quay/endpoints/v2/manifest.py

import logging
import json
import hashlib

from collections import namedtuple, OrderedDict
from datetime import datetime
from functools import wraps

import jwt.utils

from peewee import IntegrityError
from flask import make_response, request, url_for
from jwkest.jws import SIGNER_ALGS, keyrep

import features

from app import docker_v2_signing_key, app
from auth.registry_jwt_auth import process_registry_jwt_auth
from endpoints.common import parse_repository_name
from endpoints.decorators import anon_protect
from endpoints.v2 import v2_bp, require_repo_read, require_repo_write
from endpoints.v2.errors import (BlobUnknown, ManifestInvalid, ManifestUnknown, TagInvalid,
                                 NameInvalid, TagAlreadyExists)
from endpoints.trackhelper import track_and_log
from endpoints.notificationhelper import spawn_notification
from util.registry.replication import queue_storage_replication
from util.names import VALID_TAG_PATTERN
from digest import digest_tools
from data import model
from data.model import TagAlreadyCreatedException
from data.database import RepositoryTag

logger = logging.getLogger(__name__)


BASE_MANIFEST_ROUTE = '/<repopath:repository>/manifests/<regex("{0}"):manifest_ref>'
MANIFEST_DIGEST_ROUTE = BASE_MANIFEST_ROUTE.format(digest_tools.DIGEST_PATTERN)
MANIFEST_TAGNAME_ROUTE = BASE_MANIFEST_ROUTE.format(VALID_TAG_PATTERN)

# From: https://github.com/docker/distribution/blob/47a064d4195a9b56133891bbb13620c3ac83a827/manifest/schema1/manifest.go#L18
MANIFEST_CONTENT_TYPE = 'application/vnd.docker.distribution.manifest.v1+prettyjws'
MANIFEST2_SCHEMA2_CONTENT_TYPES = ['application/vnd.docker.distribution.manifest.v2+json',
                                   'application/vnd.docker.distribution.manifest.list.v2+json']

ISO_DATETIME_FORMAT_ZULU = '%Y-%m-%dT%H:%M:%SZ'
JWS_ALGORITHM = 'RS256'


ImageMetadata = namedtuple('ImageMetadata', ['digest', 'v1_metadata', 'v1_metadata_str'])
ExtractedV1Metadata = namedtuple('ExtractedV1Metadata', ['docker_id', 'parent', 'created',
                                                         'comment', 'command', 'labels'])


_SIGNATURES_KEY = 'signatures'
_PROTECTED_KEY = 'protected'
_FORMAT_LENGTH_KEY = 'formatLength'
_FORMAT_TAIL_KEY = 'formatTail'
_REPO_NAME_KEY = 'name'
_REPO_TAG_KEY = 'tag'
_FS_LAYERS_KEY = 'fsLayers'
_HISTORY_KEY = 'history'
_BLOB_SUM_KEY = 'blobSum'
_V1_COMPAT_KEY = 'v1Compatibility'
_ARCH_KEY = 'architecture'
_SCHEMA_VER = 'schemaVersion'


class SignedManifest(object):

  def __init__(self, manifest_bytes, validate=True):
    self._bytes = manifest_bytes

    self._parsed = json.loads(manifest_bytes)
    self._signatures = self._parsed[_SIGNATURES_KEY]
    self._tag = self._parsed[_REPO_TAG_KEY]

    repo_name_tuple = self._parsed[_REPO_NAME_KEY].split('/')
    if len(repo_name_tuple) > 1:
      self._namespace, self._repo_name = repo_name_tuple
    elif len(repo_name_tuple) == 1:
      self._namespace = ''
      self._repo_name = repo_name_tuple[0]
    else:
      raise ValueError('repo_name has too many or too few pieces')

    if validate:
      self._validate()

  def _validate(self):
    for signature in self._signatures:
      bytes_to_verify = '{0}.{1}'.format(signature['protected'],
                                         jwt.utils.base64url_encode(self.payload))
      signer = SIGNER_ALGS[signature['header']['alg']]
      key = keyrep(signature['header']['jwk'])
      gk = key.get_key()
      sig = jwt.utils.base64url_decode(signature['signature'].encode('utf-8'))
      verified = signer.verify(bytes_to_verify, sig, gk)
      if not verified:
        raise ValueError('manifest file failed signature verification')

