quay/formats/squashed.py

from app import app
from util.registry.gzipwrap import GZIP_BUFFER_SIZE
from util.registry.streamlayerformat import StreamLayerMerger
from formats.tarimageformatter import TarImageFormatter

import copy
import json
import math

class FileEstimationException(Exception):
  """ Exception raised by build_docker_load_stream if the estimated size of the layer TAR
      was lower than the actual size. This means the sent TAR header is wrong, and we have
      to fail.
  """
  pass


class SquashedDockerImage(TarImageFormatter):
  """ Image formatter which produces a squashed image compatible with the `docker load`
      command.
  """

  # Multiplier against the image size reported by Docker to account for the TAR metadata.
  # Note: This multiplier was not formally calculated in anyway and should be adjusted overtime
  # if/when we encounter issues with it. Unfortunately, we cannot make it too large or the Docker
  # daemon dies when trying to load the entire TAR into memory.
  SIZE_MULTIPLIER = 1.2

  def stream_generator(self, namespace, repository, tag, synthetic_image_id,
                       layer_json, get_image_iterator, get_layer_iterator, get_image_json):

    # Docker import V1 Format (.tar):
    #  repositories - JSON file containing a repo -> tag -> image map
    #  {image ID folder}:
    #     json - The layer JSON
    #     layer.tar - The TARed contents of the layer
    #     VERSION - The docker import version: '1.0'
    layer_merger = StreamLayerMerger(get_layer_iterator)

    # Yield the repositories file:
    synthetic_layer_info = {}
    synthetic_layer_info[tag + '.squash'] = synthetic_image_id

    hostname = app.config['SERVER_HOSTNAME']
    repositories = {}
    repositories[hostname + '/' + namespace + '/' + repository] = synthetic_layer_info

    yield self.tar_file('repositories', json.dumps(repositories))

    # Yield the image ID folder.
    yield self.tar_folder(synthetic_image_id)

    # Yield the JSON layer data.
    layer_json = SquashedDockerImage._build_layer_json(layer_json, synthetic_image_id)
    yield self.tar_file(synthetic_image_id + '/json', json.dumps(layer_json))

    # Yield the VERSION file.
    yield self.tar_file(synthetic_image_id + '/VERSION', '1.0')

    # Yield the merged layer data's header.
    estimated_file_size = 0
    for image in get_image_iterator():
      # In V1 we have the actual uncompressed size, which is needed for back compat with
      # older versions of Docker.
      # In V2, we use the size given in the image JSON.
      if image.storage.uncompressed_size:
        estimated_file_size += image.storage.uncompressed_size
      else:
        image_json = get_image_json(image)
        estimated_file_size += image_json.get('Size', 0) * SquashedDockerImage.SIZE_MULTIPLIER

    # Make sure the estimated file size is an integer number of bytes.
    estimated_file_size = int(math.ceil(estimated_file_size))

    yield self.tar_file_header(synthetic_image_id + '/layer.tar', estimated_file_size)

    # Yield the contents of the merged layer.
    yielded_size = 0
    for entry in layer_merger.get_generator():
      yield entry
      yielded_size += len(entry)

    # If the yielded size is more than the estimated size (which is unlikely but possible), then
    # raise an exception since the tar header will be wrong.
    if yielded_size > estimated_file_size:
      message = "Expected %s bytes, found %s bytes" % (estimated_file_size, yielded_size)
      raise FileEstimationException(message)

    # If the yielded size is less than the estimated size (which is likely), fill the rest with
    # zeros.
    if yielded_size < estimated_file_size:
      to_yield = estimated_file_size - yielded_size
      while to_yield > 0:
        yielded = min(to_yield, GZIP_BUFFER_SIZE)
        yield '\0' * yielded
        to_yield -= yielded

    # Yield any file padding to 512 bytes that is necessary.
    yield self.tar_file_padding(estimated_file_size)

    # Last two records are empty in TAR spec.
    yield '\0' * 512
    yield '\0' * 512


  @staticmethod
  def _build_layer_json(layer_json, synthetic_image_id):
    updated_json = copy.deepcopy(layer_json)
    updated_json['id'] = synthetic_image_id

    if 'parent' in updated_json:
      del updated_json['parent']

    if 'config' in updated_json and 'Image' in updated_json['config']:
      updated_json['config']['Image'] = synthetic_image_id

    if 'container_config' in updated_json and 'Image' in updated_json['container_config']:
      updated_json['container_config']['Image'] = synthetic_image_id

    return updated_json