diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/invidious/helpers/i18next.cr | 70 | ||||
| -rw-r--r-- | src/invidious/helpers/webvtt.cr | 67 | ||||
| -rw-r--r-- | src/invidious/routes/api/v1/videos.cr | 39 | ||||
| -rw-r--r-- | src/invidious/routes/channels.cr | 25 | ||||
| -rw-r--r-- | src/invidious/routing.cr | 37 | ||||
| -rw-r--r-- | src/invidious/videos/caption.cr | 41 | ||||
| -rw-r--r-- | src/invidious/videos/transcript.cr | 40 |
7 files changed, 204 insertions, 115 deletions
diff --git a/src/invidious/helpers/i18next.cr b/src/invidious/helpers/i18next.cr index e84f88fb..252af6b9 100644 --- a/src/invidious/helpers/i18next.cr +++ b/src/invidious/helpers/i18next.cr @@ -35,19 +35,27 @@ module I18next::Plurals Special_Slovenian = 21 Special_Hebrew = 22 Special_Odia = 23 + + # Mixed v3/v4 rules in Weblate + # `es`, `pt` and `pt-PT` doesn't seem to have been refreshed + # by weblate yet, but I suspect it will happen one day. + # See: https://github.com/translate/translate/issues/4873 + Special_French_Portuguese + Special_Hungarian_Serbian + Special_Spanish_Italian end private PLURAL_SETS = { PluralForms::Single_gt_one => [ - "ach", "ak", "am", "arn", "br", "fil", "fr", "gun", "ln", "mfe", "mg", - "mi", "oc", "pt", "pt-BR", "tg", "tl", "ti", "tr", "uz", "wa", + "ach", "ak", "am", "arn", "br", "fil", "gun", "ln", "mfe", "mg", + "mi", "oc", "pt", "tg", "tl", "ti", "tr", "uz", "wa", ], PluralForms::Single_not_one => [ "af", "an", "ast", "az", "bg", "bn", "ca", "da", "de", "dev", "el", "en", "eo", "es", "et", "eu", "fi", "fo", "fur", "fy", "gl", "gu", "ha", "hi", - "hu", "hy", "ia", "it", "kk", "kn", "ku", "lb", "mai", "ml", "mn", "mr", + "hu", "hy", "ia", "kk", "kn", "ku", "lb", "mai", "ml", "mn", "mr", "nah", "nap", "nb", "ne", "nl", "nn", "no", "nso", "pa", "pap", "pms", - "ps", "pt-PT", "rm", "sco", "se", "si", "so", "son", "sq", "sv", "sw", + "ps", "rm", "sco", "se", "si", "so", "son", "sq", "sv", "sw", "ta", "te", "tk", "ur", "yo", ], PluralForms::None => [ @@ -55,7 +63,7 @@ module I18next::Plurals "lo", "ms", "sah", "su", "th", "tt", "ug", "vi", "wo", "zh", ], PluralForms::Dual_Slavic => [ - "be", "bs", "cnr", "dz", "hr", "ru", "sr", "uk", + "be", "bs", "cnr", "dz", "ru", "uk", ], } @@ -81,6 +89,12 @@ module I18next::Plurals "ro" => PluralForms::Special_Romanian, "sk" => PluralForms::Special_Czech_Slovak, "sl" => PluralForms::Special_Slovenian, + # Mixed v3/v4 rules + "fr" => PluralForms::Special_French_Portuguese, + "hr" => PluralForms::Special_Hungarian_Serbian, + "it" => PluralForms::Special_Spanish_Italian, + "pt-BR" => PluralForms::Special_French_Portuguese, + "sr" => PluralForms::Special_Hungarian_Serbian, } # These are the v1 and v2 compatible suffixes. @@ -150,9 +164,8 @@ module I18next::Plurals end def get_plural_form(locale : String) : PluralForms - # Extract the ISO 639-1 or 639-2 code from an RFC 5646 language code, - # except for pt-BR and pt-PT which needs to be kept as-is. - if !locale.matches?(/^pt-(BR|PT)$/) + # Extract the ISO 639-1 or 639-2 code from an RFC 5646 language code + if !locale.matches?