Add rake task to report on a regex in editionable content

Sometimes we want to identify which published content matches a regular
expression (e.g. if we make a change to govspeak and need to republish
affected content).

Therefore adding a rake task that will report on the content that
includes a given regex in the currently published edition.

This is being broken down into batches of 1000, as our infrastructure
does not support large queries being made on a Rails console.

lib/tasks/reporting.rake

@@ -12,4 +12,17 @@ namespace :reporting do
   task published_attachments_report: :environment do
     Reports::PublishedAttachmentsReport.new.report
   end
+
+  desc "Prints a list of content IDs that documents whose live edition contains a given regular expression"
+  task :find_docs, [:regex] => :environment do |_, args|
+    regex = Regexp.new(/#{args[:regex]}/)
+
+    Document.where.not(live_edition_id: nil).find_in_batches(batch_size: 1000) do |batch|
+      batch.each do |document|
+        next unless document.editions.published.any?
+
+        puts document.content_id if regex.match?(document.editions.published.last.body)
+      end
+    end
+  end
 end

test/unit/lib/tasks/reporting_test.rb

@@ -0,0 +1,26 @@
+require "test_helper"
+require "rake"
+
+class ReportingRake < ActiveSupport::TestCase
+  setup do
+    @document_1 = create(:published_edition, body: "Some text 1")
+    @document_2 = create(:draft_edition, body: "Some text 2")
+    @document_3 = create(:published_edition, body: "Some other text 1")
+  end
+
+  teardown do
+    Rake::Task["reporting:matching_docs"].reenable
+  end
+
+  test "it prints the content IDs of the matching documents from published editions" do
+    assert_output(/#{@document_1.document.content_id}/) { Rake.application.invoke_task "reporting:matching_docs[Some text]" }
+  end
+
+  test "it does not print the content IDs of the matching documents from draft editions" do
+    assert_output(/^(?!.*#{@document_2.document.content_id}).*$/) { Rake.application.invoke_task "reporting:matching_docs[Some text]" }
+  end
+
+  test "it does not print the content IDs of the non-matching documents from published editions" do
+    assert_output(/^(?!.*#{@document_3.document.content_id}).*$/) { Rake.application.invoke_task "reporting:matching_docs[Some text]" }
+  end
+end