  @property
  def signatures(self):
    return self._signatures

  @property
  def namespace(self):
    return self._namespace

  @property
  def repo_name(self):
    return self._repo_name

  @property
  def tag(self):
    return self._tag

  @property
  def bytes(self):
    return self._bytes

  @property
  def digest(self):
    return digest_tools.sha256_digest(self.payload)

  @property
  def layers(self):
    """ Returns a generator of objects that have the blobSum and v1Compatibility keys in them,
        starting from the base image and working toward the leaf node.
    """
    for blob_sum_obj, history_obj in reversed(zip(self._parsed[_FS_LAYERS_KEY],
                                                  self._parsed[_HISTORY_KEY])):

      try:
        image_digest = digest_tools.Digest.parse_digest(blob_sum_obj[_BLOB_SUM_KEY])
      except digest_tools.InvalidDigestException:
        err_message = 'could not parse manifest digest: %s' % blob_sum_obj[_BLOB_SUM_KEY]
        raise ManifestInvalid(detail={'message': err_message})

      metadata_string = history_obj[_V1_COMPAT_KEY]

      v1_metadata = json.loads(metadata_string)
      command_list = v1_metadata.get('container_config', {}).get('Cmd', None)
      command = json.dumps(command_list) if command_list else None

      if not 'id' in v1_metadata:
        raise ManifestInvalid(detail={'message': 'invalid manifest v1 history'})

      labels = v1_metadata.get('config', {}).get('Labels', {}) or {}
      extracted = ExtractedV1Metadata(v1_metadata['id'], v1_metadata.get('parent'),
                                      v1_metadata.get('created'), v1_metadata.get('comment'),
                                      command, labels)
      yield ImageMetadata(image_digest, extracted, metadata_string)

  @property
  def payload(self):
    protected = str(self._signatures[0][_PROTECTED_KEY])
    parsed_protected = json.loads(jwt.utils.base64url_decode(protected))
    signed_content_head = self._bytes[:parsed_protected[_FORMAT_LENGTH_KEY]]
    signed_content_tail = jwt.utils.base64url_decode(str(parsed_protected[_FORMAT_TAIL_KEY]))
    return signed_content_head + signed_content_tail


class SignedManifestBuilder(object):
  """ Class which represents a manifest which is currently being built.
  """
  def __init__(self, namespace_name, repo_name, tag, architecture='amd64', schema_ver=1):
    repo_name_key = '{0}/{1}'.format(namespace_name, repo_name)
    if namespace_name == '':
      repo_name_key = repo_name

    self._base_payload = {
      _REPO_TAG_KEY: tag,
      _REPO_NAME_KEY: repo_name_key,
      _ARCH_KEY: architecture,
      _SCHEMA_VER: schema_ver,
    }

    self._fs_layer_digests = []
    self._history = []

  def add_layer(self, layer_digest, v1_json_metadata):
    self._fs_layer_digests.append({
      _BLOB_SUM_KEY: layer_digest,
    })
    self._history.append({
      _V1_COMPAT_KEY: v1_json_metadata,
    })
    return self


  def build(self, json_web_key):
    """ Build the payload and sign it, returning a SignedManifest object.
    """
    payload = OrderedDict(self._base_payload)
    payload.update({
      _HISTORY_KEY: self._history,
      _FS_LAYERS_KEY: self._fs_layer_digests,
    })

    payload_str = json.dumps(payload, indent=3)

    split_point = payload_str.rfind('\n}')

    protected_payload = {
      'formatTail': jwt.utils.base64url_encode(payload_str[split_point:]),
      'formatLength': split_point,
      'time': datetime.utcnow().strftime(ISO_DATETIME_FORMAT_ZULU),
    }
    protected = jwt.utils.base64url_encode(json.dumps(protected_payload))
    logger.debug('Generated protected block: %s', protected)

    bytes_to_sign = '{0}.{1}'.format(protected, jwt.utils.base64url_encode(payload_str))

