Add API endpoint to fetch YouTube transcripts

3 files changed, 101 insertions, 0 deletions

diff --git a/src/invidious/routes/api/v1/videos.cr b/src/invidious/routes/api/v1/videos.cr
index faff2f59..03fdc49b 100644
--- a/src/invidious/routes/api/v1/videos.cr
+++ b/src/invidious/routes/api/v1/videos.cr
@@ -411,4 +411,69 @@ module Invidious::Routes::API::V1::Videos
       end
     end
   end
+
+  # Fetches transcripts from YouTube
+  #
+  # Use the `lang` and `autogen` query parameter to select which transcript to fetch
+  # Request without any URL parameters to see all the available transcripts.
+  def self.transcripts(env)
+    env.response.content_type = "application/json"
+
+    id = env.params.url["id"]
+    lang = env.params.query["lang"]?
+    auto_generated = env.params.query["autogen"]? ? true : false
+
+    # Return all available transcript options when none is given
+    if !lang
+      begin
+        video = get_video(id)
+      rescue ex : NotFoundException
+        return error_json(404, ex)
+      rescue ex
+        return error_json(500, ex)
+      end
+
+      response = JSON.build do |json|
+        # The amount of transcripts available to fetch is the
+        # same as the amount of captions available.
+        available_transcripts = video.captions
+
+        json.object do
+          json.field "transcripts" do
+            json.array do
+              available_transcripts.each do |transcript|
+                json.object do
+                  json.field "label", transcript.name
+                  json.field "languageCode", transcript.language_code
+                  json.field "autoGenerated", transcript.auto_generated
+
+                  if transcript.auto_generated
+                    json.field "url", "/api/v1/transcripts/#{id}?lang=#{URI.encode_www_form(transcript.language_code)}&autogen"
+                  else
+                    json.field "url", "/api/v1/transcripts/#{id}?lang=#{URI.encode_www_form(transcript.language_code)}"
+                  end
+                end
+              end
+            end
+          end
+        end
+      end
+
+      return response
+    end
+
+    params = Invidious::Videos::Transcript.generate_param(id, lang, auto_generated)
+
+    begin
+      transcript = Invidious::Videos::Transcript.from_raw(
+        YoutubeAPI.get_transcript(params), lang, auto_generated
+      )
+    rescue ex : NotFoundException
+      return error_json(404, ex)
+    rescue ex
+      return error_json(500, ex)
+    end
+
+    return transcript.to_json
+  end
 end
diff --git a/src/invidious/routing.cr b/src/invidious/routing.cr
index ba05da19..125bfefc 100644
--- a/src/invidious/routing.cr
+++ b/src/invidious/routing.cr
@@ -236,6 +236,7 @@ module Invidious::Routing
       get "/api/v1/annotations/:id", {{namespace}}::Videos, :annotations
       get "/api/v1/comments/:id", {{namespace}}::Videos, :comments
       get "/api/v1/clips/:id", {{namespace}}::Videos, :clips
+      get "/api/v1/transcripts/:id", {{namespace}}::Videos, :transcripts
 
       # Feeds
       get "/api/v1/trending", {{namespace}}::Feeds, :trending
diff --git a/src/invidious/videos/transcript.cr b/src/invidious/videos/transcript.cr
index 9cd064c5..95965446 100644
--- a/src/invidious/videos/transcript.cr
+++ b/src/invidious/videos/transcript.cr
@@ -122,5 +122,40 @@ module Invidious::Videos
 
       return vtt
     end
+
+    def to_json(json : JSON::Builder)
+      json.field "languageCode", @language_code
+      json.field "autoGenerated", @auto_generated
+      json.field "label", @label
+      json.field "body" do
+        json.array do
+          @lines.each do |line|
+            json.object do
+              if line.is_a? HeadingLine
+                json.field "type", "heading"
+              else
+                json.field "type", "regular"
+              end
+
+              json.field "startMs", line.start_ms.total_milliseconds
+              json.field "endMs", line.end_ms.total_milliseconds
+              json.field "line", line.line
+            end
+          end
+        end
+      end
+    end
+
+    def to_json
+      JSON.build do |json|
+        json.object do
+          json.field "transcript" do
+            json.object do
+              to_json(json)
+            end
+          end
+        end
+      end
+    end
   end
 end