Disable language detection for texts shorter than 140 characters (#8010)

If the input text is blank after preparation (only mention, or
only URL, or empty as in a media post), then use nil as language,
since it's OK to show to everyone.

Otherwise, always fall back to the server's default locale
This commit is contained in:
Eugen Rochko 2018-07-14 04:05:36 +02:00 committed by GitHub
parent 0180037dfb
commit 38e9662d78
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 14 additions and 5 deletions

View File

@ -3,12 +3,16 @@
class LanguageDetector class LanguageDetector
include Singleton include Singleton
CHARACTER_THRESHOLD = 140
def initialize def initialize
@identifier = CLD3::NNetLanguageIdentifier.new(1, 2048) @identifier = CLD3::NNetLanguageIdentifier.new(1, 2048)
end end
def detect(text, account) def detect(text, account)
detect_language_code(text) || default_locale(account) input_text = prepare_text(text)
return if input_text.blank?
detect_language_code(input_text) || default_locale(account)
end end
def language_names def language_names
@ -23,8 +27,13 @@ class LanguageDetector
simplify_text(text).strip simplify_text(text).strip
end end
def unreliable_input?(text)
text.size < CHARACTER_THRESHOLD
end
def detect_language_code(text) def detect_language_code(text)
result = @identifier.find_language(prepare_text(text)) return if unreliable_input?(text)
result = @identifier.find_language(text)
iso6391(result.language.to_s).to_sym if result.reliable? iso6391(result.language.to_s).to_sym if result.reliable?
end end
@ -66,6 +75,6 @@ class LanguageDetector
end end
def default_locale(account) def default_locale(account)
account.user_locale&.to_sym account.user_locale&.to_sym || I18n.default_locale
end end
end end

View File

@ -57,7 +57,7 @@ describe LanguageDetector do
end end
it 'detects spanish language' do it 'detects spanish language' do
string = 'Obtener un Hola y bienvenidos a Mastodon' string = 'Obtener un Hola y bienvenidos a Mastodon. Obtener un Hola y bienvenidos a Mastodon. Obtener un Hola y bienvenidos a Mastodon. Obtener un Hola y bienvenidos a Mastodon'
result = described_class.instance.detect(string, account_without_user_locale) result = described_class.instance.detect(string, account_without_user_locale)
expect(result).to eq :es expect(result).to eq :es
@ -86,7 +86,7 @@ describe LanguageDetector do
account = double(user_locale: 'fr') account = double(user_locale: 'fr')
result = described_class.instance.detect('', account) result = described_class.instance.detect('', account)
expect(result).to eq :fr expect(result).to eq nil
end end
it 'uses nil when account is present but has no locale' do it 'uses nil when account is present but has no locale' do