Add export:live_document(s)_and_assets rake tasks

I've created two rake tasks:

1. `export:live_document_and_assets`, which exports a single
   document, given its content ID. By default it outputs to STDOUT
   but will write to a JSON file if a path is provided.
2. `export:live_documents_and_assets`, which exports all of the
   live documents and assets. By default it outputs to STDOUT
   but will write a separate JSON file for each exported document
   if an output directory is provided.

Usage (respectively):

```
rake export:live_document_and_assets["4c2ff601-4494-458b-bdc3-7d358122c2ae","./chris/foo.json"]
```

```
rake export:live_documents_and_assets["./chris"]
```

Author: ChrisBAshton

lib/tasks/export.rake

@@ -1,16 +1,32 @@
+require "json"
+
 namespace :export do
-  desc "Export live documents and assets"
-  task live_documents_and_assets: :environment do
+  desc "Export a specific live document and its assets, by its content ID"
+  task :live_document_and_assets, %i[content_id output_file] => :environment do |_, args|
+    document = Document.find_by(content_id: args[:content_id])
+    hash = WhitehallMigration::DocumentExport.export_to_hash(document)
+
+    if args[:output_file]
+      File.write(args[:output_file], JSON.pretty_generate(hash))
+    else
+      pp hash
+    end
+  end
+
+  desc "Export all live documents and assets"
+  task :live_documents_and_assets, %i[output_directory] => :environment do |_, args|
     documents = WhitehallMigration::DocumentExport.exportable_documents
 
     puts "Exporting #{documents.count} live editions"
 
-    # TODO: delete the override below. We're temporarily only looking
-    # at a pre-selected few for testing.
-    documents = [Document.find_by(content_id: "4c2ff601-4494-458b-bdc3-7d358122c2ae")]
     documents.each do |document|
-      pp WhitehallMigration::DocumentExport.export_to_hash(document)
+      hash = WhitehallMigration::DocumentExport.export_to_hash(document)
+
+      if args[:output_directory]
+        File.write("#{args[:output_directory]}/#{hash[:base_path].split("/").last}.json", JSON.pretty_generate(hash))
+      else
+        pp hash
+      end
     end
-    # TODO: think about how to avoid triggering an update to subscribers when republishing these docs.
   end
 end

spec/lib/tasks/export_spec.rb

@@ -0,0 +1,83 @@
+RSpec.describe "Export tasks" do
+  include ActiveJob::TestHelper
+
+  describe "export:live_document_and_assets" do
+    before do
+      Rake::Task["export:live_document_and_assets"].reenable
+    end
+
+    it "calls WhitehallMigration::DocumentExport.export_to_hash with correct arguments" do
+      document = create(:document, :with_live_edition)
+      allow(WhitehallMigration::DocumentExport).to receive(:export_to_hash)
+      Rake::Task["export:live_document_and_assets"].invoke(document.content_id)
+      expect(WhitehallMigration::DocumentExport).to have_received(:export_to_hash).with(document)
+    end
+
+    it "pretty-prints the result to STDOUT if no output_file is specified" do
+      document = create(:document, :with_live_edition)
+      allow(WhitehallMigration::DocumentExport).to receive(:export_to_hash).and_return({ foo: "bar" })
+      expect { Rake::Task["export:live_document_and_assets"].invoke(document.content_id) }.to output("{:foo=>\"bar\"}\n").to_stdout
+    end
+
+    it "writes the result as JSON to the given output_file if specified" do
+      document = create(:document, :with_live_edition)
+      allow(WhitehallMigration::DocumentExport).to receive(:export_to_hash).and_return({ foo: "bar", baz: "qux" })
+
+      output_file = Tempfile.new("export")
+      Rake::Task["export:live_document_and_assets"].invoke(document.content_id, output_file.path)
+
+      expected = <<~JSON
+        {
+          "foo": "bar",
+          "baz": "qux"
+        }
+      JSON
+      expect(File.read(output_file.path)).to match(expected.strip)
+    end
+  end
+
+  describe "export:live_documents_and_assets" do
+    before do
+      allow($stdout).to receive(:puts) # suppress output for cleanliness
+      Rake::Task["export:live_documents_and_assets"].reenable
+
+      # create 3 documents
+      Document.find_each(&:destroy) # Clean slate
+      @documents = [
+        create(:document, :with_live_edition),
+        create(:document, :with_live_edition),
+        create(:document, :with_live_edition),
+      ]
+      allow(WhitehallMigration::DocumentExport).to receive(:exportable_documents).and_return(@documents)
+    end
+
+    it "lists how many documents it is about to export" do
+      expect { Rake::Task["export:live_documents_and_assets"].invoke }.to output(/^Exporting 3 live editions/).to_stdout
+    end
+
+    it "calls WhitehallMigration::DocumentExport.export_to_hash with correct arguments" do
+      allow(WhitehallMigration::DocumentExport).to receive(:export_to_hash)
+      Rake::Task["export:live_documents_and_assets"].invoke
+      expect(WhitehallMigration::DocumentExport).to have_received(:export_to_hash).with(@documents[0])
+      expect(WhitehallMigration::DocumentExport).to have_received(:export_to_hash).with(@documents[1])
+      expect(WhitehallMigration::DocumentExport).to have_received(:export_to_hash).with(@documents[2])
+    end
+
+    it "pretty-prints the result to STDOUT if no output_directory is specified" do
+      allow(WhitehallMigration::DocumentExport).to receive(:export_to_hash).and_return({ foo: "bar" })
+      expect { Rake::Task["export:live_documents_and_assets"].invoke }.to output(/{:foo=>"bar"}\n{:foo=>"bar"}\n{:foo=>"bar"}\n$/).to_stdout
+    end
+
+    it "writes the result as JSON files to the given output_directory if specified" do
+      allow(WhitehallMigration::DocumentExport).to receive(:export_to_hash) do |document|
+        { base_path: "/news/example-path-#{document.id}" }
+      end
+      output_directory = Dir.mktmpdir
+      Rake::Task["export:live_documents_and_assets"].invoke(output_directory)
+
+      expected_files = @documents.map { |doc| "#{output_directory}/example-path-#{doc.id}.json" }
+      actual_files = Dir.glob("#{output_directory}/*.json").sort
+      expect(actual_files).to match_array(expected_files)
+    end
+  end
+end