https://github.com/fsubal/pdfgate

pdfgate_client.rb

class PdfgateClient
  class Error < StandardError
    attr_reader :code

    def initialize(code, message)
      super("PdfgateClient Error (#{code}): #{message}")

      @code = code
    end
  end

  Issue = Data.define(:severity, :code, :message) do
    def deny?
      severity == "deny"
    end

    def warn?
      severity == "warn"
    end
  end

  EXECUTABLE_PATH = File.expand_path(File.join(File.dirname(__FILE__), '..', 'bin', 'pdfgate'))
  POLICY_FILE_PATH = File.expand_path(File.join(File.dirname(__FILE__), '..', 'policy', 'default.json'))

  def initialize(pdf_file_path, policy_file_path: POLICY_FILE_PATH, timeout: 20.seconds)
    @pdf_file_path = pdf_file_path
    @policy_file_path = policy_file_path if File.exist?(policy_file_path)
    @timeout = timeout
  end

  def validate!
    case result
    in [{ ok: true, issues: }, 0]
      issues.map { |issue| Issue.new(**issue) }
    in [{ ok: false, issues: }, 1]
      issues.map { |issue| Issue.new(**issue) }
    in [{ error: { code:, message: } }, 2]
      raise Error.new(code, message)
    else
      error = Error.new("parse-failure", "Failed to parse the output of the pdfgate command.")
      Sentry.capture_exception(error, extra: { stdout: result[0], exitstatus: result[1] })
      raise error
    end
  end

  private

  def result
    stdout, status = run_with_timeout

    [JSON.parse(stdout, symbolize_names: true), status&.exitstatus]
  rescue JSON::ParserError
    [nil, status&.exitstatus]
  end

  def pdfgate_arguments
    [EXECUTABLE_PATH, "validate", @pdf_file_path].tap do |args|
      args.push("--policy", @policy_file_path.to_s) if @policy_file_path
    end
  end

  def run_with_timeout
    Open3.popen2(*pdfgate_arguments) do |stdin, stdout, wait_thr|
      stdin.close
      reader = Thread.new { stdout.read }
      if wait_thr.join(@timeout)
        [reader.value, wait_thr.value]
      else
        begin
          Process.kill("KILL", wait_thr.pid)
        rescue Errno::ESRCH
          nil
        end
        wait_thr.join
        [nil, nil]
      end
    end
  end
end