Use InnerTube to fetch channel videos

This commit is contained in:
leeweisa3-maker 2026-08-12 21:40:12 +08:00
parent 6865cf208e
commit fc52b5fd8b
2 changed files with 20 additions and 73 deletions

View File

@ -159,36 +159,12 @@ def fetch_channel(ucid, pull_all_videos : Bool)
LOGGER.debug("fetch_channel: #{ucid}") LOGGER.debug("fetch_channel: #{ucid}")
LOGGER.trace("fetch_channel: #{ucid} : pull_all_videos = #{pull_all_videos}") LOGGER.trace("fetch_channel: #{ucid} : pull_all_videos = #{pull_all_videos}")
namespaces = { LOGGER.trace("fetch_channel: #{ucid} : Downloading channel information")
"yt" => "http://www.youtube.com/xml/schemas/2015", channel_info = get_about_info(ucid, nil)
"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(rss)
author = rss.xpath_node("//default:feed/default:title", namespaces)
if !author
raise InfoException.new("Deleted or invalid channel")
end
author = author.content
# Auto-generated channels
# https://support.google.com/youtube/answer/2579942
if author.ends_with?(" - Topic") ||
{"Popular on YouTube", "Music", "Sports", "Gaming"}.includes? author
auto_generated = true
end
LOGGER.trace("fetch_channel: #{ucid} : author = #{author}, auto_generated = #{auto_generated}")
channel = InvidiousChannel.new({ channel = InvidiousChannel.new({
id: ucid, id: channel_info.ucid,
author: author, author: channel_info.author,
updated: Time.utc, updated: Time.utc,
deleted: false, deleted: false,
subscribed: nil, subscribed: nil,
@ -197,61 +173,32 @@ def fetch_channel(ucid, pull_all_videos : Bool)
LOGGER.trace("fetch_channel: #{ucid} : Downloading channel videos page") LOGGER.trace("fetch_channel: #{ucid} : Downloading channel videos page")
videos, continuation = IV::Channel::Tabs.get_videos(channel) videos, continuation = IV::Channel::Tabs.get_videos(channel)
LOGGER.trace("fetch_channel: #{ucid} : Extracting videos from channel RSS feed") LOGGER.trace("fetch_channel: #{ucid} : Extracting videos from channel response")
rss.xpath_nodes("//default:feed/default:entry", namespaces).each do |entry| videos.select(SearchVideo).each do |channel_video|
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)
.select(&.id.== video_id)[0]?
length_seconds = channel_video.try &.length_seconds
length_seconds ||= 0
live_now = channel_video.try &.badges.live_now?
live_now ||= false
premiere_timestamp = channel_video.try &.premiere_timestamp
video = ChannelVideo.new({ video = ChannelVideo.new({
id: video_id, id: channel_video.id,
title: title, title: channel_video.title,
published: published, published: channel_video.published,
updated: updated, updated: Time.utc,
ucid: ucid, ucid: channel.id,
author: author, author: channel.author,
length_seconds: length_seconds, length_seconds: channel_video.length_seconds,
live_now: live_now, live_now: channel_video.badges.live_now?,
premiere_timestamp: premiere_timestamp, premiere_timestamp: channel_video.premiere_timestamp,
views: views, views: channel_video.views,
}) })
LOGGER.trace("fetch_channel: #{ucid} : video #{video_id} : Updating or inserting video") LOGGER.trace("fetch_channel: #{ucid} : video #{channel_video.id} : Updating or inserting video")
# We don't include the 'premiere_timestamp' here because channel pages don't include them, # We don't include the 'premiere_timestamp' here because channel pages don't include them,
# meaning the above timestamp is always null # meaning the above timestamp is always null
was_insert = Invidious::Database::ChannelVideos.insert(video) was_insert = Invidious::Database::ChannelVideos.insert(video)
if was_insert if was_insert
LOGGER.trace("fetch_channel: #{ucid} : video #{video_id} : Inserted, updating subscriptions") LOGGER.trace("fetch_channel: #{ucid} : video #{channel_video.id} : Inserted, updating subscriptions")
NOTIFICATION_CHANNEL.send(VideoNotification.from_video(video)) NOTIFICATION_CHANNEL.send(VideoNotification.from_video(video))
else else
LOGGER.trace("fetch_channel: #{ucid} : video #{video_id} : Updated") LOGGER.trace("fetch_channel: #{ucid} : video #{channel_video.id} : Updated")
end end
end end

View File

@ -43,7 +43,7 @@ class Invidious::Jobs::RefreshChannelsJob < Invidious::Jobs::BaseJob
end end
rescue ex rescue ex
LOGGER.error("RefreshChannelsJob: #{id} : #{ex.message}") LOGGER.error("RefreshChannelsJob: #{id} : #{ex.message}")
if ex.message == "Deleted or invalid channel" if {"Deleted or invalid channel", "This channel does not exist."}.includes?(ex.message)
Invidious::Database::Channels.update_mark_deleted(id) Invidious::Database::Channels.update_mark_deleted(id)
else else
lim_fibers = 1 lim_fibers = 1