From 839e90aeff93a18d59cb4fc53eb25cc5c152b44a Mon Sep 17 00:00:00 2001
From: Samantaz Fox <coding@samantaz.fr>
Date: Sat, 15 Jul 2023 15:41:04 +0200
Subject: [PATCH] Extractors: Add module for 'hashtagTileRenderer'

---
 src/invidious/helpers/serialized_yt_data.cr | 21 +++++++-
 src/invidious/yt_backend/extractors.cr      | 53 ++++++++++++++++++++-
 2 files changed, 72 insertions(+), 2 deletions(-)

diff --git a/src/invidious/helpers/serialized_yt_data.cr b/src/invidious/helpers/serialized_yt_data.cr
index 7c12ad0e..e0bd7279 100644
--- a/src/invidious/helpers/serialized_yt_data.cr
+++ b/src/invidious/helpers/serialized_yt_data.cr
@@ -232,6 +232,25 @@ struct SearchChannel
   end
 end
 
+struct SearchHashtag
+  include DB::Serializable
+
+  property title : String
+  property url : String
+  property video_count : Int64
+  property channel_count : Int64
+
+  def to_json(locale : String?, json : JSON::Builder)
+    json.object do
+      json.field "type", "hashtag"
+      json.field "title", self.title
+      json.field "url", self.url
+      json.field "videoCount", self.video_count
+      json.field "channelCount", self.channel_count
+    end
+  end
+end
+
 class Category
   include DB::Serializable
 
@@ -274,4 +293,4 @@ struct Continuation
   end
 end
 
-alias SearchItem = SearchVideo | SearchChannel | SearchPlaylist | Category
+alias SearchItem = SearchVideo | SearchChannel | SearchPlaylist | SearchHashtag | Category
diff --git a/src/invidious/yt_backend/extractors.cr b/src/invidious/yt_backend/extractors.cr
index e5029dc5..8456313b 100644
--- a/src/invidious/yt_backend/extractors.cr
+++ b/src/invidious/yt_backend/extractors.cr
@@ -11,15 +11,16 @@ private ITEM_CONTAINER_EXTRACTOR = {
 }
 
 private ITEM_PARSERS = {
+  Parsers::RichItemRendererParser,
   Parsers::VideoRendererParser,
   Parsers::ChannelRendererParser,
   Parsers::GridPlaylistRendererParser,
   Parsers::PlaylistRendererParser,
   Parsers::CategoryRendererParser,
-  Parsers::RichItemRendererParser,
   Parsers::ReelItemRendererParser,
   Parsers::ItemSectionRendererParser,
   Parsers::ContinuationItemRendererParser,
+  Parsers::HashtagRendererParser,
 }
 
 private alias InitialData = Hash(String, JSON::Any)
@@ -210,6 +211,56 @@ private module Parsers
     end
   end
 
+  # Parses an Innertube `hashtagTileRenderer` into a `SearchHashtag`.
+  # Returns `nil` when the given object is not a `hashtagTileRenderer`.
+  #
+  # A `hashtagTileRenderer` is a kind of search result.
+  # It can be found when searching for any hashtag (e.g "#hi" or "#shorts")
+  module HashtagRendererParser
+    def self.process(item : JSON::Any, author_fallback : AuthorFallback)
+      if item_contents = item["hashtagTileRenderer"]?
+        return self.parse(item_contents)
+      end
+    end
+
+    private def self.parse(item_contents)
+      title = extract_text(item_contents["hashtag"]).not_nil! # E.g "#hi"
+
+      # E.g "/hashtag/hi"
+      url = item_contents.dig?("onTapCommand", "commandMetadata", "webCommandMetadata", "url").try &.as_s
+      url ||= URI.encode_path("/hashtag/#{title.lchop('#')}")
+
+      video_count_txt = extract_text(item_contents["hashtagVideoCount"]?)     # E.g "203K videos"
+      channel_count_txt = extract_text(item_contents["hashtagChannelCount"]?) # E.g "81K channels"
+
+      # Fallback for video/channel counts
+      if channel_count_txt.nil? || video_count_txt.nil?
+        # E.g: "203K videos • 81K channels"
+        info_text = extract_text(item_contents["hashtagInfoText"]?).try &.split(" • ")
+
+        if info_text && info_text.size == 2
+          video_count_txt ||= info_text[0]
+          channel_count_txt ||= info_text[1]
+        end
+      end
+
+      return SearchHashtag.new({
+        title:         title,
+        url:           url,
+        video_count:   short_text_to_number(video_count_txt || ""),
+        channel_count: short_text_to_number(channel_count_txt || ""),
+      })
+    rescue ex
+      LOGGER.debug("HashtagRendererParser: Failed to extract renderer.")
+      LOGGER.debug("HashtagRendererParser: Got exception: #{ex.message}")
+      return nil
+    end
+
+    def self.parser_name
+      return {{@type.name}}
+    end
+  end
+
   # Parses a InnerTube gridPlaylistRenderer into a SearchPlaylist. Returns nil when the given object isn't a gridPlaylistRenderer
   #
   # A gridPlaylistRenderer renders a playlist, that is located in a grid, to click on within the YouTube and Invidious UI.