Add language metadata for AO3.

This commit is contained in:
Jim Miller 2015-02-25 11:05:54 -06:00
parent 1d8b065261
commit b64db3524e
2 changed files with 60 additions and 1 deletions

View file

@ -222,6 +222,10 @@ class ArchiveOfOurOwnOrgAdapter(BaseSiteAdapter):
if a != None:
self.story.setMetadata('rating',stripHTML(a.text))
d = metasoup.find('dd',{'class':"language"})
if d != None:
self.story.setMetadata('language',stripHTML(d.text))
a = metasoup.find('dd',{'class':"fandom tags"})
fandoms = a.findAll('a',{'class':"tag"})
for fandom in fandoms:

View file

@ -222,8 +222,63 @@ langs = {
"Vietnamese":"vi",
"Thai":"th",
"Devanagari":"hi",
}
## These are from/for AO3:
u'العربية':'ar',
u'беларуская':'be',
u'Български език':'bg',
u'Català':'ca',
u'Čeština':'cs',
u'Cymraeg':'cy',
u'Dansk':'da',
u'Deutsch':'de',
u'Ελληνικά':'el',
u'English':'en',
u'Esperanto':'eo',
u'Español':'es',
u'eesti keel':'et',
u'فارسی':'fa',
u'Suomi':'fi',
u'Wikang Filipino':'fil',
u'Français':'fr',
u'Gaeilge':'ga',
u'Gàidhlig':'gd',
u'עִבְרִית':'he',
u'हिन्दी':'hi',
u'Hrvatski':'hr',
u'Magyar':'hu',
u'Bahasa Indonesia':'id',
u'Íslenska':'is',
u'Italiano':'it',
u'日本語':'ja',
u'한국말':'ko',
u'Lingua latina':'la',
u'Lietuvių':'lt',
u'Latviešu valoda':'lv',
u'मराठी':'mr',
u'بهاس ملايو ':'ms',
u'Nederlands':'nl',
u'Norsk':'no',
u'ਪੰਜਾਬੀ':'pa',
u'Polski':'pl',
u'Português':'pt',
u'Quenya':'qya',
u'Română':'ro',
u'Русский':'ru',
u'Slovenčina':'sk',
u'Shqip':'sq',
u'српски':'sr',
u'Svenska':'sv',
u'ไทย':'th',
u'tlhIngan-Hol':'tlh', # Klingon. Has a real ISO 639-2 code.
#'Thermian':'', # Alien language from Galaxy Quest.
u'Türkçe':'fr',
u'українська':'uk',
u'Tiếng Việt':'vi',
u'中文':'zh',
u'Bahasa Malaysia':'zsm',
}
def re_compile(regex,line):
try:
return re.compile(regex)