diff --git a/Gemfile b/Gemfile
index 1ddb6f383a..7713bfb2b1 100644
--- a/Gemfile
+++ b/Gemfile
@@ -29,9 +29,7 @@ gem 'addressable', '~> 2.8'
gem 'bootsnap', '~> 1.10.2', require: false
gem 'browser'
gem 'charlock_holmes', '~> 0.7.7'
-gem 'iso-639'
gem 'chewy', '~> 7.2'
-gem 'cld3', '~> 3.4.4'
gem 'devise', '~> 4.8'
gem 'devise-two-factor', '~> 4.0'
diff --git a/Gemfile.lock b/Gemfile.lock
index 18459a1593..2baa12038a 100644
--- a/Gemfile.lock
+++ b/Gemfile.lock
@@ -152,8 +152,6 @@ GEM
elasticsearch (>= 7.12.0, < 7.14.0)
elasticsearch-dsl
chunky_png (1.4.0)
- cld3 (3.4.4)
- ffi (>= 1.1.0, < 1.16.0)
climate_control (0.2.0)
coderay (1.1.3)
color_diff (0.1)
@@ -303,7 +301,6 @@ GEM
terminal-table (>= 1.5.1)
idn-ruby (0.1.4)
ipaddress (0.8.3)
- iso-639 (0.3.5)
jmespath (1.5.0)
json (2.5.1)
json-canonicalization (0.3.0)
@@ -701,7 +698,6 @@ DEPENDENCIES
capybara (~> 3.36)
charlock_holmes (~> 0.7.7)
chewy (~> 7.2)
- cld3 (~> 3.4.4)
climate_control (~> 0.2)
color_diff (~> 0.1)
concurrent-ruby
@@ -729,7 +725,6 @@ DEPENDENCIES
httplog (~> 1.5.0)
i18n-tasks (~> 0.9)
idn-ruby
- iso-639
json-ld
json-ld-preloaded (~> 3.2)
kaminari (~> 1.2)
diff --git a/app/controllers/activitypub/replies_controller.rb b/app/controllers/activitypub/replies_controller.rb
index fde6c861f2..4ff7cfa080 100644
--- a/app/controllers/activitypub/replies_controller.rb
+++ b/app/controllers/activitypub/replies_controller.rb
@@ -63,15 +63,29 @@ class ActivityPub::RepliesController < ActivityPub::BaseController
end
def next_page
- only_other_accounts = !(@replies&.last&.account_id == @account.id && @replies.size == DESCENDANTS_LIMIT)
+ if only_other_accounts?
+ # Only consider remote accounts
+ return nil if @replies.size < DESCENDANTS_LIMIT
- account_status_replies_url(
- @account,
- @status,
- page: true,
- min_id: only_other_accounts && !only_other_accounts? ? nil : @replies&.last&.id,
- only_other_accounts: only_other_accounts
- )
+ account_status_replies_url(
+ @account,
+ @status,
+ page: true,
+ min_id: @replies&.last&.id,
+ only_other_accounts: true
+ )
+ else
+ # For now, we're serving only self-replies, but next page might be other accounts
+ next_only_other_accounts = @replies&.last&.account_id != @account.id || @replies.size < DESCENDANTS_LIMIT
+
+ account_status_replies_url(
+ @account,
+ @status,
+ page: true,
+ min_id: next_only_other_accounts ? nil : @replies&.last&.id,
+ only_other_accounts: next_only_other_accounts
+ )
+ end
end
def page_params
diff --git a/app/controllers/concerns/localized.rb b/app/controllers/concerns/localized.rb
index fe1142f345..f7b62f09c1 100644
--- a/app/controllers/concerns/localized.rb
+++ b/app/controllers/concerns/localized.rb
@@ -7,27 +7,24 @@ module Localized
around_action :set_locale
end
- def set_locale
- locale = current_user.locale if respond_to?(:user_signed_in?) && user_signed_in?
- locale ||= session[:locale] ||= default_locale
- locale = default_locale unless I18n.available_locales.include?(locale.to_sym)
-
- I18n.with_locale(locale) do
- yield
- end
+ def set_locale(&block)
+ I18n.with_locale(requested_locale || I18n.default_locale, &block)
end
private
- def default_locale
- if ENV['DEFAULT_LOCALE'].present?
- I18n.default_locale
- else
- request_locale || I18n.default_locale
- end
+ def requested_locale
+ requested_locale_name = available_locale_or_nil(params[:locale])
+ requested_locale_name ||= available_locale_or_nil(current_user.locale) if respond_to?(:user_signed_in?) && user_signed_in?
+ requested_locale_name ||= http_accept_language if ENV['DEFAULT_LOCALE'].blank?
+ requested_locale_name
end
- def request_locale
- http_accept_language.language_region_compatible_from(I18n.available_locales)
+ def http_accept_language
+ HttpAcceptLanguage::Parser.new(request.headers.fetch('Accept-Language')).language_region_compatible_from(I18n.available_locales) if request.headers.key?('Accept-Language')
+ end
+
+ def available_locale_or_nil(locale_name)
+ locale_name.to_sym if locale_name.present? && I18n.available_locales.include?(locale_name.to_sym)
end
end
diff --git a/app/helpers/languages_helper.rb b/app/helpers/languages_helper.rb
index 7307242086..f3ed7b3140 100644
--- a/app/helpers/languages_helper.rb
+++ b/app/helpers/languages_helper.rb
@@ -1,94 +1,237 @@
# frozen_string_literal: true
module LanguagesHelper
- HUMAN_LOCALES = {
- af: 'Afrikaans',
- ar: 'العربية',
- ast: 'Asturianu',
- bg: 'Български',
- bn: 'বাংলা',
- br: 'Breton',
- ca: 'Català',
- co: 'Corsu',
- cs: 'Čeština',
- cy: 'Cymraeg',
- da: 'Dansk',
- de: 'Deutsch',
- el: 'Ελληνικά',
- en: 'English',
- eo: 'Esperanto',
+ ISO_639_1 = {
+ aa: ['Afar', 'Afaraf'].freeze,
+ ab: ['Abkhaz', 'аҧсуа бызшәа'].freeze,
+ ae: ['Avestan', 'avesta'].freeze,
+ af: ['Afrikaans', 'Afrikaans'].freeze,
