tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[fix] update language to English when using English names in CJK countries

Browse Source
This commit is contained in:
Al
2016-12-25 01:18:54 -05:00
parent 51802035de
commit 6da092e144
1 changed files with 6 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
scripts/geodata/addresses/components.py
View File

@@ -54,7 +54,10 @@ PARSER_DEFAULT_CONFIG = os.path.join(this_dir, os.pardir, os.pardir, os.pardir,
'resources', 'parser', 'default.yaml') 'resources', 'parser', 'default.yaml')
CHINA = 'cn'
JAPAN = 'jp' JAPAN = 'jp'
KOREA = 'kr'
JAPANESE_ROMAJI = 'ja_rm' JAPANESE_ROMAJI = 'ja_rm'
ENGLISH = 'en' ENGLISH = 'en'
@@ -63,6 +66,7 @@ CHINESE = 'zh'
KOREAN = 'ko' KOREAN = 'ko'
CJK_LANGUAGES = set([CHINESE, JAPANESE, KOREAN]) CJK_LANGUAGES = set([CHINESE, JAPANESE, KOREAN])
CJK_COUNTRIES = set([CHINA, JAPAN, KOREA])
class AddressComponents(object): class AddressComponents(object):
@@ -1661,6 +1665,8 @@ class AddressComponents(object):
lang, script = language.split('_', 1) lang, script = language.split('_', 1)
if lang not in CJK_LANGUAGES and script.lower() not in self.valid_scripts: if lang not in CJK_LANGUAGES and script.lower() not in self.valid_scripts:
language = lang language = lang
elif country in CJK_COUNTRIES and (non_local_language == ENGLISH or (language_suffix or '').lstrip(':').lower() == ENGLISH):
language = ENGLISH
return address_components, country, language return address_components, country, language