quay/endpoints/verbs/__init__.py

import logging
import hashlib

from flask import redirect, Blueprint, abort, send_file, make_response, request

import features

from app import app, signer, storage, metric_queue
from auth.auth import process_auth
from auth.auth_context import get_authenticated_user
from auth.permissions import ReadRepositoryPermission
from data import database
from data.interfaces.verbs import pre_oci_model as model
from endpoints.common import route_show_if, parse_repository_name
from endpoints.decorators import anon_protect
from endpoints.trackhelper import track_and_log
from endpoints.v2.blob import BLOB_DIGEST_ROUTE
from image.appc import AppCImageFormatter
from image.docker.squashed import SquashedDockerImageFormatter
from storage import Storage
from util.registry.filelike import wrap_with_handler
from util.registry.queuefile import QueueFile
from util.registry.queueprocess import QueueProcess
from util.registry.torrent import (make_torrent, per_user_torrent_filename, public_torrent_filename,
                                   PieceHasher)


verbs = Blueprint('verbs', __name__)
logger = logging.getLogger(__name__)


def _open_stream(formatter, namespace, repository, tag, derived_image_id, repo_image, handlers):
  """
  This method generates a stream of data which will be replicated and read from the queue files.
  This method runs in a separate process.
  """
  # For performance reasons, we load the full image list here, cache it, then disconnect from
  # the database.
  with database.UseThenDisconnect(app.config):
    image_list = list(model.get_manifest_layers_with_blobs(repo_image))

  def get_next_image():
    for current_image in image_list:
      yield current_image

  def get_next_layer():
    # Re-Initialize the storage engine because some may not respond well to forking (e.g. S3)
    store = Storage(app, metric_queue)
    for current_image in image_list:
      current_image_path = model.get_blob_path(current_image.blob)
      current_image_stream = store.stream_read_file(current_image.blob.locations,
                                                    current_image_path)

      logger.debug('Returning image layer %s: %s', current_image.image_id, current_image_path)
      yield current_image_stream

  stream = formatter.build_stream(namespace, repository, tag, repo_image, derived_image_id,
                                  get_next_image, get_next_layer)

  for handler_fn in handlers:
    stream = wrap_with_handler(stream, handler_fn)

  return stream.read


def _sign_derived_image(verb, derived_image, queue_file):
  """ Read from the queue file and sign the contents which are generated. This method runs in a
      separate process. """
  signature = None
  try:
    signature = signer.detached_sign(queue_file)
  except:
    logger.exception('Exception when signing %s deriving image %s', verb, derived_image.ref)
    return

  # Setup the database (since this is a new process) and then disconnect immediately
  # once the operation completes.
  if not queue_file.raised_exception:
    with database.UseThenDisconnect(app.config):
      model.set_derived_image_signature(derived_image, signer.name, signature)


def _write_derived_image_to_storage(verb, derived_image, queue_file):
  """ Read from the generated stream and write it back to the storage engine. This method runs in a
      separate process.
  """
  def handle_exception(ex):
    logger.debug('Exception when building %s derived image %s: %s', verb, derived_image.ref, ex)

    with database.UseThenDisconnect(app.config):
      model.delete_derived_image(derived_image)

  queue_file.add_exception_handler(handle_exception)

  # Re-Initialize the storage engine because some may not respond well to forking (e.g. S3)
  store = Storage(app, metric_queue)
  image_path = model.get_blob_path(derived_image.blob)
  store.stream_write(derived_image.blob.locations, image_path, queue_file)
  queue_file.close()


def _torrent_for_blob(blob, is_public):
  """ Returns a response containing the torrent file contents for the given blob. May abort
      with an error if the state is not valid (e.g. non-public, non-user request).
  """
  # Make sure the storage has a size.
  if not blob.size:
    abort(404)

  # Lookup the torrent information for the storage.
  torrent_info = model.get_torrent_info(blob)
  if torrent_info is None:
    abort(404)

  # Lookup the webseed path for the storage.
  path = model.get_blob_path(blob)
  webseed = storage.get_direct_download_url(blob.locations, path,
                                            expires_in=app.config['BITTORRENT_WEBSEED_LIFETIME'])
  if webseed is None:
    # We cannot support webseeds for storages that cannot provide direct downloads.
    abort(make_response('Storage engine does not support seeding.', 501))

  # Build the filename for the torrent.
  if is_public:
    name = public_torrent_filename(blob.uuid)
  else:
    user = get_authenticated_user()
    if not user:
      abort(403)

    name = per_user_torrent_filename(user.uuid, blob.uuid)