+ ak: ['Akan', 'Akan'].freeze,
+ am: ['Amharic', 'አማርኛ'].freeze,
+ an: ['Aragonese', 'aragonés'].freeze,
+ ar: ['Arabic', 'اللغة العربية'].freeze,
+ as: ['Assamese', 'অসমীয়া'].freeze,
+ av: ['Avaric', 'авар мацӀ'].freeze,
+ ay: ['Aymara', 'aymar aru'].freeze,
+ az: ['Azerbaijani', 'azərbaycan dili'].freeze,
+ ba: ['Bashkir', 'башҡорт теле'].freeze,
+ be: ['Belarusian', 'беларуская мова'].freeze,
+ bg: ['Bulgarian', 'български език'].freeze,
+ bh: ['Bihari', 'भोजपुरी'].freeze,
+ bi: ['Bislama', 'Bislama'].freeze,
+ bm: ['Bambara', 'bamanankan'].freeze,
+ bn: ['Bengali', 'বাংলা'].freeze,
+ bo: ['Tibetan', 'བོད་ཡིག'].freeze,
+ br: ['Breton', 'brezhoneg'].freeze,
+ bs: ['Bosnian', 'bosanski jezik'].freeze,
+ ca: ['Catalan', 'Català'].freeze,
+ ce: ['Chechen', 'нохчийн мотт'].freeze,
+ ch: ['Chamorro', 'Chamoru'].freeze,
+ co: ['Corsican', 'corsu'].freeze,
+ cr: ['Cree', 'ᓀᐦᐃᔭᐍᐏᐣ'].freeze,
+ cs: ['Czech', 'čeština'].freeze,
+ cu: ['Old Church Slavonic', 'ѩзыкъ словѣньскъ'].freeze,
+ cv: ['Chuvash', 'чӑваш чӗлхи'].freeze,
+ cy: ['Welsh', 'Cymraeg'].freeze,
+ da: ['Danish', 'dansk'].freeze,
+ de: ['German', 'Deutsch'].freeze,
+ dv: ['Divehi', 'Dhivehi'].freeze,
+ dz: ['Dzongkha', 'རྫོང་ཁ'].freeze,
+ ee: ['Ewe', 'Eʋegbe'].freeze,
+ el: ['Greek', 'Ελληνικά'].freeze,
+ en: ['English', 'English'].freeze,
+ eo: ['Esperanto', 'Esperanto'].freeze,
+ es: ['Spanish', 'Español'].freeze,
+ et: ['Estonian', 'eesti'].freeze,
+ eu: ['Basque', 'euskara'].freeze,
+ fa: ['Persian', 'فارسی'].freeze,
+ ff: ['Fula', 'Fulfulde'].freeze,
+ fi: ['Finnish', 'suomi'].freeze,
+ fj: ['Fijian', 'Vakaviti'].freeze,
+ fo: ['Faroese', 'føroyskt'].freeze,
+ fr: ['French', 'Français'].freeze,
+ fy: ['Western Frisian', 'Frysk'].freeze,
+ ga: ['Irish', 'Gaeilge'].freeze,
+ gd: ['Scottish Gaelic', 'Gàidhlig'].freeze,
+ gl: ['Galician', 'galego'].freeze,
+ gu: ['Gujarati', 'ગુજરાતી'].freeze,
+ gv: ['Manx', 'Gaelg'].freeze,
+ ha: ['Hausa', 'هَوُسَ'].freeze,
+ he: ['Hebrew', 'עברית'].freeze,
+ hi: ['Hindi', 'हिन्दी'].freeze,
+ ho: ['Hiri Motu', 'Hiri Motu'].freeze,
+ hr: ['Croatian', 'Hrvatski'].freeze,
+ ht: ['Haitian', 'Kreyòl ayisyen'].freeze,
+ hu: ['Hungarian', 'magyar'].freeze,
+ hy: ['Armenian', 'Հայերեն'].freeze,
+ hz: ['Herero', 'Otjiherero'].freeze,
+ ia: ['Interlingua', 'Interlingua'].freeze,
+ id: ['Indonesian', 'Bahasa Indonesia'].freeze,
+ ie: ['Interlingue', 'Interlingue'].freeze,
+ ig: ['Igbo', 'Asụsụ Igbo'].freeze,
+ ii: ['Nuosu', 'ꆈꌠ꒿ Nuosuhxop'].freeze,
+ ik: ['Inupiaq', 'Iñupiaq'].freeze,
+ io: ['Ido', 'Ido'].freeze,
+ is: ['Icelandic', 'Íslenska'].freeze,
+ it: ['Italian', 'Italiano'].freeze,
+ iu: ['Inuktitut', 'ᐃᓄᒃᑎᑐᑦ'].freeze,
+ ja: ['Japanese', '日本語'].freeze,
+ jv: ['Javanese', 'basa Jawa'].freeze,
+ ka: ['Georgian', 'ქართული'].freeze,
+ kg: ['Kongo', 'Kikongo'].freeze,
+ ki: ['Kikuyu', 'Gĩkũyũ'].freeze,
+ kj: ['Kwanyama', 'Kuanyama'].freeze,
+ kk: ['Kazakh', 'қазақ тілі'].freeze,
+ kl: ['Kalaallisut', 'kalaallisut'].freeze,
+ km: ['Khmer', 'ខេមរភាសា'].freeze,
+ kn: ['Kannada', 'ಕನ್ನಡ'].freeze,
+ ko: ['Korean', '한국어'].freeze,
+ kr: ['Kanuri', 'Kanuri'].freeze,
+ ks: ['Kashmiri', 'कश्मीरी'].freeze,
+ ku: ['Kurdish', 'Kurdî'].freeze,
+ kv: ['Komi', 'коми кыв'].freeze,
+ kw: ['Cornish', 'Kernewek'].freeze,
+ ky: ['Kyrgyz', 'Кыргызча'].freeze,
+ la: ['Latin', 'latine'].freeze,
+ lb: ['Luxembourgish', 'Lëtzebuergesch'].freeze,
+ lg: ['Ganda', 'Luganda'].freeze,
+ li: ['Limburgish', 'Limburgs'].freeze,
+ ln: ['Lingala', 'Lingála'].freeze,
+ lo: ['Lao', 'ພາສາ'].freeze,
+ lt: ['Lithuanian', 'lietuvių kalba'].freeze,
+ lu: ['Luba-Katanga', 'Tshiluba'].freeze,
+ lv: ['Latvian', 'latviešu valoda'].freeze,
+ mg: ['Malagasy', 'fiteny malagasy'].freeze,
+ mh: ['Marshallese', 'Kajin M̧ajeļ'].freeze,
+ mi: ['Māori', 'te reo Māori'].freeze,
+ mk: ['Macedonian', 'македонски јазик'].freeze,
+ ml: ['Malayalam', 'മലയാളം'].freeze,
+ mn: ['Mongolian', 'Монгол хэл'].freeze,
+ mr: ['Marathi', 'मराठी'].freeze,
+ ms: ['Malay', 'Bahasa Malaysia'].freeze,
+ mt: ['Maltese', 'Malti'].freeze,
+ my: ['Burmese', 'ဗမာစာ'].freeze,
+ na: ['Nauru', 'Ekakairũ Naoero'].freeze,
+ nb: ['Norwegian Bokmål', 'Norsk bokmål'].freeze,
+ nd: ['Northern Ndebele', 'isiNdebele'].freeze,
+ ne: ['Nepali', 'नेपाली'].freeze,
+ ng: ['Ndonga', 'Owambo'].freeze,
+ nl: ['Dutch', 'Nederlands'].freeze,
+ nn: ['Norwegian Nynorsk', 'Norsk nynorsk'].freeze,
+ no: ['Norwegian', 'Norsk'].freeze,
+ nr: ['Southern Ndebele', 'isiNdebele'].freeze,
+ nv: ['Navajo', 'Diné bizaad'].freeze,
+ ny: ['Chichewa', 'chiCheŵa'].freeze,
+ oc: ['Occitan', 'occitan'].freeze,
+ oj: ['Ojibwe', 'ᐊᓂᔑᓈᐯᒧᐎᓐ'].freeze,
+ om: ['Oromo', 'Afaan Oromoo'].freeze,
+ or: ['Oriya', 'ଓଡ଼ିଆ'].freeze,
+ os: ['Ossetian', 'ирон æвзаг'].freeze,
