diff options
Diffstat (limited to 'src')
25 files changed, 304 insertions, 259 deletions
diff --git a/src/invidious/channels/channels.cr b/src/invidious/channels/channels.cr index 63dd2194..c3d6124f 100644 --- a/src/invidious/channels/channels.cr +++ b/src/invidious/channels/channels.cr @@ -159,12 +159,18 @@ def fetch_channel(ucid, pull_all_videos : Bool) LOGGER.debug("fetch_channel: #{ucid}") LOGGER.trace("fetch_channel: #{ucid} : pull_all_videos = #{pull_all_videos}") + namespaces = { + "yt" => "http://www.youtube.com/xml/schemas/2015", + "media" => "http://search.yahoo.com/mrss/", + "default" => "http://www.w3.org/2005/Atom", + } + LOGGER.trace("fetch_channel: #{ucid} : Downloading RSS feed") rss = YT_POOL.client &.get("/feeds/videos.xml?channel_id=#{ucid}").body LOGGER.trace("fetch_channel: #{ucid} : Parsing RSS feed") - rss = XML.parse_html(rss) + rss = XML.parse(rss) - author = rss.xpath_node(%q(//feed/title)) + author = rss.xpath_node("//default:feed/default:title", namespaces) if !author raise InfoException.new("Deleted or invalid channel") end @@ -192,15 +198,23 @@ def fetch_channel(ucid, pull_all_videos : Bool) videos, continuation = IV::Channel::Tabs.get_videos(channel) LOGGER.trace("fetch_channel: #{ucid} : Extracting videos from channel RSS feed") - rss.xpath_nodes("//feed/entry").each do |entry| - video_id = entry.xpath_node("videoid").not_nil!.content - title = entry.xpath_node("title").not_nil!.content - published = Time.parse_rfc3339(entry.xpath_node("published").not_nil!.content) - updated = Time.parse_rfc3339(entry.xpath_node("updated").not_nil!.content) - author = entry.xpath_node("author/name").not_nil!.content - ucid = entry.xpath_node("channelid").not_nil!.content - views = entry.xpath_node("group/community/statistics").try &.["views"]?.try &.to_i64? - views ||= 0_i64 + rss.xpath_nodes("//default:feed/default:entry", namespaces).each do |entry| + video_id = entry.xpath_node("yt:videoId", namespaces).not_nil!.content + title = entry.xpath_node("default:title", namespaces).not_nil!.content + + published = Time.parse_rfc3339( + entry.xpath_node("default:published", namespaces).not_nil!.content + ) + updated = Time.parse_rfc3339( + entry.xpath_node("default:updated", namespaces).not_nil!.content + ) + + author = entry.xpath_node("default:author/default:name", namespaces).not_nil!.content + ucid = entry.xpath_node("yt:channelId", namespaces).not_nil!.content + + views = entry + .xpath_node("media:group/media:community/media:statistics", namespaces) + .try &.["views"]?.try &.to_i64? || 0_i64 channel_video = videos .select(SearchVideo) diff --git a/src/invidious/channels/community.cr b/src/invidious/channels/community.cr index ce34ff82..2c7b9fec 100644 --- a/src/invidious/channels/community.cr +++ b/src/invidious/channels/community.cr @@ -31,18 +31,16 @@ def fetch_channel_community(ucid, continuation, locale, format, thin_mode) session_token: session_token, } - response = YT_POOL.client &.post("/comment_service_ajax?action_get_comments=1&ctoken=#{continuation}&continuation=#{continuation}&hl=en&gl=US", headers, form: post_req) - body = JSON.parse(response.body) + body = YoutubeAPI.browse(continuation) - body = body["response"]["continuationContents"]["itemSectionContinuation"]? || - body["response"]["continuationContents"]["backstageCommentsContinuation"]? + body = body.dig?("continuationContents", "itemSectionContinuation") || + body.dig?("continuationContents", "backstageCommentsContinuation") if !body raise InfoException.new("Could not extract continuation.") end end - continuation = body["continuations"]?.try &.[0]["nextContinuationData"]["continuation"].as_s posts = body["contents"].as_a if message = posts[0]["messageRenderer"]? @@ -125,49 +123,13 @@ def fetch_channel_community(ucid, continuation, locale, format, thin_mode) if attachment = post["backstageAttachment"]? json.field "attachment" do - json.object do - case attachment.as_h - when .has_key?("videoRenderer") - attachment = attachment["videoRenderer"] - json.field "type", "video" - - if !attachment["videoId"]? - error_message = (attachment["title"]["simpleText"]? || - attachment["title"]["runs"]?.try &.[0]?.try &.["text"]?) - - json.field "error", error_message - else - video_id = attachment["videoId"].as_s - - video_title = attachment["title"]["simpleText"]? || attachment["title"]["runs"]?.try &.[0]?.try &.