PrivateCoffee/invidious
4
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 2

Fix parsing shorts on channel page

Browse source
This commit is contained in:
ChunkyProgrammer 2023-03-21 23:47:52 -04:00
parent 8a44bd11d2
commit 5767344746
2 changed files with 35 additions and 59 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

31
src/invidious/channels/videos.cr
View file

@ -127,38 +127,15 @@ module Invidious::Channel::Tabs
# Shorts # Shorts
# ------------------- # -------------------
private def fetch_shorts_data(ucid : String, continuation : String? = nil) def get_shorts(channel : AboutChannel, continuation : String? = nil)
if continuation.nil? if continuation.nil?
# EgZzaG9ydHPyBgUKA5oBAA%3D%3D is the protobuf object to load "shorts" # EgZzaG9ydHPyBgUKA5oBAA%3D%3D is the protobuf object to load "shorts"
# TODO: try to extract the continuation tokens that allows other sorting options # 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 else
return YoutubeAPI.browse(continuation: continuation) initial_data = 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
end end
return extract_items(initial_data, channel.author, channel.ucid)
end end
# ------------------- # -------------------

63
src/invidious/yt_backend/extractors.cr
View file

@ -423,42 +423,41 @@ private module Parsers
"overlay", "reelPlayerOverlayRenderer" "overlay", "reelPlayerOverlayRenderer"
) )
# Sometimes, the "reelPlayerOverlayRenderer" object is missing the if video_details_container = reel_player_overlay.dig?(
# important part of the response. We use this exception to tell "reelPlayerHeaderSupportedRenderers",
# the calling function to fetch the content again. "reelPlayerHeaderRenderer"
if !reel_player_overlay.as_h.has_key?("reelPlayerHeaderSupportedRenderers") )
raise RetryOnceException.new # Author infos
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
# Title & publication date
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
# 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 end
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
ucid = video_details_container
.dig?("channelNavigationEndpoint", "browseEndpoint", "browseId")
.try &.as_s || author_fallback.id
# Title & publication date
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
# View count # View count
# View count used to be in the reelWatchEndpoint, but that changed? # View count used to be in the reelWatchEndpoint, but that changed?
view_count_text = item_contents.dig?("viewCountText", "simpleText") view_count_text ||= item_contents.dig?("viewCountText", "simpleText")
view_count_text ||= video_details_container.dig?("viewCountText", "simpleText")
view_count = view_count_text.try &.as_s.gsub(/\D+/, "").to_i64? || 0_i64 view_count = view_count_text.try &.as_s.gsub(/\D+/, "").to_i64? || 0_i64