+ pa: ['Panjabi', 'ਪੰਜਾਬੀ'].freeze,
+ pi: ['Pāli', 'पाऴि'].freeze,
+ pl: ['Polish', 'Polski'].freeze,
+ ps: ['Pashto', 'پښتو'].freeze,
+ pt: ['Portuguese', 'Português'].freeze,
+ qu: ['Quechua', 'Runa Simi'].freeze,
+ rm: ['Romansh', 'rumantsch grischun'].freeze,
+ rn: ['Kirundi', 'Ikirundi'].freeze,
+ ro: ['Romanian', 'Română'].freeze,
+ ru: ['Russian', 'Русский'].freeze,
+ rw: ['Kinyarwanda', 'Ikinyarwanda'].freeze,
+ sa: ['Sanskrit', 'संस्कृतम्'].freeze,
+ sc: ['Sardinian', 'sardu'].freeze,
+ sd: ['Sindhi', 'सिन्धी'].freeze,
+ se: ['Northern Sami', 'Davvisámegiella'].freeze,
+ sg: ['Sango', 'yângâ tî sängö'].freeze,
+ si: ['Sinhala', 'සිංහල'].freeze,
+ sk: ['Slovak', 'slovenčina'].freeze,
+ sl: ['Slovenian', 'slovenščina'].freeze,
+ sn: ['Shona', 'chiShona'].freeze,
+ so: ['Somali', 'Soomaaliga'].freeze,
+ sq: ['Albanian', 'Shqip'].freeze,
+ sr: ['Serbian', 'српски језик'].freeze,
+ ss: ['Swati', 'SiSwati'].freeze,
+ st: ['Southern Sotho', 'Sesotho'].freeze,
+ su: ['Sundanese', 'Basa Sunda'].freeze,
+ sv: ['Swedish', 'Svenska'].freeze,
+ sw: ['Swahili', 'Kiswahili'].freeze,
+ ta: ['Tamil', 'தமிழ்'].freeze,
+ te: ['Telugu', 'తెలుగు'].freeze,
+ tg: ['Tajik', 'тоҷикӣ'].freeze,
+ th: ['Thai', 'ไทย'].freeze,
+ ti: ['Tigrinya', 'ትግርኛ'].freeze,
+ tk: ['Turkmen', 'Türkmen'].freeze,
+ tl: ['Tagalog', 'Wikang Tagalog'].freeze,
+ tn: ['Tswana', 'Setswana'].freeze,
+ to: ['Tonga', 'faka Tonga'].freeze,
+ tr: ['Turkish', 'Türkçe'].freeze,
+ ts: ['Tsonga', 'Xitsonga'].freeze,
+ tt: ['Tatar', 'татар теле'].freeze,
+ tw: ['Twi', 'Twi'].freeze,
+ ty: ['Tahitian', 'Reo Tahiti'].freeze,
+ ug: ['Uyghur', 'ئۇيغۇرچە'].freeze,
+ uk: ['Ukrainian', 'Українська'].freeze,
+ ur: ['Urdu', 'اردو'].freeze,
+ uz: ['Uzbek', 'Ўзбек'].freeze,
+ ve: ['Venda', 'Tshivenḓa'].freeze,
+ vi: ['Vietnamese', 'Tiếng Việt'].freeze,
+ vo: ['Volapük', 'Volapük'].freeze,
+ wa: ['Walloon', 'walon'].freeze,
+ wo: ['Wolof', 'Wollof'].freeze,
+ xh: ['Xhosa', 'isiXhosa'].freeze,
+ yi: ['Yiddish', 'ייִדיש'].freeze,
+ yo: ['Yoruba', 'Yorùbá'].freeze,
+ za: ['Zhuang', 'Saɯ cueŋƅ'].freeze,
+ zh: ['Chinese', '中文'].freeze,
+ zu: ['Zulu', 'isiZulu'].freeze,
+ }.freeze
+
+ ISO_639_3 = {
+ ast: ['Asturian', 'Asturianu'].freeze,
+ kab: ['Kabyle', 'Taqbaylit'].freeze,
+ kmr: ['Northern Kurdish', 'Kurmancî'].freeze,
+ zgh: ['Standard Moroccan Tamazight', 'ⵜⴰⵎⴰⵣⵉⵖⵜ'].freeze,
+ }.freeze
+
+ SUPPORTED_LOCALES = {}.merge(ISO_639_1).merge(ISO_639_3).freeze
+
+ # For ISO-639-1 and ISO-639-3 language codes, we have their official
+ # names, but for some translations, we need the names of the
+ # regional variants specifically
+ REGIONAL_LOCALE_NAMES = {
'es-AR': 'Español (Argentina)',
'es-MX': 'Español (México)',
- es: 'Español',
- et: 'Eesti',
- eu: 'Euskara',
- fa: 'فارسی',
- fi: 'Suomi',
- fr: 'Français',
- ga: 'Gaeilge',
- gd: 'Gàidhlig',
- gl: 'Galego',
- he: 'עברית',
- hi: 'हिन्दी',
- hr: 'Hrvatski',
- hu: 'Magyar',
- hy: 'Հայերեն',
- id: 'Bahasa Indonesia',
- io: 'Ido',
- is: 'Íslenska',
- it: 'Italiano',
- ja: '日本語',
- ka: 'ქართული',
- kab: 'Taqbaylit',
- kk: 'Қазақша',
- kmr: 'Kurmancî',
- kn: 'ಕನ್ನಡ',
- ko: '한국어',
- ku: 'سۆرانی',
- lt: 'Lietuvių',
- lv: 'Latviešu',
- mk: 'Македонски',
- ml: 'മലയാളം',
- mr: 'मराठी',
- ms: 'Bahasa Melayu',
- nl: 'Nederlands',
- nn: 'Nynorsk',
- no: 'Norsk',
- oc: 'Occitan',
- pl: 'Polski',
'pt-BR': 'Português (Brasil)',
'pt-PT': 'Português (Portugal)',
- pt: 'Português',
- ro: 'Română',
- ru: 'Русский',
- sa: 'संस्कृतम्',
- sc: 'Sardu',
- si: 'සිංහල',
- sk: 'Slovenčina',
- sl: 'Slovenščina',
- sq: 'Shqip',
'sr-Latn': 'Srpski (latinica)',
- sr: 'Српски',
- sv: 'Svenska',
- ta: 'தமிழ்',
- te: 'తెలుగు',
- th: 'ไทย',
- tr: 'Türkçe',
- uk: 'Українська',
- ur: 'اُردُو',
- vi: 'Tiếng Việt',
- zgh: 'ⵜⴰⵎⴰⵣⵉⵖⵜ',
'zh-CN': '简体中文',
'zh-HK': '繁體中文（香港）',
'zh-TW': '繁體中文（臺灣）',
- zh: '中文',
}.freeze
def human_locale(locale)
if locale == 'und'
I18n.t('generic.none')
+ elsif (supported_locale = SUPPORTED_LOCALES[locale.to_sym])
+ supported_locale[1]
+ elsif (regional_locale = REGIONAL_LOCALE_NAMES[locale.to_sym])
+ regional_locale
else
- HUMAN_LOCALES[locale.to_sym] || locale
+ locale
end
end
+
+ def valid_locale_or_nil(str)
+ return if str.blank?
+
+ code, = str.to_s.split(/[_-]/) # Strip out the region from e.g. en_US or ja-JP
+
+ return unless valid_locale?(code)
+
+ code
+ end
+
+ def valid_locale?(locale)
+ SUPPORTED_LOCALES.key?(locale.to_sym)
+ end
end
diff --git a/app/helpers/settings_helper.rb b/app/helpers/settings_helper.rb
index 23739d1cd4..3d5592867c 100644
--- a/app/helpers/settings_helper.rb
+++ b/app/helpers/settings_helper.rb
@@ -2,7 +2,7 @@
module SettingsHelper
def filterable_languages
- LanguageDetector.instance.language_names.select(&LanguagesHelper::HUMAN_LOCALES.method(:key?))
