Parse hashtag header when getting the first hashtag page

This commit is contained in:
Chunky programmer 2023-06-06 19:31:55 -04:00
parent 545a5937d8
commit 6c8f9a6991
5 changed files with 129 additions and 16 deletions

View file

@ -1,15 +1,42 @@
module Invidious::Hashtag module Invidious::Hashtag
extend self extend self
def fetch(hashtag : String, page : Int, region : String? = nil) : Array(SearchItem) def fetch(hashtag : String, page : Int, region : String? = nil) : HashtagPage
cursor = (page - 1) * 60 cursor = (page - 1) * 60
ctoken = generate_continuation(hashtag, cursor) header = nil
client_config = YoutubeAPI::ClientConfig.new(region: region) client_config = YoutubeAPI::ClientConfig.new(region: region)
# for any page besides the first page, get the list of videos
if cursor > 0
ctoken = generate_continuation(hashtag, cursor)
response = YoutubeAPI.browse(continuation: ctoken, client_config: client_config) response = YoutubeAPI.browse(continuation: ctoken, client_config: client_config)
else
# get first page + header info
response = YoutubeAPI.browse("FEhashtag", params: get_hashtag_first_page(hashtag), client_config: client_config)
if header = response.dig?("header")
header = parse_item(header).as(HashtagHeader)
end
end
items, _ = extract_items(response) items, _ = extract_items(response)
return items return HashtagPage.new({
videos: items,
header: header,
})
end
def get_hashtag_first_page(hashtag : String)
object = {
"93:embedded" => {
"1:string" => hashtag,
"2:varint" => 0_i64,
"3:varint" => 1_i64,
},
}
return object.try { |i| Protodec::Any.cast_json(i) }
.try { |i| Protodec::Any.from_json(i) }
.try { |i| Base64.urlsafe_encode(i) }
.try { |i| URI.encode_www_form(i) }
end end
def generate_continuation(hashtag : String, cursor : Int) def generate_continuation(hashtag : String, cursor : Int)

View file

@ -274,4 +274,63 @@ struct Continuation
end end
end end
struct HashtagPage
include DB::Serializable
property videos : Array(SearchItem) | Array(Video)
property header : HashtagHeader?
def to_json(locale : String?, json : JSON::Builder)
json.object do
json.field "type", "hashtag"
if self.header != nil
json.field "header" do
self.header.to_json(json)
end
end
json.field "results" do
json.array do
self.videos.each do |item|
item.to_json(locale, json)
end
end
end
end
end
# TODO: remove the locale and follow the crystal convention
def to_json(locale : String?, _json : Nil)
JSON.build do |json|
to_json(locale, json)
end
end
def to_json(json : JSON::Builder)
to_json(nil, json)
end
end
struct HashtagHeader
include DB::Serializable
property tag : String
property channel_count : Int64
property video_count : Int64
def to_json(json : JSON::Builder)
json.object do
json.field "type", "hashtagHeader"
json.field "hashtag", self.tag
json.field "channelCount", self.channel_count
json.field "videoCount", self.video_count
end
end
def to_json(_json : Nil)
JSON.build do |json|
to_json(json)
end
end
end
alias SearchItem = SearchVideo | SearchChannel | SearchPlaylist | Category alias SearchItem = SearchVideo | SearchChannel | SearchPlaylist | Category

View file

@ -66,21 +66,13 @@ module Invidious::Routes::API::V1::Search
env.response.content_type = "application/json" env.response.content_type = "application/json"
begin begin
results = Invidious::Hashtag.fetch(hashtag, page, region) hashtagPage = Invidious::Hashtag.fetch(hashtag, page, region)
rescue ex rescue ex
return error_json(400, ex) return error_json(400, ex)
end end
JSON.build do |json| JSON.build do |json|
json.object do hashtagPage.to_json(locale, json)
json.field "results" do
json.array do
results.each do |item|
item.to_json(locale, json)
end
end
end
end
end end
end end
end end

View file

@ -91,7 +91,8 @@ module Invidious::Routes::Search
end end
begin begin
videos = Invidious::Hashtag.fetch(hashtag, page) hashtagPage = Invidious::Hashtag.fetch(hashtag, page)
videos = hashtagPage.videos
rescue ex rescue ex
return error_template(500, ex) return error_template(500, ex)
end end

View file

@ -20,6 +20,7 @@ private ITEM_PARSERS = {
Parsers::ReelItemRendererParser, Parsers::ReelItemRendererParser,
Parsers::ItemSectionRendererParser, Parsers::ItemSectionRendererParser,
Parsers::ContinuationItemRendererParser, Parsers::ContinuationItemRendererParser,
Parsers::HashtagHeaderRenderer,
} }
private alias InitialData = Hash(String, JSON::Any) private alias InitialData = Hash(String, JSON::Any)
@ -550,6 +551,39 @@ private module Parsers
return {{@type.name}} return {{@type.name}}
end end
end end
# Parses an InnerTube hashtagHeaderRender into a HashtagHeaderRender.
# Returns nil when the given object isn't a hashtagHeaderRender.
#
# hashtagHeaderRender contains metadate of the hashtag page such as video count and channel count
#
module HashtagHeaderRenderer
def self.process(item : JSON::Any, author_fallback : AuthorFallback)
if item_contents = item["hashtagHeaderRenderer"]?
return self.parse(item_contents)
end
end
private def self.parse(item_contents)
info = extract_text(item_contents.dig?("hashtagInfoText")) || ""
regex_match = /(?<videos>\d+\S)\D+(?<channels>\d+\S)/.match(info)
hashtag = extract_text(item_contents.dig?("hashtag")) || ""
videos = short_text_to_number(regex_match.try &.["videos"]?.try &.to_s || "0")
channels = short_text_to_number(regex_match.try &.["channels"]?.try &.to_s || "0")
return HashtagHeader.new({
tag: hashtag,
channel_count: channels,
video_count: videos,
})
end
def self.parser_name
return {{@type.name}}
end
end
end end
# The following are the extractors for extracting an array of items from # The following are the extractors for extracting an array of items from