    signer = SIGNER_ALGS[JWS_ALGORITHM]
    signature = jwt.utils.base64url_encode(signer.sign(bytes_to_sign, json_web_key.get_key()))
    logger.debug('Generated signature: %s', signature)

    public_members = set(json_web_key.public_members)
    public_key = {comp: value for comp, value in json_web_key.to_dict().items()
                  if comp in public_members}

    signature_block = {
      'header': {
        'jwk': public_key,
        'alg': JWS_ALGORITHM,
      },
      'signature': signature,
      _PROTECTED_KEY: protected,
    }

    logger.debug('Encoded signature block: %s', json.dumps(signature_block))

    payload.update({
      _SIGNATURES_KEY: [signature_block],
    })

    return SignedManifest(json.dumps(payload, indent=3))


@v2_bp.route(MANIFEST_TAGNAME_ROUTE, methods=['GET'])
@parse_repository_name()
@process_registry_jwt_auth(scopes=['pull'])
@require_repo_read
@anon_protect
def fetch_manifest_by_tagname(namespace_name, repo_name, manifest_ref):
  try:
    manifest = model.tag.load_tag_manifest(namespace_name, repo_name, manifest_ref)
  except model.InvalidManifestException:
    try:
      model.tag.get_active_tag(namespace_name, repo_name, manifest_ref)
    except RepositoryTag.DoesNotExist:
      raise ManifestUnknown()

    try:
      manifest = _generate_and_store_manifest(namespace_name, repo_name, manifest_ref)
    except model.DataModelException:
      logger.exception('Exception when generating manifest for %s/%s:%s', namespace_name, repo_name,
                       manifest_ref)
      raise ManifestUnknown()

  repo = model.repository.get_repository(namespace_name, repo_name)
  if repo is not None:
    track_and_log('pull_repo', repo, analytics_name='pull_repo_100x', analytics_sample=0.01)

  response = make_response(manifest.json_data, 200)
  response.headers['Content-Type'] = MANIFEST_CONTENT_TYPE
  response.headers['Docker-Content-Digest'] = manifest.digest
  return response


@v2_bp.route(MANIFEST_DIGEST_ROUTE, methods=['GET'])
@parse_repository_name()
@process_registry_jwt_auth(scopes=['pull'])
@require_repo_read
@anon_protect
def fetch_manifest_by_digest(namespace_name, repo_name, manifest_ref):
  try:
    manifest = model.tag.load_manifest_by_digest(namespace_name, repo_name, manifest_ref)
  except model.InvalidManifestException:
    # Without a tag name to reference, we can't make an attempt to generate the manifest
    raise ManifestUnknown()

  repo = model.repository.get_repository(namespace_name, repo_name)
  if repo is not None:
    track_and_log('pull_repo', repo)

  response = make_response(manifest.json_data, 200)
  response.headers['Content-Type'] = MANIFEST_CONTENT_TYPE
  response.headers['Docker-Content-Digest'] = manifest.digest
  return response


def _reject_manifest2_schema2(func):
  @wraps(func)
  def wrapped(*args, **kwargs):
    if request.content_type in MANIFEST2_SCHEMA2_CONTENT_TYPES:
      raise ManifestInvalid(detail={'message': 'manifest schema version not supported'},
                            http_status_code=415)
    return func(*args, **kwargs)
  return wrapped


@v2_bp.route(MANIFEST_TAGNAME_ROUTE, methods=['PUT'])
@_reject_manifest2_schema2
@parse_repository_name()
@process_registry_jwt_auth(scopes=['pull', 'push'])
@require_repo_write
@anon_protect
def write_manifest_by_tagname(namespace_name, repo_name, manifest_ref):
  try:
    manifest = SignedManifest(request.data)
  except ValueError:
    raise ManifestInvalid(detail={'message': 'could not parse manifest'})

  if manifest.tag != manifest_ref:
    raise TagInvalid()

  return _write_manifest(namespace_name, repo_name, manifest)


