Refactor ocr jobs

Gemfile

@@ -33,7 +33,6 @@ gem "ezid-client", "1.9.4" # v1.9.0 introduces response errors in our tests/stub
 gem "faker"
 gem "ffi", "~> 1.16.0"
 gem "filewatcher", "~> 1.0"
-gem "flipper"
 gem "flutie"
 gem "font-awesome-rails"
 gem "google-cloud-pubsub"
@@ -68,8 +67,8 @@ gem "mime-types"
 gem "mini_magick"
 gem "modernizr-rails"
 # Pin because capistrano raises an error at >= 7.2
-gem "net-ssh", "~> 7.1.0"
 gem "net-sftp"
+gem "net-ssh", "~> 7.1.0"
 gem "normalize-rails"
 gem "oai"
 gem "omniauth", "1.9.2"

Gemfile.lock

@@ -439,8 +439,6 @@ GEM
     ffi (1.16.3)
     filewatcher (1.1.1)
       optimist (~> 3.0)
-    flipper (1.3.2)
-      concurrent-ruby (< 2)
     flutie (2.2.0)
     font-awesome-rails (4.7.0.8)
       railties (>= 3.2, < 8.0)
@@ -1291,7 +1289,6 @@ DEPENDENCIES
   faker
   ffi (~> 1.16.0)
   filewatcher (~> 1.0)
-  flipper
   flutie
   font-awesome-rails
   foreman

app/jobs/create_ocr_request_job.rb

@@ -10,15 +10,7 @@ def perform(file_path:)
     ocr_request = OcrRequest.new(filename: filename, state: "Enqueued")
     ocr_request.save!
     ocr_request.pdf.attach(io: File.open(file_path), filename: filename, content_type: "application/pdf")
-    out_path = File.join(ocr_out_dir, filename)
-    PdfOcrJob.perform_later(resource: ocr_request, out_path: out_path)
+    PdfOcrJob.perform_later(resource: ocr_request)
     File.delete(file_path)
   end
-
-  def ocr_out_dir
-    out_dir = Figgy.config["ocr_out_path"]
-    FileUtils.mkdir_p(out_dir) unless File.directory?(out_dir)
-
-    out_dir
-  end
 end

app/jobs/pdf_ocr_job.rb

@@ -2,12 +2,11 @@
 
 class PdfOcrJob < ApplicationJob
   queue_as :high
-  attr_reader :blob, :out_path, :resource
+  attr_reader :blob, :resource
 
-  def perform(resource:, out_path:)
+  def perform(resource:)
     logger.info("PDF OCR job initiated for: #{resource.filename}")
     @resource = resource
-    @out_path = out_path
     @blob = resource.pdf # Required for ActiveStorage blob to tempfile method.
     update_state(state: "Processing")
     return unless pdf_attached?
@@ -40,4 +39,16 @@ def update_state(state:, message: nil)
     resource.note = message if message
     resource.save
   end
+
+  def out_path
+    File.join(ocr_out_dir, resource.filename)
+  end
+
+  def ocr_out_dir
+    @ocr_out_dir ||= begin
+                       path = Figgy.config["ocr_out_path"]
+                       FileUtils.mkdir_p(path) unless File.directory?(path)
+                       path
+                     end
+  end
 end

spec/jobs/pdf_ocr_job_spec.rb

@@ -4,7 +4,7 @@
 RSpec.describe PdfOcrJob do
   describe "#perform" do
     let(:out_dir) { Figgy.config["ocr_out_path"] }
-    let(:out_path) { File.join(out_dir, "ocr-sample.pdf") }
+    let(:out_path) { File.join(out_dir, "sample.pdf") }
     let(:resource) { FactoryBot.create(:ocr_request, file: fixture_path) }
 
     before do
@@ -21,7 +21,7 @@
       let(:fixture_path) { Rails.root.join("spec", "fixtures", "files", "sample.pdf") }
 
       it "creates on OCRed PDF in an output directory and deletes the attached PDF" do
-        expect { described_class.perform_now(resource: resource, out_path: out_path) }
+        expect { described_class.perform_now(resource: resource) }
           .to change { File.exist?(out_path) }
           .from(false).to(true)
         ocr_request = OcrRequest.all.first
@@ -34,7 +34,7 @@
       let(:fixture_path) { Rails.root.join("spec", "fixtures", "files", "bad.pdf") }
 
       it "saves error on the ocr request resource and copies original file to out path" do
-        described_class.perform_now(resource: resource, out_path: out_path)
+        described_class.perform_now(resource: resource)
         ocr_request = OcrRequest.all.first
         expect(ocr_request.state).to eq "Error"
         expect(ocr_request.note).to include "PDF OCR job failed"
@@ -47,7 +47,7 @@
       let(:resource) { FactoryBot.create(:ocr_request) }
 
       it "adds an error message to the ocr request resource" do
-        described_class.perform_now(resource: resource, out_path: out_path)
+        described_class.perform_now(resource: resource)
         ocr_request = OcrRequest.all.first
         expect(ocr_request.state).to eq "Error"
         expect(ocr_request.note).to include "Resource has no attached PDF"