["text"]? - json.field "title", video_title - json.field "videoId", video_id - json.field "videoThumbnails" do - Invidious::JSONify::APIv1.thumbnails(json, video_id) - end - - json.field "lengthSeconds", decode_length_seconds(attachment["lengthText"]["simpleText"].as_s) - - author_info = attachment["ownerText"]["runs"][0].as_h - - json.field "author", author_info["text"].as_s - json.field "authorId", author_info["navigationEndpoint"]["browseEndpoint"]["browseId"] - json.field "authorUrl", author_info["navigationEndpoint"]["commandMetadata"]["webCommandMetadata"]["url"] - - # TODO: json.field "authorThumbnails", "channelThumbnailSupportedRenderers" - # TODO: json.field "authorVerified", "ownerBadges" - - published = decode_date(attachment["publishedTimeText"]["simpleText"].as_s) - - json.field "published", published.to_unix - json.field "publishedText", translate(locale, "`x` ago", recode_date(published, locale)) - - view_count = attachment["viewCountText"]?.try &.["simpleText"].as_s.gsub(/\D/, "").to_i64? || 0_i64 - - json.field "viewCount", view_count - json.field "viewCountText", translate_count(locale, "generic_views_count", view_count, NumberFormatting::Short) - end - when .has_key?("backstageImageRenderer") + case attachment.as_h + when .has_key?("videoRenderer") + parse_item(attachment) + .as(SearchVideo) + .to_json(locale, json) + when .has_key?("backstageImageRenderer") + json.object do attachment = attachment["backstageImageRenderer"] json.field "type", "image" @@ -188,7 +150,9 @@ def fetch_channel_community(ucid, continuation, locale, format, thin_mode) end end end - when .has_key?("pollRenderer") + end + when .has_key?("pollRenderer") + json.object do attachment = attachment["pollRenderer"] json.field "type", "poll" json.field "totalVotes", short_text_to_number(attachment["totalVotes"]["simpleText"].as_s.split(" ")[0]) @@ -221,7 +185,9 @@ def fetch_channel_community(ucid, continuation, locale, format, thin_mode) end end end - when .has_key?("postMultiImageRenderer") + end + when .has_key?("postMultiImageRenderer") + json.object do attachment = attachment["postMultiImageRenderer"] json.field "type", "multiImage" json.field "images" do @@ -245,7 +211,13 @@ def fetch_channel_community(ucid, continuation, locale, format, thin_mode) end end end - else + end + when .has_key?("playlistRenderer") + parse_item(attachment) + .as(SearchPlaylist) + .to_json(locale, json) + else + json.object do json.field "type", "unknown" json.field "error", "Unrecognized attachment type." end @@ -270,10 +242,8 @@ def fetch_channel_community(ucid, continuation, locale, format, thin_mode) end end end - - if body["continuations"]? - continuation = body["continuations"][0]["nextContinuationData"]["continuation"].as_s - json.field "continuation", extract_channel_community_cursor(continuation) + if cont = posts.dig?(-1, "continuationItemRenderer", "continuationEndpoint", "continuationCommand", "token") + json.field "continuation", extract_channel_community_cursor(cont.as_s) end end end diff --git a/src/invidious/channels/videos.cr b/src/invidious/channels/videos.cr index 3d53f2ab..12ed4a7d 100644 --- a/src/invidious/channels/videos.cr +++ b/src/invidious/channels/videos.cr @@ -129,38 +129,15 @@ module Invidious::Channel::Tabs # Shorts # ------------------- - private def fetch_shorts_data(ucid : String, continuation : String? = nil) + 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 - return YoutubeAPI.browse(ucid, params: "EgZzaG9ydHPyBgUKA5oBAA%3D%3D") + initial_data = YoutubeAPI.browse(channel.ucid, params: "EgZzaG9ydHPyBgUKA5oBAA%3D%3D") else - return YoutubeAPI.browse(continuation: continuation) - end - end - - def get_shorts(channel : AboutChannel, continuation : String? = nil) - initial_data = self.fetch_shorts_data(channel.ucid, continuation) - - begin - # Try to parse the initial data fetched above - return extract_items(initial_data, channel.author, channel.ucid) - rescue ex : RetryOnceException - # Sometimes, for a completely unknown reason, the "reelItemRenderer" - # object is missing some critical information (it happens once in about - # 20 subsequent requests). Refreshing the page is required to properly - # show the "shorts" tab. - # - # In order to make the experience smoother for the user, we simulate - # said page refresh by fetching again the JSON. If that still doesn't - # work, we raise a BrokenTubeException, as something is really broken. - begin - initial_data = self.fetch_shorts_data(channel.ucid, continuation) - return extract_items(initial_data, channel.author, channel.ucid) - rescue ex : RetryOnceException - raise BrokenTubeException.new "reelPlayerHeaderSupportedRenderers" - end + initial_data = YoutubeAPI.browse(continuation: continuation) end + return extract_items(initial_data, channel.author, channel.ucid) end # ------------------- diff --git a/src/invidious/comments.cr b/src/invidious/comments.cr index b15d63d4..01556099 100644 --- a/src/invidious/comments.cr +++ b/src/invidious/comments.cr @@ -346,7 +346,7 @@ def template_youtube_comments(comments, locale, thin_mode, is_replies = false) html << <<-END_HTML <div class="pure-g" style="width:100%"> <div class="channel-profile pure-u-4-24 pure-u-md-2-24"> - <img loading="lazy" style="margin-right:1em;margin-top:1em;width:90%" src="#{author_thumbnail}"> + <img