  # Return the torrent file.
  torrent_file = make_torrent(name, webseed, blob.size, torrent_info.piece_length,
                              torrent_info.pieces)

  headers = {'Content-Type': 'application/x-bittorrent',
             'Content-Disposition': 'attachment; filename={0}.torrent'.format(name)}

  return make_response(torrent_file, 200, headers)


def _torrent_repo_verb(repo_image, tag, verb, **kwargs):
  """ Handles returning a torrent for the given verb on the given image and tag. """
  if not features.BITTORRENT:
    # Torrent feature is not enabled.
    abort(406)

  # Lookup an *existing* derived storage for the verb. If the verb's image storage doesn't exist,
  # we cannot create it here, so we 406.
  derived_image = model.lookup_derived_image(repo_image, verb, varying_metadata={'tag': tag})
  if derived_image is None:
    abort(406)

  # Return the torrent.
  public_repo = model.repository_is_public(repo_image.repository.namespace_name,
                                           repo_image.repository.name)
  torrent = _torrent_for_blob(derived_image.blob, public_repo)

  # Log the action.
  track_and_log('repo_verb', repo_image.repository, tag=tag, verb=verb, torrent=True, **kwargs)
  return torrent


def _verify_repo_verb(_, namespace, repository, tag, verb, checker=None):
  permission = ReadRepositoryPermission(namespace, repository)
  if not permission.can() and not model.repository_is_public(namespace, repository):
    abort(403)

  # Lookup the requested tag.
  tag_image = model.get_tag_image(namespace, repository, tag)
  if tag_image is None:
    abort(404)

  # If there is a data checker, call it first.
  if checker is not None:
    if not checker(tag_image):
      logger.debug('Check mismatch on %s/%s:%s, verb %s', namespace, repository, tag, verb)
      abort(404)

  return tag_image


def _repo_verb_signature(namespace, repository, tag, verb, checker=None, **kwargs):
  # Verify that the image exists and that we have access to it.
  repo_image = _verify_repo_verb(storage, namespace, repository, tag, verb, checker)

  # derived_image the derived image storage for the verb.
  derived_image = model.lookup_derived_image(repo_image, verb, varying_metadata={'tag': tag})
  if derived_image is None or derived_image.blob.uploading:
    return make_response('', 202)

  # Check if we have a valid signer configured.
  if not signer.name:
    abort(404)

  # Lookup the signature for the verb.
  signature_value = model.get_derived_image_signature(derived_image, signer.name)
  if signature_value is None:
    abort(404)

  # Return the signature.
  return make_response(signature_value)


def _repo_verb(namespace, repository, tag, verb, formatter, sign=False, checker=None, **kwargs):
  # Verify that the image exists and that we have access to it.
  repo_image = _verify_repo_verb(storage, namespace, repository, tag, verb, checker)

  # Check for torrent. If found, we return a torrent for the repo verb image (if the derived
  # image already exists).
  if request.accept_mimetypes.best == 'application/x-bittorrent':
    metric_queue.repository_pull.Inc(labelvalues=[namespace, repository, verb + '+torrent'])
    return _torrent_repo_verb(repo_image, tag, verb, **kwargs)

  # Log the action.
  track_and_log('repo_verb', repo_image.repository, tag=tag, verb=verb, **kwargs)
  metric_queue.repository_pull.Inc(labelvalues=[namespace, repository, verb])

  # Lookup/create the derived image for the verb and repo image.
  derived_image = model.lookup_or_create_derived_image(repo_image, verb,
                                                       storage.preferred_locations[0],
                                                       varying_metadata={'tag': tag})
  if not derived_image.blob.uploading:
    logger.debug('Derived %s image %s exists in storage', verb, derived_image.ref)
    derived_layer_path = model.get_blob_path(derived_image.blob)
    is_head_request = request.method == 'HEAD'
    download_url = storage.get_direct_download_url(derived_image.blob.locations, derived_layer_path,
                                                   head=is_head_request)
    if download_url:
      logger.debug('Redirecting to download URL for derived %s image %s', verb, derived_image.ref)
      return redirect(download_url)

    # Close the database handle here for this process before we send the long download.
    database.close_db_filter(None)

    logger.debug('Sending cached derived %s image %s', verb, derived_image.ref)
    return send_file(storage.stream_read_file(derived_image.blob.locations, derived_layer_path))

  logger.debug('Building and returning derived %s image %s', verb, derived_image.ref)

