adds the ability to fetch transcript from youtube

.github/workflows/ci.yml

@@ -1,12 +1,6 @@
 name: linters
 
-on:
-  pull_request:
-    branches:
-      - "*"
-  push:
-    branches:
-      - main
+on: [push]
 
 concurrency: ci-${{ github.ref }}
 
@@ -43,6 +37,7 @@ jobs:
     runs-on: ubuntu-latest
     env:
       RAILS_ENV: test
+      MEILI_MASTER_KEY: masterKey
     steps:
       - uses: actions/checkout@v4
 
@@ -61,6 +56,9 @@ jobs:
       - name: Install dependencies
         run: yarn install --frozen-lockfile
 
+      - name: Meilisearch setup with Docker
+        run: docker run -d -p 7700:7700 getmeili/meilisearch:v1.1 meilisearch --master-key=masterKey --no-analytics
+
       - name: Build assets
         run: bin/vite build --clear --mode=test
 

Gemfile

@@ -112,3 +112,9 @@ gem "view_component", "~> 3.7"
 gem "dry-initializer-rails"
 
 gem "dry-types", "~> 1.7"
+
+gem "google-protobuf", require: false
+
+gem "webvtt-ruby"
+
+gem "active_job-performs", "~> 0.3.1"

Gemfile.lock

@@ -50,6 +50,8 @@ GEM
       erubi (~> 1.11)
       rails-dom-testing (~> 2.2)
       rails-html-sanitizer (~> 1.6)
+    active_job-performs (0.3.1)
+      activejob (>= 6.1)
     activejob (7.1.3.4)
       activesupport (= 7.1.3.4)
       globalid (>= 0.3.6)
@@ -170,6 +172,18 @@ GEM
       raabro (~> 1.4)
     globalid (1.2.1)
       activesupport (>= 6.1)
+    google-protobuf (4.27.1-aarch64-linux)
+      bigdecimal
+      rake (>= 13)
+    google-protobuf (4.27.1-arm64-darwin)
+      bigdecimal
+      rake (>= 13)
+    google-protobuf (4.27.1-x86_64-darwin)
+      bigdecimal
+      rake (>= 13)
+    google-protobuf (4.27.1-x86_64-linux)
+      bigdecimal
+      rake (>= 13)
     groupdate (6.4.0)
       activesupport (>= 6.1)
     hashdiff (1.1.0)
@@ -182,7 +196,7 @@ GEM
       activesupport (>= 3.0)
       nokogiri (>= 1.6)
     io-console (0.7.2)
-    irb (1.13.1)
+    irb (1.14.0)
       rdoc (>= 4.0.0)
       reline (>= 0.4.2)
     jbuilder (2.12.0)
@@ -223,7 +237,7 @@ GEM
       actionpack (>= 6.0.0, < 7.2)
     method_source (1.1.0)
     mini_mime (1.1.5)
-    minitest (5.23.1)
+    minitest (5.24.1)
     msgpack (1.7.2)
     multi_xml (0.7.1)
       bigdecimal (~> 3.1)
@@ -266,7 +280,7 @@ GEM
       nio4r (~> 2.0)
     raabro (1.4.0)
     racc (1.8.0)
-    rack (3.1.3)
+    rack (3.1.6)
     rack-mini-profiler (3.3.1)
       rack (>= 1.2.0)
     rack-proxy (0.7.7)
@@ -427,9 +441,10 @@ GEM
     websocket-driver (0.7.6)
       websocket-extensions (>= 0.1.0)
     websocket-extensions (0.1.5)
+    webvtt-ruby (0.4.2)
     xpath (3.2.0)
       nokogiri (~> 1.8)
-    zeitwerk (2.6.15)
+    zeitwerk (2.6.16)
 
 PLATFORMS
   aarch64-linux
@@ -439,6 +454,7 @@ PLATFORMS
   x86_64-linux
 
 DEPENDENCIES
+  active_job-performs (~> 0.3.1)
   activerecord-enhancedsqlite3-adapter
   ahoy_matey (~> 4.2)
   annotate
@@ -456,6 +472,7 @@ DEPENDENCIES
   dry-types (~> 1.7)
   erb_lint (~> 0.4.0)
   error_highlight (>= 0.4.0)
+  google-protobuf
   groupdate (~> 6.2)
   inline_svg (~> 1.9)
   jbuilder
@@ -484,6 +501,7 @@ DEPENDENCIES
   vite_rails
   web-console
   webmock
+  webvtt-ruby
 
