Merge pull request #9628 from alphagov/remove-csv-rake-task

Remove rake task that relies on CSV input

Gemfile

@@ -54,7 +54,6 @@ gem "record_tag_helper", require: false
 gem "redis"
 gem "responders"
 gem "rinku", require: "rails_rinku"
-gem "ruby-progressbar", require: false
 gem "rubyzip"
 gem "sentry-sidekiq"
 gem "sidekiq-scheduler"

Gemfile.lock

@@ -1093,7 +1093,6 @@ DEPENDENCIES
   responders
   rinku
   rubocop-govuk
-  ruby-progressbar
   rubyzip
   sentry-sidekiq
   sidekiq-scheduler

lib/tasks/data_migrator.rake

@@ -1,5 +1,3 @@
-require "csv"
-
 namespace :db do
   namespace :data do
     desc "Run all data migrations, or a specific version assigned to environment variable VERSION"

lib/tasks/publishing_api.rake

@@ -123,135 +123,6 @@ namespace :publishing_api do
     end
   end
 
-  namespace :bulk_republish do
-    desc "Republish all About pages"
-    task all_about_pages: :environment do
-      about_us_pages = Organisation.all.map(&:about_us).compact
-      count = about_us_pages.count
-      puts "# Sending #{count} 'about us' pages to Publishing API"
-      about_us_pages.each_with_index do |about_us_page, i|
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue(
-          "bulk_republishing",
-          about_us_page.document_id,
-          true,
-        )
-        puts "Queuing #{i}-#{i + 99} of #{count} items" if (i % 100).zero?
-      end
-      puts "Finished queuing items for Publishing API"
-    end
-
-    desc "Republish all documents with draft editions"
-    task all_drafts: :environment do
-      editions = Edition.in_pre_publication_state.includes(:document)
-
-      puts "Enqueueing #{editions.count} documents"
-      editions.find_each do |edition|
-        document_id = edition.document.id
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", document_id, true)
-      end
-      puts "Finished enqueueing items for Publishing API"
-    end
-
-    desc "Republish all editions which have attachments to the Publishing API"
-    task editions_with_attachments: :environment do
-      editions = Edition.publicly_visible.where(
-        id: Attachment.where(attachable_type: "Edition").select("attachable_id"),
-      )
-
-      editions.joins(:document).distinct.pluck("documents.id").each do |document_id|
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", document_id, true)
-      end
-    end
-
-    desc "⚠️  WARNING: this rake task republishes **all** documents with HTML attachments (this can block publishing for > 1 hour) ⚠️. Republish all documents with HTML attachments to the Publishing API."
-    task html_attachments: :environment do
-      document_ids = Edition
-        .publicly_visible
-        .where(id: HtmlAttachment.where(attachable_type: "Edition").select(:attachable_id))
-        .pluck(:document_id)
-      document_ids.each do |document_id|
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", document_id, true)
-      end
-    end
-
-    desc "Republish all draft editions with HTML attachments to the Publishing API"
-    task drafts_with_html_attachments: :environment do
-      document_ids = Edition
-        .in_pre_publication_state
-        .where(id: HtmlAttachment.where(attachable_type: "Edition").select(:attachable_id))
-        .pluck(:document_id)
-      document_ids.each do |document_id|
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", document_id, true)
-      end
-    end
-
-    desc "Republish all documents of a given type, e.g. 'NewsArticle'"
-    task :document_type, [:document_type] => :environment do |_, args|
-      begin
-        document_type = args[:document_type].constantize
-      rescue NameError
-        abort "Unknown document type #{args[:document_type]}\nCheck the GOV.UK developer documentation for a list of acceptable document types: https://docs.publishing.service.gov.uk/manual/republishing-content.html#whitehall"
-      end
-
-      documents = document_type.all
-      puts "Enqueueing #{documents.count} documents"
-      documents.find_each do |document|
-        if document.respond_to?(:publish_to_publishing_api)
-          Whitehall::PublishingApi.bulk_republish_async(document) if document.can_publish_to_publishing_api?
-        else
-          PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", document.document_id, true)
-        end
-      end
-      puts "Finished enqueueing items for Publishing API"
-    end
-
-    desc "Republish all documents of a given organisation"
-    task :by_organisation, [:organisation_slug] => :environment do |_, args|
-      org = Organisation.find_by(slug: args[:organisation_slug])
-      editions = Edition.latest_edition.in_organisation(org)
-      puts "Enqueueing #{editions.count} documents"
-      editions.find_each do |edition|
-        document = edition.document
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", document.id, true)
-      end
-      puts "Finished enqueueing items for Publishing API"
-    end
-
-    desc "Republish documents by content id"
-    task :documents_by_content_ids, %w[content_ids] => :environment do |_, args|
-      content_ids =  args[:content_ids].split
-      document_ids = Document.where(content_id: content_ids).pluck(:id)
-
-      puts "Bulk republishing #{document_ids.count} documents"
-
-      document_ids.each do |id|
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", id, true)
-      end
-    end
-
-    desc "Republish documents by content ids from CSV"
-    task :documents_by_content_ids_from_csv, [:csv_file_name] => :environment do |_, args|
-      csv = CSV.read(Rails.root.join("lib/tasks/#{args[:csv_file_name]}.csv"), headers: true)
-      content_ids = csv["content_id"].uniq
-      document_ids = Document.where(content_id: content_ids).pluck(:id)
-
-      puts "Bulk republishing #{document_ids.count} documents"
-
-      document_ids.each do |id|
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", id, true)
-      end
-    end
-
-    desc "Republish all documents"
-    task all_documents: :environment do
-      puts "Enqueueing #{Document.count} documents"
-      Document.find_each do |document|
-        PublishingApiDocumentRepublishingWorker.perform_async_in_queue("bulk_republishing", document.id, true)
-      end
-      puts "Finished enqueueing items for Publishing API"
-    end
-  end
-
   namespace :unpublish do
     desc "Manually unpublish content with a redirect"
     # This task is for unpublishing Whitehall managed content where