1
0
mirror of https://github.com/ryanoasis/nerd-fonts.git synced 2024-12-19 20:12:52 +02:00
nerd-fonts/bin/scripts/name_parser/FontnameTools.py
Fini Jastrow 99688c40c7 font-patcher: Introduce weight check
[why]
Windows seems to construct the font names including the PS weight.
We have some sourcefonts that are broken (i.e. have in fact different
weights but have the same PS weight and/or OS2 weight.

That raises problems with the fonts on Windows.

[how]
Check and compare all weight metadata (except CID) and issue a warning
if they differ too much. That might fail with unusual weight names,
though.

See Issue #1333 and PR #1358.

Reported-by: LeoniePhiline
Signed-off-by: Fini Jastrow <ulf.fini.jastrow@desy.de>
2023-09-27 05:46:33 +02:00

421 lines
17 KiB
Python

#!/usr/bin/env python
# coding=utf8
import re
import sys
class FontnameTools:
"""Deconstruct a fontname to get standardized name parts"""
@staticmethod
def front_upper(word):
"""Capitalize a string (but keep case of subsequent chars)"""
return word[:1].upper() + word[1:]
@staticmethod
def camel_casify(word):
"""Remove blanks and use CamelCase for the new word"""
return ''.join(map(FontnameTools.front_upper, word.split(' ')))
@staticmethod
def camel_explode(word):
"""Explode CamelCase -> Camel Case"""
# But do not explode "JetBrains" etc at string start...
excludes = [
'JetBrains',
'DejaVu',
'OpenDyslexicAlta',
'OpenDyslexicMono',
'OpenDyslexic',
'DaddyTimeMono',
'InconsolataGo',
'ProFontWindows',
'ProFont',
'ProggyClean',
]
m = re.match('(' + '|'.join(excludes) + ')(.*)', word)
(prefix, word) = m.group(1,2) if m != None else ('', word)
if len(word) == 0:
return prefix
parts = re.split('(?<=[a-z0-9])(?=[A-Z])', word)
if len(prefix):
parts.insert(0, prefix)
return ' '.join(parts)
@staticmethod
def drop_empty(l):
"""Remove empty strings from list of strings"""
return [x for x in l if len(x) > 0]
@staticmethod
def concat(*all_things):
"""Flatten list of (strings or lists of strings) to a blank-separated string"""
all = []
for thing in all_things:
if type(thing) is not list:
all.append(thing)
else:
all += thing
return ' '.join(FontnameTools.drop_empty(all))
@staticmethod
def unify_style_names(style_name):
"""Substitude some known token with standard wording"""
known_names = {
# Source of the table is the current sourcefonts
# Left side needs to be lower case
'book': '',
'ce': 'CE',
'normal': 'Regular',
}
return known_names.get(style_name.lower(), style_name)
@staticmethod
def find_in_dicts(key, dicts):
"""Find an entry in a list of dicts, return entry and in which list it was"""
for i, d in enumerate(dicts):
if key in d:
return ( d[key], i )
return (None, 0)
@staticmethod
def get_shorten_form_idx(aggressive, prefix, form_if_prefixed):
"""Get the tuple index of known_* data tables"""
if aggressive:
return 0
if len(prefix):
return form_if_prefixed
return 1
@staticmethod
def shorten_style_name(name, aggressive):
"""Substitude some known styles to short form"""
# If aggressive is False create the mild short form
# aggressive == True: Always use first form of everything
# aggressive == False:
# - has no modifier: use the second form
# - has modifier: use second form of mod plus first form of weights2
# - has modifier: use second form of mod plus second form of widths
name_rest = name
name_pre = ''
form = FontnameTools.get_shorten_form_idx(aggressive, '', 0)
for mod in FontnameTools.known_modifiers:
if name.startswith(mod) and len(name) > len(mod): # Second condition specifically for 'Demi'
name_pre = FontnameTools.known_modifiers[mod][form]
name_rest = name[len(mod):]
break
subst, i = FontnameTools.find_in_dicts(name_rest, [ FontnameTools.known_weights2, FontnameTools.known_widths ])
form = FontnameTools.get_shorten_form_idx(aggressive, name_pre, i)
if isinstance(subst, tuple):
return name_pre + subst[form]
if not len(name_pre):
# The following sets do not allow modifiers
subst, _ = FontnameTools.find_in_dicts(name_rest, [ FontnameTools.known_weights1, FontnameTools.known_slopes ])
if isinstance(subst, tuple):
return subst[form]
return name
@staticmethod
def short_styles(lists, aggressive):
"""Shorten all style names in a list or a list of lists"""
if not len(lists) or not isinstance(lists[0], list):
return list(map(lambda x: FontnameTools.shorten_style_name(x, aggressive), lists))
return [ list(map(lambda x: FontnameTools.shorten_style_name(x, aggressive), styles)) for styles in lists ]
@staticmethod
def make_oblique_style(weights, styles):
"""Move "Oblique" from weights to styles for font naming purposes"""
if 'Oblique' in weights:
weights = list(weights)
weights.remove('Oblique')
styles = list(styles)
styles.append('Oblique')
return (weights, styles)
@staticmethod
def get_name_token(name, tokens):
"""Try to find any case insensitive token from tokens in the name, return tuple with found token-list and rest"""
