diff options
| -rw-r--r-- | src/invidious/channels/videos.cr | 192 | ||||
| -rw-r--r-- | src/invidious/routes/channels.cr | 16 | ||||
| -rw-r--r-- | src/invidious/yt_backend/extractors.cr | 141 |
3 files changed, 252 insertions, 97 deletions
diff --git a/src/invidious/channels/videos.cr b/src/invidious/channels/videos.cr index 6cc30142..96400f47 100644 --- a/src/invidious/channels/videos.cr +++ b/src/invidious/channels/videos.cr @@ -1,78 +1,3 @@ -def produce_channel_content_continuation(ucid, content_type, page = 1, auto_generated = nil, sort_by = "newest", v2 = false) - object_inner_2 = { - "2:0:embedded" => { - "1:0:varint" => 0_i64, - }, - "5:varint" => 50_i64, - "6:varint" => 1_i64, - "7:varint" => (page * 30).to_i64, - "9:varint" => 1_i64, - "10:varint" => 0_i64, - } - - object_inner_2_encoded = object_inner_2 - .try { |i| Protodec::Any.cast_json(i) } - .try { |i| Protodec::Any.from_json(i) } - .try { |i| Base64.urlsafe_encode(i) } - .try { |i| URI.encode_www_form(i) } - - content_type_numerical = - case content_type - when "videos" then 15 - when "livestreams" then 14 - else 15 # Fallback to "videos" - end - - sort_by_numerical = - case sort_by - when "newest" then 1_i64 - when "popular" then 2_i64 - when "oldest" then 4_i64 - else 1_i64 # Fallback to "newest" - end - - object_inner_1 = { - "110:embedded" => { - "3:embedded" => { - "#{content_type_numerical}:embedded" => { - "1:embedded" => { - "1:string" => object_inner_2_encoded, - }, - "2:embedded" => { - "1:string" => "00000000-0000-0000-0000-000000000000", - }, - "3:varint" => sort_by_numerical, - }, - }, - }, - } - - object_inner_1_encoded = object_inner_1 - .try { |i| Protodec::Any.cast_json(i) } - .try { |i| Protodec::Any.from_json(i) } - .try { |i| Base64.urlsafe_encode(i) } - .try { |i| URI.encode_www_form(i) } - - object = { - "80226972:embedded" => { - "2:string" => ucid, - "3:string" => object_inner_1_encoded, - "35:string" => "browse-feed#{ucid}videos102", - }, - } - - continuation = object.try { |i| Protodec::Any.cast_json(i) } - .try { |i| Protodec::Any.from_json(i) } - .try { |i| Base64.urlsafe_encode(i) } - .try { |i| URI.encode_www_form(i) } - - return continuation -end - -def make_initial_content_ctoken(ucid, content_type, sort_by) : String - return produce_channel_content_continuation(ucid, content_type, sort_by: sort_by) -end - module Invidious::Channel::Tabs extend self @@ -101,7 +26,7 @@ module Invidious::Channel::Tabs end def get_videos(author : String, ucid : String, *, continuation : String? = nil, sort_by = "newest") - continuation ||= make_initial_content_ctoken(ucid, "videos", sort_by) + continuation ||= make_initial_videos_ctoken(ucid, sort_by) initial_data = YoutubeAPI.browse(continuation: continuation) return extract_items(initial_data, author, ucid) @@ -130,14 +55,10 @@ module Invidious::Channel::Tabs # Shorts # ------------------- - def get_shorts(channel : AboutChannel, continuation : String? = nil) - if continuation.nil? - # EgZzaG9ydHPyBgUKA5oBAA%3D%3D is the protobuf object to load "shorts" - # TODO: try to extract the continuation tokens that allows other sorting options - initial_data = YoutubeAPI.browse(channel.ucid, params: "EgZzaG9ydHPyBgUKA5oBAA%3D%3D") - else - initial_data = YoutubeAPI.browse(continuation: continuation) - end + def get_shorts(channel : AboutChannel, *, continuation : String? = nil, sort_by = "newest") + continuation ||= make_initial_shorts_ctoken(channel.ucid, sort_by) + initial_data = YoutubeAPI.browse(continuation: continuation) + return extract_items(initial_data, channel.author, channel.ucid) end @@ -145,9 +66,8 @@ module Invidious::Channel::Tabs # Livestreams # ------------------- - def get_livestreams(channel : AboutChannel, continuation : String? = nil, sort_by = "newest") - continuation ||= make_initial_content_ctoken(channel.ucid, "livestreams", sort_by) - + def get_livestreams(channel : AboutChannel, *, continuation : String? = nil, sort_by = "newest") + continuation ||= make_initial_livestreams_ctoken(channel.ucid, sort_by) initial_data = YoutubeAPI.browse(continuation: continuation) return extract_items(initial_data, channel.author, channel.ucid) @@ -171,4 +91,102 @@ module Invidious::Channel::Tabs return items, next_continuation end + + # ------------------- + # C-tokens + # ------------------- + + private def sort_options_videos_short(sort_by : String) + case sort_by + when "newest" then return 4_i64 + when "popular" then return 2_i64 + when "oldest" then return 5_i64 + else return 4_i64 # Fallback to "newest" + end + end + + # Generate the initial "continuation token" to get the first page of the + # "videos" tab. The following page requires the ctoken provided in that + # first page, and so on. + private def make_initial_videos_ctoken(ucid : String, sort_by = "newest") + object = { + "15:embedded" => { + "2:embedded" => { + "1:string" => "00000000-0000-0000-0000-000000000000", + }, + "4:varint" => sort_options_videos_short(sort_by), + }, + } + + return channel_ctoken_wrap(ucid, object) + end + + # Generate the initial "continuation token" to get the first page of the + # "shorts" tab. The following page requires the ctoken provided in that + # first page, and so on. + private def make_initial_shorts_ctoken(ucid : String, sort_by = "newest") + object = { + "10:embedded" => { + "2:embedded" => { + "1:string" => "00000000-0000-0000-0000-000000000000", + }, + "4:varint" => sort_options_videos_short(sort_by), + }, + } + + return channel_ctoken_wrap(ucid, object) + end + + # Generate the initial "continuation token" to get the first page of the + # "livestreams" tab. The following page requires the ctoken provided in that + # first page, and so on. + private def make_initial_livestreams_ctoken(ucid : String, sort_by = "newest") + sort_by_numerical = + case sort_by + when "newest" then 12_i64 + when "popular" then 14_i64 + when "oldest" then 13_i64 + else 12_i64 # Fallback to "newest" + end + + object = { + "14:embedded" => { + "2:embedded" => { + "1:string" => "00000000-0000-0000-0000-000000000000", + }, + "5:varint" => sort_by_numerical, + }, + } + + return channel_ctoken_wrap(ucid, object) + end + + # The protobuf structure common between videos/shorts/livestreams + private def channel_ctoken_wrap(ucid : String, object) + object_inner = { + "110:embedded" => { + "3:embedded" => object, + }, + } + + object_inner_encoded = object_inner + .try { |i| Protodec::Any.cast_json(i) } + .try { |i| Protodec::Any.from_json(i) } + .try { |i| Base64.urlsafe_encode(i) } + .try { |i| URI.encode_www_form(i) } + + object = { + "80226972:embedded" => { + "2:string" => ucid, + "3:string" => object_inner_encoded, + }, + } + + continuation = object.try { |i| Protodec::Any.cast_json(i) } + .try { |i| Protodec::Any.from_json(i) } + .try { |i| Base64.urlsafe_encode(i) } + .try { |i| URI.encode_www_form(i) } + + return continuation + end end diff --git a/src/invidious/routes/channels.cr b/src/invidious/routes/channels.cr index 952098e0..7d634cbb 100644 --- a/src/invidious/routes/channels.cr +++ b/src/invidious/routes/channels.cr @@ -20,10 +20,11 @@ module Invidious::Routes::Channels sort_by = env.params.query["sort_by"]?.try &.downcase if channel.auto_generated + sort_by ||= "last" sort_options = {"last", "oldest", "newest"} items, next_continuation = fetch_channel_playlists( - channel.ucid, channel.author, continuation, (sort_by || "last") + channel.ucid, channel.author, continuation, sort_by ) items.uniq! do |item| @@ -49,9 +50,11 @@ module Invidious::Routes::Channels end next_continuation = nil else + sort_by ||= "newest" sort_options = {"newest", "oldest", "popular"} - items, next_continuation = Channel::Tabs.get_videos( - channel, continuation: continuation, sort_by: (sort_by || "newest") + + items, next_continuation = Channel::Tabs.get_60_videos( + channel, continuation: continuation, sort_by: sort_by ) end end @@ -82,13 +85,12 @@ module Invidious::Routes::Channels end next_continuation = nil else - # TODO: support sort option for shorts - sort_by = "" - sort_options = [] of String + sort_by = env.params.query["sort_by"]?.try &.downcase || "newest" + sort_options = {"newest", "oldest", "popular"} # Fetch items and continuation token items, next_continuation = Channel::Tabs.get_shorts( - channel, continuation: continuation + channel, continuation: continuation, sort_by: sort_by ) end diff --git a/src/invidious/yt_backend/extractors.cr b/src/invidious/yt_backend/extractors.cr index 4074de86..2631b62a 100644 --- a/src/invidious/yt_backend/extractors.cr +++ b/src/invidious/yt_backend/extractors.cr @@ -21,6 +21,7 @@ private ITEM_PARSERS = { Parsers::ItemSectionRendererParser, Parsers::ContinuationItemRendererParser, Parsers::HashtagRendererParser, + Parsers::LockupViewModelParser, } private alias InitialData = Hash(String, JSON::Any) @@ -467,9 +468,9 @@ private module Parsers # Parses an InnerTube richItemRenderer into a SearchVideo. # Returns nil when the given object isn't a RichItemRenderer # - # A richItemRenderer seems to be a simple wrapper for a videoRenderer, used - # by the result page for hashtags and for the podcast tab on channels. - # It is located inside a continuationItems container for hashtags. + # A richItemRenderer seems to be a simple wrapper for a various other types, + # used on the hashtags result page and the channel podcast tab. It is located + # itself inside a richGridRenderer container. # module RichItemRendererParser def self.process(item : JSON::Any, author_fallback : AuthorFallback) @@ -482,6 +483,8 @@ private module Parsers child = VideoRendererParser.process(item_contents, author_fallback) child ||= ReelItemRendererParser.process(item_contents, author_fallback) child ||= PlaylistRendererParser.process(item_contents, author_fallback) + child ||= LockupViewModelParser.process(item_contents, author_fallback) + child ||= ShortsLockupViewModelParser.process(item_contents, author_fallback) return child end @@ -496,6 +499,9 @@ private module Parsers # reelItemRenderer items are used in the new (2022) channel layout, # in the "shorts" tab. # + # NOTE: As of 10/2024, it might have been fully replaced by shortsLockupViewModel + # TODO: Confirm that hypothesis + # module ReelItemRendererParser def self.process(item : JSON::Any, author_fallback : AuthorFallback) if item_contents = item["reelItemRenderer"]? @@ -582,6 +588,135 @@ private module Parsers end end + # Parses an InnerTube lockupViewModel into a SearchPlaylist. + # Returns nil when the given object is not a lockupViewModel. + # + # This structure is present since November 2024 on the "podcasts" and + # "playlists" tabs of the channel page. It is usually encapsulated in either + # a richItemRenderer or a richGridRenderer. + # + module LockupViewModelParser + def self.process(item : JSON::Any, author_fallback : AuthorFallback) + if item_contents = item["lockupViewModel"]? + return self.parse(item_contents, author_fallback) + end + end + + private def self.parse(item_contents, author_fallback) + playlist_id = item_contents["contentId"].as_s + + thumbnail_view_model = item_contents.dig( + "contentImage", "collectionThumbnailViewModel", + "primaryThumbnail", "thumbnailViewModel" + ) + + thumbnail = thumbnail_view_model.dig("image", "sources", 0, "url").as_s + + # This complicated sequences tries to extract the following data structure: + # "overlays": [{ + # "thumbnailOverlayBadgeViewModel": { + # "thumbnailBadges": [{ + # "thumbnailBadgeViewModel": { + # "text": "430 episodes", + # "badgeStyle": "THUMBNAIL_OVERLAY_BADGE_STYLE_DEFAULT" + # } + # }] + # } + # }] + # + # NOTE: this simplistic `.to_i` conversion might not work on larger + # playlists and hasn't been tested. + video_count = thumbnail_view_model.dig("overlays").as_a + .compact_map(&.dig?("thumbnailOverlayBadgeViewModel", "thumbnailBadges").try &.as_a) + .flatten + .find(nil, &.dig?("thumbnailBadgeViewModel", "text").try { |node| + {"episodes", "videos"}.any? { |str| node.as_s.ends_with?(str) } + }) + .try &.dig("thumbnailBadgeViewModel", "text").as_s.to_i(strict: false) + + metadata = item_contents.dig("metadata", "lockupMetadataViewModel") + title = metadata.dig("title", "content").as_s + + # TODO: Retrieve "updated" info from metadata parts + # rows = metadata.dig("metadata", "contentMetadataViewModel", "metadataRows").as_a + # parts_text = rows.map(&.dig?("metadataParts", "text", "content").try &.as_s) + # One of these parts should contain a string like: "Updated 2 days ago" + + # TODO: Maybe add a button to access the first video of the playlist? + # item_contents.dig("rendererContext", "commandContext", "onTap", "innertubeCommand", "watchEndpoint") + # Available fields: "videoId", "playlistId", "params" + + return SearchPlaylist.new({ + title: title, + id: playlist_id, + author: author_fallback.name, + ucid: author_fallback.id, + video_count: video_count || -1, + videos: [] of SearchPlaylistVideo, + thumbnail: thumbnail, + author_verified: false, + }) + end + + def self.parser_name + return {{@type.name}} + end + end + + # Parses an InnerTube shortsLockupViewModel into a SearchVideo. + # Returns nil when the given object is not a shortsLockupViewModel. + # + # This structure is present since around October 2024 on the "shorts" tab of + # the channel page and likely replaces the reelItemRenderer structure. It is + # usually (always?) encapsulated in a richItemRenderer. + # + module ShortsLockupViewModelParser + def self.process(item : JSON::Any, author_fallback : AuthorFallback) + if item_contents = item["shortsLockupViewModel"]? + return self.parse(item_contents, author_fallback) + end + end + + private def self.parse(item_contents, author_fallback) + # TODO: Maybe add support for "oardefault.jpg" thumbnails? + # thumbnail = item_contents.dig("thumbnail", "sources", 0, "url").as_s + # Gives: https://i.ytimg.com/vi/{video_id}/oardefault.jpg?... + + video_id = item_contents.dig( + "onTap", "innertubeCommand", "reelWatchEndpoint", "videoId" + ).as_s + + title = item_contents.dig("overlayMetadata", "primaryText", "content").as_s + + view_count = short_text_to_number( + item_contents.dig("overlayMetadata", "secondaryText", "content").as_s + ) + + # Approximate to one minute, as "shorts" generally don't exceed that. + # NOTE: The actual duration is not provided by Youtube anymore. + # TODO: Maybe use -1 as an error value and handle that on the frontend? + duration = 60_i32 + + SearchVideo.new({ + title: title, + id: video_id, + author: author_fallback.name, + ucid: author_fallback.id, + published: Time.unix(0), + views: view_count, + description_html: "", + length_seconds: duration, + premiere_timestamp: Time.unix(0), + author_verified: false, + badges: VideoBadges::None, + }) + end + + def self.parser_name + return {{@type.name}} + end + end + # Parses an InnerTube continuationItemRenderer into a Continuation. # Returns nil when the given object isn't a continuationItemRenderer. # |