fix: fetch channel videos through Innertube

This commit is contained in:
Agastya 2026-09-26 09:11:45 +05:30
parent c88230067b
commit b5fcc9bdbc
4 changed files with 43 additions and 78 deletions

View File

@ -19,6 +19,11 @@ record AboutChannel,
verified : Bool, verified : Bool,
is_age_gated : Bool is_age_gated : Bool
private DELETED_CHANNEL_ERROR_MESSAGES = {
"This channel does not exist.",
"This account has been terminated for a violation of YouTube's Terms of Service.",
}
def get_about_info(ucid) : AboutChannel def get_about_info(ucid) : AboutChannel
begin begin
# Fetch channel information from channel home page # Fetch channel information from channel home page
@ -29,7 +34,7 @@ def get_about_info(ucid) : AboutChannel
if initdata.dig?("alerts", 0, "alertRenderer", "type") == "ERROR" if initdata.dig?("alerts", 0, "alertRenderer", "type") == "ERROR"
error_message = initdata["alerts"][0]["alertRenderer"]["text"]["simpleText"].as_s error_message = initdata["alerts"][0]["alertRenderer"]["text"]["simpleText"].as_s
if error_message == "This channel does not exist." if DELETED_CHANNEL_ERROR_MESSAGES.includes?(error_message)
raise NotFoundException.new(error_message) raise NotFoundException.new(error_message)
else else
raise InfoException.new(error_message) raise InfoException.new(error_message)

View File

@ -159,36 +159,13 @@ 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 = { channel_info = get_about_info(ucid)
"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") LOGGER.trace("fetch_channel: #{ucid} : author = #{channel_info.author}, auto_generated = #{channel_info.auto_generated}")
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: ucid,
author: author, author: channel_info.author,
updated: Time.utc, updated: Time.utc,
deleted: false, deleted: false,
subscribed: nil, subscribed: nil,
@ -197,61 +174,32 @@ 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")
rss.xpath_nodes("//default:feed/default:entry", namespaces).each do |entry| videos.select(SearchVideo).each do |video|
video_id = entry.xpath_node("yt:videoId", namespaces).not_nil!.content channel_video = ChannelVideo.new({
title = entry.xpath_node("default:title", namespaces).not_nil!.content id: video.id,
title: video.title,
published = Time.parse_rfc3339( published: video.published,
entry.xpath_node("default:published", namespaces).not_nil!.content updated: Time.utc,
) ucid: video.ucid,
updated = Time.parse_rfc3339( author: video.author,
entry.xpath_node("default:updated", namespaces).not_nil!.content length_seconds: video.length_seconds,
) live_now: video.badges.live_now?,
premiere_timestamp: video.premiere_timestamp,
author = entry.xpath_node("default:author/default:name", namespaces).not_nil!.content views: video.views,
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({
id: video_id,
title: title,
published: published,
updated: updated,
ucid: ucid,
author: author,
length_seconds: length_seconds,
live_now: live_now,
premiere_timestamp: premiere_timestamp,
views: views,
}) })
LOGGER.trace("fetch_channel: #{ucid} : video #{video_id} : Updating or inserting video") LOGGER.trace("fetch_channel: #{ucid} : video #{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(channel_video, preserve_timestamps_on_conflict: true)
if was_insert if was_insert
LOGGER.trace("fetch_channel: #{ucid} : video #{video_id} : Inserted, updating subscriptions") LOGGER.trace("fetch_channel: #{ucid} : video #{video.id} : Inserted, updating subscriptions")
NOTIFICATION_CHANNEL.send(VideoNotification.from_video(video)) NOTIFICATION_CHANNEL.send(VideoNotification.from_video(channel_video))
else else
LOGGER.trace("fetch_channel: #{ucid} : video #{video_id} : Updated") LOGGER.trace("fetch_channel: #{ucid} : video #{video.id} : Updated")
end end
end end
@ -279,7 +227,7 @@ def fetch_channel(ucid, pull_all_videos : Bool)
# We are notified of Red videos elsewhere (PubSub), which includes a correct published date, # We are notified of Red videos elsewhere (PubSub), which includes a correct published date,
# so since they don't provide a published date here we can safely ignore them. # so since they don't provide a published date here we can safely ignore them.
if Time.utc - video.published > 1.minute if Time.utc - video.published > 1.minute
was_insert = Invidious::Database::ChannelVideos.insert(video) was_insert = Invidious::Database::ChannelVideos.insert(video, preserve_timestamps_on_conflict: true)
if was_insert if was_insert
NOTIFICATION_CHANNEL.send(VideoNotification.from_video(video)) NOTIFICATION_CHANNEL.send(VideoNotification.from_video(video))
end end

View File

@ -98,18 +98,30 @@ module Invidious::Database::ChannelVideos
# ------------------- # -------------------
# This function returns the status of the query (i.e: success?) # This function returns the status of the query (i.e: success?)
def insert(video : ChannelVideo, with_premiere_timestamp : Bool = false) : Bool def insert(
video : ChannelVideo,
with_premiere_timestamp : Bool = false,
preserve_timestamps_on_conflict : Bool = false,
) : Bool
if with_premiere_timestamp if with_premiere_timestamp
last_items = "premiere_timestamp = $9, views = $10" last_items = "premiere_timestamp = $9, views = $10"
else else
last_items = "views = $10" last_items = "views = $10"
end end
if preserve_timestamps_on_conflict
published_on_conflict = "published = COALESCE(channel_videos.published, EXCLUDED.published)"
updated_on_conflict = "updated = CASE WHEN channel_videos.title IS DISTINCT FROM EXCLUDED.title OR channel_videos.ucid IS DISTINCT FROM EXCLUDED.ucid OR channel_videos.author IS DISTINCT FROM EXCLUDED.author OR channel_videos.length_seconds IS DISTINCT FROM EXCLUDED.length_seconds OR channel_videos.live_now IS DISTINCT FROM EXCLUDED.live_now THEN EXCLUDED.updated ELSE channel_videos.updated END"
else
published_on_conflict = "published = $3"
updated_on_conflict = "updated = $4"
end
request = <<-SQL request = <<-SQL
INSERT INTO channel_videos INSERT INTO channel_videos
VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10) VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10)
ON CONFLICT (id) DO UPDATE ON CONFLICT (id) DO UPDATE
SET title = $2, published = $3, updated = $4, ucid = $5, SET title = $2, #{published_on_conflict}, #{updated_on_conflict}, ucid = $5,
author = $6, length_seconds = $7, live_now = $8, #{last_items} author = $6, length_seconds = $7, live_now = $8, #{last_items}
RETURNING (xmax=0) AS was_insert RETURNING (xmax=0) AS was_insert
SQL SQL

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 ex.is_a?(NotFoundException)
Invidious::Database::Channels.update_mark_deleted(id) Invidious::Database::Channels.update_mark_deleted(id)
else else
lim_fibers = 1 lim_fibers = 1