Improvements to caption extraction

This commit also includes a minor change to some testing parameters.
When running the tests duration for files missing metadata checks were
failing because the return was a second longer than previous. This could
be from an update to mediainfo or ffmpeg that is rounding differently?

Author: masaball

lib/active_encode/base.rb

@@ -4,6 +4,7 @@
 require 'active_encode/errors'
 require 'active_encode/status'
 require 'active_encode/technical_metadata'
+require 'active_encode/subtitle_technical_metadata'
 require 'active_encode/input'
 require 'active_encode/output'
 require 'active_encode/callbacks'

lib/active_encode/engine_adapters/ffmpeg_adapter.rb

@@ -92,14 +92,14 @@ def create(input_url, options = {})
           new_encode.input.duration = fixed_duration(working_path("duration_input_metadata", new_encode.id))
         end
 
-        options[:subtitle_count] = new_encode.input.subtitle_count if new_encode.input.subtitle_count&.positive?
+        subtitle_count = new_encode.input.subtitle_count if new_encode.input.subtitle_count&.positive?
 
         new_encode.state = :running
         new_encode.percent_complete = 1
         new_encode.errors = []
 
         # Run the ffmpeg command and save its pid
-        command = ffmpeg_command(input_url, new_encode.id, options)
+        command = ffmpeg_command(input_url, new_encode.id, options, subtitle_count)
         # Capture the exit status in a file in order to differentiate warning output in stderr between real process failure
         exit_status_file = working_path("exit_status.code", new_encode.id)
         command = "#{command}; echo $? > #{exit_status_file}"
@@ -264,21 +264,32 @@ def build_supplemental_outputs(encode)
           file.id = "#{encode.input.id}-#{File.basename(file_path)}"
           file.created_at = encode.created_at
           file.updated_at = File.mtime file_path
+          # TODO: Add handling for label and language if they are included in stream's metadata
+          # file.label = tech_metadata
+          # file.language = tech_metadata
+
+          # Subtitle metadata has to be pulled from input because subtitle tracks
+          # are not encoded into outputs.
+          metadata_path = working_path("input_metadata", encode.id)
+          subtitle_track = /(?:caption)(\d)(?:\.vtt)/.match(file_path)[1].to_i
+          file.assign_subtitle_tech_metadata(get_subtitle_tech_metadata(metadata_path, subtitle_track))
 
           files << file
         end
 
         files
       end
 
-      def ffmpeg_command(input_url, id, opts)
+      def ffmpeg_command(input_url, id, opts, subtitle_count = nil)
         sanitized_filename = ActiveEncode.sanitize_base input_url
         output_opt = opts[:outputs].collect do |output|
           file_name = "outputs/#{sanitized_filename}-#{output[:label]}.#{output[:extension]}"
           " #{output[:ffmpeg_opt]} #{working_path(file_name, id)}"
         end.join(" ")
 
-        supplemental_file_opt = caption_extraction_options(sanitized_filename, opts[:subtitle_count], id) if opts[:subtitle_count]&.positive?
+        supplemental_file_opt = if opts[:extract_subtitles] && subtitle_count.present?
+                                  caption_extraction_options(sanitized_filename, subtitle_count, id)
+                                end
 
         header_opt = Array(opts[:headers]).map do |k, v|
           "#{k}: #{v}\r\n"
@@ -365,6 +376,20 @@ def get_xpath_text(doc, xpath, cast_method)
         doc.xpath(xpath).first&.text&.send(cast_method)
       end
 
+      def get_subtitle_tech_metadata(file_path, track_number)
+        doc = Nokogiri::XML File.read(file_path)
+        doc.remove_namespaces!
+
+        # XPath starts counting from 1, not 0
+        {
+          format: get_xpath_text(doc, "//track[@type='Text'][#{track_number + 1}]/Format/text()", :to_s),
+          codec: get_xpath_text(doc, "//track[@type='Text'][#{track_number + 1}]/CodecID/text()", :to_s)
+          # duration: ,
+          # language: ,
+          # label:
+        }
+      end
+
       def fixed_duration(path)
         get_tech_metadata(path)[:duration]
       end

lib/active_encode/output.rb

@@ -3,6 +3,7 @@ module ActiveEncode
   class Output
     include Status
     include TechnicalMetadata
+    include SubtitleTechnicalMetadata
 
     attr_accessor :id
     attr_accessor :url

lib/active_encode/subtitle_technical_metadata.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+require 'active_support'
+
+module ActiveEncode
+  module SubtitleTechnicalMetadata
+    extend ActiveSupport::Concern
+
+    included do
+      attr_accessor :language
+      attr_accessor :codec
+      attr_accessor :format
+    end
+
+    def assign_subtitle_tech_metadata(metadata)
+      [:language, :codec, :format].each do |field|
+        send("#{field}=", metadata[field]) if metadata.key?(field)
+      end
+    end
+  end
+end

spec/integration/ffmpeg_adapter_spec.rb

@@ -21,7 +21,7 @@
   let!(:work_dir) { stub_const "ActiveEncode::EngineAdapters::FfmpegAdapter::WORK_DIR", @dir }
   let(:file) { "file://" + Rails.root.join('..', 'spec', 'fixtures', 'fireworks.mp4').to_s }
   let(:created_job) do
-    ActiveEncode::Base.create(file, outputs: [{ label: "low", ffmpeg_opt: "-s 640x480", extension: "mp4" }, { label: "high", ffmpeg_opt: "-s 1280x720", extension: "mp4" }])
+    ActiveEncode::Base.create(file, outputs: [{ label: "low", ffmpeg_opt: "-s 640x480", extension: "mp4" }, { label: "high", ffmpeg_opt: "-s 1280x720", extension: "mp4" }], extract_subtitles: true)
   end
   let(:running_job) do
     allow(Process).to receive(:getpgid).and_return 8888
@@ -141,8 +141,8 @@ def find_encode(id)
       end
 
       it "assigns the correct duration to the encode" do
-        expect(create_without_metadata_job.input.duration).to eq 4_640
-        expect(find_without_metadata_job.input.duration).to eq 4_640
+        expect(create_without_metadata_job.input.duration).to eq 4_641.0
+        expect(find_without_metadata_job.input.duration).to eq 4_641.0
       end
 
       context 'when uri encoded' do
@@ -162,8 +162,8 @@ def find_encode(id)
         end
 
         it "assigns the correct duration to the encode" do
-          expect(create_without_metadata_job.input.duration).to eq 4_640
-          expect(find_without_metadata_job.input.duration).to eq 4_640
+          expect(create_without_metadata_job.input.duration).to eq 4_641.0
+          expect(find_without_metadata_job.input.duration).to eq 4_641.0
         end
       end
     end

spec/integration/pass_through_adapter_spec.rb

@@ -134,8 +134,8 @@ def touch_fixture(id, filename)
       end
 
       it "assigns the correct duration to the encode" do
-        expect(create_without_metadata_job.input.duration).to eq 4_640
-        expect(find_without_metadata_job.input.duration).to eq 4_640
+        expect(create_without_metadata_job.input.duration).to eq 4_641.0
+        expect(find_without_metadata_job.input.duration).to eq 4_641.0
       end
 
       context 'when uri encoded' do
@@ -151,8 +151,8 @@ def touch_fixture(id, filename)
         end
 
         it "assigns the correct duration to the encode" do
-          expect(create_without_metadata_job.input.duration).to eq 4_640
-          expect(find_without_metadata_job.input.duration).to eq 4_640
+          expect(create_without_metadata_job.input.duration).to eq 4_641.0
+          expect(find_without_metadata_job.input.duration).to eq 4_641.0
         end
       end
     end