Rewrite extract_item and extract_items functions

This commit completely rewrites the extract_item and extract_items
function. Before this commit these two function were an unreadable
mess. The extract_item function was a lengthy if-elsif chain
while the extract_items function contained an incomprehensible
mess of .try, else and ||.

With this commit both of these functions have been pulled into a
separate file with the internal logic being moved to a few classes.

This significantly reduces the size of these two methods, enhances
readability and makes adding new extraction/parse rules much simpler.

See diff for details.
This commit is contained in:
syeopite 2021-05-04 01:48:51 -07:00
parent 48a518b252
commit a027fbf7af
No known key found for this signature in database
GPG key ID: 6FA616E5A5294A82
5 changed files with 341 additions and 182 deletions

View file

@ -385,7 +385,7 @@ def fetch_channel_featured_channels(ucid, tab_data, params = nil, continuation =
initial_data = request_youtube_api_browse(continuation)
channels_tab_content = initial_data["onResponseReceivedActions"][0]["appendContinuationItemsAction"]["continuationItems"]
return process_featured_channels([channels_tab_content,], nil, title, continuation_items=true)
return process_featured_channels([channels_tab_content], nil, title, continuation_items = true)
else
if params.is_a?(String)
initial_data = request_youtube_api_browse(ucid, params)
@ -964,7 +964,6 @@ def get_about_info(ucid, locale)
tab_names << node["tabRenderer"]["title"].as_s.downcase
tab_data << {i, node["tabRenderer"]["endpoint"]["browseEndpoint"]["params"].as_s}
end
end
tabs = Hash.zip(tab_names, tab_data)
end

View file

@ -94,7 +94,7 @@ def _extract_channel_data(channel)
author_thumbnail: author_thumbnail,
subscriber_count: subscriber_count,
video_count: video_count,
description_html: description_html
description_html: description_html,
})
end
@ -129,7 +129,7 @@ def process_featured_channels(data, submenu_data, title=nil, continuation_items=
title: category_title,
contents: contents,
browse_endpoint_param: browse_endpoint_param,
continuation_token: nil
continuation_token: nil,
})
end
else
@ -162,7 +162,7 @@ def process_featured_channels(data, submenu_data, title=nil, continuation_items=
title: category_title,
contents: contents,
browse_endpoint_param: browse_endpoint_param,
continuation_token: continuation_token
continuation_token: continuation_token,
})
end