loading="lazy" style="margin-right:1em;margin-top:1em;width:90%" src="#{author_thumbnail}" alt="" /> </div> <div class="pure-u-20-24 pure-u-md-22-24"> <p> @@ -367,37 +367,30 @@ def template_youtube_comments(comments, locale, thin_mode, is_replies = false) html << <<-END_HTML <div class="pure-g"> <div class="pure-u-1 pure-u-md-1-2"> - <img loading="lazy" style="width:100%" src="/ggpht#{URI.parse(attachment["url"].as_s).request_target}"> + <img loading="lazy" style="width:100%" src="/ggpht#{URI.parse(attachment["url"].as_s).request_target}" alt="" /> </div> </div> END_HTML when "video" - html << <<-END_HTML - <div class="pure-g"> - <div class="pure-u-1 pure-u-md-1-2"> - <div style="position:relative;width:100%;height:0;padding-bottom:56.25%;margin-bottom:5px"> - END_HTML - if attachment["error"]? html << <<-END_HTML + <div class="pure-g video-iframe-wrapper"> <p>#{attachment["error"]}</p> + </div> END_HTML else html << <<-END_HTML - <iframe id='ivplayer' style='position:absolute;width:100%;height:100%;left:0;top:0' src='/embed/#{attachment["videoId"]?}?autoplay=0' style='border:none;'></iframe> + <div class="pure-g video-iframe-wrapper"> + <iframe class="video-iframe" src='/embed/#{attachment["videoId"]?}?autoplay=0'></iframe> + </div> END_HTML end - - html << <<-END_HTML - </div> - </div> - </div> - END_HTML else nil # Ignore end end html << <<-END_HTML + <p> <span title="#{Time.unix(child["published"].as_i64).to_s(translate(locale, "%A %B %-d, %Y"))}">#{translate(locale, "`x` ago", recode_date(Time.unix(child["published"].as_i64), locale))} #{child["isEdited"] == true ? translate(locale, "(edited)") : ""}</span> | END_HTML @@ -416,6 +409,7 @@ def template_youtube_comments(comments, locale, thin_mode, is_replies = false) html << <<-END_HTML <i class="icon ion-ios-thumbs-up"></i> #{number_with_separator(child["likeCount"])} + </p> END_HTML if child["creatorHeart"]? @@ -428,7 +422,7 @@ def template_youtube_comments(comments, locale, thin_mode, is_replies = false) html << <<-END_HTML <span class="creator-heart-container" title="#{translate(locale, "`x` marked it with a ❤", child["creatorHeart"]["creatorName"].as_s)}"> <div class="creator-heart"> - <img loading="lazy" class="creator-heart-background-hearted" src="#{creator_thumbnail}"></img> + <img loading="lazy" class="creator-heart-background-hearted" src="#{creator_thumbnail}" alt="" /> <div class="creator-heart-small-hearted"> <div class="icon ion-ios-heart creator-heart-small-container"></div> </div> @@ -604,7 +598,7 @@ def text_to_parsed_content(text : String) : JSON::Any currentNode = {"text" => urlMatch[0], "navigationEndpoint" => {"urlEndpoint" => {"url" => urlMatch[0]}}} currentNodes << (JSON.parse(currentNode.to_json)) # If text remain after match create new simple node with text after match - afterNode = {"text" => splittedLastNode.size > 0 ? splittedLastNode[1] : ""} + afterNode = {"text" => splittedLastNode.size > 1 ? splittedLastNode[1] : ""} currentNodes << (JSON.parse(afterNode.to_json)) end @@ -635,55 +629,8 @@ def content_to_comment_html(content, video_id : String? = "") text = HTML.escape(run["text"].as_s) - if run["navigationEndpoint"]? - if url = run["navigationEndpoint"]["urlEndpoint"]?.try &.["url"].as_s - url = URI.parse(url) - displayed_url = text - - if url.host == "youtu.be" - url = "/watch?v=#{url.request_target.lstrip('/')}" - elsif url.host.nil? || url.host.not_nil!.ends_with?("youtube.com") - if url.path == "/redirect" - # Sometimes, links can be corrupted (why?) so make sure to fallback - # nicely. See https://github.com/iv-org/invidious/issues/2682 - url = url.query_params["q"]? || "" - displayed_url = url - else - url = url.request_target - displayed_url = "youtube.com#{url}" - end - end - - text = %(<a href="#{url}">#{reduce_uri(displayed_url)}</a>) - elsif watch_endpoint = run["navigationEndpoint"]["watchEndpoint"]? - start_time = watch_endpoint["startTimeSeconds"]?.try &.as_i - link_video_id = watch_endpoint["videoId"].as_s - - url = "/watch?v=#{link_video_id}" - url += "&t=#{start_time}" if !start_time.nil? - - # If the current video ID (passed through from the caller function) - # is the same as the video ID in the link, add HTML attributes for - # the JS handler function that bypasses page reload. - # - # See: https://github.com/iv-org/invidious/issues/3063 - if link_video_id == video_id - start_time ||= 0 - text = %(<a href="#{url}" data-onclick="jump_to_time" data-jump-time="#{start_time}">#{reduce_uri(text)}</a>) - else - text = %(<a href="#{url}">#{text}</a>) - end - elsif url = run.dig?("navigationEndpoint", "commandMetadata", "webCommandMetadata", "url").try &.as_s - if text.starts_with?(/\s?[@#]/) - # Handle "pings" in comments and hasthags differently - # See: - # - https://github.com/iv-org/invidious/issues/3038 - # - https://github.com/iv-org/invidious/issues/3062 - text = %(<a href="#{url}">#{text}</a>) - else - text = %(<a href="#{url}">#{reduce_uri(url)}</a>) - end - end + if navigationEndpoint = run.dig?