From cd1abb43901e12d0184dd45d66d9c0b51d204ffe Mon Sep 17 00:00:00 2001 From: Chunky programmer <78101139+ChunkyProgrammer@users.noreply.github.com> Date: Tue, 6 Jun 2023 19:31:55 -0400 Subject: [PATCH] Parse hashtag header when getting the first hashtag page --- src/invidious/hashtag.cr | 142 +++++++++++++++++++++----- src/invidious/routes/api/v1/search.cr | 12 +-- src/invidious/routes/search.cr | 3 +- 3 files changed, 123 insertions(+), 34 deletions(-) diff --git a/src/invidious/hashtag.cr b/src/invidious/hashtag.cr index d9d584c9..15faba7c 100644 --- a/src/invidious/hashtag.cr +++ b/src/invidious/hashtag.cr @@ -1,42 +1,138 @@ module Invidious::Hashtag extend self - def fetch(hashtag : String, page : Int, region : String? = nil) : Array(SearchItem) + struct HashtagPage + include DB::Serializable + + property videos : Array(SearchItem) | Array(Video) + property header : HashtagHeader? + property has_next_continuation : Bool + + 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 + json.field "hasNextPage", self.has_next_continuation + 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 "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 + + def fetch(hashtag : String, page : Int, region : String? = nil) : HashtagPage cursor = (page - 1) * 60 - ctoken = generate_continuation(hashtag, cursor) - + header = nil client_config = YoutubeAPI::ClientConfig.new(region: region) - response = YoutubeAPI.browse(continuation: ctoken, client_config: client_config) + item = generate_continuation(hashtag, cursor) + # item is a ctoken + if cursor > 0 + response = YoutubeAPI.browse(continuation: item, client_config: client_config) + else + # item browses the first page (including metadata) + response = YoutubeAPI.browse("FEhashtag", params: item, client_config: client_config) + if item_contents = response.dig?("header", "hashtagHeaderRenderer") + header = parse_hashtag_renderer(item_contents) + end + end - items, _ = extract_items(response) - return items + items, next_continuation = extract_items(response) + return HashtagPage.new({ + videos: items, + header: header, + has_next_continuation: next_continuation != nil, + }) end def generate_continuation(hashtag : String, cursor : Int) object = { - "80226972:embedded" => { - "2:string" => "FEhashtag", - "3:base64" => { - "1:varint" => 60_i64, # result count - "15:base64" => { - "1:varint" => cursor.to_i64, - "2:varint" => 0_i64, - }, - "93:2:embedded" => { - "1:string" => hashtag, - "2:varint" => 0_i64, - "3:varint" => 1_i64, - }, - }, - "35:string" => "browse-feedFEhashtag", + "93:2:embedded" => { + "1:string" => hashtag, + "2:varint" => 0_i64, + "3:varint" => 1_i64, }, } + if cursor > 0 + object = { + "80226972:embedded" => { + "2:string" => "FEhashtag", + "3:base64" => { + "1:varint" => 60_i64, # result count + "15:base64" => { + "1:varint" => cursor.to_i64, + "2:varint" => 0_i64, + }, + "93:2:embedded" => { + "1:string" => hashtag, + "2:varint" => 0_i64, + "3:varint" => 1_i64, + }, + }, + "35:string" => "browse-feedFEhashtag", + }, + } + end - continuation = object.try { |i| Protodec::Any.cast_json(i) } + 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 - return continuation + def parse_hashtag_renderer(item_contents) + info = extract_text(item_contents.dig?("hashtagInfoText")) || "" + + regex_match = /(?\d+\S)\D+(?\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 end diff --git a/src/invidious/routes/api/v1/search.cr b/src/invidious/routes/api/v1/search.cr index 2922b060..1777060a 100644 --- a/src/invidious/routes/api/v1/search.cr +++ b/src/invidious/routes/api/v1/search.cr @@ -69,21 +69,13 @@ module Invidious::Routes::API::V1::Search env.response.content_type = "application/json" begin - results = Invidious::Hashtag.fetch(hashtag, page, region) + hashtagPage = Invidious::Hashtag.fetch(hashtag, page, region) rescue ex return error_json(400, ex) end JSON.build do |json| - json.object do - json.field "results" do - json.array do - results.each do |item| - item.to_json(locale, json) - end - end - end - end + hashtagPage.to_json(locale, json) end end end diff --git a/src/invidious/routes/search.cr b/src/invidious/routes/search.cr index 44970922..0a6665b4 100644 --- a/src/invidious/routes/search.cr +++ b/src/invidious/routes/search.cr @@ -101,7 +101,8 @@ module Invidious::Routes::Search end begin - items = Invidious::Hashtag.fetch(hashtag, page) + hashtagPage = Invidious::Hashtag.fetch(hashtag, page) + items = hashtagPage.videos rescue ex return error_template(500, ex) end