Hacky script that can export data from a Publishing API instance into local environment - it tends to fail a lot.

publishing-api-import.rb

require "bundler/inline"

# To run this you need to be connected to the Publsihing API via a port forward
# I set this up by adding myself as a user in: https://signon.integration.publishing.service.gov.uk/api_users
# then storing the bearer token in ~/.publishing-api-token
#
# Then I logged onto the VPN and port forwarded Publishing API to 9000 with:
# ssh $(ssh integration "govuk_node_list --single-node -c publishing_api").integration -CNL 9000:127.0.0.1:3093
#
# and ran: PUBLISHING_API_BEARER_TOKEN=$(cat ~/.publishing-api-token) ruby import-publishing-api.rb
gemfile do
  source "https://rubygems.org"
  gem "gds-api-adapters"
  gem "byebug"
end

publishing_api_integration = GdsApi::PublishingApiV2.new(
  "http://localhost:9000",
  { bearer_token: ENV['PUBLISHING_API_BEARER_TOKEN'], timeout: 60 },
)

publishing_api_local = GdsApi.publishing_api_v2

# Change these based on which document type to import
types = %w[organisation taxon]

pairs = types.map do |type|
  enum = publishing_api_integration.get_content_items_enum(document_type: type,
                                                           states: %w[published])
  [type, enum]
end

pairs.each do |(name, enum)|
  puts "loading #{name}s"
  imported = 0
  skipped = 0
  enum.each.with_index do |result, i|
    puts "#{imported} #{name}s imported, #{skipped} skipped" if i > 0 && i % 10 == 0

    content_id = result["content_id"]

    begin
      publishing_api_local.get_content(content_id)
      skipped += 1
      next
    rescue GdsApi::HTTPNotFound
    end

    fields = %w[
      analytics_identifier
      base_path
      description
      details
      document_type
      first_published_at
      last_edited_at
      phase
      public_updated_at
      publishing_app
      rendering_app
      routes
      schema_name
      title
      update_type
      links
    ]

    payload = result.slice(*fields).tap do |r|
      r["update_type"] ||= "major"
    end

    begin
      publishing_api_local.put_content(content_id, payload)
      publishing_api_local.publish(content_id)
      links = publishing_api_integration.get_links(content_id)
      publishing_api_local.patch_links(content_id, links: links["links"]) unless links["links"].empty?
      imported += 1
    rescue GdsApi::HTTPUnprocessableEntity
      skipped += 1
    end
  end
end