Merge pull request #2856 from SamantazFox/fix-related-videos

Fix related videos
This commit is contained in:
Samantaz Fox 2022-02-03 19:23:32 +01:00 committed by GitHub
commit bd221b7b2c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 122 additions and 42 deletions

View File

@ -0,0 +1,8 @@
# Exception used to hold the name of the missing item
# Should be used in all parsing functions
class BrokenTubeException < InfoException
getter element : String
def initialize(@element)
end
end

View File

@ -446,7 +446,7 @@ struct Video
end end
json.field "author", rv["author"] json.field "author", rv["author"]
json.field "authorUrl", rv["author_url"]? json.field "authorUrl", "/channel/#{rv["ucid"]?}"
json.field "authorId", rv["ucid"]? json.field "authorId", rv["ucid"]?
if rv["author_thumbnail"]? if rv["author_thumbnail"]?
json.field "authorThumbnails" do json.field "authorThumbnails" do
@ -455,7 +455,7 @@ struct Video
qualities.each do |quality| qualities.each do |quality|
json.object do json.object do
json.field "url", rv["author_thumbnail"]?.try &.gsub(/s\d+-/, "s#{quality}-") json.field "url", rv["author_thumbnail"].gsub(/s\d+-/, "s#{quality}-")
json.field "width", quality json.field "width", quality
json.field "height", quality json.field "height", quality
end end
@ -465,7 +465,7 @@ struct Video
end end
json.field "lengthSeconds", rv["length_seconds"]?.try &.to_i json.field "lengthSeconds", rv["length_seconds"]?.try &.to_i
json.field "viewCountText", rv["short_view_count_text"]? json.field "viewCountText", rv["short_view_count"]?
json.field "viewCount", rv["view_count"]?.try &.empty? ? nil : rv["view_count"].to_i64 json.field "viewCount", rv["view_count"]?.try &.empty? ? nil : rv["view_count"].to_i64
end end
end end
@ -802,23 +802,50 @@ class VideoRedirect < Exception
end end
end end
def parse_related(r : JSON::Any) : JSON::Any? # Use to parse both "compactVideoRenderer" and "endScreenVideoRenderer".
# TODO: r["endScreenPlaylistRenderer"], etc. # The former is preferred as it has more videos in it. The second has
return if !r["endScreenVideoRenderer"]? # the same 11 first entries as the compact rendered.
r = r["endScreenVideoRenderer"].as_h #
# TODO: "compactRadioRenderer" (Mix) and
def parse_related_video(related : JSON::Any) : Hash(String, JSON::Any)?
return nil if !related["videoId"]?
return if !r["lengthInSeconds"]? # The compact renderer has video length in seconds, where the end
# screen rendered has a full text version ("42:40")
length = related["lengthInSeconds"]?.try &.as_i.to_s
length ||= related.dig?("lengthText", "simpleText").try do |box|
decode_length_seconds(box.as_s).to_s
end
rv = {} of String => JSON::Any # Both have "short", so the "long" option shouldn't be required
rv["author"] = r["shortBylineText"]["runs"][0]?.try &.["text"] || JSON::Any.new("") channel_info = (related["shortBylineText"]? || related["longBylineText"]?)
rv["ucid"] = r["shortBylineText"]["runs"][0]?.try &.["navigationEndpoint"]["browseEndpoint"]["browseId"] || JSON::Any.new("") .try &.dig?("runs", 0)
rv["author_url"] = JSON::Any.new("/channel/#{rv["ucid"]}")
rv["length_seconds"] = JSON::Any.new(r["lengthInSeconds"].as_i.to_s) author = channel_info.try &.dig?("text")
rv["title"] = r["title"]["simpleText"] ucid = channel_info.try { |ci| HelperExtractors.get_browse_id(ci) }
rv["short_view_count_text"] = JSON::Any.new(r["shortViewCountText"]?.try &.["simpleText"]?.try &.as_s || "")
rv["view_count"] = JSON::Any.new(r["title"]["accessibility"]?.try &.["accessibilityData"]["label"].as_s.match(/(?<views>[1-9](\d+,?)*) views/).try &.["views"].gsub(/\D/, "") || "") # "4,088,033 views", only available on compact renderer
rv["id"] = r["videoId"] # and when video is not a livestream
JSON::Any.new(rv) view_count = related.dig?("viewCountText", "simpleText")
.try &.as_s.gsub(/\D/, "")
short_view_count = related.try do |r|
HelperExtractors.get_short_view_count(r).to_s
end
LOGGER.trace("parse_related_video: Found \"watchNextEndScreenRenderer\" container")
# TODO: when refactoring video types, make a struct for related videos
# or reuse an existing type, if that fits.
return {
"id" => related["videoId"],
"title" => related["title"]["simpleText"],
"author" => author || JSON::Any.new(""),
"ucid" => JSON::Any.new(ucid || ""),
"length_seconds" => JSON::Any.new(length || "0"),
"view_count" => JSON::Any.new(view_count || "0"),
"short_view_count" => JSON::Any.new(short_view_count || "0"),
}
end end
def extract_video_info(video_id : String, proxy_region : String? = nil, context_screen : String? = nil) def extract_video_info(video_id : String, proxy_region : String? = nil, context_screen : String? = nil)
@ -871,31 +898,62 @@ def extract_video_info(video_id : String, proxy_region : String? = nil, context_
params[f] = player_response[f] if player_response[f]? params[f] = player_response[f] if player_response[f]?
end end
params["relatedVideos"] = (
player_response
.dig?("playerOverlays", "playerOverlayRenderer", "endScreen", "watchNextEndScreenRenderer", "results")
.try &.as_a.compact_map { |r| parse_related r } || \
player_response
.dig?("webWatchNextResponseExtensionData", "relatedVideoArgs")
.try &.as_s.split(",").map { |r|
r = HTTP::Params.parse(r).to_h
JSON::Any.new(Hash.zip(r.keys, r.values.map { |v| JSON::Any.new(v) }))
}
).try { |a| JSON::Any.new(a) } || JSON::Any.new([] of JSON::Any)
# Top level elements # Top level elements
primary_results = player_response main_results = player_response.dig?("contents", "twoColumnWatchNextResults")
.dig?("contents", "twoColumnWatchNextResults", "results", "results", "contents")
raise BrokenTubeException.new("twoColumnWatchNextResults") if !main_results
primary_results = main_results.dig?("results", "results", "contents")
secondary_results = main_results
.dig?("secondaryResults", "secondaryResults", "results")
raise BrokenTubeException.new("results") if !primary_results
raise BrokenTubeException.new("secondaryResults") if !secondary_results
video_primary_renderer = primary_results video_primary_renderer = primary_results
.try &.as_a.find(&.["videoPrimaryInfoRenderer"]?) .as_a.find(&.["videoPrimaryInfoRenderer"]?)
.try &.["videoPrimaryInfoRenderer"] .try &.["videoPrimaryInfoRenderer"]
video_secondary_renderer = primary_results video_secondary_renderer = primary_results
.try &.as_a.find(&.["videoSecondaryInfoRenderer"]?) .as_a.find(&.["videoSecondaryInfoRenderer"]?)
.try &.["videoSecondaryInfoRenderer"] .try &.["videoSecondaryInfoRenderer"]
raise BrokenTubeException.new("videoPrimaryInfoRenderer") if !video_primary_renderer
raise BrokenTubeException.new("videoSecondaryInfoRenderer") if !video_secondary_renderer
# Related videos
LOGGER.debug("extract_video_info: parsing related videos...")
related = [] of JSON::Any
# Parse "compactVideoRenderer" items (under secondary results)
secondary_results.as_a.each do |element|
if item = element["compactVideoRenderer"]?
related_video = parse_related_video(item)
related << JSON::Any.new(related_video) if related_video
end
end
# If nothing was found previously, fall back to end screen renderer
if related.empty?
# Container for "endScreenVideoRenderer" items
player_overlays = player_response.dig?(
"playerOverlays", "playerOverlayRenderer",
"endScreen", "watchNextEndScreenRenderer", "results"
)
player_overlays.try &.as_a.each do |element|
if item = element["endScreenVideoRenderer"]?
related_video = parse_related_video(item)
related << JSON::Any.new(related_video) if related_video
end
end
end
params["relatedVideos"] = JSON::Any.new(related)
# Likes/dislikes # Likes/dislikes
toplevel_buttons = video_primary_renderer toplevel_buttons = video_primary_renderer