# The default mode (allow_regex_token = False) will try to find any verbatim string in the
# tokens list (case insensitive matching) and give that tokens list item back with
# unchanged case (i.e. [ 'Bold' ] will match "bold" and return it as [ 'Bold', ]
# In the regex mode (allow_regex_token = True) it will use the tokens elements as
# regexes and return the original (i.e. from name) case.
#
# Token are always used in a regex and may not capture, use non capturing
# grouping if needed (?: ... )
lower_tokens = [ t.lower() for t in tokens ]
not_matched = ""
all_tokens = []
j = 1
token_regex = '|'.join(tokens)
# Allow a dash between CamelCase token word parts, i.e. Camel-Case
# This allows for styles like Extra-Bold
token_regex = re.sub(r'(?<=[a-z])(?=[A-Z])', '-?', token_regex)
regex = re.compile('(.*?)(' + token_regex + ')(.*)', re.IGNORECASE)
while j:
j = regex.match(name)
if not j:
break
if len(j.groups()) != 3:
sys.exit('Malformed regex in FontnameTools.get_name_token()')
not_matched += ' ' + j.groups()[0] # Blanc prevents unwanted concatenation of unmatched substrings
tok = j.groups()[1].lower()
tok = tok.replace('-', '') # Remove dashes between CamelCase token words
if tok in lower_tokens:
tok = tokens[lower_tokens.index(tok)]
tok = FontnameTools.unify_style_names(tok)
if len(tok):
all_tokens.append(tok)
name = j.groups()[2] # Recurse rest
not_matched += ' ' + name
return ( not_matched.strip(), all_tokens )
@staticmethod
def postscript_char_filter(name):
"""Filter out characters that are not allowed in Postscript names"""
# The name string must be restricted to the printable ASCII subset, codes 33 to 126,
# except for the 10 characters '[', ']', '(', ')', '{', '}', '<', '>', '/', '%'
out = ""
for c in name:
if c in '[](){}<>/%' or ord(c) < 33 or ord(c) > 126:
continue
out += c
return out
SIL_TABLE = [
( '(a)nka/(c)oder', r'\1na\2onder' ),
( '(a)nonymous', r'\1nonymice' ),
( '(b)itstream( ?)(v)era( ?sans ?mono)?', r'\1itstrom\2Wera' ),
( '(c)ascadia( ?)(c)ode', r'\1askaydia\2\3ove' ),
( '(c)ascadia( ?)(m)ono', r'\1askaydia\2\3ono' ),
( 'Gohufont', r'GohuFont'), # Correct to CamelCase
( '(h)ermit', r'\1urmit' ),
( '(h)asklig', r'\1asklug' ),
( 'iA([- ]?)writer', r'iM\1Writing' ),
( 'IBM[- ]?plex', r'Blex' ), # We do not keep the case here
( '(i)ntel( ?)(o)ne', r'\1ntone' ),
( '(l)iberation', r'\1iteration' ),
( '(m)( ?)plus', r'\1+'), # Added this, because they use a plus symbol :->
( '(s)hare', r'\1hure' ),
( '(s)ource', r'\1auce' ),
( '(t)erminus', r'\1erminess' ),
# Noone cares that font names starting with a digit are forbidden:
( 'IBM 3270', r'3270'), # for historical reasons and 'IBM' is a TM or something
# Some name parts that are too long for us
( '(.*sans ?m)ono', r'\1'), # Various SomenameSansMono fonts
( '(.*code ?lat)in Expanded', r'\1X'), # for 'M PLUS Code Latin Expanded'
( '(.*code ?lat)in', r'\1'), # for 'M PLUS Code Latin'
( '(b)ig( ?)(b)lue( ?)(t)erminal', r'\1ig\3lue\5erm'), # Shorten BigBlueTerminal
( '(.*)437TT', r'\g<1>437'), # Shorten BigBlueTerminal 437 TT even further
( '(.*dyslexic ?alt)a', r'\1'), # Open Dyslexic Alta -> Open Dyslexic Alt
( '(.*dyslexic ?m)ono', r'\1'), # Open Dyslexic Mono -> Open Dyslexic M
( '(overpass ?m)ono', r'\1'), # Overpass Mono -> Overpass M
( '(proggyclean) ?tt', r'\1'), # Remove TT from ProggyClean
( '(terminess) ?\(ttf\)', r'\1'), # Remove TTF from Terminus (after renamed to Terminess)
( '(im ?writing ?q)uattro', r'\1uat'), # Rename iM Writing Quattro to Quat
( '(im ?writing ?(mono|duo|quat)) ?s', r'\1'), # Remove S from all iM Writing styles
]
