Page MenuHomePhabricator
Paste P7116

_altlang dictionaries
ActivePublic
Actions

Authored by Dalba on May 11 2018, 7:01 AM.
Tags
Referenced Files
F18167626: _altlang dictionaries
May 11 2018, 1:29 PM
F18167623: _altlang dictionaries
May 11 2018, 1:28 PM
F18166828: _altlang dictionaries
May 11 2018, 11:33 AM
F18166820: _altlang dictionaries
May 11 2018, 11:33 AM
F18166701: _altlang dictionaries
May 11 2018, 11:17 AM
F18166688: _altlang dictionaries
May 11 2018, 11:12 AM
F18160482: _altlang dictionaries
May 11 2018, 7:01 AM
Subscribers
None
from ast import *
parsed = parse('''def _altlang(lang):
# Akan
if lang in ['ak', 'tw']:
return ['ak', 'tw']
# Amharic
if lang in ['aa', 'ti']:
return ['am']
# Arab
if lang in ['arc', 'arz', 'so']:
return ['ar']
if lang == 'kab':
return ['ar', 'fr']
# Bulgarian
if lang in ['cu', 'mk']:
return ['bg', 'sr', 'sh']
# Czech
if lang in ['cs', 'sk']:
return ['cs', 'sk']
# German
if lang in ['bar', 'frr', 'ksh', 'pdc', 'pfl']:
return ['de']
if lang == 'lb':
return ['de', 'fr']
if lang in ['als', 'gsw']:
return ['als', 'gsw', 'de']
if lang == 'nds':
return ['nds-nl', 'de']
if lang in ['dsb', 'hsb']:
return ['hsb', 'dsb', 'de']
if lang == 'sli':
return ['de', 'pl']
if lang == 'rm':
return ['de', 'it']
if lang == 'stq':
return ['nds', 'de']
# Greek
if lang in ['grc', 'pnt']:
return ['el']
# Esperanto
if lang in ['io', 'nov']:
return ['eo']
# Spanish
if lang in ['an', 'arn', 'ast', 'ay', 'ext', 'lad', 'nah', 'nv', 'qu',
'yua']:
return ['es']
if lang == 'ca':
return ['oc', 'es']
if lang in ['gl', 'gn']:
return ['es', 'pt']
if lang == 'eu':
return ['es', 'fr']
if lang == 'cbk-zam':
return ['es', 'tl']
# Estonian
if lang in ['fiu-vro', 'vro']:
return ['fiu-vro', 'vro', 'et']
if lang == 'liv':
return ['et', 'lv']
# Persian (Farsi)
if lang in ['azb', 'lrc', 'ps']:
return ['fa']
if lang in ['glk', 'mzn']:
return ['glk', 'mzn', 'fa', 'ar']
# Finnish
if lang == 'vep':
return ['et', 'fi', 'ru']
if lang == 'fit':
return ['fi', 'sv']
if lang == 'olo':
return ['fi']
# French
if lang in ['atj', 'bm', 'br', 'ff', 'ht', 'kbp', 'kg', 'ln', 'mg', 'nrm',
'pcd', 'rw', 'sg', 'ty', 'wa', 'wo']:
return ['fr']
if lang == 'oc':
return ['fr', 'ca', 'es']
if lang in ['co', 'frp']:
return ['fr', 'it']
# Hindi
if lang in ['mai', 'sa']:
return ['hi']
if lang in ['ne', 'new']:
return ['ne', 'new', 'hi']
if lang == 'dty':
return ['ne']
if lang in ['bh', 'bho']:
return ['bh', 'bho']
# Indonesian and Malay
if lang in ['ace', 'bug', 'bjn', 'id', 'jv', 'ms', 'su']:
return ['id', 'ms', 'jv']
if lang == 'map-bms':
return ['jv', 'id', 'ms']
if lang == 'min':
return ['id']
# Inuit languages
if lang in ['ik', 'iu']:
return ['iu', 'kl']
if lang == 'kl':
return ['da', 'iu', 'no', 'nb']
# Italian
if lang in ['eml', 'fur', 'lij', 'lmo', 'nap', 'pms', 'roa-tara', 'sc',
'scn', 'vec']:
return ['it']
# Lithuanian
if lang in ['bat-smg', 'sgs']:
return ['bat-smg', 'sgs', 'lt']
# Latvian
if lang == 'ltg':
return ['lv']
# Dutch
if lang in ['af', 'fy', 'li', 'pap', 'srn', 'vls', 'zea']:
return ['nl']
if lang == 'nds-nl':
return ['nds', 'nl']
# Polish
if lang in ['csb', 'szl']:
return ['pl']
# Portuguese
if lang in ['fab', 'mwl', 'tet']:
return ['pt']
# Romanian
if lang in ['roa-rup', 'rup']:
return ['roa-rup', 'rup', 'ro']
if lang in ['mo', 'rmy']:
return ['ro']
# Russian and Belarusian
if lang in ['ab', 'av', 'ba', 'bxr', 'ce', 'cv', 'inh', 'kk', 'koi', 'krc',
'kv', 'ky', 'lbe', 'lez', 'mdf', 'mhr', 'mn', 'mrj', 'myv',
'os', 'sah', 'tg', 'tyv', 'udm', 'uk', 'xal']:
return ['ru']
if lang in ['kbd', 'ady']:
return ['kbd', 'ady', 'ru']
if lang == 'tt':