View file

@ -0,0 +1,320 @@
# This file contains helper methods to parse the Youtube API json data into
# neat little packages we can use
# Tuple of Parsers/Extractors so we can easily cycle through them.
private ITEM_CONTAINER_EXTRACTOR = {
YoutubeTabsExtractor.new,
SearchResultsExtractor.new,
ContinuationExtractor.new
}
private ITEM_PARSERS = {
VideoParser.new,
ChannelParser.new,
GridPlaylistParser.new,
PlaylistParser.new,
}
private struct AuthorFallback
property name, id
def initialize(@name : String? = nil, @id : String? = nil)
end
end
# The following are the parsers for parsing raw item data into neatly packaged structs.
# They're accessed through the process() method which validates the given data as applicable
# to their specific struct and then use the internal parse() method to assemble the struct
# specific to their category.
private class ItemParser
# Base type for all item parsers.
def process(item : JSON::Any, author_fallback : AuthorFallback)
end
private def parse(item_contents : JSON::Any, author_fallback : AuthorFallback)
end
end
private class VideoParser < ItemParser
def process(item, author_fallback)
if item_contents = (item["videoRenderer"]? || item["gridVideoRenderer"]?)
return self.parse(item_contents, author_fallback)
end
end
private def parse(item_contents, author_fallback)
video_id = item_contents["videoId"].as_s
title = item_contents["title"].try { |t| t["simpleText"]?.try &.as_s || t["runs"]?.try &.as_a.map(&.["text"].as_s).join("") } || ""
author_info = item_contents["ownerText"]?.try &.["runs"]?.try &.as_a?.try &.[0]?
author = author_info.try &.["text"].as_s || author_fallback.name || ""
author_id = author_info.try &.["navigationEndpoint"]?.try &.["browseEndpoint"]["browseId"].as_s || author_fallback.id || ""
published = item_contents["publishedTimeText"]?.try &.["simpleText"]?.try { |t| decode_date(t.as_s) } || Time.local
view_count = item_contents["viewCountText"]?.try &.["simpleText"]?.try &.as_s.gsub(/\D+/, "").to_i64? || 0_i64
description_html = item_contents["descriptionSnippet"]?.try { |t| parse_content(t) } || ""
length_seconds = item_contents["lengthText"]?.try &.["simpleText"]?.try &.as_s.try { |t| decode_length_seconds(t) } ||
item_contents["thumbnailOverlays"]?.try &.as_a.find(&.["thumbnailOverlayTimeStatusRenderer"]?).try &.["thumbnailOverlayTimeStatusRenderer"]?
.try &.["text"]?.try &.["simpleText"]?.try &.as_s.try { |t| decode_length_seconds(t) } || 0
live_now = false
paid = false
premium = false
premiere_timestamp = item_contents["upcomingEventData"]?.try &.["startTime"]?.try { |t| Time.unix(t.as_s.to_i64) }
item_contents["badges"]?.try &.as_a.each do |badge|
b = badge["metadataBadgeRenderer"]
case b["label"].as_s
when "LIVE NOW"
live_now = true
when "New", "4K", "CC"
# TODO
when "Premium"
paid = true
# TODO: Potentially available as item_contents["topStandaloneBadge"]["metadataBadgeRenderer"]
premium = true
else nil # Ignore
end
end
SearchVideo.new({
title: title,
id: video_id,
author: author,
ucid: author_id,
published: published,
views: view_count,
description_html: description_html,
length_seconds: length_seconds,
live_now: live_now,
paid: paid,
premium: premium,
premiere_timestamp: premiere_timestamp,
})
end
end
private class ChannelParser < ItemParser
def process(item, author_fallback)
if item_contents = item["channelRenderer"]?
return self.parse(item_contents, author_fallback)
end
end
private def parse(item_contents, author_fallback)
author = item_contents["title"]["simpleText"]?.try &.as_s || author_fallback.name || ""
author_id = item_contents["channelId"]?.try &.as_s || author_fallback.id || ""
author_thumbnail = item_contents["thumbnail"]["thumbnails"]?.try &.as_a[0]?.try &.["url"]?.try &.as_s || ""
subscriber_count = item_contents["subscriberCountText"]?.try &.["simpleText"]?.try &.as_s.try { |s| short_text_to_number(s.split(" ")[0]) } || 0
auto_generated = false
auto_generated = true if !item_contents["videoCountText"]?
video_count = item_contents["videoCountText"]?.try &.["runs"].as_a[0]?.try &.["text"].as_s.gsub(/\D/, "").to_i || 0
description_html = item_contents["descriptionSnippet"]?.try { |t| parse_content(t) } || ""
SearchChannel.new({
author: author,
ucid: author_id,
author_thumbnail: author_thumbnail,
subscriber_count: subscriber_count,
video_count: video_count,
description_html: description_html,
auto_generated: auto_generated,
})
end
end
private class GridPlaylistParser < ItemParser
def process(item, author_fallback)
if item_contents = item["gridPlaylistRenderer"]?
return self.parse(item_contents, author_fallback)
end
end
private def parse(item_contents, author_fallback)
title = item_contents["title"]["runs"].as_a[0]?.try &.["text"].as_s || ""
plid = item_contents["playlistId"]?.try &.as_s || ""
video_count = item_contents["videoCountText"]["runs"].as_a[0]?.try &.["text"].as_s.gsub(/\D/, "").to_i || 0
playlist_thumbnail = item_contents["thumbnail"]["thumbnails"][0]?.try &.["url"]?.try &.as_s || ""
SearchPlaylist.new({
title: title,
id: plid,
author: author_fallback.name || "",
ucid: author_fallback.id || "",
video_count: video_count,
videos: [] of SearchPlaylistVideo,