+ LanguagesHelper::SUPPORTED_LOCALES.keys
end
def hash_to_object(hash)
diff --git a/app/lib/activitypub/activity/create.rb b/app/lib/activitypub/activity/create.rb
index ad273c20bb..cf31b6ff62 100644
--- a/app/lib/activitypub/activity/create.rb
+++ b/app/lib/activitypub/activity/create.rb
@@ -112,7 +112,7 @@ class ActivityPub::Activity::Create < ActivityPub::Activity
url: @status_parser.url || @status_parser.uri,
account: @account,
text: converted_object_type? ? converted_text : (@status_parser.text || ''),
- language: @status_parser.language || detected_language,
+ language: @status_parser.language,
spoiler_text: converted_object_type? ? '' : (@status_parser.spoiler_text || ''),
created_at: @status_parser.created_at,
edited_at: @status_parser.edited_at,
@@ -370,10 +370,6 @@ class ActivityPub::Activity::Create < ActivityPub::Activity
Formatter.instance.linkify([@status_parser.title.presence, @status_parser.spoiler_text.presence, @status_parser.url || @status_parser.uri].compact.join("\n\n"))
end
- def detected_language
- LanguageDetector.instance.detect(@status_parser.text, @account) if supported_object_type?
- end
-
def unsupported_media_type?(mime_type)
mime_type.present? && !MediaAttachment.supported_mime_types.include?(mime_type)
end
diff --git a/app/lib/language_detector.rb b/app/lib/language_detector.rb
deleted file mode 100644
index 40452eddc9..0000000000
--- a/app/lib/language_detector.rb
+++ /dev/null
@@ -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
diff --git a/app/lib/link_details_extractor.rb b/app/lib/link_details_extractor.rb
index 56ad0717b7..fabbd244df 100644
--- a/app/lib/link_details_extractor.rb
+++ b/app/lib/link_details_extractor.rb
@@ -2,6 +2,20 @@
class LinkDetailsExtractor
include ActionView::Helpers::TagHelper
+ include LanguagesHelper
+
+ # Some publications wrap their JSON-LD data in their <script> tags
+ # in commented-out CDATA blocks, they need to be removed before
+ # attempting to parse JSON
+ CDATA_JUNK_PATTERN = %r{^[\s]*(
+ (/\*[\s]*<!\[CDATA\[[\s]*\*/) # Block comment style opening
+ |
+ (//[\s]*<!\[CDATA\[) # Single-line comment style opening
+ |
+ (/\*[\s]*\]\]>[\s]*\*/) # Block comment style closing
+ |
+ (//[\s]*\]\]>) # Single-line comment style closing
+ )[\s]*$}x
class StructuredData
SUPPORTED_TYPES = %w(
@@ -61,6 +75,10 @@ class LinkDetailsExtractor
publisher.dig('logo', 'url')
end
+ def valid?
+ json.present?
+ end
+
private
def author
@@ -134,11 +152,11 @@ class LinkDetailsExtractor
end
def title
- structured_data&.headline || opengraph_tag('og:title') || document.xpath('//title').map(&:content).first
+ html_entities.decode(structured_data&.headline || opengraph_tag('og:title') || document.xpath('//title').map(&:content).first)
end
def description
- structured_data&.description || opengraph_tag('og:description') || meta_tag('description')
+ html_entities.decode(structured_data&.description || opengraph_tag('og:description') || meta_tag('description'))
end
def image
@@ -146,11 +164,11 @@ class LinkDetailsExtractor
end
def canonical_url
- valid_url_or_nil(opengraph_tag('og:url') || link_tag('canonical'), same_origin_only: true) || @original_url.to_s
+ valid_url_or_nil(link_tag('canonical') || opengraph_tag('og:url'), same_origin_only: true) || @original_url.to_s
end
def provider_name
- structured_data&.publisher_name || opengraph_tag('og:site_name')
+ html_entities.decode(structured_data&.publisher_name || opengraph_tag('og:site_name'))
end
def provider_url
@@ -158,7 +176,7 @@ class LinkDetailsExtractor
end
def author_name
- structured_data&.author_name || opengraph_tag('og:author') || opengraph_tag('og:author:username')
+ html_entities.decode(structured_data&.author_name || opengraph_tag('og:author') || opengraph_tag('og:author:username'))
end
def author_url
@@ -201,14 +219,6 @@ class LinkDetailsExtractor
nil
end
- def valid_locale_or_nil(str)
- return nil if str.blank?
-
- code, = str.split(/_-/) # Strip out the region from e.g. en_US or ja-JA
- locale = ISO_639.find(code)
- locale&.alpha2
- end
-
def link_tag(name)
document.xpath("//link[@rel=\"#{name}\"]").map { |link| link['href'] }.first
end
@@ -223,10 +233,24 @@ class LinkDetailsExtractor
def structured_data
@structured_data ||= begin
- json_ld = document.xpath('//script[@type="application/ld+json"]').map(&:content).first
- json_ld.present? ? StructuredData.new(json_ld) : nil
- rescue Oj::ParseError
- nil
+ # Some publications have more than one JSON-LD definition on the page,
+ # and some of those definitions aren't valid JSON either, so we have
+ # to loop through here until we find something that is the right type
+ # and doesn't break
+ document.xpath('//script[@type="application/ld+json"]').filter_map do |element|
+ json_ld = element.content&.gsub(CDATA_JUNK_PATTERN, '')
+
+ next if json_ld.blank?
+
+ structured_data = StructuredData.new(html_entities.decode(json_ld))
+
+ next unless structured_data.valid?
+
+ structured_data
+ rescue Oj::ParseError, EncodingError
+ Rails.logger.debug("Invalid JSON-LD in #{@original_url}")
+ next
+ end.first
end
end
@@ -246,4 +270,8 @@ class LinkDetailsExtractor
detector.strip_tags = true
end
end
+
+ def html_entities
+ @html_entities ||= HTMLEntities.new
+ end
end
diff --git a/app/models/account_suggestions/global_source.rb b/app/models/account_suggestions/global_source.rb
index ac764de50f..03ed1b6c2a 100644
--- a/app/models/account_suggestions/global_source.rb
+++ b/app/models/account_suggestions/global_source.rb
@@ -6,7 +6,7 @@ class AccountSuggestions::GlobalSource < AccountSuggestions::Source
end
def get(account, skip_account_ids: [], limit: 40)
- account_ids = account_ids_for_locale(account.user_locale) - [account.id] - skip_account_ids
+ account_ids = account_ids_for_locale(I18n.locale.to_str.split(/[_-]/).first) - [account.id] - skip_account_ids
as_ordered_suggestions(
scope(account).where(id: account_ids),
diff --git a/app/models/user.rb b/app/models/user.rb
index 9afdc481d7..ee20e293e8 100644
--- a/app/models/user.rb
+++ b/app/models/user.rb
@@ -245,6 +245,10 @@ class User < ApplicationRecord
save!
end
+ def preferred_posting_language
+ settings.default_language || locale
+ end
+
def setting_default_privacy
settings.default_privacy || (account.locked? ? 'private' : 'public')
end
diff --git a/app/services/activitypub/process_status_update_service.rb b/app/services/activitypub/process_status_update_service.rb
index 9779281272..b1cea1cdfc 100644
--- a/app/services/activitypub/process_status_update_service.rb
+++ b/app/services/activitypub/process_status_update_service.rb
@@ -120,7 +120,7 @@ class ActivityPub::ProcessStatusUpdateService < BaseService
@status.text = @status_parser.text || ''
@status.spoiler_text = @status_parser.spoiler_text || ''
@status.sensitive = @account.sensitized? || @status_parser.sensitive || false
- @status.language = @status_parser.language || detected_language
+ @status.language = @status_parser.language
@status.edited_at = @status_parser.edited_at || Time.now.utc if significant_changes?
