Add API endpoint for fetching transcripts from YouTube (#4788)

This commit is contained in:
syeopite
2025-02-26 13:56:39 -08:00
3 changed files with 122 additions and 0 deletions

View File

@@ -429,4 +429,90 @@ 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"]?
label = env.params.query["label"]?
auto_generated = env.params.query["autogen"]? ? true : false
# Return all available transcript options when none is given
if !label && !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
# If lang is not given then we attempt to fetch
# the transcript through the given label
if lang.nil?
begin
video = get_video(id)
rescue ex : NotFoundException
return error_json(404, ex)
rescue ex
return error_json(500, ex)
end
target_transcript = video.captions.select(&.name.== label)
if target_transcript.empty?
return error_json(404, NotFoundException.new("Requested transcript does not exist"))
else
target_transcript = target_transcript[0]
lang, auto_generated = target_transcript.language_code, target_transcript.auto_generated
end
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