thumbnail: playlist_thumbnail,
})
end
end
private class PlaylistParser < ItemParser
def process(item, author_fallback)
if item_contents = item["playlistRenderer"]?
return self.parse(item_contents, author_fallback)
end
end
def parse(item_contents, author_fallback)
title = item_contents["title"]["simpleText"]?.try &.as_s || ""
plid = item_contents["playlistId"]?.try &.as_s || ""
video_count = item_contents["videoCount"]?.try &.as_s.to_i || 0
playlist_thumbnail = item_contents["thumbnails"].as_a[0]?.try &.["thumbnails"]?.try &.as_a[0]?.try &.["url"].as_s || ""
author_info = item_contents["shortBylineText"]?.try &.["runs"]?.try &.as_a?.try &.[0]?
author = author_info.try &.["text"].as_s || author_fallback.name || ""
author_id = author_info.try &.["navigationEndpoint"]?.try &.["browseEndpoint"]["browseId"].as_s || author_fallback.id || ""
videos = item_contents["videos"]?.try &.as_a.map do |v|
v = v["childVideoRenderer"]
v_title = v["title"]["simpleText"]?.try &.as_s || ""
v_id = v["videoId"]?.try &.as_s || ""
v_length_seconds = v["lengthText"]?.try &.["simpleText"]?.try { |t| decode_length_seconds(t.as_s) } || 0
SearchPlaylistVideo.new({
title: v_title,
id: v_id,
length_seconds: v_length_seconds,
})
end || [] of SearchPlaylistVideo
# TODO: item_contents["publishedTimeText"]?
SearchPlaylist.new({
title: title,
id: plid,
author: author,
ucid: author_id,
video_count: video_count,
videos: videos,
thumbnail: playlist_thumbnail,
})
end
end
# The following are the extractors for extracting an array of items from
# the internal Youtube API's JSON response. The result is then packaged into
# a structure we can more easily use via the parsers above. Their internals are
# identical to the item parsers.
private class ItemsContainerExtractor
def process(item : Hash(String, JSON::Any))
end
private def extract(target : JSON::Any)
end
end
private class YoutubeTabsExtractor < ItemsContainerExtractor
def process(initial_data)
if target = initial_data["twoColumnBrowseResultsRenderer"]?
self.extract(target)
end
end
private def extract(target)
raw_items = [] of JSON::Any
selected_tab = extract_selected_tab(target["tabs"])
content = selected_tab["tabRenderer"]["content"]
content["sectionListRenderer"]["contents"].as_a.each do | renderer_container |
renderer_container = renderer_container["itemSectionRenderer"]
renderer_container_contents = renderer_container["contents"].as_a[0]
# Shelf renderer usually refer to a category and would need special handling once
# An extractor for categories are added. But for now it is just used to
# extract items for the trending page
if items_container = renderer_container_contents["shelfRenderer"]?
if items_container["content"]["expandedShelfContentsRenderer"]?
items_container = items_container["content"]["expandedShelfContentsRenderer"]
end
elsif items_container = renderer_container_contents["gridRenderer"]?
else
items_container = renderer_container_contents
end
items_container["items"].as_a.each do | item |
raw_items << item
end
end
return raw_items
end
end
private class SearchResultsExtractor < ItemsContainerExtractor
def process(initial_data)
if target = initial_data["twoColumnSearchResultsRenderer"]?
self.extract(target)
end
end
private def extract(target)
raw_items = [] of JSON::Any
content = target["primaryContents"]
renderer = content["sectionListRenderer"]["contents"].as_a[0]["itemSectionRenderer"]
raw_items = renderer["contents"].as_a
return raw_items
end
end
private class ContinuationExtractor < ItemsContainerExtractor
def process(initial_data)
if target = initial_data["continuationContents"]?
self.extract(target)
end
end
private def extract(target)
raw_items = [] of JSON::Any
if content = target["gridContinuation"]?
raw_items = content["items"].as_a
end
return raw_items
end
end
def extract_item(item : JSON::Any, author_fallback : String? = nil, author_id_fallback : String? = nil)
# Parses an item from Youtube's JSON response into a more usable structure.
# The end result can either be a SearchVideo, SearchPlaylist or SearchChannel.
author_fallback = AuthorFallback.new(author_fallback, author_id_fallback)
# Cycles through all of the item parsers and attempt to parse the raw YT JSON data.
# Each parser automatically validates the data given to see if the data is
# applicable to itself. If not nil is returned and the next parser is attemped.
ITEM_PARSERS.each do |parser|
result = parser.process(item, author_fallback)
if !result.nil?
return result
end
end
# TODO radioRenderer, showRenderer, shelfRenderer, horizontalCardListRenderer, searchPyvRenderer
end
def extract_items(initial_data : Hash(String, JSON::Any), author_fallback : String? = nil, author_id_fallback : String? = nil)
items = [] of SearchItem
initial_data = initial_data["contents"]?.try &.as_h || initial_data["response"]?.try &.as_h || initial_data
# This is identicial to the parser cyling of extract_item().
ITEM_CONTAINER_EXTRACTOR.each do | extractor |
results = extractor.process(initial_data)
if !results.nil?
results.each do | item |
parsed_result = extract_item(item, author_fallback, author_id_fallback)
if !parsed_result.nil?
items << parsed_result
end
end
end
end
return items
end