 RUBY VERSION
    ruby 3.3.1p55

app/clients/youtube/transcript.rb

@@ -0,0 +1,74 @@
+require "message_pb"
+
+module Youtube
+  class Transcript
+    attr_reader :response
+
+    def get_vtt(video_id)
+      message = {one: "asr", two: "en"}
+      typedef = MessageType
+      two = get_base64_protobuf(message, typedef)
+
+      message = {one: video_id, two: two}
+      params = get_base64_protobuf(message, typedef)
+
+      url = "https://www.youtube.com/youtubei/v1/get_transcript"
+      headers = {"Content-Type" => "application/json"}
+      body = {
+        context: {
+          client: {
+            clientName: "WEB",
+            clientVersion: "2.20240313"
+          }
+        },
+        params: params
+      }
+
+      @response = HTTParty.post(url, headers: headers, body: body.to_json)
+      convert_to_vtt(JSON.parse(response.body))
+    end
+
+    def self.get_vtt(video_id)
+      new.get_vtt(video_id)
+    end
+
+    private
+
+    def encode_message(message, typedef)
+      encoded_message = typedef.new(message)
+      encoded_message.to_proto
+    end
+
+    def get_base64_protobuf(message, typedef)
+      encoded_data = encode_message(message, typedef)
+      Base64.encode64(encoded_data).delete("\n")
+    end
+
+    def convert_to_vtt(transcript)
+      vtt_content = "WEBVTT\n\n"
+      events = transcript.dig("actions", 0, "updateEngagementPanelAction", "content", "transcriptRenderer", "content", "transcriptSearchPanelRenderer", "body", "transcriptSegmentListRenderer", "initialSegments")
+      if events
+        events.each_with_index do |event, index|
+          segment = event["transcriptSegmentRenderer"]
+          start_time = format_time(segment["startMs"].to_i)
+          end_time = format_time(segment["endMs"].to_i)
+          text = segment.dig("snippet", "runs")&.map { |run| run["text"] }&.join || ""
+          vtt_content += "#{index + 1}\n"
+          vtt_content += "#{start_time} --> #{end_time}\n"
+          vtt_content += "#{text}\n\n"
+        end
+      else
+        vtt_content += "NOTE No transcript data available\n"
+      end
+      vtt_content
+    end
+
+    def format_time(ms)
+      hours = ms / (1000 * 60 * 60)
+      minutes = (ms % (1000 * 60 * 60)) / (1000 * 60)
+      seconds = (ms % (1000 * 60)) / 1000
+      milliseconds = ms % 1000
+      format("%02d:%02d:%02d.%03d", hours, minutes, seconds, milliseconds)
+    end
+  end
+end

app/components/application_component.rb

@@ -5,8 +5,8 @@ class ApplicationComponent < ViewComponent::Base
   attr_accessor :attributes
   option :display, default: proc { true }
 
-  def initialize(*args, **options)
-    super(*args, **options)
+  def initialize(*, **options)
+    super
     defined_option_keys = self.class.dry_initializer.options.map(&:source)
     self.attributes = options.except(*defined_option_keys)
   end

app/models/talk.rb

@@ -27,6 +27,8 @@ class Talk < ApplicationRecord
   include Sluggable
   include Suggestable
   slug_from :title
+
+  # include MeiliSearch
   include MeiliSearch::Rails
   ActiveRecord_Relation.include Pagy::Meilisearch
   extend Pagy::Meilisearch
@@ -36,12 +38,19 @@ class Talk < ApplicationRecord
   has_many :speaker_talks, dependent: :destroy, inverse_of: :talk, foreign_key: :talk_id
   has_many :speakers, through: :speaker_talks
 
+  serialize :transcript, coder: WebVTTSerializer
+
   # validations
   validates :title, presence: true
 
   # delegates
   delegate :name, to: :event, prefix: true, allow_nil: true
 
+  # jobs
+  performs :udpate_transcript!, queue_as: :low do
+    retry_on StandardError, wait: :polynomially_longer
+  end
+
   # search
   meilisearch do
     attribute :title
@@ -118,4 +127,8 @@ def thumbnail_xl
   def related_talks(limit: 6)
     Talk.order("RANDOM()").excluding(self).limit(limit)
   end
+
+  def update_transcript!
+    update!(transcript: Youtube::Transcript.get_vtt(video_id))
+  end
 end

app/serializers/webvtt_serializer.rb

@@ -0,0 +1,32 @@
+require "webvtt"
+
+class WebVTTSerializer
+  def self.dump(transcript)
+    return "" if transcript.blank?
+
+    # If transcript is a raw VTT string, convert it to cues first
+    transcript = self.load(transcript) if transcript.is_a?(String)
+
+    webvtt = "WEBVTT\n\n"
+    transcript.each do |cue|
+      webvtt += "#{cue[:start_time]} --> #{cue[:end_time]}\n#{cue[:text]}\n\n"
+    end
+    webvtt.strip
+  end
+
+  def self.load(transcript)
+    return [] if transcript.blank?
+
+    cues = []
+    # Split transcript by blank lines
+    transcript.split("\n\n").each do |block|
+      lines = block.split("\n")
+      next if lines.size < 2
+
+      timecodes = lines[0].split(" --> ")
+      text = lines[1..].join("\n")
+      cues << {start_time: timecodes[0], end_time: timecodes[1], text: text}
+    end
+    cues
+  end
+end

config/initializers/inflections.rb

@@ -11,6 +11,6 @@
 # end
 
 # These inflection rules are supported but not enabled by default:
-# ActiveSupport::Inflector.inflections(:en) do |inflect|
-#   inflect.acronym "RESTful"
-# end
+ActiveSupport::Inflector.inflections(:en) do |inflect|
+  inflect.acronym "WebVTT"
+end

config/initializers/meilisearch.rb

@@ -1,4 +1,5 @@
 MeiliSearch::Rails.configuration = {
   meilisearch_url: Rails.env.local? ? "http://localhost:7700" : "http://91.107.208.207:7700", # example: http://localhost:7700
-  meilisearch_api_key: ENV["MEILI_MASTER_KEY"]
+  meilisearch_api_key: ENV["MEILI_MASTER_KEY"],
+  per_environment: true
 }

db/migrate/20240611113918_add_transcript_to_talk.rb

@@ -0,0 +1,5 @@
+class AddTranscriptToTalk < ActiveRecord::Migration[7.1]
+  def change
+    add_column :talks, :transcript, :text, default: "", null: false
+  end
+end