("navigationEndpoint") + text = parse_link_endpoint(navigationEndpoint, text, video_id) end text = "<b>#{text}</b>" if run["bold"]? @@ -702,7 +649,7 @@ def content_to_comment_html(content, video_id : String? = "") str << %(title=") << emojiAlt << "\" " str << %(width=") << emojiThumb["width"] << "\" " str << %(height=") << emojiThumb["height"] << "\" " - str << %(class="channel-emoji"/>) + str << %(class="channel-emoji" />) end else # Hide deleted channel emoji diff --git a/src/invidious/database/users.cr b/src/invidious/database/users.cr index 0a4a4fd8..d54e6a76 100644 --- a/src/invidious/database/users.cr +++ b/src/invidious/database/users.cr @@ -52,7 +52,7 @@ module Invidious::Database::Users def mark_watched(user : User, vid : String) request = <<-SQL UPDATE users - SET watched = array_append(watched, $1) + SET watched = array_append(array_remove(watched, $1), $1) WHERE email = $2 SQL diff --git a/src/invidious/hashtag.cr b/src/invidious/hashtag.cr index bc329205..d9d584c9 100644 --- a/src/invidious/hashtag.cr +++ b/src/invidious/hashtag.cr @@ -17,21 +17,18 @@ module Invidious::Hashtag "80226972:embedded" => { "2:string" => "FEhashtag", "3:base64" => { - "1:varint" => cursor.to_i64, - }, - "7:base64" => { - "325477796:embedded" => { - "1:embedded" => { - "2:0:embedded" => { - "2:string" => '#' + hashtag, - "4:varint" => 0_i64, - "11:string" => "", - }, - "4:string" => "browse-feedFEhashtag", - }, - "2:string" => hashtag, + "1:varint" => 60_i64, # result count + "15:base64" => { + "1:varint" => cursor.to_i64, + "2:varint" => 0_i64, + }, + "93:2:embedded" => { + "1:string" => hashtag, + "2:varint" => 0_i64, + "3:varint" => 1_i64, }, }, + "35:string" => "browse-feedFEhashtag", }, } diff --git a/src/invidious/helpers/serialized_yt_data.cr b/src/invidious/helpers/serialized_yt_data.cr index c1874780..7c12ad0e 100644 --- a/src/invidious/helpers/serialized_yt_data.cr +++ b/src/invidious/helpers/serialized_yt_data.cr @@ -84,6 +84,7 @@ struct SearchVideo json.field "descriptionHtml", self.description_html json.field "viewCount", self.views + json.field "viewCountText", translate_count(locale, "generic_views_count", self.views, NumberFormatting::Short) json.field "published", self.published.to_unix json.field "publishedText", translate(locale, "`x` ago", recode_date(self.published, locale)) json.field "lengthSeconds", self.length_seconds diff --git a/src/invidious/helpers/utils.cr b/src/invidious/helpers/utils.cr index 500a2582..bcf7c963 100644 --- a/src/invidious/helpers/utils.cr +++ b/src/invidious/helpers/utils.cr @@ -389,3 +389,56 @@ def reduce_uri(uri : URI | String, max_length : Int32 = 50, suffix : String = " end return str end + +# Get the html link from a NavigationEndpoint or an innertubeCommand +def parse_link_endpoint(endpoint : JSON::Any, text : String, video_id : String) + if url = endpoint.dig?("urlEndpoint", "url").try &.as_s + url = URI.parse(url) + displayed_url = text + + if url.host == "youtu.be" + url = "/watch?v=#{url.request_target.lstrip('/')}" + elsif url.host.nil? || url.host.not_nil!.ends_with?("youtube.com") + if url.path == "/redirect" + # Sometimes, links can be corrupted (why?) so make sure to fallback + # nicely. See https://github.com/iv-org/invidious/issues/2682 + url = url.query_params["q"]? || "" + displayed_url = url + else + url = url.request_target + displayed_url = "youtube.com#{url}" + end + end + + text = %(<a href="#{url}">#{reduce_uri(displayed_url)}</a>) + elsif watch_endpoint = endpoint.dig?("watchEndpoint") + start_time = watch_endpoint["startTimeSeconds"]?.try &.as_i + link_video_id = watch_endpoint["videoId"].as_s + + url = "/watch?v=#{link_video_id}" + url += "&t=#{start_time}" if !start_time.nil? + + # If the current video ID (passed through from the caller function) + # is the same as the video ID in the link, add HTML attributes for + # the JS handler function that bypasses page reload. + # + # See: https://github.com/iv-org/invidious/issues/3063 + if link_video_id == video_id + start_time ||= 0 + text = %(<a href="#{url}" data-onclick="jump_to_time" data-jump-time="#{start_time}">#{reduce_uri(text)}</a>) + else + text = %(<a href="#{url}">#{text}</a>) + end + elsif url = endpoint.dig?("commandMetadata", "webCommandMetadata", "url").try &.as_s + if text.starts_with?(/\s?