View file

@ -251,169 +251,9 @@ def extract_videos(initial_data : Hash(String, JSON::Any), author_fallback : Str
extract_items(initial_data, author_fallback, author_id_fallback).select(&.is_a?(SearchVideo)).map(&.as(SearchVideo))
end
def extract_item(item : JSON::Any, author_fallback : String? = nil, author_id_fallback : String? = nil)
if i = (item["videoRenderer"]? || item["gridVideoRenderer"]?)
video_id = i["videoId"].as_s
title = i["title"].try { |t| t["simpleText"]?.try &.as_s || t["runs"]?.try &.as_a.map(&.["text"].as_s).join("") } || ""
author_info = i["ownerText"]?.try &.["runs"]?.try &.as_a?.try &.[0]?
author = author_info.try &.["text"].as_s || author_fallback || ""
author_id = author_info.try &.["navigationEndpoint"]?.try &.["browseEndpoint"]["browseId"].as_s || author_id_fallback || ""
published = i["publishedTimeText"]?.try &.["simpleText"]?.try { |t| decode_date(t.as_s) } || Time.local
view_count = i["viewCountText"]?.try &.["simpleText"]?.try &.as_s.gsub(/\D+/, "").to_i64? || 0_i64
description_html = i["descriptionSnippet"]?.try { |t| parse_content(t) } || ""
length_seconds = i["lengthText"]?.try &.["simpleText"]?.try &.as_s.try { |t| decode_length_seconds(t) } ||
i["thumbnailOverlays"]?.try &.as_a.find(&.["thumbnailOverlayTimeStatusRenderer"]?).try &.["thumbnailOverlayTimeStatusRenderer"]?
.try &.["text"]?.try &.["simpleText"]?.try &.as_s.try { |t| decode_length_seconds(t) } || 0
live_now = false
paid = false
premium = false
premiere_timestamp = i["upcomingEventData"]?.try &.["startTime"]?.try { |t| Time.unix(t.as_s.to_i64) }
i["badges"]?.try &.as_a.each do |badge|
b = badge["metadataBadgeRenderer"]
case b["label"].as_s
when "LIVE NOW"
live_now = true
when "New", "4K", "CC"
# TODO
when "Premium"
paid = true
# TODO: Potentially available as i["topStandaloneBadge"]["metadataBadgeRenderer"]
premium = true
else nil # Ignore
end
end
SearchVideo.new({
title: title,
id: video_id,
author: author,
ucid: author_id,
published: published,
views: view_count,
description_html: description_html,
length_seconds: length_seconds,
live_now: live_now,
paid: paid,
premium: premium,
premiere_timestamp: premiere_timestamp,
})
elsif i = item["channelRenderer"]?
author = i["title"]["simpleText"]?.try &.as_s || author_fallback || ""
author_id = i["channelId"]?.try &.as_s || author_id_fallback || ""
author_thumbnail = i["thumbnail"]["thumbnails"]?.try &.as_a[0]?.try &.["url"]?.try &.as_s || ""
subscriber_count = i["subscriberCountText"]?.try &.["simpleText"]?.try &.as_s.try { |s| short_text_to_number(s.split(" ")[0]) } || 0
auto_generated = false
auto_generated = true if !i["videoCountText"]?
video_count = i["videoCountText"]?.try &.["runs"].as_a[0]?.try &.["text"].as_s.gsub(/\D/, "").to_i || 0
description_html = i["descriptionSnippet"]?.try { |t| parse_content(t) } || ""
SearchChannel.new({
author: author,
ucid: author_id,
author_thumbnail: author_thumbnail,
subscriber_count: subscriber_count,
video_count: video_count,
description_html: description_html,
auto_generated: auto_generated,
})
elsif i = item["gridPlaylistRenderer"]?
title = i["title"]["runs"].as_a[0]?.try &.["text"].as_s || ""
plid = i["playlistId"]?.try &.as_s || ""