@status.save!
@@ -210,10 +210,6 @@ class ActivityPub::ProcessStatusUpdateService < BaseService
{ redis: Redis.current, key: "create:#{@uri}", autorelease: 15.minutes.seconds }
end
- def detected_language
- LanguageDetector.instance.detect(@status_parser.text, @account)
- end
-
def create_previous_edit!
# We only need to create a previous edit when no previous edits exist, e.g.
# when the status has never been edited. For other cases, we always create
diff --git a/app/services/post_status_service.rb b/app/services/post_status_service.rb
index 9d26e0f5bf..c5061dd635 100644
--- a/app/services/post_status_service.rb
+++ b/app/services/post_status_service.rb
@@ -2,6 +2,7 @@
class PostStatusService < BaseService
include Redisable
+ include LanguagesHelper
MIN_SCHEDULE_OFFSET = 5.minutes.freeze
@@ -118,10 +119,6 @@ class PostStatusService < BaseService
raise Mastodon::ValidationError, I18n.t('media_attachments.validations.not_ready') if @media.any?(&:not_processed?)
end
- def language_from_option(str)
- ISO_639.find(str)&.alpha2
- end
-
def process_mentions_service
ProcessMentionsService.new
end
@@ -174,7 +171,7 @@ class PostStatusService < BaseService
sensitive: @sensitive,
spoiler_text: @options[:spoiler_text] || '',
visibility: @visibility,
- language: language_from_option(@options[:language]) || @account.user&.setting_default_language&.presence || LanguageDetector.instance.detect(@text, @account),
+ language: valid_locale_or_nil(@options[:language].presence || @account.user&.preferred_posting_language || I18n.default_locale),
application: @options[:application],
content_type: @options[:content_type] || @account.user&.setting_default_content_type,
rate_limit: @options[:with_rate_limit],
diff --git a/app/validators/import_validator.rb b/app/validators/import_validator.rb
index a182abfa50..9f19aee2ae 100644
--- a/app/validators/import_validator.rb
+++ b/app/validators/import_validator.rb
@@ -1,5 +1,7 @@
# frozen_string_literal: true
+require 'csv'
+
class ImportValidator < ActiveModel::Validator
KNOWN_HEADERS = [
'Account address',
diff --git a/app/views/admin/follow_recommendations/show.html.haml b/app/views/admin/follow_recommendations/show.html.haml
index 2878f07d76..d1c160bd28 100644
--- a/app/views/admin/follow_recommendations/show.html.haml
+++ b/app/views/admin/follow_recommendations/show.html.haml
@@ -10,7 +10,7 @@
.filter-subset.filter-subset--with-select
%strong= t('admin.follow_recommendations.language')
.input.select.optional
- = select_tag :language, options_for_select(I18n.available_locales.map { |key| [human_locale(key), key]}, @language)
+ = select_tag :language, options_for_select(I18n.available_locales.map { |key| key.to_s.split(/[_-]/).first.to_sym }.uniq.map { |key| [human_locale(key), key]}, @language)
.filter-subset
%strong= t('admin.follow_recommendations.status')
diff --git a/app/views/settings/preferences/other/show.html.haml b/app/views/settings/preferences/other/show.html.haml
index 3b5c7016d8..76ff2bcbcf 100644
--- a/app/views/settings/preferences/other/show.html.haml
+++ b/app/views/settings/preferences/other/show.html.haml
@@ -27,7 +27,7 @@
= f.input :setting_default_privacy, collection: Status.selectable_visibilities, wrapper: :with_label, include_blank: false, label_method: lambda { |visibility| safe_join([I18n.t("statuses.visibilities.#{visibility}"), I18n.t("statuses.visibilities.#{visibility}_long")], ' - ') }, required: false, hint: false
.fields-group.fields-row__column.fields-row__column-6
- = f.input :setting_default_language, collection: [nil] + filterable_languages.sort, wrapper: :with_label, label_method: lambda { |locale| locale.nil? ? I18n.t('statuses.language_detection') : human_locale(locale) }, required: false, include_blank: false, hint: false
+ = f.input :setting_default_language, collection: [nil] + filterable_languages, wrapper: :with_label, label_method: lambda { |locale| locale.nil? ? I18n.t('statuses.default_language') : human_locale(locale) }, required: false, include_blank: false, hint: false
.fields-group
= f.input :setting_default_sensitive, as: :boolean, wrapper: :with_label
@@ -41,7 +41,7 @@
%h4= t 'preferences.public_timelines'
.fields-group
- = f.input :chosen_languages, collection: filterable_languages.sort, wrapper: :with_block_label, include_blank: false, label_method: lambda { |locale| human_locale(locale) }, required: false, as: :check_boxes, collection_wrapper_tag: 'ul', item_wrapper_tag: 'li'
+ = f.input :chosen_languages, collection: filterable_languages, wrapper: :with_block_label, include_blank: false, label_method: lambda { |locale| human_locale(locale) }, required: false, as: :check_boxes, collection_wrapper_tag: 'ul', item_wrapper_tag: 'li'
.actions
= f.button :button, t('generic.save_changes'), type: :submit
diff --git a/app/workers/activitypub/processing_worker.rb b/app/workers/activitypub/processing_worker.rb
index cef5953194..37e316354b 100644
--- a/app/workers/activitypub/processing_worker.rb
+++ b/app/workers/activitypub/processing_worker.rb
@@ -6,7 +6,10 @@ class ActivityPub::ProcessingWorker
sidekiq_options backtrace: true, retry: 8
def perform(account_id, body, delivered_to_account_id = nil)
- ActivityPub::ProcessCollectionService.new.call(body, Account.find(account_id), override_timestamps: true, delivered_to_account_id: delivered_to_account_id, delivery: true)
+ account = Account.find_by(id: account_id)
+ return if account.nil?