[@#]/) + # Handle "pings" in comments and hasthags differently + # See: + # - https://github.com/iv-org/invidious/issues/3038 + # - https://github.com/iv-org/invidious/issues/3062 + text = %(<a href="#{url}">#{text}</a>) + else + text = %(<a href="#{url}">#{reduce_uri(url)}</a>) + end + end + return text +end diff --git a/src/invidious/mixes.cr b/src/invidious/mixes.cr index 3f342b92..823ca85b 100644 --- a/src/invidious/mixes.cr +++ b/src/invidious/mixes.cr @@ -97,7 +97,7 @@ def template_mix(mix) <li class="pure-menu-item"> <a href="/watch?v=#{video["videoId"]}&list=#{mix["mixId"]}"> <div class="thumbnail"> - <img loading="lazy" class="thumbnail" src="/vi/#{video["videoId"]}/mqdefault.jpg"> + <img loading="lazy" class="thumbnail" src="/vi/#{video["videoId"]}/mqdefault.jpg" alt="" /> <p class="length">#{recode_length_seconds(video["lengthSeconds"].as_i)}</p> </div> <p style="width:100%">#{video["title"]}</p> diff --git a/src/invidious/playlists.cr b/src/invidious/playlists.cr index 57f1f53e..013be268 100644 --- a/src/invidious/playlists.cr +++ b/src/invidious/playlists.cr @@ -507,7 +507,7 @@ def template_playlist(playlist) <li class="pure-menu-item" id="#{video["videoId"]}"> <a href="/watch?v=#{video["videoId"]}&list=#{playlist["playlistId"]}&index=#{video["index"]}"> <div class="thumbnail"> - <img loading="lazy" class="thumbnail" src="/vi/#{video["videoId"]}/mqdefault.jpg"> + <img loading="lazy" class="thumbnail" src="/vi/#{video["videoId"]}/mqdefault.jpg" alt="" /> <p class="length">#{recode_length_seconds(video["lengthSeconds"].as_i)}</p> </div> <p style="width:100%">#{video["title"]}</p> diff --git a/src/invidious/routes/api/v1/search.cr b/src/invidious/routes/api/v1/search.cr index 21451d33..9fb283c2 100644 --- a/src/invidious/routes/api/v1/search.cr +++ b/src/invidious/routes/api/v1/search.cr @@ -55,4 +55,32 @@ module Invidious::Routes::API::V1::Search return error_json(500, ex) end end + + def self.hashtag(env) + hashtag = env.params.url["hashtag"] + + page = env.params.query["page"]?.try &.to_i? || 1 + + locale = env.get("preferences").as(Preferences).locale + region = env.params.query["region"]? + env.response.content_type = "application/json" + + begin + results = Invidious::Hashtag.fetch(hashtag, page, region) + rescue ex + return error_json(400, ex) + end + + JSON.build do |json| + json.object do + json.field "results" do + json.array do + results.each do |item| + item.to_json(locale, json) + end + end + end + end + end + end end diff --git a/src/invidious/routes/watch.cr b/src/invidious/routes/watch.cr index 5d3845c3..813cb0f4 100644 --- a/src/invidious/routes/watch.cr +++ b/src/invidious/routes/watch.cr @@ -76,7 +76,7 @@ module Invidious::Routes::Watch end env.params.query.delete_all("iv_load_policy") - if watched && preferences.watch_history && !watched.includes? id + if watched && preferences.watch_history Invidious::Database::Users.mark_watched(user.as(User), id) end @@ -259,9 +259,7 @@ module Invidious::Routes::Watch case action when "action_mark_watched" - if !user.watched.includes? id - Invidious::Database::Users.mark_watched(user, id) - end + Invidious::Database::Users.mark_watched(user, id) when "action_mark_unwatched" Invidious::Database::Users.mark_unwatched(user, id) else diff --git a/src/invidious/routing.cr b/src/invidious/routing.cr index 9e2ade3d..72ee9194 100644 --- a/src/invidious/routing.cr +++ b/src/invidious/routing.cr @@ -243,6 +243,7 @@ module Invidious::Routing # Search get "/api/v1/search", {{namespace}}::Search, :search get "/api/v1/search/suggestions", {{namespace}}::Search, :search_suggestions + get "/api/v1/hashtag/:hashtag", {{namespace}}::Search, :hashtag # Authenticated diff --git a/src/invidious/search/processors.cr b/src/invidious/search/processors.cr index 7e909590..25edb936 100644 --- a/src/invidious/search/processors.cr +++ b/src/invidious/search/processors.cr @@ -10,7 +10,7 @@ module Invidious::Search initial_data = YoutubeAPI.search(query.text, search_params, client_config: client_config) items, _ = extract_items(initial_data) - return items + return items.reject!(Category) end # Search a youtube channel @@ -32,7 +32,7 @@ module Invidious::Search response_json = YoutubeAPI.browse(continuation) items, _ = extract_items(response_json, "", ucid) - return items + return items.reject!(Category) end # Search inside of user subscriptions diff --git a/src/invidious/search/query.cr b/src/invidious/search/query.cr index 24e79609..e38845d9 100644 --- a/src/invidious/search/query.cr +++ b/src/invidious/search/query.cr @@ -113,7 +113,7 @@ module Invidious::Search case @type when .regular?, .playlist? - items = unnest_items(Processors.regular(self)) + items = Processors.regular(self) # when .channel? items = Processors.channel(self) @@ -136,26 +136,5 @@ module Invidious::Search return params end - - # TODO: clean code - private def unnest_items(all_items) : Array(SearchItem) - items = [] of SearchItem - - # Light processing to flatten search results out of Categories. - # They should ideally be supported in the future. - all_items.each do |i| - if i.is_a? Category - i.contents.each do |nest_i| - if !nest_i.is_a? Video - items << nest_i - end - end - else - items << i - end - end - - return items - end end end diff --git a/src/invidious/trending.cr b/src/invidious/trending.cr index 134eb437..2d9f8a83 100644 --- a/src/invidious/trending.cr +++ b/src/invidious/trending.cr @@ -17,7 +17,24 @@ def fetch_trending(trending_type, region, locale) client_config = YoutubeAPI::ClientConfig.new(region: region) initial_data = YoutubeAPI.browse("FEtrending", params: params, client_config: client_config) - trending = extract_videos(initial_data) - return {trending, plid} + items, _ = extract_items(initial_data) + + extracted = [] of SearchItem + + items.each do |itm| + if itm.is_a?