# From https://adobe-type-tools.github.io/font-tech-notes/pdfs/5088.FontNames.pdf
# The first short variant is from the linked table.
# The second (longer) short variant is from diverse fonts like Noto.
# We can
# - use the long form
# - use the very short form (first)
# - use mild short form:
# - has no modifier: use the second form
# - has modifier: use second form of mod plus first form of weights2
# - has modifier: use second form of mod plus second form of widths
# This is encoded in get_shorten_form_idx()
known_weights1 = { # can not take modifiers
'Medium': ('Md', 'Med'),
'Nord': ('Nd', 'Nord'),
'Book': ('Bk', 'Book'),
'Text': ('Txt', 'Text'),
'Poster': ('Po', 'Poster'),
'Demi': ('Dm', 'Demi'), # Demi is sometimes used as a weight, sometimes as a modifier
'Regular': ('Rg', 'Reg'),
'Display': ('DS', 'Disp'),
'Super': ('Su', 'Sup'),
'Retina': ('Rt', 'Ret'),
}
known_weights2 = { # can take modifiers
'Black': ('Blk', 'Black'),
'Bold': ('Bd', 'Bold'),
'Heavy': ('Hv', 'Heavy'),
'Thin': ('Th', 'Thin'),
'Light': ('Lt', 'Light'),
' ': (), # Just for CodeClimate :-/
}
known_styles = [ # Keywords that end up as style (i.e. a RIBBI set)
'Bold', 'Italic', 'Regular', 'Normal'
]
known_widths = { # can take modifiers
'Compressed': ('Cm', 'Comp'),
'Extended': ('Ex', 'Extd'),
'Condensed': ('Cn', 'Cond'),
'Narrow': ('Nr', 'Narrow'),
'Compact': ('Ct', 'Compact'),
}
known_slopes = { # can not take modifiers
'Inclined': ('Ic', 'Incl'),
'Oblique': ('Obl', 'Obl'),
'Italic': ('It', 'Italic'),
'Upright': ('Up', 'Uprght'),
'Kursiv': ('Ks', 'Kurs'),
'Sloped': ('Sl', 'Slop'),
}
known_modifiers = {
'Demi': ('Dm', 'Dem'),
'Ultra': ('Ult', 'Ult'),
'Semi': ('Sm', 'Sem'),
'Extra': ('X', 'Ext'),
}
equivalent_weights = {
100: ('thin', 'hairline'),
200: ('extralight', 'ultralight'),
300: ('light', ),
350: ('semilight', ),
400: ('regular', 'normal', 'book', 'text', 'nord', 'retina'),
500: ('medium', ),
600: ('semibold', 'demibold', 'demi'),
700: ('bold', ),
800: ('extrabold', 'ultrabold'),
900: ('black', 'heavy', 'poster', 'extrablack', 'ultrablack'),
}
@staticmethod
def weight_string_to_number(w):
""" Convert a common string approximation to a PS/2 weight value """
if not len(w):
return 400
for num, strs in FontnameTools.equivalent_weights.items():
if w.lower() in strs:
return num
return None
@staticmethod
def weight_to_string(w):
""" Convert a PS/2 weight value to the common string approximation """
if w < 150:
str = 'Thin'
elif w < 250:
str = 'Extra-Light'
elif w < 350:
str = 'Light'
elif w < 450:
str = 'Regular'
elif w < 550:
str = 'Medium'
elif w < 650:
str = 'Semi-Bold'
elif w < 750:
str = 'Bold'
elif w < 850:
str = 'Extra-Bold'
else:
str = 'Black'
return str
@staticmethod
def is_keep_regular(basename):
"""This has been decided by the font designers, we need to mimic that (for comparison purposes)"""
KEEP_REGULAR = [
'Agave',
'Arimo',
'Aurulent',
'Cascadia',
'Cousine',
'Fantasque',
'Fira',
'Overpass',
'Lilex',
'Inconsolata$', # not InconsolataGo
'IAWriter',
'Meslo',
'Monoid',
'Mononoki',
'Hack',
'JetBrains Mono',
'Noto Sans',
'Noto Serif',
'Victor',
]
for kr in KEEP_REGULAR:
if (basename.rstrip() + '$').startswith(kr): return True
return False
@staticmethod
def _parse_simple_font_name(name):
"""Parse a fontname that does not follow the 'FontFamilyName-FontStyle' pattern"""
# This is the usual case, because the font-patcher usually uses the fullname and
# not the PS name
if ' ' in name:
return FontnameTools.parse_font_name(name.replace(' ', '-'))