  # Calculate a derived image ID.
  derived_image_id = hashlib.sha256(repo_image.image_id + ':' + verb).hexdigest()

  def _cleanup():
    # Close any existing DB connection once the process has exited.
    database.close_db_filter(None)

  hasher = PieceHasher(app.config['BITTORRENT_PIECE_SIZE'])

  def _store_metadata_and_cleanup():
    with database.UseThenDisconnect(app.config):
      model.set_torrent_info(derived_image.blob, app.config['BITTORRENT_PIECE_SIZE'],
                             hasher.final_piece_hashes())
      model.set_blob_size(derived_image.blob, hasher.hashed_bytes)

  # Create a queue process to generate the data. The queue files will read from the process
  # and send the results to the client and storage.
  handlers = [hasher.update]
  args = (formatter, namespace, repository, tag, derived_image_id, repo_image, handlers)
  queue_process = QueueProcess(_open_stream,
                               8 * 1024, 10 * 1024 * 1024, # 8K/10M chunk/max
                               args, finished=_store_metadata_and_cleanup)

  client_queue_file = QueueFile(queue_process.create_queue(), 'client')
  storage_queue_file = QueueFile(queue_process.create_queue(), 'storage')

  # If signing is required, add a QueueFile for signing the image as we stream it out.
  signing_queue_file = None
  if sign and signer.name:
    signing_queue_file = QueueFile(queue_process.create_queue(), 'signing')

  # Start building.
  queue_process.run()

  # Start the storage saving.
  storage_args = (verb, derived_image, storage_queue_file)
  QueueProcess.run_process(_write_derived_image_to_storage, storage_args, finished=_cleanup)

  if sign and signer.name:
    signing_args = (verb, derived_image, signing_queue_file)
    QueueProcess.run_process(_sign_derived_image, signing_args, finished=_cleanup)

  # Close the database handle here for this process before we send the long download.
  database.close_db_filter(None)

  # Return the client's data.
  return send_file(client_queue_file)


def os_arch_checker(os, arch):
  def checker(repo_image):
    image_json = repo_image.compat_metadata

    # Verify the architecture and os.
    operating_system = image_json.get('os', 'linux')
    if operating_system != os:
      return False

    architecture = image_json.get('architecture', 'amd64')

    # Note: Some older Docker images have 'x86_64' rather than 'amd64'.
    # We allow the conversion here.
    if architecture == 'x86_64' and operating_system == 'linux':
      architecture = 'amd64'

    if architecture != arch:
      return False

    return True

  return checker


@route_show_if(features.ACI_CONVERSION)
@anon_protect
@verbs.route('/aci/<server>/<namespace>/<repository>/<tag>/sig/<os>/<arch>/', methods=['GET'])
@verbs.route('/aci/<server>/<namespace>/<repository>/<tag>/aci.asc/<os>/<arch>/', methods=['GET'])
@process_auth
def get_aci_signature(server, namespace, repository, tag, os, arch):
  return _repo_verb_signature(namespace, repository, tag, 'aci', checker=os_arch_checker(os, arch),
                              os=os, arch=arch)


@route_show_if(features.ACI_CONVERSION)
@anon_protect
@verbs.route('/aci/<server>/<namespace>/<repository>/<tag>/aci/<os>/<arch>/', methods=['GET', 'HEAD'])
@process_auth
def get_aci_image(server, namespace, repository, tag, os, arch):
  return _repo_verb(namespace, repository, tag, 'aci', AppCImageFormatter(),
                    sign=True, checker=os_arch_checker(os, arch), os=os, arch=arch)


@anon_protect
@verbs.route('/squash/<namespace>/<repository>/<tag>', methods=['GET'])
@process_auth
def get_squashed_tag(namespace, repository, tag):
  return _repo_verb(namespace, repository, tag, 'squash', SquashedDockerImageFormatter())


@route_show_if(features.BITTORRENT)
@anon_protect
@verbs.route('/torrent{0}'.format(BLOB_DIGEST_ROUTE), methods=['GET'])
@process_auth
@parse_repository_name()
def get_tag_torrent(namespace_name, repo_name, digest):
  permission = ReadRepositoryPermission(namespace_name, repo_name)
  public_repo = model.repository_is_public(namespace_name, repo_name)
  if not permission.can() and not public_repo:
    abort(403)

  user = get_authenticated_user()
  if user is None and not public_repo:
    # We can not generate a private torrent cluster without a user uuid (e.g. token auth)
    abort(403)

  blob = model.get_repo_blob_by_digest(namespace_name, repo_name, digest)
  if blob is None:
    abort(404)

  metric_queue.repository_pull.Inc(labelvalues=[namespace_name, repo_name, 'torrent'])
  return _torrent_for_blob(blob, public_repo)