table)
else:
for table_tag in {
b'cmap', b'hhea', b'head', b'hmtx', b'maxp', b'name',
b'OS/2', b'post', b'cvt ', b'fpgm', b'glyf', b'loca',
b'prep', b'CFF ', b'VORG', b'EBDT', b'EBLC', b'EBSC',
b'BASE', b'GSUB', b'GPOS', b'GDEF', b'JSTF', b'gasp',
b'hdmx', b'kern', b'LTSH', b'PCLT', b'VDMX', b'vhea',
b'vmtx', b'MATH'
}:
table = bytes(raw_or_get_table(table_tag))
if table:
self.tables[table_tag] = self.TABLE_MAP.get(
table_tag, UnknownTable)(table)
if not self.tables:
raise UnsupportedFont('This font has no tables')
self.sfnt_version = (b'\0\x01\0\0'
if b'glyf' in self.tables else b'OTTO')
def subset_truetype(sfnt, character_map, extra_glyphs):
loca = sfnt[b'loca']
glyf = sfnt[b'glyf']
try:
head, maxp = sfnt[b'head'], sfnt[b'maxp']
except KeyError:
raise UnsupportedFont(
'This font does not contain head and/or maxp tables')
loca.load_offsets(head, maxp)
resolved_glyphs = resolve_glyphs(loca, glyf, character_map, extra_glyphs)
if not resolved_glyphs or set(resolved_glyphs) == {0}:
raise NoGlyphs('This font has no glyphs for the specified character '
'set, subsetting it is pointless')
# Keep only character codes that have resolved glyphs
for code, glyph_id in tuple(character_map.items()):
if glyph_id not in resolved_glyphs:
del character_map[code]
# Update the glyf table
glyph_offset_map = glyf.update(resolved_glyphs)
# Update the loca table
loca.subset(glyph_offset_map)
head.index_to_loc_format = 0 if loca.fmt == 'H' else 1
head.update()
maxp.num_glyphs = len(loca.offset_map) - 1
def ExtensionSubstitution(raw, offset, subtable_map={}):
data = Unpackable(raw, offset)
subst_format, extension_lookup_type, offset = data.unpack('2HL')
if subst_format != 1:
raise UnsupportedFont(
'ExtensionSubstitution has unknown format: 0x%x' % subst_format)
return subtable_map[extension_lookup_type](raw, offset + data.start_pos)
def read_extra_header(self, data):
self.feature_params = data.unpack('H')
if False and self.feature_params != 0:
# Source code pro sets this to non NULL
raise UnsupportedFont(
'This FeatureTable has non NULL FeatureParams: 0x%x' %
self.feature_params)
def read_data(self, vmtx):
if hasattr(self, 'ascender'):
return
field_types = (
'_version_number',
'l',
'ascender',
'h',
'descender',
'h',
'line_gap',
'h',
'advance_height_max',
'H',
'min_top_side_bearing',
'h',
'min_bottom_side_bearing',
'h',
'y_max_extent',
'h',
'caret_slope_rise',
'h',
'caret_slop_run',
'h',
'caret_offset',
'h',
'r1',
'h',
'r2',
'h',
'r3',
'h',
'r4',
'h',
'metric_data_format',
'h',
'number_of_v_metrics',
'H',
)
self._fmt = ('>%s' % (''.join(field_types[1::2]))).encode('ascii')
self._fields = field_types[0::2]
for f, val in zip(self._fields, unpack_from(self._fmt, self.raw)):
setattr(self, f, val)
raw = vmtx.raw
num = self.number_of_v_metrics
if len(raw) < 4 * num:
raise UnsupportedFont('The vmtx table has insufficient data')
long_hor_metric = raw[:4 * num]
long_hor_metric = raw[:4 * num]
a = read_array(long_hor_metric)
self.advance_heights = a[0::2]
a = read_array(long_hor_metric, 'h')
self.top_side_bearings = a[1::2]
def get_glyph_map(self, glyph_ids):
'''
Get a mapping of character codes to glyph ids for the specified glyph
ids.
'''
if self.bmp_table is None:
raise UnsupportedFont('This font has no Windows BMP cmap subtable.'
' Most likely a special purpose font.')
glyph_ids = frozenset(glyph_ids)
return self.bmp_table.get_glyph_map(glyph_ids)
def __init__(self, raw, offset):
data = Unpackable(raw, offset)
self.format = data.unpack('H')
if self.format not in self.formats:
raise UnsupportedFont(
'Unknown format for Lookup Subtable %s: 0x%x' %
(self.__class__.__name__, self.format))
if self.has_initial_coverage:
coverage_offset = data.unpack('H') + data.start_pos
self.coverage = Coverage(raw, coverage_offset,
self.__class__.__name__)
self.initialize(data)
def get_character_map(self, chars):
'''
Get a mapping of character codes to glyph ids in the font.
'''
if self.bmp_table is None:
raise UnsupportedFont('This font has no Windows BMP cmap subtable.'