(Category) + # Ignore the smaller categories, as they generally contain a sponsored + # channel, which brings a lot of noise on the trending page. + # See: https://github.com/iv-org/invidious/issues/2989 + next if itm.contents.size < 24 + + extracted.concat extract_category(itm) + else + extracted << itm + end + end + + # Deduplicate items before returning results + return extracted.select(SearchVideo).uniq!(&.id), plid end diff --git a/src/invidious/user/imports.cr b/src/invidious/user/imports.cr index 673991f7..e4b25156 100644 --- a/src/invidious/user/imports.cr +++ b/src/invidious/user/imports.cr @@ -102,7 +102,7 @@ struct Invidious::User if data["watch_history"]? user.watched += data["watch_history"].as_a.map(&.as_s) - user.watched.uniq! + user.watched.reverse!.uniq!.reverse! Invidious::Database::Users.update_watch_history(user) end diff --git a/src/invidious/videos/description.cr b/src/invidious/videos/description.cr new file mode 100644 index 00000000..542cb416 --- /dev/null +++ b/src/invidious/videos/description.cr @@ -0,0 +1,64 @@ +require "json" +require "uri" + +private def copy_string(str : String::Builder, iter : Iterator, count : Int) : Int + copied = 0 + while copied < count + cp = iter.next + break if cp.is_a?(Iterator::Stop) + + str << cp.chr + + # A codepoint from the SMP counts twice + copied += 1 if cp > 0xFFFF + copied += 1 + end + + return copied +end + +def parse_description(desc, video_id : String) : String? + return "" if desc.nil? + + content = desc["content"].as_s + return "" if content.empty? + + commands = desc["commandRuns"]?.try &.as_a + return content if commands.nil? + + # Not everything is stored in UTF-8 on youtube's side. The SMP codepoints + # (0x10000 and above) are encoded as UTF-16 surrogate pairs, which are + # automatically decoded by the JSON parser. It means that we need to count + # copied byte in a special manner, preventing the use of regular string copy. + iter = content.each_codepoint + + index = 0 + + return String.build do |str| + commands.each do |command| + cmd_start = command["startIndex"].as_i + cmd_length = command["length"].as_i + + # Copy the text chunk between this command and the previous if needed. + length = cmd_start - index + index += copy_string(str, iter, length) + + # We need to copy the command's text using the iterator + # and the special function defined above. + cmd_content = String.build(cmd_length) do |str2| + copy_string(str2, iter, cmd_length) + end + + link = cmd_content + if on_tap = command.dig?("onTap", "innertubeCommand") + link = parse_link_endpoint(on_tap, cmd_content, video_id) + end + str << link + index += cmd_length + end + + # Copy the end of the string (past the last command). + remaining_length = content.size - index + copy_string(str, iter, remaining_length) if remaining_length > 0 + end +end diff --git a/src/invidious/videos/parser.cr b/src/invidious/videos/parser.cr index 13ee5f65..2e8eecc3 100644 --- a/src/invidious/videos/parser.cr +++ b/src/invidious/videos/parser.cr @@ -284,8 +284,10 @@ def parse_video_info(video_id : String, player_response : Hash(String, JSON::Any description = microformat.dig?("description", "simpleText").try &.as_s || "" short_description = player_response.dig?("videoDetails", "shortDescription") - description_html = video_secondary_renderer.try &.dig?("description", "runs") - .try &.as_a.try { |t| content_to_comment_html(t, video_id) } + # description_html = video_secondary_renderer.try &.dig?("description", "runs") + # .try &.as_a.try { |t| content_to_comment_html(t, video_id) } + + description_html = parse_description(video_secondary_renderer.try &.dig?