(/^pt-BR$/) locale = locale.split('-')[0] end @@ -246,6 +259,10 @@ module I18next::Plurals when .special_slovenian? then return special_slovenian(count) when .special_hebrew? then return special_hebrew(count) when .special_odia? then return special_odia(count) + # Mixed v3/v4 forms + when .special_spanish_italian? then return special_cldr_Spanish_Italian(count) + when .special_french_portuguese? then return special_cldr_French_Portuguese(count) + when .special_hungarian_serbian? then return special_cldr_Hungarian_Serbian(count) else # default, if nothing matched above return 0_u8 @@ -507,5 +524,42 @@ module I18next::Plurals def self.special_odia(count : Int) : UInt8 return (count == 1) ? 0_u8 : 1_u8 end + + # ------------------- + # "v3.5" rules + # ------------------- + + # Plural form for Spanish & Italian languages + # + # This rule is mostly compliant to CLDR v42 + # + def self.special_cldr_Spanish_Italian(count : Int) : UInt8 + return 0_u8 if (count == 1) # one + return 1_u8 if (count != 0 && count % 1_000_000 == 0) # many + return 2_u8 # other + end + + # Plural form for French and Portuguese + # + # This rule is mostly compliant to CLDR v42 + # + def self.special_cldr_French_Portuguese(count : Int) : UInt8 + return 0_u8 if (count == 0 || count == 1) # one + return 1_u8 if (count % 1_000_000 == 0) # many + return 2_u8 # other + end + + # Plural form for Hungarian and Serbian + # + # This rule is mostly compliant to CLDR v42 + # + def self.special_cldr_Hungarian_Serbian(count : Int) : UInt8 + n_mod_10 = count % 10 + n_mod_100 = count % 100 + + return 0_u8 if (n_mod_10 == 1 && n_mod_100 != 11) # one + return 1_u8 if (2 <= n_mod_10 <= 4 && (n_mod_100 < 12 || 14 < n_mod_100)) # few + return 2_u8 # other + end end end diff --git a/src/invidious/helpers/webvtt.cr b/src/invidious/helpers/webvtt.cr new file mode 100644 index 00000000..56f761ed --- /dev/null +++ b/src/invidious/helpers/webvtt.cr @@ -0,0 +1,67 @@ +# Namespace for logic relating to generating WebVTT files +# +# Probably not compliant to WebVTT's specs but it is enough for Invidious. +module WebVTT + # A WebVTT builder generates WebVTT files + private class Builder + def initialize(@io : IO) + end + + # Writes an vtt cue with the specified time stamp and contents + def cue(start_time : Time::Span, end_time : Time::Span, text : String) + timestamp(start_time, end_time) + @io << text + @io << "\n\n" + end + + private def timestamp(start_time : Time::Span, end_time : Time::Span) + timestamp_component(start_time) + @io << " --> " + timestamp_component(end_time) + + @io << '\n' + end + + private def timestamp_component(timestamp : Time::Span) + @io << timestamp.hours.to_s.rjust(2, '0') + @io << ':' << timestamp.minutes.to_s.rjust(2, '0') + @io << ':' << timestamp.seconds.to_s.rjust(2, '0') + @io << '.' << timestamp.milliseconds.to_s.rjust(3, '0') + end + + def document(setting_fields : Hash(String, String)? = nil, &) + @io << "WEBVTT\n" + + if setting_fields + setting_fields.each do |name, value| + @io << name << ": " << value << '\n' + end + end + + @io << '\n' + + yield + end + end + + # Returns the resulting `String` of writing WebVTT to the yielded `WebVTT::Builder` + # + # ``` + # string = WebVTT.build do |vtt| + # vtt.cue(Time::Span.new(seconds: 1), Time::Span.new(seconds: 2), "Line 1") + # vtt.cue(Time::Span.new(seconds: 2), Time::Span.new(seconds: 3), "Line 2") + # end + # + # string # => "WEBVTT\n\n00:00:01.000 --> 00:00:02.000\nLine 1\n\n00:00:02.000 --> 00:00:03.000\nLine 2\n\n" + # ``` + # + # Accepts an optional settings fields hash to add settings attribute to the resulting vtt file. + def self.build(setting_fields : Hash(String, String)? = nil, &) + String.build