' Most likely a special purpose font.')
chars = sorted(set(chars))
ans = OrderedDict()
for i, glyph_id in enumerate(self.bmp_table.get_glyph_ids(chars)):
if glyph_id > 0:
ans[chars[i]] = glyph_id
return ans
def pdf_subset(sfnt, glyphs):
for tag in tuple(sfnt.tables):
if tag not in {
b'hhea', b'head', b'hmtx', b'maxp', b'OS/2', b'post', b'cvt ',
b'fpgm', b'glyf', b'loca', b'prep', b'CFF ', b'VORG'
}:
# Remove non core tables since they are unused in PDF rendering
del sfnt[tag]
if b'loca' in sfnt and b'glyf' in sfnt:
# TrueType Outlines
subset_truetype(sfnt, {}, glyphs)
elif b'CFF ' in sfnt:
# PostScript Outlines
subset_postscript(sfnt, {}, glyphs)
else:
raise UnsupportedFont('This font does not contain TrueType '
'or PostScript outlines')
def decompile(self):
(self._version, self.scriptlist_offset, self.featurelist_offset,
self.lookuplist_offset) = unpack_from(b'>L3H', self.raw)
if self._version != 0x10000:
raise UnsupportedFont('The GSUB table has unknown version: 0x%x'%
self._version)
self.script_list_table = ScriptListTable(self.raw,
self.scriptlist_offset)
# self.script_list_table.dump()
self.feature_list_table = FeatureListTable(self.raw,
self.featurelist_offset)
# self.feature_list_table.dump()
self.lookup_list_table = LookupListTable(self.raw,
self.lookuplist_offset)
def __init__(self, raw, offset, strings, num_glyphs, is_CID):
super(Charset, self).__init__()
self.standard_charset = offset if offset in {0, 1, 2} else None
if is_CID and self.standard_charset is not None:
raise ValueError("CID font must not use a standard charset")
if self.standard_charset is None:
self.append(b'.notdef')
fmt = unpack_from(b'>B', raw, offset)[0]
offset += 1
f = {
0: self.parse_fmt0,
1: self.parse_fmt1,
2: partial(self.parse_fmt1, is_two_byte=True)
}.get(fmt, None)
if f is None:
raise UnsupportedFont('This font uses unsupported charset '
'table format: %d' % fmt)
f(raw, offset, strings, num_glyphs, is_CID)
def __init__(self, raw, offset, parent_table_name):
data = Unpackable(raw, offset)
self.format, count = data.unpack('2H')
if self.format not in {1, 2}:
raise UnsupportedFont('Unknown Coverage format: 0x%x in %s' %
(self.format, parent_table_name))
if self.format == 1:
self.glyph_ids = data.unpack('%dH' % count, single_special=False)
self.glyph_ids_map = {
gid: i
for i, gid in enumerate(self.glyph_ids)
}
else:
self.ranges = []
ranges = data.unpack('%dH' % (3 * count), single_special=False)
for i in range(count):
start, end, start_coverage_index = ranges[i * 3:(i + 1) * 3]
self.ranges.append(
CoverageRange(start, end, start_coverage_index))
def restrict_format_0(self, raw, glyph_ids):
if self._version == 0:
version, length, coverage, npairs = unpack_from(b'>4H', raw)
headerfmt = b'>3H'
else:
length, coverage, tuple_index, npairs = unpack_from(b'>L3H', raw)
headerfmt = b'>L2H'
offset = calcsize(headerfmt + b'4H')
entries = []
entrysz = calcsize(b'>2Hh')
for i in range(npairs):
try:
left, right, value = unpack_from(b'>2Hh', raw, offset)
except struct_error:
offset = len(raw)
break # Buggy kern table
if left in glyph_ids and right in glyph_ids:
entries.append(pack(b'>2Hh', left, right, value))
offset += entrysz
if offset != len(raw):
raise UnsupportedFont('This font has extra data at the end of'
' a Format 0 kern subtable')
npairs = len(entries)
if npairs == 0:
return b''
entry_selector = max_power_of_two(npairs)
search_range = (2**entry_selector) * 6
range_shift = (npairs - (2**entry_selector)) * 6
entries = b''.join(entries)
length = calcsize(headerfmt + b'4H') + len(entries)
if self._version == 0:
header = pack(headerfmt, version, length, coverage)
else:
header = pack(headerfmt, length, coverage, tuple_index)
return header + pack(b'>4H', npairs, search_range, entry_selector,
range_shift) + entries
def restrict_to_glyphs(self, glyph_ids):
if self._version not in {0, 0x10000}:
raise UnsupportedFont('kern table has version: %x' % self._version)
offset = 4 if (self._version == 0) else 8
tables = []
for i in range(self.num_tables):
if self._version == 0:
version, length, coverage = unpack_from(
b'>3H', self.raw, offset)
table_format = version
else:
length, coverage = unpack_from(b'>LH', self.raw, offset)
table_format = coverage & 0xff
raw = self.raw[offset:offset + length]
if table_format == 0:
raw = self.restrict_format_0(raw, glyph_ids)
if not raw:
continue
tables.append(raw)
offset += length
self.raw = pack(self.headerfmt, self._version,
len(tables)) + b''.join(tables)
def read_extra_header(self, data):
self.lookup_order, self.required_feature_index = data.unpack('2H')
if self.lookup_order != 0:
raise UnsupportedFont('This LanguageSystemTable has an unknown'
' lookup order: 0x%x' % self.lookup_order)
def __init__(self, raw):
(self.major_version, self.minor_version, self.header_size,
self.offset_size) = unpack_from(b'>4B', raw)
if (self.major_version, self.minor_version) != (1, 0):
raise UnsupportedFont('The CFF table has unknown version: '
'(%d, %d)' %
(self.major_version, self.minor_version))
offset = self.header_size
# Read Names Index
self.font_names = Index(raw, offset)
offset = self.font_names.pos
if len(self.font_names) > 1:
raise UnsupportedFont('CFF table has more than one font.')
