dlp.rb

require "logger"
require 'pry-byebug'
require "google/cloud/dlp/v2"
include Google::Cloud::Dlp::V2

ROW_SCAN_LIMIT = 10000

# Doc: https://googleapis.dev/ruby/google-cloud-dlp-v2/latest/Google/Cloud/Dlp/V2.html
client = DlpService::Client.new


target_tables = {
  'pj_name': {
    'dataset_name': [
      'table_name'
}

def inspect_config
  InspectConfig.new(
    # see: https://cloud.google.com/dlp/docs/infotypes-reference
    info_types: [
      InfoType.new(name: 'EMAIL_ADDRESS'),
      InfoType.new(name: 'PERSON_NAME'),
      InfoType.new(name: 'PHONE_NUMBER')
    ],
    min_likelihood: 'POSSIBLE',
    # content_options: ,
    # custom_info_types: ,
    # exclude_info_types: ,
    include_quote: true,
    # limits: ,
    # rule_set: ,
  )
end

def storage_config(project_name, dataset_name, table_name)
  bq_options = BigQueryOptions.new(
    rows_limit: ROW_SCAN_LIMIT,
    table_reference: BigQueryTable.new(
      project_id: project_name,
      dataset_id: dataset_name,
      table_id: table_name
    ),
    # excluded_fields: ,
    # identifying_fields: ,
    # rows_limit_percent: ,
    sample_method: BigQueryOptions::SampleMethod::RANDOM_START,
  )

  StorageConfig.new(big_query_options: bq_options)

end

def inspect_job(project_name, dataset_name, table_name)
  inspect_job = InspectJobConfig.new(
    actions: [
      Action.new(
        save_findings: Action::SaveFindings.new(
          output_config: OutputStorageConfig.new(
            table: BigQueryTable.new(
              project_id: 'data-infra-secure',
              dataset_id: 'dlp',
              table_id: 'organon-summary'
            )
          )
        )
      )
    ],
    inspect_config: inspect_config,
    # inspect_template_name: '',
    storage_config: storage_config(project_name, dataset_name, table_name),
  )
end

target_tables.each do |pj, datasets|
  datasets.each do |dataset, tables|
    tables.each do |table|
      puts "Start Inspect Job: #{pj}-#{dataset}-#{table}"

      request = CreateDlpJobRequest.new(
        job_id: "#{pj}-#{dataset}-#{table}-#{Time.now.to_i}",
        parent: 'projects/data-infra-secure/locations/asia-northeast1',
        inspect_job: inspect_job(pj, dataset, table)
      )

      result = client.create_dlp_job request
      p result
    end
  end
end

module MyLogger
  LOGGER = Logger.new $stderr, level: Logger::WARN
  def logger
    LOGGER
  end
end

# Define a gRPC module-level logger method before grpc/logconfig.rb loads.
module GRPC
  extend MyLogger
end