do |str| + builder = Builder.new(str) + builder.document(setting_fields) do + yield builder + end + end + end +end diff --git a/src/invidious/routes/api/v1/videos.cr b/src/invidious/routes/api/v1/videos.cr index 25e766d2..449c9f9b 100644 --- a/src/invidious/routes/api/v1/videos.cr +++ b/src/invidious/routes/api/v1/videos.cr @@ -101,20 +101,17 @@ module Invidious::Routes::API::V1::Videos if caption.name.includes? "auto-generated" caption_xml = YT_POOL.client &.get(url).body + settings_field = { + "Kind" => "captions", + "Language" => "#{tlang || caption.language_code}", + } + if caption_xml.starts_with?("<?xml") webvtt = caption.timedtext_to_vtt(caption_xml, tlang) else caption_xml = XML.parse(caption_xml) - webvtt = String.build do |str| - str << <<-END_VTT - WEBVTT - Kind: captions - Language: #{tlang || caption.language_code} - - - END_VTT - + webvtt = WebVTT.build(settings_field) do |webvtt| caption_nodes = caption_xml.xpath_nodes("//transcript/text") caption_nodes.each_with_index do |node, i| start_time = node["start"].to_f.seconds @@ -127,9 +124,6 @@ module Invidious::Routes::API::V1::Videos end_time = start_time + duration end - start_time = "#{start_time.hours.to_s.rjust(2, '0')}:#{start_time.minutes.to_s.rjust(2, '0')}:#{start_time.seconds.to_s.rjust(2, '0')}.#{start_time.milliseconds.to_s.rjust(3, '0')}" - end_time = "#{end_time.hours.to_s.rjust(2, '0')}:#{end_time.minutes.to_s.rjust(2, '0')}:#{end_time.seconds.to_s.rjust(2, '0')}.#{end_time.milliseconds.to_s.rjust(3, '0')}" - text = HTML.unescape(node.content) text = text.gsub(/<font color="#[a-fA-F0-9]{6}">/, "") text = text.gsub(/<\/font>/, "") @@ -137,12 +131,7 @@ module Invidious::Routes::API::V1::Videos text = "<v #{md["name"]}>#{md["text"]}</v>" end - str << <<-END_CUE - #{start_time} --> #{end_time} - #{text} - - - END_CUE + webvtt.cue(start_time, end_time, text) end end end @@ -215,11 +204,7 @@ module Invidious::Routes::API::V1::Videos storyboard = storyboard[0] end - String.build do |str| - str << <<-END_VTT - WEBVTT - END_VTT - + WebVTT.build do |vtt| start_time = 0.milliseconds end_time = storyboard[:interval].milliseconds @@ -231,12 +216,8 @@ module Invidious::Routes::API::V1::Videos storyboard[:storyboard_height].times do |j| storyboard[:storyboard_width].times do |k| - str << <<-END_CUE - #{start_time}.000 --> #{end_time}.000 - #{url}#xywh=#{storyboard[:width] * k},#{storyboard[:height] * j},#{storyboard[:width] - 2},#{storyboard[:height]} - - - END_CUE + current_cue_url = "#{url}#xywh=#{storyboard[:width] * k},#{storyboard[:height] * j},#{storyboard[:width] - 2},#{storyboard[:height]}" + vtt.cue(start_time, end_time, current_cue_url) start_time += storyboard[:interval].milliseconds end_time += storyboard[:interval].milliseconds diff --git a/src/invidious/routes/channels.cr b/src/invidious/routes/channels.cr index 9892ae2a..d744712d 100644 --- a/src/invidious/routes/channels.cr +++ b/src/invidious/routes/channels.cr @@ -1,6 +1,12 @@ {% skip_file if flag?