Retry failed transcription jobs as en

Stephen Paul Weber created 3 years ago

In case language_id guessed wrong

Change summary

lib/rev_ai.rb | 23 ++++++++++++++++++++---
lib/statsd.rb |  5 +++++
sgx_jmp.rb    |  2 +-
web.rb        |  6 +++---
4 files changed, 29 insertions(+), 7 deletions(-)

Detailed changes

lib/rev_ai.rb

@@ -7,8 +7,9 @@ require "json"
 class RevAi
 	class Failed < StandardError; end
 
-	def initialize(token: CONFIG[:rev_ai_token])
+	def initialize(token: CONFIG[:rev_ai_token], logger: log)
 		@token = token
+		@log = logger
 	end
 
 	def stt(language, media_url, callback_url, **kwargs)
@@ -24,9 +25,9 @@ class RevAi
 		)
 	end
 
-	def stt_result(job)
+	def stt_result(job, url)
 		job = job["job"]
-		return failed(job) if job["status"] == "failed"
+		return maybe_retry_as_en(job, url) if job["status"] == "failed"
 
 		req(
 			:get,
@@ -79,4 +80,20 @@ protected
 	def failed(job)
 		EMPromise.reject(Failed.new("#{job['failure']} #{job['failure_detail']}"))
 	end
+
+	def maybe_retry_as_en(job, url)
+		return failed(job) if job["language"] == "en"
+
+		metadata = JSON.parse(job["metadata"]).transform_keys(&:to_sym)
+		return failed(job) if metadata[:retry]
+
+		retry_as_en(metadata, url)
+	end
+
+	def retry_as_en(metadata, url)
+		@log.info "Retry transcription as EN"
+		stt("en", metadata["media_url"], url, **metadata.merge(retry: true)).then do
+			{}
+		end
+	end
 end

lib/statsd.rb

@@ -30,4 +30,9 @@ Registration::Payment::Mail.statsd_count :write, "registration.payment.mail"
 Registration::Finish.extend StatsD::Instrument
 Registration::Finish.statsd_count :write, "registration.finish"
 
+RevAi.extend StatsD::Instrument
+RevAi.statsd_count :stt, "rev_ai.stt"
+RevAi.statsd_count :language_id, "rev_ai.language_id"
+RevAi.statsd_count :retry_as_en, "rev_ai.retry_as_en"
+
 # rubocop:enable Layout/LineLength

sgx_jmp.rb

@@ -100,8 +100,8 @@ require_relative "lib/registration"
 require_relative "lib/transaction"
 require_relative "lib/tel_selections"
 require_relative "lib/snikket"
-require_relative "lib/statsd"
 require_relative "web"
+require_relative "lib/statsd"
 
 ELECTRUM = Electrum.new(**CONFIG[:electrum])
 EM::Hiredis::Client.load_scripts_from("./redis_lua")

web.rb

@@ -122,7 +122,7 @@ class Web < Roda
 	end
 
 	def rev_ai
-		RevAi.new
+		RevAi.new(logger: log.child(loggable_params))
 	end
 
 	TEL_CANDIDATES = {
@@ -270,8 +270,8 @@ class Web < Roda
 						end
 
 						r.post "transcription" do
-							rev_ai.stt_result(params).then { |result|
-								next "OK" if result["text"].empty?
+							rev_ai.stt_result(params, request.url).then { |result|
+								next "OK" if result["text"].to_s.empty?
 
 								customer_repo.find(
 									result.dig("metadata", "customer_id")