forked from berserker/microblog
commit
692963d43b
26 changed files with 601 additions and 567 deletions
@ -1,101 +0,0 @@ |
||||
# frozen_string_literal: true |
||||
|
||||
class LanguageDetector |
||||
include Singleton |
||||
|
||||
WORDS_THRESHOLD = 4 |
||||
RELIABLE_CHARACTERS_RE = /[\p{Hebrew}\p{Arabic}\p{Syriac}\p{Thaana}\p{Nko}\p{Han}\p{Katakana}\p{Hiragana}\p{Hangul}\p{Thai}]+/m |
||||
|
||||
def initialize |
||||
@identifier = CLD3::NNetLanguageIdentifier.new(1, 2048) |
||||
end |
||||
|
||||
def detect(text, account) |
||||
input_text = prepare_text(text) |
||||
|
||||
return if input_text.blank? |
||||
|
||||
detect_language_code(input_text) || default_locale(account) |
||||
end |
||||
|
||||
def language_names |
||||
@language_names = CLD3::TaskContextParams::LANGUAGE_NAMES.map { |name| iso6391(name.to_s).to_sym }.uniq |
||||
end |
||||
|
||||
private |
||||
|
||||
def prepare_text(text) |
||||
simplify_text(text).strip |
||||
end |
||||
|
||||
def unreliable_input?(text) |
||||
!reliable_input?(text) |
||||
end |
||||
|
||||
def reliable_input?(text) |
||||
sufficient_text_length?(text) || language_specific_character_set?(text) |
||||
end |
||||
|
||||
def sufficient_text_length?(text) |
||||
text.split(/\s+/).size >= WORDS_THRESHOLD |
||||
end |
||||
|
||||
def language_specific_character_set?(text) |
||||
words = text.scan(RELIABLE_CHARACTERS_RE) |
||||
|
||||
if words.present? |
||||
words.reduce(0) { |acc, elem| acc + elem.size }.to_f / text.size > 0.3 |
||||
else |
||||
false |
||||
end |
||||
end |
||||
|
||||
def detect_language_code(text) |
||||
return if unreliable_input?(text) |
||||
|
||||
result = @identifier.find_language(text) |
||||
|
||||
iso6391(result.language.to_s).to_sym if result&.reliable? |
||||
end |
||||
|
||||
def iso6391(bcp47) |
||||
iso639 = bcp47.split('-').first |
||||
|
||||
# CLD3 returns grandfathered language code for Hebrew |
||||
return 'he' if iso639 == 'iw' |
||||
|
||||
ISO_639.find(iso639).alpha2 |
||||
end |
||||
|
||||
def simplify_text(text) |
||||
new_text = remove_html(text) |
||||
new_text.gsub!(FetchLinkCardService::URL_PATTERN, '\1') |
||||
new_text.gsub!(Account::MENTION_RE, '') |
||||
new_text.gsub!(Tag::HASHTAG_RE) { |string| string.gsub(/[#_]/, '#' => '', '_' => ' ').gsub(/[a-z][A-Z]|[a-zA-Z][\d]/) { |s| s.insert(1, ' ') }.downcase } |
||||
new_text.gsub!(/:#{CustomEmoji::SHORTCODE_RE_FRAGMENT}:/, '') |
||||
new_text.gsub!(/\s+/, ' ') |
||||
new_text |
||||
end |
||||
|
||||
def new_scrubber |
||||
scrubber = Rails::Html::PermitScrubber.new |
||||
scrubber.tags = %w(br p) |
||||
scrubber |
||||
end |
||||
|
||||
def scrubber |
||||
@scrubber ||= new_scrubber |
||||
end |
||||
|
||||
def remove_html(text) |
||||
text = Loofah.fragment(text).scrub!(scrubber).to_s |
||||
text.gsub!('<br>', "\n") |
||||
text.gsub!('</p><p>', "\n\n") |
||||
text.gsub!(/(^<p>|<\/p>$)/, '') |
||||
text |
||||
end |
||||
|
||||
def default_locale(account) |
||||
account.user_locale&.to_sym || I18n.default_locale if account.local? |
||||
end |
||||
end |
@ -1,134 +0,0 @@ |
||||
# frozen_string_literal: true |
||||
|
||||
require 'rails_helper' |
||||
|
||||
describe LanguageDetector do |
||||
describe 'prepare_text' do |
||||
it 'returns unmodified string without special cases' do |
||||
string = 'just a regular string' |
||||
result = described_class.instance.send(:prepare_text, string) |
||||
|
||||
expect(result).to eq string |
||||
end |
||||
|
||||
it 'collapses spacing in strings' do |
||||
string = 'The formatting in this is very odd' |
||||
|
||||
result = described_class.instance.send(:prepare_text, string) |
||||
expect(result).to eq 'The formatting in this is very odd' |
||||
end |
||||
|
||||
it 'strips usernames from strings before detection' do |