(:api_only) %} module Invidious::Routes::Channels + # Redirection for unsupported routes ("tabs") + def self.redirect_home(env) + ucid = env.params.url["ucid"] + return env.redirect "/channel/#{URI.encode_www_form(ucid)}" + end + def self.home(env) self.videos(env) end @@ -217,6 +223,11 @@ module Invidious::Routes::Channels env.redirect "/channel/#{ucid}" end + private KNOWN_TABS = { + "home", "videos", "shorts", "streams", "podcasts", + "releases", "playlists", "community", "channels", "about", + } + # Redirects brand url channels to a normal /channel/:ucid route def self.brand_redirect(env) locale = env.get("preferences").as(Preferences).locale @@ -227,7 +238,10 @@ module Invidious::Routes::Channels yt_url_params = URI::Params.encode(env.params.query.to_h.select(["a", "u", "user"])) # Retrieves URL params that only Invidious uses - invidious_url_params = URI::Params.encode(env.params.query.to_h.select!(["a", "u", "user"])) + invidious_url_params = env.params.query.dup + invidious_url_params.delete_all("a") + invidious_url_params.delete_all("u") + invidious_url_params.delete_all("user") begin resolved_url = YoutubeAPI.resolve_url("https://youtube.com#{env.request.path}#{yt_url_params.size > 0 ? "?#{yt_url_params}" : ""}") @@ -236,14 +250,17 @@ module Invidious::Routes::Channels return error_template(404, translate(locale, "This channel does not exist.")) end - selected_tab = env.request.path.split("/")[-1] - if {"home", "videos", "shorts", "streams", "playlists", "community", "channels", "about"}.includes? selected_tab + selected_tab = env.params.url["tab"]? + + if KNOWN_TABS.includes? selected_tab url = "/channel/#{ucid}/#{selected_tab}" else url = "/channel/#{ucid}" end - env.redirect url + url += "?#{invidious_url_params}" if !invidious_url_params.empty? + + return env.redirect url end # Handles redirects for the /profile endpoint diff --git a/src/invidious/routing.cr b/src/invidious/routing.cr index 9c43171c..8c38aaed 100644 --- a/src/invidious/routing.cr +++ b/src/invidious/routing.cr @@ -124,28 +124,41 @@ module Invidious::Routing get "/channel/:ucid/community", Routes::Channels, :community get "/channel/:ucid/channels", Routes::Channels, :channels get "/channel/:ucid/about", Routes::Channels, :about + get "/channel/:ucid/live", Routes::Channels, :live get "/user/:user/live", Routes::Channels, :live get "/c/:user/live", Routes::Channels, :live - {"", "/videos", "/shorts", "/streams", "/playlists", "/community", "/about"}.each do |path| - # /c/LinusTechTips - get "/c/:user#{path}", Routes::Channels, :brand_redirect - # /user/linustechtips | Not always the same as /c/ - get "/user/:user#{path}", Routes::Channels, :brand_redirect - # /@LinusTechTips | Handle - get "/@:user#{path}", Routes::Channels, :brand_redirect - # /attribution_link?a=anything&u=/channel/UCZYTClx2T1of7BRZ86-8fow - get "/attribution_link#{path}", Routes::Channels, :brand_redirect - # /profile?user=linustechtips - get "/profile/#{path}", Routes::Channels, :profile - end + # Channel catch-all, to redirect future routes to the channel's home + # NOTE: defined last in order to be processed after the other routes + get "/channel/:ucid/*", Routes::Channels, :redirect_home + + # /c/LinusTechTips + get "/c/:user", Routes::Channels, :brand_redirect + get "/c/:user/:tab", Routes::Channels, :brand_redirect + + # /user/linustechtips (Not always the same as /c/) + get "/user/:user", Routes::Channels, :brand_redirect + get "/user/:user/:tab", Routes::Channels, :brand_redirect + + # /@LinusTechTips (Handle) + get "/@:user", Routes::Channels, :brand_redirect + get "/@:user/:tab", Routes::Channels, :brand_redirect + + # /attribution_link?a=anything&u=/channel/UCZYTClx2T1of7BRZ86-8fow + get "/attribution_link", Routes::Channels, :brand_redirect + get "/attribution_link/:tab", Routes::Channels, :brand_redirect + + # /profile?user=linustechtips + get "/profile", Routes::Channels, :profile + get "/profile/*", Routes::Channels, :profile end def register_watch_routes get "/watch", Routes::Watch, :handle post "/watch_ajax", Routes::Watch, :mark_watched get "/watch/:id", Routes::Watch, :redirect + get "/live/:id", Routes::Watch, :redirect get "/shorts/:id", Routes::Watch, :redirect get "/clip/:clip", Routes::Watch, :clip get "/w/:id", Routes::Watch, :redirect diff --git a/src/invidious/videos/caption.cr b/src/invidious/videos/caption.cr index 256dfcc0..484e61d2 100644 --- a/src/invidious/videos/caption.cr +++ b/src/invidious/videos/caption.cr @@ -52,17 +52,13 @@ module Invidious::Videos break end end - result = String.build do |result| - result << <<-END_VTT - WEBVTT - Kind: captions - Language: #{tlang || @language_code} + settings_field = { + "Kind" => "captions", + "Language" => "#{tlang || @language_code}", + } - END_VTT - - result << "\n\n" - + result = WebVTT.build(settings_field) do |vtt| cues.each_with_index do |node, i| start_time = node["t"].to_f.milliseconds @@ -76,29 +72,16 @@ module Invidious::Videos end_time = start_time + duration end - # start_time - result << start_time.hours.to_s.rjust(2, '0') - result << ':' << start_time.minutes.to_s.rjust(2, '0') - result << ':' << start_time.seconds.to_s.rjust(2, '0') - result << '.' << start_time.milliseconds.to_s.rjust(3, '0') - - result << " --> " - - # end_time - result << end_time.hours.to_s.rjust(2, '0') - result << ':' << end_time.minutes.to_s.rjust(2, '0') - result << ':' << end_time.seconds.to_s.rjust(2, '0') - result << '.' << end_time.milliseconds.to_s.rjust(3, '0') - - result << "\n" - - node.children.each do |s| - result << s.content + text = String.build do |io| + node.children.each do |s| + io << s.content + end end - result << "\n" - result << "\n" + + vtt.cue(start_time, end_time, text) end end + return result end end diff --git a/src/invidious/videos/transcript.cr b/src/invidious/videos/transcript.cr index f3360a52..dac00eea 100644 --- a/src/invidious/videos/transcript.cr +++ b/src/invidious/videos/transcript.cr @@ -34,41 +34,15 @@ module Invidious::Videos # Convert into array of TranscriptLine lines = self.parse(initial_data) - # Taken from Invidious::Videos::Captions::Metadata.timedtext_to_vtt() - vtt = String.build do |vtt| - vtt << <<-END_VTT - WEBVTT - Kind: captions - Language: #{target_language} - - - END_VTT - - vtt << "\n\n" + settings_field = { + "Kind" => "captions", + "Language" => target_language, + } + # Taken from Invidious::Videos::Captions::Metadata.timedtext_to_vtt() + vtt = WebVTT.build(settings_field) do |vtt| lines.each do |line| - start_time = line.start_ms - end_time = line.end_ms - - # start_time - vtt << start_time.hours.to_s.rjust(2, '0') - vtt << ':' << start_time.minutes.to_s.rjust(2, '0') - vtt << ':' << start_time.seconds.to_s.rjust(2, '0') - vtt << '.' << start_time.milliseconds.to_s.rjust(3, '0') - - vtt << " --> " - - # end_time - vtt << end_time.hours.to_s.rjust(2, '0') - vtt << ':' << end_time.minutes.to_s.rjust(2, '0') - vtt << ':' << end_time.seconds.to_s.rjust(2, '0') - vtt << '.' << end_time.milliseconds.to_s.rjust(3, '0') - - vtt << "\n" - vtt << line.line - - vtt << "\n" - vtt << "\n" + vtt.cue(line.start_ms, line.end_ms, line.line) end end |