+
+ ActivityPub::ProcessCollectionService.new.call(body, account, override_timestamps: true, delivered_to_account_id: delivered_to_account_id, delivery: true)
rescue ActiveRecord::RecordInvalid => e
Rails.logger.debug "Error processing incoming ActivityPub object: #{e}"
end
diff --git a/app/workers/scheduler/accounts_statuses_cleanup_scheduler.rb b/app/workers/scheduler/accounts_statuses_cleanup_scheduler.rb
index f42d4bca6a..7195f0ff97 100644
--- a/app/workers/scheduler/accounts_statuses_cleanup_scheduler.rb
+++ b/app/workers/scheduler/accounts_statuses_cleanup_scheduler.rb
@@ -66,7 +66,7 @@ class Scheduler::AccountsStatusesCleanupScheduler
end
def compute_budget
- threads = Sidekiq::ProcessSet.new.filter { |x| x['queues'].include?('push') }.map { |x| x['concurrency'] }.sum
+ threads = Sidekiq::ProcessSet.new.select { |x| x['queues'].include?('push') }.map { |x| x['concurrency'] }.sum
[PER_THREAD_BUDGET * threads, MAX_BUDGET].min
end
diff --git a/app/workers/scheduler/follow_recommendations_scheduler.rb b/app/workers/scheduler/follow_recommendations_scheduler.rb
index effc63e598..084619cbd9 100644
--- a/app/workers/scheduler/follow_recommendations_scheduler.rb
+++ b/app/workers/scheduler/follow_recommendations_scheduler.rb
@@ -16,28 +16,33 @@ class Scheduler::FollowRecommendationsScheduler
AccountSummary.refresh
FollowRecommendation.refresh
- fallback_recommendations = FollowRecommendation.order(rank: :desc).limit(SET_SIZE).index_by(&:account_id)
+ fallback_recommendations = FollowRecommendation.order(rank: :desc).limit(SET_SIZE)
- I18n.available_locales.each do |locale|
+ I18n.available_locales.map { |locale| locale.to_s.split(/[_-]/).first }.uniq.each do |locale|
recommendations = begin
if AccountSummary.safe.filtered.localized(locale).exists? # We can skip the work if no accounts with that language exist
- FollowRecommendation.localized(locale).order(rank: :desc).limit(SET_SIZE).index_by(&:account_id)
+ FollowRecommendation.localized(locale).order(rank: :desc).limit(SET_SIZE).map { |recommendation| [recommendation.account_id, recommendation.rank] }
else
- {}
+ []
end
end
# Use language-agnostic results if there are not enough language-specific ones
- missing = SET_SIZE - recommendations.keys.size
+ missing = SET_SIZE - recommendations.size
+
+ if missing.positive? && fallback_recommendations.size.positive?
+ max_fallback_rank = fallback_recommendations.first.rank || 0
+
+ # Language-specific results should be above language-agnostic ones,
+ # otherwise language-agnostic ones will always overshadow them
+ recommendations.map! { |(account_id, rank)| [account_id, rank + max_fallback_rank] }
- if missing.positive?
added = 0
- # Avoid duplicate results
- fallback_recommendations.each_value do |recommendation|
- next if recommendations.key?(recommendation.account_id)
+ fallback_recommendations.each do |recommendation|
+ next if recommendations.any? { |(account_id, _)| account_id == recommendation.account_id }
- recommendations[recommendation.account_id] = recommendation
+ recommendations << [recommendation.account_id, recommendation.rank]
added += 1
break if added >= missing
@@ -47,8 +52,8 @@ class Scheduler::FollowRecommendationsScheduler
redis.pipelined do
redis.del(key(locale))
- recommendations.each_value do |recommendation|
- redis.zadd(key(locale), recommendation.rank, recommendation.account_id)
+ recommendations.each do |(account_id, rank)|
+ redis.zadd(key(locale), rank, account_id)
end
end
end
diff --git a/config/application.rb b/config/application.rb
index 5617228849..c6f7751622 100644
--- a/config/application.rb
+++ b/config/application.rb
@@ -149,10 +149,14 @@ module Mastodon
:'zh-TW',
]
- config.i18n.default_locale = ENV['DEFAULT_LOCALE']&.to_sym
+ config.i18n.default_locale = begin
+ custom_default_locale = ENV['DEFAULT_LOCALE']&.to_sym
- unless config.i18n.available_locales.include?(config.i18n.default_locale)
- config.i18n.default_locale = :en
+ if config.i18n.available_locales.include?(custom_default_locale)
+ custom_default_locale
+ else
+ :en
+ end
end
# config.paths.add File.join('app', 'api'), glob: File.join('**', '*.rb')
@@ -169,7 +173,6 @@ module Mastodon
Doorkeeper::Application.send :include, ApplicationExtension
Doorkeeper::AccessToken.send :include, AccessTokenExtension
Devise::FailureApp.send :include, AbstractController::Callbacks
- Devise::FailureApp.send :include, HttpAcceptLanguage::EasyAccess
Devise::FailureApp.send :include, Localized
end
end
diff --git a/config/locales/en.yml b/config/locales/en.yml
index 85aa87c7a6..600090e784 100644
--- a/config/locales/en.yml
+++ b/config/locales/en.yml
@@ -1307,13 +1307,13 @@ en:
other: "%{count} videos"
boosted_from_html: Boosted from %{acct_link}
content_warning: 'Content warning: %{warning}'
+ default_language: Same as interface language
disallowed_hashtags:
one: 'contained a disallowed hashtag: %{tags}'
other: 'contained the disallowed hashtags: %{tags}'
edited_at: Edited %{date}
errors:
in_reply_not_found: The post you are trying to reply to does not appear to exist.
- language_detection: Automatically detect language
open_in_web: Open in web
over_character_limit: character limit of %{max} exceeded
pin_errors:
diff --git a/lib/tasks/repo.rake b/lib/tasks/repo.rake
index bbf7f20ee7..795b54c59c 100644
--- a/lib/tasks/repo.rake
+++ b/lib/tasks/repo.rake
@@ -96,7 +96,8 @@ namespace :repo do
end.uniq.compact
missing_available_locales = locales_in_files - I18n.available_locales
- missing_locale_names = I18n.available_locales.reject { |locale| LanguagesHelper::HUMAN_LOCALES.key?(locale) }
+ supported_locale_codes = Set.new(LanguagesHelper::SUPPORTED_LOCALES.keys + LanguagesHelper::REGIONAL_LOCALE_NAMES.keys)
+ missing_locale_names = I18n.available_locales.reject { |locale| supported_locale_codes.include?(locale) }
critical = false
@@ -123,7 +124,7 @@ namespace :repo do
unless missing_locale_names.empty?
puts pastel.yellow("You are missing human-readable names for these locales: #{pastel.bold(missing_locale_names.join(', '))}")
- puts pastel.yellow("Add them to #{pastel.bold('HUMAN_LOCALES')} in app/helpers/settings_helper.rb or remove the locales from #{pastel.bold('I18n.available_locales')} in config/application.rb")
+ puts pastel.yellow("Add them to app/helpers/languages_helper.rb or remove the locales from #{pastel.bold('I18n.available_locales')} in config/application.rb")
end
if critical
diff --git a/spec/controllers/activitypub/replies_controller_spec.rb b/spec/controllers/activitypub/replies_controller_spec.rb
index bf82fd0207..a2c7f336ff 100644
--- a/spec/controllers/activitypub/replies_controller_spec.rb
+++ b/spec/controllers/activitypub/replies_controller_spec.rb
@@ -4,8 +4,9 @@ require 'rails_helper'
RSpec.describe ActivityPub::RepliesController, type: :controller do
let(:status) { Fabricate(:status, visibility: parent_visibility) }
- let(:remote_reply_id) { nil }
- let(:remote_account) { nil }
+ let(:remote_account) { Fabricate(:account, domain: 'foobar.com') }
+ let(:remote_reply_id) { 'https://foobar.com/statuses/1234' }
+ let(:remote_querier) { nil }
shared_examples 'cachable response' do
it 'does not set cookies' do
@@ -23,8 +24,151 @@ RSpec.describe ActivityPub::RepliesController, type: :controller do
end
end
+ shared_examples 'common behavior' do
+ context 'when status is private' do
+ let(:parent_visibility) { :private }
+
+ it 'returns http not found' do
+ expect(response).to have_http_status(404)
+ end
+ end
+
+ context 'when status is direct' do
+ let(:parent_visibility) { :direct }
+
+ it 'returns http not found' do
+ expect(response).to have_http_status(404)
+ end
+ end
+ end
+
+ shared_examples 'disallowed access' do
+ context 'when status is public' do
+ let(:parent_visibility) { :public }
+
+ it 'returns http not found' do
+ expect(response).to have_http_status(404)
+ end
+ end
+
+ it_behaves_like 'common behavior'
+ end
+
+ shared_examples 'allowed access' do
+ context 'when account is permanently suspended' do
+ let(:parent_visibility) { :public }
+
+ before do
+ status.account.suspend!