||||
string = '@username Yeah, very surreal...! also @friend' |
||||
|
||||
result = described_class.instance.send(:prepare_text, string) |
||||
expect(result).to eq 'Yeah, very surreal...! also' |
||||
end |
||||
|
||||
it 'strips URLs from strings before detection' do |
||||
string = 'Our website is https://example.com and also http://localhost.dev' |
||||
|
||||
result = described_class.instance.send(:prepare_text, string) |
||||
expect(result).to eq 'Our website is and also' |
||||
end |
||||
|
||||
it 'converts #hashtags back to normal text before detection' do |
||||
string = 'Hey look at all the #animals and #FishAndChips' |
||||
|
||||
result = described_class.instance.send(:prepare_text, string) |
||||
expect(result).to eq 'Hey look at all the animals and fish and chips' |
||||
end |
||||
end |
||||
|
||||
describe 'detect' do |
||||
let(:account_without_user_locale) { Fabricate(:user, locale: nil).account } |
||||
let(:account_remote) { Fabricate(:account, domain: 'joinmastodon.org') } |
||||
|
||||
it 'detects english language for basic strings' do |
||||
strings = [ |
||||
"Hello and welcome to mastodon how are you today?", |
||||
"I'd rather not!", |
||||
"a lot of people just want to feel righteous all the time and that's all that matters", |
||||
] |
||||
strings.each do |string| |
||||
result = described_class.instance.detect(string, account_without_user_locale) |
||||
|
||||
expect(result).to eq(:en), string |
||||
end |
||||
end |
||||
|
||||
it 'detects spanish language' do |
||||
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) |
||||
|
||||
expect(result).to eq :es |
||||
end |
||||
|
||||
describe 'when language can\'t be detected' do |
||||
it 'uses nil when sent an empty document' do |
||||
result = described_class.instance.detect('', account_without_user_locale) |
||||
expect(result).to eq nil |
||||
end |
||||
|
||||
describe 'because of a URL' do |
||||
it 'uses nil when sent just a URL' do |
||||
string = 'http://example.com/media/2kFTgOJLXhQf0g2nKB4' |
||||
cld_result = CLD3::NNetLanguageIdentifier.new(0, 2048).find_language(string) |
||||
expect(cld_result).not_to eq :en |
||||
|
||||
result = described_class.instance.detect(string, account_without_user_locale) |
||||
|
||||
expect(result).to eq nil |
||||
end |
||||
end |
||||
|
||||
describe 'with an account' do |
||||
it 'uses the account locale when present' do |
||||
account = double(user_locale: 'fr') |
||||
result = described_class.instance.detect('', account) |
||||
|
||||
expect(result).to eq nil |
||||
end |
||||
|
||||
it 'uses nil when account is present but has no locale' do |
||||
result = described_class.instance.detect('', account_without_user_locale) |
||||
|
||||
expect(result).to eq nil |
||||
end |
||||
end |
||||
|
||||
describe 'with an `en` default locale' do |
||||
it 'uses nil for undetectable string' do |
||||
result = described_class.instance.detect('', account_without_user_locale) |
||||
|
||||
expect(result).to eq nil |
||||
end |
||||
end |
||||
|
||||
describe 'remote user' do |
||||
it 'detects Korean language' do |
||||
string = '안녕하세요' |
||||
result = described_class.instance.detect(string, account_remote) |
||||
|
||||
expect(result).to eq :ko |
||||
end |
||||
end |
||||
|
||||
describe 'with a non-`en` default locale' do |
||||
around(:each) do |example| |
||||
before = I18n.default_locale |
||||
I18n.default_locale = :ja |
||||
example.run |
||||
I18n.default_locale = before |
||||
end |
||||
|
||||
it 'uses nil for undetectable string' do |
||||
string = '' |
||||
result = described_class.instance.detect(string, account_without_user_locale) |
||||
|
||||
expect(result).to eq nil |
||||
end |
||||
end |
||||
end |
||||
end |
||||
end |
Loading…
Reference in new issue