# Do we have a number-name boundary?
p = re.split('(?<=[0-9])(?=[a-zA-Z])', name)
if len(p) > 1:
return FontnameTools.parse_font_name('-'.join(p))
# Or do we have CamelCase?
n = FontnameTools.camel_explode(name)
if n != name:
return FontnameTools.parse_font_name(n.replace(' ', '-'))
return (False, FontnameTools.camel_casify(name), [], [], [], '')
@staticmethod
def parse_font_name(name):
"""Expects a fontname following the 'FontFamilyName-FontStyle' pattern and returns ... parts"""
# This could parse filenames in the beginning but that was never used in production; code removed with this commit
for special in [
('ExtLt', 'ExtraLight'), # IBM-Plex
('Medm', 'Medium'), # IBM-Plex
('Semi-Condensed', 'SemiCondensed'), # 3270
('SmBld', 'SemiBold'), # IBM-Plex
]:
name = re.sub(r'\b' + special[0] + r'\b', special[1], name, 1, re.IGNORECASE)
name = re.sub('[_\s]+', ' ', name)
matches = re.match(r'([^-]+)(?:-(.*))?', name)
familyname = FontnameTools.camel_casify(matches.group(1))
style = matches.group(2)
if not style:
return FontnameTools._parse_simple_font_name(name)
# These are the FontStyle keywords we know, in three categories
# Weights end up as Typographic Family parts ('after the dash')
# Styles end up as Family parts (for classic grouping of four)
# Others also end up in Typographic Family ('before the dash')
weights = [ m + s
for s in list(FontnameTools.known_weights2) + list(FontnameTools.known_widths)
for m in list(FontnameTools.known_modifiers) + [''] if m != s
] + list(FontnameTools.known_weights1) + list(FontnameTools.known_slopes)
weights = [ w for w in weights if w not in FontnameTools.known_styles ]
# Some font specialities:
other = [
'-', 'Book', 'For', 'Powerline',
'IIx', # Profont IIx
'LGC', # Inconsolata LGC
r'\bCE\b', # ProggycleanTT CE
r'[12][cmp]n?', # MPlus
r'(?:uni-)?1[14]', # GohuFont uni
]
( style, weight_token ) = FontnameTools.get_name_token(style, weights)
( style, style_token ) = FontnameTools.get_name_token(style, FontnameTools.known_styles)
( style, other_token ) = FontnameTools.get_name_token(style, other)
while 'Regular' in style_token and len(style_token) > 1:
# Correct situation where "Regular" and something else is given
style_token.remove('Regular')
# Recurse to see if unmatched stuff between dashes can belong to familyname
matches2 = re.match(r'(\w+)-(.*)', style)
if matches2:
return FontnameTools.parse_font_name(familyname + matches2.group(1) + '-' + matches2.group(2))
style = re.sub(r'(^|\s)\d+(\.\d+)+(\s|$)', r'\1\3', style) # Remove (free standing) version numbers
style_parts = FontnameTools.drop_empty(style.split(' '))
style = ' '.join(map(FontnameTools.front_upper, style_parts))
familyname = FontnameTools.camel_explode(familyname)
return (True, familyname, weight_token, style_token, other_token, style)