+ status.account.deletion_request.destroy
+ end
+
+ it 'returns http gone' do
+ expect(response).to have_http_status(410)
+ end
+ end
+
+ context 'when account is temporarily suspended' do
+ let(:parent_visibility) { :public }
+
+ before do
+ status.account.suspend!
+ end
+
+ it 'returns http forbidden' do
+ expect(response).to have_http_status(403)
+ end
+ end
+
+ context 'when status is public' do
+ let(:parent_visibility) { :public }
+ let(:json) { body_as_json }
+ let(:page_json) { json[:first] }
+
+ it 'returns http success' do
+ expect(response).to have_http_status(200)
+ end
+
+ it 'returns application/activity+json' do
+ expect(response.media_type).to eq 'application/activity+json'
+ end
+
+ it_behaves_like 'cachable response'
+
+ context 'without only_other_accounts' do
+ it "returns items with thread author's replies" do
+ expect(page_json).to be_a Hash
+ expect(page_json[:items]).to be_an Array
+ expect(page_json[:items].size).to eq 1
+ expect(page_json[:items].all? { |item| item[:to].include?(ActivityPub::TagManager::COLLECTIONS[:public]) || item[:cc].include?(ActivityPub::TagManager::COLLECTIONS[:public]) }).to be true
+ end
+
+ context 'when there are few self-replies' do
+ it 'points next to replies from other people' do
+ expect(page_json).to be_a Hash
+ expect(Addressable::URI.parse(page_json[:next]).query.split('&')).to include('only_other_accounts=true', 'page=true')
+ end
+ end
+
+ context 'when there are many self-replies' do
+ before do
+ 10.times { Fabricate(:status, account: status.account, thread: status, visibility: :public) }
+ end
+
+ it 'points next to other self-replies' do
+ expect(page_json).to be_a Hash
+ expect(Addressable::URI.parse(page_json[:next]).query.split('&')).to include('only_other_accounts=false', 'page=true')
+ end
+ end
+ end
+
+ context 'with only_other_accounts' do
+ let(:only_other_accounts) { 'true' }
+
+ it 'returns items with other public or unlisted replies' do
+ expect(page_json).to be_a Hash
+ expect(page_json[:items]).to be_an Array
+ expect(page_json[:items].size).to eq 3
+ end
+
+ it 'only inlines items that are local and public or unlisted replies' do
+ inlined_replies = page_json[:items].select { |x| x.is_a?(Hash) }
+ public_collection = ActivityPub::TagManager::COLLECTIONS[:public]
+ expect(inlined_replies.all? { |item| item[:to].include?(public_collection) || item[:cc].include?(public_collection) }).to be true
+ expect(inlined_replies.all? { |item| ActivityPub::TagManager.instance.local_uri?(item[:id]) }).to be true
+ end
+
+ it 'uses ids for remote toots' do
+ remote_replies = page_json[:items].select { |x| !x.is_a?(Hash) }
+ expect(remote_replies.all? { |item| item.is_a?(String) && !ActivityPub::TagManager.instance.local_uri?(item) }).to be true
+ end
+
+ context 'when there are few replies' do
+ it 'does not have a next page' do
+ expect(page_json).to be_a Hash
+ expect(page_json[:next]).to be_nil
+ end
+ end
+
+ context 'when there are many replies' do
+ before do
+ 10.times { Fabricate(:status, thread: status, visibility: :public) }
+ end
+
+ it 'points next to other replies' do
+ expect(page_json).to be_a Hash
+ expect(Addressable::URI.parse(page_json[:next]).query.split('&')).to include('only_other_accounts=true', 'page=true')
+ end
+ end
+ end
+ end
+
+ it_behaves_like 'common behavior'
+ end
+
before do
- allow(controller).to receive(:signed_request_account).and_return(remote_account)
+ stub_const 'ActivityPub::RepliesController::DESCENDANTS_LIMIT', 5
+ allow(controller).to receive(:signed_request_account).and_return(remote_querier)
Fabricate(:status, thread: status, visibility: :public)
Fabricate(:status, thread: status, visibility: :public)
@@ -32,215 +176,36 @@ RSpec.describe ActivityPub::RepliesController, type: :controller do
Fabricate(:status, account: status.account, thread: status, visibility: :public)
Fabricate(:status, account: status.account, thread: status, visibility: :private)
- Fabricate(:status, account: remote_account, thread: status, visibility: :public, uri: remote_reply_id) if remote_reply_id
+ Fabricate(:status, account: remote_account, thread: status, visibility: :public, uri: remote_reply_id)
end
describe 'GET #index' do
+ subject(:response) { get :index, params: { account_username: status.account.username, status_id: status.id, only_other_accounts: only_other_accounts } }
+ let(:only_other_accounts) { nil }
+
context 'with no signature' do
- subject(:response) { get :index, params: { account_username: status.account.username, status_id: status.id } }
- subject(:body) { body_as_json }
-
- context 'when account is permanently suspended' do
- let(:parent_visibility) { :public }
-
- before do
- status.account.suspend!
- status.account.deletion_request.destroy
- end
-
- it 'returns http gone' do
- expect(response).to have_http_status(410)
- end
- end
-
- context 'when account is temporarily suspended' do
- let(:parent_visibility) { :public }
-
- before do
- status.account.suspend!