@v2_bp.route(MANIFEST_DIGEST_ROUTE, methods=['PUT'])
@_reject_manifest2_schema2
@parse_repository_name()
@process_registry_jwt_auth(scopes=['pull', 'push'])
@require_repo_write
@anon_protect
def write_manifest_by_digest(namespace_name, repo_name, manifest_ref):
  try:
    manifest = SignedManifest(request.data)
  except ValueError:
    raise ManifestInvalid(detail={'message': 'could not parse manifest'})

  if manifest.digest != manifest_ref:
    raise ManifestInvalid(detail={'message': 'manifest digest mismatch'})

  return _write_manifest(namespace_name, repo_name, manifest)


def _updated_v1_metadata(v1_metadata_json, updated_id_map):
  parsed = json.loads(v1_metadata_json)
  parsed['id'] = updated_id_map[parsed['id']]

  if parsed.get('parent') and parsed['parent'] in updated_id_map:
    parsed['parent'] = updated_id_map[parsed['parent']]

  if parsed.get('container_config', {}).get('Image'):
    existing_image = parsed['container_config']['Image']
    if existing_image in updated_id_map:
      parsed['container_config']['image'] = updated_id_map[existing_image]

  return json.dumps(parsed)


def _write_manifest_itself(namespace_name, repo_name, manifest):
  # Ensure that the manifest is for this repository. If the manifest's namespace is empty, then
  # it is for the library namespace and we need an extra check.
  if (manifest.namespace == '' and features.LIBRARY_SUPPORT and
      namespace_name == app.config['LIBRARY_NAMESPACE']):
    # This is a library manifest. All good.
    pass
  elif manifest.namespace != namespace_name:
    raise NameInvalid()

  if manifest.repo_name != repo_name:
    raise NameInvalid()

  # Ensure that the repository exists.
  repo = model.repository.get_repository(namespace_name, repo_name)
  if repo is None:
    raise NameInvalid()

  # Lookup all the images and their parent images (if any) inside the manifest. This will let us
  # know which V1 images we need to synthesize and which ones are invalid.
  layers = list(manifest.layers)

  docker_image_ids = {mdata.v1_metadata.docker_id for mdata in layers}
  parent_image_ids = {mdata.v1_metadata.parent for mdata in layers
                      if mdata.v1_metadata.parent}
  all_image_ids = list(docker_image_ids | parent_image_ids)

  images_query = model.image.lookup_repository_images(repo, all_image_ids)
  images_map = {image.docker_image_id: image for image in images_query}

  # Lookup the storages associated with each blob in the manifest.
  checksums = list({str(mdata.digest) for mdata in manifest.layers})
  storage_query = model.storage.lookup_repo_storages_by_content_checksum(repo, checksums)
  storage_map = {storage.content_checksum: storage for storage in storage_query}

  # Ensure that we have valid V1 docker IDs. If Docker gives us a V1 layer ID pointing to
  # a storage with a content checksum different from the existing, then we need to rewrite
  # the Docker ID to ensure consistency.
  tag_name = manifest.tag
  has_rewritten_ids = False
  updated_id_map = {}

  # Synthesized image id hash. Can be used to pull a "content addressable" image id out of thin air.
  digest_history = hashlib.sha256()

  for mdata in layers:
    digest_str = str(mdata.digest)
    v1_mdata = mdata.v1_metadata
    working_docker_id = v1_mdata.docker_id

    # Update our digest_history hash for the new layer data.
    digest_history.update(digest_str)
    digest_history.update("@")
    digest_history.update(mdata.v1_metadata_str.encode('utf-8'))
    digest_history.update("|")

    # Ensure that all blobs exist.
    blob_storage = storage_map.get(digest_str)
    if blob_storage is None:
      raise BlobUnknown(detail={'digest': digest_str})

    # Ensure that the V1 image's storage matches the V2 blob. If not, we've found
    # a data inconsistency and need to create a new layer ID for the V1 image, and all images
    # that follow it in the ancestry chain.
    if ((v1_mdata.docker_id in images_map and
         images_map[v1_mdata.docker_id].storage.content_checksum != digest_str) or
        has_rewritten_ids):

      working_docker_id = digest_history.hexdigest()
      logger.warning('Rewriting docker_id %s/%s %s -> %s', namespace_name, repo_name,
                     v1_mdata.docker_id, working_docker_id)
      has_rewritten_ids = True