("attributedDescription"), video_id) # Video metadata diff --git a/src/invidious/views/components/channel_info.ecr b/src/invidious/views/components/channel_info.ecr index f216359f..59888760 100644 --- a/src/invidious/views/components/channel_info.ecr +++ b/src/invidious/views/components/channel_info.ecr @@ -1,6 +1,6 @@ <% if channel.banner %> <div class="h-box"> - <img style="width:100%" src="/ggpht<%= URI.parse(channel.banner.not_nil!.gsub("=w1060-", "=w1280-")).request_target %>"> + <img style="width:100%" src="/ggpht<%= URI.parse(channel.banner.not_nil!.gsub("=w1060-", "=w1280-")).request_target %>" alt="" /> </div> <div class="h-box"> @@ -11,7 +11,7 @@ <div class="pure-g h-box"> <div class="pure-u-2-3"> <div class="channel-profile"> - <img src="/ggpht<%= channel_profile_pic %>"> + <img src="/ggpht<%= channel_profile_pic %>" alt="" /> <span><%= author %></span><% if !channel.verified.nil? && channel.verified %> <i class="icon ion ion-md-checkmark-circle"></i><% end %> </div> </div> diff --git a/src/invidious/views/components/item.ecr b/src/invidious/views/components/item.ecr index fa12374f..7cfd38db 100644 --- a/src/invidious/views/components/item.ecr +++ b/src/invidious/views/components/item.ecr @@ -7,7 +7,7 @@ <a href="/channel/<%= item.ucid %>"> <% if !env.get("preferences").as(Preferences).thin_mode %> <center> - <img loading="lazy" tabindex="-1" style="width:56.25%" src="/ggpht<%= URI.parse(item.author_thumbnail).request_target.gsub(/=s\d+/, "=s176") %>"/> + <img loading="lazy" tabindex="-1" style="width:56.25%" src="/ggpht<%= URI.parse(item.author_thumbnail).request_target.gsub(/=s\d+/, "=s176") %>" alt="" /> </center> <% end %> <p dir="auto"><%= HTML.escape(item.author) %><% if !item.author_verified.nil? && item.author_verified %> <i class="icon ion ion-md-checkmark-circle"></i><% end %></p> @@ -25,7 +25,7 @@ <a style="width:100%" href="<%= url %>"> <% if !env.get("preferences").as(Preferences).thin_mode %> <div class="thumbnail"> - <img loading="lazy" tabindex="-1" class="thumbnail" src="<%= URI.parse(item.thumbnail || "/").request_target %>"/> + <img loading="lazy" tabindex="-1" class="thumbnail" src="<%= URI.parse(item.thumbnail || "/").request_target %>" alt="" /> <p class="length"><%= translate_count(locale, "generic_videos_count", item.video_count, NumberFormatting::Separator) %></p> </div> <% end %> @@ -38,7 +38,7 @@ <a href="/watch?v=<%= item.id %>&list=<%= item.rdid %>"> <% if !env.get("preferences").as(Preferences).thin_mode %> <div class="thumbnail"> - <img loading="lazy" tabindex="-1" class="thumbnail" src="/vi/<%= item.id %>/mqdefault.jpg"/> + <img loading="lazy" tabindex="-1" class="thumbnail" src="/vi/<%= item.id %>/mqdefault.jpg" alt="" /> <% if item.length_seconds != 0 %> <p class="length"><%= recode_length_seconds(item.length_seconds) %></p> <% end %> @@ -58,7 +58,7 @@ <a style="width:100%" href="/watch?v=<%= item.id %>&list=<%= item.plid %>&index=<%= item.index %>"> <% if !env.get("preferences").as(Preferences).thin_mode %> <div class="thumbnail"> - <img loading="lazy" tabindex="-1" class="thumbnail" src="/vi/<%= item.id %>/mqdefault.jpg"/> + <img loading="lazy" tabindex="-1" class="thumbnail" src="/vi/<%= item.id %>/mqdefault.jpg" alt="" /> <% if plid_form = env.get?("remove_playlist_items") %> <form data-onsubmit="return_false" action="/playlist_ajax?action_remove_video=1&set_video_id=<%= item.index %>&playlist_id=<%= plid_form %>&referer=<%= env.get("current_page") %>" method="post"> @@ -112,7 +112,7 @@ <a style="width:100%" href="/watch?v=<%= item.id %>"> <% if !env.get("preferences").as(Preferences).thin_mode %> <div class="thumbnail"> - <img loading="lazy" tabindex="-1" class="thumbnail" src="/vi/<%= item.id %>/mqdefault.jpg"/> + <img loading="lazy" tabindex="-1" class="thumbnail" src="/vi/<%= item.id %>/mqdefault.jpg" alt="" /> <% if env.get? "show_watched" %> <form data-onsubmit="return_false" action="/watch_ajax?action_mark_watched=1&id=<%= item.id %>&referer=<%= env.get("current_page") %>" method="post"> <input type="hidden" name="csrf_token" value="<%= HTML.escape(env.get?("csrf_token").try &.as(String) || "") %>"> diff --git a/src/invidious/views/feeds/history.ecr b/src/invidious/views/feeds/history.ecr index 471d21db..2234b297 100644 --- a/src/invidious/views/feeds/history.ecr +++ b/src/invidious/views/feeds/history.ecr @@ -34,7 +34,7 @@ <a style="width:100%" href="/watch?v=<%= item %>"> <% if !env.get("preferences").as(Preferences).thin_mode %> <div class="thumbnail"> - <img class="thumbnail" src="/vi/<%= item %>/mqdefault.jpg"/> + <img class="thumbnail" src="/vi/<%= item %>/mqdefault.jpg" alt="" /> <form data-onsubmit="return_false" action="/watch_ajax?action_mark_unwatched=1&id=<%= item %>&referer=<%= env.get("current_page") %>" method="post"> <input type="hidden" name="csrf_token" value="<%= URI.encode_www_form(env.get?