# Read Top Dict
self.top_index = Index(raw, offset)
self.top_dict = TopDict()
offset = self.top_index.pos
# Read strings
self.strings = Strings(raw, offset)
offset = self.strings.pos
# Read global subroutines
self.global_subrs = Subrs(raw, offset)
offset = self.global_subrs.pos
# Decompile Top Dict
self.top_dict.decompile(self.strings, self.global_subrs,
self.top_index[0])
self.is_CID = 'ROS' in self.top_dict
if self.is_CID:
raise UnsupportedFont(
'Subsetting of CID keyed fonts is not supported')
# Read CharStrings (Glyph definitions)
try:
offset = self.top_dict['CharStrings']
except KeyError:
raise ValueError('This font has no CharStrings')
cs_type = self.top_dict.safe_get('CharstringType')
if cs_type != 2:
raise UnsupportedFont('This font has unsupported CharstringType: '
'%s' % cs_type)
self.char_strings = CharStringsIndex(raw, offset)
self.num_glyphs = len(self.char_strings)
# Read Private Dict
self.private_dict = self.private_subrs = None
pd = self.top_dict.safe_get('Private')
if pd:
size, offset = pd
self.private_dict = PrivateDict()
self.private_dict.decompile(self.strings, self.global_subrs,
raw[offset:offset + size])
if 'Subrs' in self.private_dict:
self.private_subrs = Subrs(raw,
offset + self.private_dict['Subrs'])
# Read charset (Glyph names)
self.charset = Charset(raw, self.top_dict.safe_get('charset'),
self.strings, self.num_glyphs, self.is_CID)
def subset(raw, individual_chars, ranges=(), warnings=None):
warn = partial(do_warn, warnings)
chars = set(map(safe_ord, individual_chars))
for r in ranges:
chars |= set(range(safe_ord(r[0]), safe_ord(r[1]) + 1))
# Always add the space character for ease of use from the command line
if safe_ord(' ') not in chars:
chars.add(safe_ord(' '))
sfnt = Sfnt(raw)
old_sizes = sfnt.sizes()
# Remove the Digital Signature table since it is useless in a subset
# font anyway
sfnt.pop(b'DSIG', None)
# Remove non core tables as they aren't likely to be used by renderers
# anyway
core_tables = {
b'cmap', b'hhea', b'head', b'hmtx', b'maxp', b'name', b'OS/2', b'post',
b'cvt ', b'fpgm', b'glyf', b'loca', b'prep', b'CFF ', b'VORG', b'EBDT',
b'EBLC', b'EBSC', b'BASE', b'GSUB', b'GPOS', b'GDEF', b'JSTF', b'gasp',
b'hdmx', b'kern', b'LTSH', b'PCLT', b'VDMX', b'vhea', b'vmtx', b'MATH'
}
for tag in list(sfnt):
if tag not in core_tables:
del sfnt[tag]
try:
cmap = sfnt[b'cmap']
except KeyError:
raise UnsupportedFont('This font has no cmap table')
# Get mapping of chars to glyph ids for all specified chars
character_map = cmap.get_character_map(chars)
extra_glyphs = set()
if b'GSUB' in sfnt:
# Parse all substitution rules to ensure that glyphs that can be
# substituted for the specified set of glyphs are not removed
gsub = sfnt[b'GSUB']
try:
gsub.decompile()
extra_glyphs = gsub.all_substitutions(character_map.values())
except UnsupportedFont as e:
warn('Usupported GSUB table: %s' % e)
except Exception:
warn('Failed to decompile GSUB table:', traceback.format_exc())
if b'loca' in sfnt and b'glyf' in sfnt:
# TrueType Outlines
subset_truetype(sfnt, character_map, extra_glyphs)
elif b'CFF ' in sfnt:
# PostScript Outlines
subset_postscript(sfnt, character_map, extra_glyphs)
else:
raise UnsupportedFont('This font does not contain TrueType '
'or PostScript outlines')
# Restrict the cmap table to only contain entries for the resolved glyphs
cmap.set_character_map(character_map)
if b'kern' in sfnt:
try:
sfnt[b'kern'].restrict_to_glyphs(frozenset(character_map.values()))
except UnsupportedFont as e:
warn('kern table unsupported, ignoring: %s' % e)
except Exception:
warn('Subsetting of kern table failed, ignoring:',
traceback.format_exc())
raw, new_sizes = sfnt()
return raw, old_sizes, new_sizes