return ['tt-cyrl', 'ru']
if lang in ['be', 'be-tarask']:
return ['be', 'be-tarask', 'ru']
if lang == 'kaa':
return ['uz', 'ru']
# Serbocroatian
if lang in ['bs', 'hr', 'sh']:
return ['sh', 'hr', 'bs', 'sr', 'sr-el']
if lang == 'sr':
return ['sr-el', 'sh', 'hr', 'bs']
# Tagalog
if lang in ['bcl', 'ceb', 'ilo', 'pag', 'pam', 'war']:
return ['tl']
# Turkish and Kurdish
if lang in ['diq', 'ku']:
return ['ku', 'ku-latn', 'tr']
if lang == 'gag':
return ['tr']
if lang == 'ckb':
return ['ku']
# Ukrainian
if lang in ['crh', 'crh-latn']:
return ['crh', 'crh-latn', 'uk', 'ru']
if lang in ['rue']:
return ['uk', 'ru']
# Chinese
if lang in ['zh-classical', 'lzh', 'minnan', 'nan', 'zh-tw',
'zh', 'zh-hans']:
return ['zh', 'zh-hans', 'zh-tw', 'zh-cn', 'zh-classical', 'lzh']
if lang == 'zh-min-nan':
return ['cdo', 'zh', 'zh-hans', 'zh-tw', 'zh-cn', 'zh-classical',
'lzh']
if lang in ['cdo', 'gan', 'hak', 'ii', 'wuu', 'za', 'zh-classical', 'lzh',
'zh-cn', 'zh-yue', 'yue']:
return ['zh', 'zh-hans' 'zh-cn', 'zh-tw', 'zh-classical', 'lzh']
# Scandinavian languages
if lang in ['da', 'sv']:
return ['da', 'no', 'nb', 'sv', 'nn']
if lang in ['fo', 'is']:
return ['da', 'no', 'nb', 'nn', 'sv']
if lang == 'nn':
return ['no', 'nb', 'sv', 'da']
if lang in ['no', 'nb']:
return ['no', 'nb', 'da', 'nn', 'sv']
if lang == 'se':
return ['sv', 'no', 'nb', 'nn', 'fi']
# Other languages
if lang in ['bi', 'tpi']:
return ['bi', 'tpi']
if lang == 'yi':
return ['he', 'de']
if lang in ['ia', 'ie']:
return ['ia', 'la', 'it', 'fr', 'es']
if lang == 'xmf':
return ['ka']
if lang in ['nso', 'st']:
return ['st', 'nso']
if lang in ['kj', 'ng']:
return ['kj', 'ng']
if lang in ['meu', 'hmo']:
return ['meu', 'hmo']
if lang in ['as', 'bpy']:
return ['bn']
if lang == 'tcy':
return ['kn']
# Default value
return []''')
_LANG_TO_GROUP_NAME = {}
_GROUP_NAME_TO_FALLBACKS = {}
# The last statement is the return statement, others are if-statements
for ifexp in parsed.body[0].body[:-1]:
test = ifexp.test
op = type(test.ops[0])
if op is In:
input_langs = literal_eval(test.comparators[0])
elif op is Eq:
input_langs = [literal_eval(test.comparators[0])]
else:
raise RuntimeError
return_list = literal_eval(ifexp.body[0].value)
lang_group = input_langs[0]
for lang in input_langs:
if lang in _LANG_TO_GROUP_NAME:
print('Warning"', lang, '"was found in more than one group.')
else:
_LANG_TO_GROUP_NAME[lang] = lang_group
assert lang_group not in _GROUP_NAME_TO_FALLBACKS
_GROUP_NAME_TO_FALLBACKS[lang_group] = return_list
print('{\n ' + '\n '.join(
f"'{k}': '{v}'," for k, v in sorted(_LANG_TO_GROUP_NAME.items())) + '\n}')
print("{\n '': [],\n " + '\n '.join(
f"'{k}': {v}," for k, v in sorted(_GROUP_NAME_TO_FALLBACKS.items()))
+ '\n}')

Event Timeline

Dalba created this paste.May 11 2018, 7:01 AM
Dalba edited the content of this paste. (Show Details)May 11 2018, 11:12 AM
Dalba edited the content of this paste. (Show Details)May 11 2018, 11:17 AM
Dalba edited the content of this paste. (Show Details)May 11 2018, 11:33 AM
Dalba edited the content of this paste. (Show Details)
Dalba edited the content of this paste. (Show Details)May 11 2018, 1:28 PM
Dalba edited the content of this paste. (Show Details)
Content licensed under Creative Commons Attribution-ShareAlike (CC BY-SA) 4.0 unless otherwise noted; code licensed under GNU General Public License (GPL) 2.0 or later and other open source licenses. By using this site, you agree to the Terms of Use, Privacy Policy, and Code of Conduct. · Wikimedia Foundation · Privacy Policy · Code of Conduct · Terms of Use · Disclaimer · CC-BY-SA · GPL