("csrf_token").try &.as(String) || "") %>"> <p class="watched"> diff --git a/src/invidious/views/watch.ecr b/src/invidious/views/watch.ecr index a3ec94e8..5b3190f3 100644 --- a/src/invidious/views/watch.ecr +++ b/src/invidious/views/watch.ecr @@ -208,7 +208,7 @@ we're going to need to do it here in order to allow for translations. <a href="/channel/<%= video.ucid %>" style="display:block;width:fit-content;width:-moz-fit-content"> <div class="channel-profile"> <% if !video.author_thumbnail.empty? %> - <img src="/ggpht<%= URI.parse(video.author_thumbnail).request_target %>"> + <img src="/ggpht<%= URI.parse(video.author_thumbnail).request_target %>" alt="" /> <% end %> <span id="channel-name"><%= author %><% if !video.author_verified.nil? && video.author_verified %> <i class="icon ion ion-md-checkmark-circle"></i><% end %></span> </div> @@ -298,7 +298,7 @@ we're going to need to do it here in order to allow for translations. <a href="/watch?v=<%= rv["id"] %>&listen=<%= params.listen %>"> <% if !env.get("preferences").as(Preferences).thin_mode %> <div class="thumbnail"> - <img loading="lazy" class="thumbnail" src="/vi/<%= rv["id"] %>/mqdefault.jpg"> + <img loading="lazy" class="thumbnail" src="/vi/<%= rv["id"] %>/mqdefault.jpg" alt="" /> <p class="length"><%= recode_length_seconds(rv["length_seconds"]?.try &.to_i? || 0) %></p> </div> <% end %> diff --git a/src/invidious/yt_backend/extractors.cr b/src/invidious/yt_backend/extractors.cr index 1a37d606..8ff4c1f9 100644 --- a/src/invidious/yt_backend/extractors.cr +++ b/src/invidious/yt_backend/extractors.cr @@ -268,7 +268,7 @@ private module Parsers end private def self.parse(item_contents, author_fallback) - title = item_contents["title"]["simpleText"]?.try &.as_s || "" + title = extract_text(item_contents["title"]) || "" plid = item_contents["playlistId"]?.try &.as_s || "" video_count = HelperExtractors.get_video_count(item_contents) @@ -448,44 +448,43 @@ private module Parsers "overlay", "reelPlayerOverlayRenderer" ) - # Sometimes, the "reelPlayerOverlayRenderer" object is missing the - # important part of the response. We use this exception to tell - # the calling function to fetch the content again. - if !reel_player_overlay.as_h.has_key?("reelPlayerHeaderSupportedRenderers") - raise RetryOnceException.new - end - - video_details_container = reel_player_overlay.dig( - "reelPlayerHeaderSupportedRenderers", - "reelPlayerHeaderRenderer" - ) - - # Author infos + if video_details_container = reel_player_overlay.dig?( + "reelPlayerHeaderSupportedRenderers", + "reelPlayerHeaderRenderer" + ) + # Author infos - author = video_details_container - .dig?("channelTitleText", "runs", 0, "text") - .try &.as_s || author_fallback.name + author = video_details_container + .dig?("channelTitleText", "runs", 0, "text") + .try &.as_s || author_fallback.name - ucid = video_details_container - .dig?("channelNavigationEndpoint", "browseEndpoint", "browseId") - .try &.as_s || author_fallback.id + ucid = video_details_container + .dig?("channelNavigationEndpoint", "browseEndpoint", "browseId") + .try &.as_s || author_fallback.id - # Title & publication date + # Title & publication date - title = video_details_container.dig?("reelTitleText") - .try { |t| extract_text(t) } || "" + title = video_details_container.dig?("reelTitleText") + .try { |t| extract_text(t) } || "" - published = video_details_container - .dig?("timestampText", "simpleText") - .try { |t| decode_date(t.as_s) } || Time.utc + published = video_details_container + .dig?("timestampText", "simpleText") + .try { |t| decode_date(t.as_s) } || Time.utc + # View count + view_count_text = video_details_container.dig?("viewCountText", "simpleText") + else + author = author_fallback.name + ucid = author_fallback.id + published = Time.utc + title = item_contents.dig?("headline", "simpleText").try &.as_s || "" + end # View count # View count used to be in the reelWatchEndpoint, but that changed? - view_count_text = item_contents.dig?("viewCountText", "simpleText") - view_count_text ||= video_details_container.dig?("viewCountText", "simpleText") + view_count_text ||= item_contents.dig?("viewCountText", "simpleText") - view_count = view_count_text.try &.as_s.gsub(/\D+/, "").to_i64? || 0_i64 + view_count = short_text_to_number(view_count_text.try &.as_s || "0") # Duration diff --git a/src/invidious/yt_backend/extractors_utils.cr b/src/invidious/yt_backend/extractors_utils.cr index 0cb3c079..11d95958 100644 --- a/src/invidious/yt_backend/extractors_utils.cr +++ b/src/invidious/yt_backend/extractors_utils.cr @@ -68,19 +68,17 @@ rescue ex return false end -def extract_videos(initial_data : Hash(String, JSON::Any), author_fallback : String? = nil, author_id_fallback : String? = nil) : Array(SearchVideo) - extracted, _ = extract_items(initial_data, author_fallback, author_id_fallback) +# This function extracts SearchVideo items from a Category. +# Categories are commonly returned in search results and trending pages. +def extract_category(category : Category) : Array(SearchVideo) + return category.contents.select(SearchVideo) +end - target = [] of (SearchItem | Continuation) - extracted.each do |i| - if i.is_a?(Category) - i.contents.each { |cate_i| target << cate_i if !cate_i.is_a? Video } - else - target << i - end +# :ditto: +def extract_category(category : Category, &) + category.contents.select(SearchVideo).each do |item| + yield item end - - return target.select(SearchVideo) end def extract_selected_tab(tabs) |