View File

@ -321,11 +321,11 @@ we're going to need to do it here in order to allow for translations.
</div> </div>
<div class="pure-u-10-24" style="text-align:right"> <div class="pure-u-10-24" style="text-align:right">
<% if views = rv["short_view_count_text"]?.try &.delete(", views watching") %> <b class="width:100%"><%=
<% if !views.empty? %> views = rv["view_count"]?.try &.to_i?
<b class="width:100%"><%= translate_count(locale, "generic_views_count", views.to_i? || 0) %></b> views ||= rv["view_count_short"]?.try { |x| short_text_to_number(x) }
<% end %> translate_count(locale, "generic_views_count", views || 0, NumberFormatting::Short)
<% end %> %></b>
</div> </div>
</h5> </h5>
</a> </a>

View File

@ -505,7 +505,7 @@ end
# #
# Mostly used to extract out repeated structures to deal with code # Mostly used to extract out repeated structures to deal with code
# repetition. # repetition.
private module HelperExtractors module HelperExtractors
# Retrieves the amount of videos present within the given InnerTube data. # Retrieves the amount of videos present within the given InnerTube data.
# #
# Returns a 0 when it's unable to do so # Returns a 0 when it's unable to do so
@ -519,6 +519,20 @@ private module HelperExtractors
end end
end end
# Retrieves the amount of views/viewers a video has.
# Seems to be used on related videos only
#
# Returns "0" when unable to parse
def self.get_short_view_count(container : JSON::Any) : String
box = container["shortViewCountText"]?
return "0" if !box
# Simpletext: "4M views"
# runs: {"text": "1.1K"},{"text":" watching"}
return box["simpleText"]?.try &.as_s.sub(" views", "") ||
box.dig?("runs", 0, "text").try &.as_s || "0"
end
# Retrieve lowest quality thumbnail from InnerTube data # Retrieve lowest quality thumbnail from InnerTube data
# #
# TODO allow configuration of image quality (-1 is highest) # TODO allow configuration of image quality (-1 is highest)