video_count = i["videoCountText"]["runs"].as_a[0]?.try &.["text"].as_s.gsub(/\D/, "").to_i || 0
playlist_thumbnail = i["thumbnail"]["thumbnails"][0]?.try &.["url"]?.try &.as_s || ""
SearchPlaylist.new({
title: title,
id: plid,
author: author_fallback || "",
ucid: author_id_fallback || "",
video_count: video_count,
videos: [] of SearchPlaylistVideo,
thumbnail: playlist_thumbnail,
})
elsif i = item["playlistRenderer"]?
title = i["title"]["simpleText"]?.try &.as_s || ""
plid = i["playlistId"]?.try &.as_s || ""
video_count = i["videoCount"]?.try &.as_s.to_i || 0
playlist_thumbnail = i["thumbnails"].as_a[0]?.try &.["thumbnails"]?.try &.as_a[0]?.try &.["url"].as_s || ""
author_info = i["shortBylineText"]?.try &.["runs"]?.try &.as_a?.try &.[0]?
author = author_info.try &.["text"].as_s || author_fallback || ""
author_id = author_info.try &.["navigationEndpoint"]?.try &.["browseEndpoint"]["browseId"].as_s || author_id_fallback || ""
videos = i["videos"]?.try &.as_a.map do |v|
v = v["childVideoRenderer"]
v_title = v["title"]["simpleText"]?.try &.as_s || ""
v_id = v["videoId"]?.try &.as_s || ""
v_length_seconds = v["lengthText"]?.try &.["simpleText"]?.try { |t| decode_length_seconds(t.as_s) } || 0
SearchPlaylistVideo.new({
title: v_title,
id: v_id,
length_seconds: v_length_seconds,
})
end || [] of SearchPlaylistVideo
# TODO: i["publishedTimeText"]?
SearchPlaylist.new({
title: title,
id: plid,
author: author,
ucid: author_id,
video_count: video_count,
videos: videos,
thumbnail: playlist_thumbnail,
})
elsif i = item["radioRenderer"]? # Mix
# TODO
elsif i = item["showRenderer"]? # Show
# TODO
elsif i = item["shelfRenderer"]?
elsif i = item["horizontalCardListRenderer"]?
elsif i = item["searchPyvRenderer"]? # Ad
end
end
def extract_items(initial_data : Hash(String, JSON::Any), author_fallback : String? = nil, author_id_fallback : String? = nil)
items = [] of SearchItem
channel_v2_response = initial_data
.try &.["continuationContents"]?
.try &.["gridContinuation"]?
.try &.["items"]?
if channel_v2_response
channel_v2_response.try &.as_a.each { |item|
extract_item(item, author_fallback, author_id_fallback)
.try { |t| items << t }
}
else
initial_data.try { |t| t["contents"]? || t["response"]? }
.try { |t| t["twoColumnBrowseResultsRenderer"]?.try &.["tabs"].as_a.select(&.["tabRenderer"]?.try &.["selected"].as_bool)[0]?.try &.["tabRenderer"]["content"] ||
t["twoColumnSearchResultsRenderer"]?.try &.["primaryContents"] ||
t["continuationContents"]? }
.try { |t| t["sectionListRenderer"]? || t["sectionListContinuation"]? }
.try &.["contents"].as_a
.each { |c| c.try &.["itemSectionRenderer"]?.try &.["contents"].as_a
.try { |t| t[0]?.try &.["shelfRenderer"]?.try &.["content"]["expandedShelfContentsRenderer"]?.try &.["items"].as_a ||
t[0]?.try &.["gridRenderer"]?.try &.["items"].as_a || t }
.each { |item|
extract_item(item, author_fallback, author_id_fallback)
.try { |t| items << t }
} }
end
items
def extract_selected_tab(tabs)
# Extract the selected tab from the array of tabs Youtube returns
return selected_target = tabs.as_a.select(&.["tabRenderer"]?.try &.["selected"].as_bool)[0]
end
def check_enum(db, enum_name, struct_type = nil)