- end
-
- it 'returns http forbidden' do
- expect(response).to have_http_status(403)
- end
- end
-
- context 'when status is public' do
- let(:parent_visibility) { :public }
-
- it 'returns http success' do
- expect(response).to have_http_status(200)
- end
-
- it 'returns application/activity+json' do
- expect(response.media_type).to eq 'application/activity+json'
- end
-
- it_behaves_like 'cachable response'
-
- it 'returns items with account\'s own replies' do
- expect(body[:first]).to be_a Hash
- expect(body[:first][:items]).to be_an Array
- expect(body[:first][:items].size).to eq 1
- expect(body[:first][:items].all? { |item| item[:to].include?(ActivityPub::TagManager::COLLECTIONS[:public]) || item[:cc].include?(ActivityPub::TagManager::COLLECTIONS[:public]) }).to be true
- end
- end
-
- context 'when status is private' do
- let(:parent_visibility) { :private }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
-
- context 'when status is direct' do
- let(:parent_visibility) { :direct }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
+ it_behaves_like 'allowed access'
end
context 'with signature' do
- let(:remote_account) { Fabricate(:account, domain: 'example.com') }
- let(:only_other_accounts) { nil }
+ let(:remote_querier) { Fabricate(:account, domain: 'example.com') }
- context do
- before do
- get :index, params: { account_username: status.account.username, status_id: status.id, only_other_accounts: only_other_accounts }
- end
-
- context 'when status is public' do
- let(:parent_visibility) { :public }
-
- it 'returns http success' do
- expect(response).to have_http_status(200)
- end
-
- it 'returns application/activity+json' do
- expect(response.media_type).to eq 'application/activity+json'
- end
-
- it_behaves_like 'cachable response'
-
- context 'without only_other_accounts' do
- it 'returns items with account\'s own replies' do
- json = body_as_json
-
- expect(json[:first]).to be_a Hash
- expect(json[:first][:items]).to be_an Array
- expect(json[:first][:items].size).to eq 1
- expect(json[:first][:items].all? { |item| item[:to].include?(ActivityPub::TagManager::COLLECTIONS[:public]) || item[:cc].include?(ActivityPub::TagManager::COLLECTIONS[:public]) }).to be true
- end
- end
-
- context 'with only_other_accounts' do
- let(:only_other_accounts) { 'true' }
-
- it 'returns items with other public or unlisted replies' do
- json = body_as_json
-
- expect(json[:first]).to be_a Hash
- expect(json[:first][:items]).to be_an Array
- expect(json[:first][:items].size).to eq 2
- expect(json[:first][:items].all? { |item| item[:to].include?(ActivityPub::TagManager::COLLECTIONS[:public]) || item[:cc].include?(ActivityPub::TagManager::COLLECTIONS[:public]) }).to be true
- end
-
- context 'with remote responses' do
- let(:remote_reply_id) { 'foo' }
-
- it 'returned items are all inlined local toots or are ids' do
- json = body_as_json
-
- expect(json[:first]).to be_a Hash
- expect(json[:first][:items]).to be_an Array
- expect(json[:first][:items].size).to eq 3
- expect(json[:first][:items].all? { |item| item.is_a?(Hash) ? ActivityPub::TagManager.instance.local_uri?(item[:id]) : item.is_a?(String) }).to be true
- expect(json[:first][:items]).to include remote_reply_id
- end
- end
- end
- end
-
- context 'when status is private' do
- let(:parent_visibility) { :private }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
-
- context 'when status is direct' do
- let(:parent_visibility) { :direct }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
- end
+ it_behaves_like 'allowed access'
context 'when signed request account is blocked' do
before do
- status.account.block!(remote_account)
- get :index, params: { account_username: status.account.username, status_id: status.id }
+ status.account.block!(remote_querier)
end
- context 'when status is public' do
- let(:parent_visibility) { :public }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
-
- context 'when status is private' do
- let(:parent_visibility) { :private }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
-
- context 'when status is direct' do
- let(:parent_visibility) { :direct }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
+ it_behaves_like 'disallowed access'
end
context 'when signed request account is domain blocked' do
before do
- status.account.block_domain!(remote_account.domain)
- get :index, params: { account_username: status.account.username, status_id: status.id }
+ status.account.block_domain!(remote_querier.domain)
end
- context 'when status is public' do
- let(:parent_visibility) { :public }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
-
- context 'when status is private' do
- let(:parent_visibility) { :private }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
-
- context 'when status is direct' do
- let(:parent_visibility) { :direct }
-
- it 'returns http not found' do
- expect(response).to have_http_status(404)
- end
- end
+ it_behaves_like 'disallowed access'
end
end
end
diff --git a/spec/helpers/languages_helper_spec.rb b/spec/helpers/languages_helper_spec.rb
index 6db617824d..b455cee33d 100644
--- a/spec/helpers/languages_helper_spec.rb
+++ b/spec/helpers/languages_helper_spec.rb
@@ -3,9 +3,9 @@
require 'rails_helper'
describe LanguagesHelper do
- describe 'the HUMAN_LOCALES constant' do
- it 'includes all I18n locales' do
- expect(described_class::HUMAN_LOCALES.keys).to include(*I18n.available_locales)
+ describe 'the SUPPORTED_LOCALES constant' do
+ it 'includes all i18n locales' do
+ expect(Set.new(described_class::SUPPORTED_LOCALES.keys + described_class::REGIONAL_LOCALE_NAMES.keys)).to include(*I18n.available_locales)
end
end
diff --git a/spec/lib/language_detector_spec.rb b/spec/lib/language_detector_spec.rb
deleted file mode 100644
index b7ba0f6c4f..0000000000
--- a/spec/lib/language_detector_spec.rb
+++ /dev/null
@@ -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
diff --git a/spec/lib/link_details_extractor_spec.rb b/spec/lib/link_details_extractor_spec.rb
index 850857b2dd..84bb4579ca 100644
--- a/spec/lib/link_details_extractor_spec.rb
+++ b/spec/lib/link_details_extractor_spec.rb
@@ -26,4 +26,126 @@ RSpec.describe LinkDetailsExtractor do
end
end
end
+
+ context 'when structured data is present' do
+ let(:original_url) { 'https://example.com/page.html' }
+
+ context 'and is wrapped in CDATA tags' do
+ let(:html) { <<-HTML }
+<!doctype html>
+<html>
+<head>
+ <script type="application/ld+json">
+ //<![CDATA[
+ {"@context":"http://schema.org","@type":"NewsArticle","mainEntityOfPage":"https://example.com/page.html","headline":"Foo","datePublished":"2022-01-31T19:53:00+00:00","url":"https://example.com/page.html","description":"Bar","author":{"@type":"Person","name":"Hoge"},"publisher":{"@type":"Organization","name":"Baz"}}
+ //]]>
+ </script>
+</head>
+</html>
+ HTML
+
+ describe '#title' do
+ it 'returns the title from structured data' do
+ expect(subject.title).to eq 'Foo'
+ end
+ end
+
+ describe '#description' do
+ it 'returns the description from structured data' do
+ expect(subject.description).to eq 'Bar'
+ end
+ end
+
+ describe '#provider_name' do
+ it 'returns the provider name from structured data' do
+ expect(subject.provider_name).to eq 'Baz'
+ end
+ end
+
+ describe '#author_name' do
+ it 'returns the author name from structured data' do
+ expect(subject.author_name).to eq 'Hoge'
+ end
+ end
+ end
+
+ context 'but the first tag is invalid JSON' do
+ let(:html) { <<-HTML }
+<!doctype html>
+<html>
+<body>
+ <script type="application/ld+json">
+ {
+ "@context":"https://schema.org",
+ "@type":"ItemList",
+ "url":"https://example.com/page.html",
+ "name":"Foo",
+ "description":"Bar"
+ },
+ {
+ "@context": "https://schema.org",
+ "@type": "BreadcrumbList",
+ "itemListElement":[
+ {
+ "@type":"ListItem",
+ "position":1,
+ "item":{
+ "@id":"https://www.example.com",
+ "name":"Baz"
+ }
+ }
+ ]
+ }
+ </script>
+ <script type="application/ld+json">
+ {
+ "@context":"https://schema.org",
+ "@type":"NewsArticle",
+ "mainEntityOfPage": {
+ "@type":"WebPage",
+ "@id": "http://example.com/page.html"
+ },
+ "headline": "Foo",
+ "description": "Bar",
+ "datePublished": "2022-01-31T19:46:00+00:00",
+ "author": {
+ "@type": "Organization",
+ "name": "Hoge"
+ },
+ "publisher": {
+ "@type": "NewsMediaOrganization",
+ "name":"Baz",
+ "url":"https://example.com/"
+ }
+ }
+ </script>
+</body>
+</html>
+ HTML
+
+ describe '#title' do
+ it 'returns the title from structured data' do
+ expect(subject.title).to eq 'Foo'
+ end
+ end
+
+ describe '#description' do
+ it 'returns the description from structured data' do
+ expect(subject.description).to eq 'Bar'
+ end
+ end
+
+ describe '#provider_name' do
+ it 'returns the provider name from structured data' do
+ expect(subject.provider_name).to eq 'Baz'
+ end
+ end
+
+ describe '#author_name' do
+ it 'returns the author name from structured data' do
+ expect(subject.author_name).to eq 'Hoge'
+ end
+ end
+ end
+ end
end