    # Store the new docker id in the map
    updated_id_map[v1_mdata.docker_id] = working_docker_id

    # Lookup the parent image for the layer, if any.
    parent_image = None
    if v1_mdata.parent is not None:
      parent_image = images_map.get(v1_mdata.parent)
      if parent_image is None:
        msg = 'Parent not found with docker image id {0}'.format(v1_mdata.parent)
        raise ManifestInvalid(detail={'message': msg})

    # Synthesize and store the v1 metadata in the db.
    v1_metadata_json = mdata.v1_metadata_str
    if has_rewritten_ids:
      v1_metadata_json = _updated_v1_metadata(mdata.v1_metadata_str, updated_id_map)

    image = model.image.synthesize_v1_image(repo, blob_storage, working_docker_id,
                                            v1_mdata.created, v1_mdata.comment, v1_mdata.command,
                                            v1_metadata_json, parent_image)
    images_map[v1_mdata.docker_id] = image

  if not layers:
    # The manifest doesn't actually reference any layers!
    raise ManifestInvalid(detail={'message': 'manifest does not reference any layers'})

  # Store the manifest pointing to the tag.
  manifest_digest = manifest.digest
  leaf_layer_id = images_map[layers[-1].v1_metadata.docker_id].docker_image_id

  try:
    tag_manifest, manifest_created = model.tag.store_tag_manifest(namespace_name, repo_name,
                                                                  tag_name, leaf_layer_id,
                                                                  manifest_digest, manifest.bytes)
  except TagAlreadyCreatedException:
    logger.warning('Tag %s was already created under repository %s/%s pointing to image %s',
                   tag_name, namespace_name, repo_name, leaf_layer_id)
    raise TagAlreadyExists()

  if manifest_created:
    for key, value in layers[-1].v1_metadata.labels.iteritems():
      model.label.create_manifest_label(tag_manifest, key, value, 'manifest')

  # Queue all blob manifests for replication.
  # TODO(jschorr): Find a way to optimize this insertion.
  if features.STORAGE_REPLICATION:
    for mdata in layers:
      digest_str = str(mdata.digest)
      blob_storage = storage_map.get(digest_str)
      queue_storage_replication(namespace_name, blob_storage)

  return (repo, tag_name, manifest_digest)


def _write_manifest(namespace_name, repo_name, manifest):
  (repo, tag_name, manifest_digest) = _write_manifest_itself(namespace_name, repo_name, manifest)

  # Spawn the repo_push event.
  event_data = {
    'updated_tags': [tag_name],
  }

  track_and_log('push_repo', repo, tag=tag_name)
  spawn_notification(repo, 'repo_push', event_data)

  response = make_response('OK', 202)
  response.headers['Docker-Content-Digest'] = manifest_digest
  response.headers['Location'] = url_for('v2.fetch_manifest_by_digest',
                                         repository='%s/%s' % (namespace_name, repo_name),
                                         manifest_ref=manifest_digest)
  return response


@v2_bp.route(MANIFEST_DIGEST_ROUTE, methods=['DELETE'])
@parse_repository_name()
@process_registry_jwt_auth(scopes=['pull', 'push'])
@require_repo_write
@anon_protect
def delete_manifest_by_digest(namespace_name, repo_name, manifest_ref):
  """ Delete the manifest specified by the digest. Note: there is no equivalent
      method for deleting by tag name because it is forbidden by the spec.
  """
  try:
    manifest = model.tag.load_manifest_by_digest(namespace_name, repo_name, manifest_ref)
  except model.InvalidManifestException:
    # Without a tag name to reference, we can't make an attempt to generate the manifest
    raise ManifestUnknown()

  # Mark the tag as no longer alive.
  try:
    model.tag.delete_tag(namespace_name, repo_name, manifest.tag.name)
  except model.DataModelException:
    # Tag is not alive.
    raise ManifestUnknown()

  track_and_log('delete_tag', manifest.tag.repository,
                tag=manifest.tag.name, digest=manifest_ref)

  return make_response('', 202)


def _generate_and_store_manifest(namespace_name, repo_name, tag_name):
  # First look up the tag object and its ancestors
  image = model.tag.get_tag_image(namespace_name, repo_name, tag_name, include_storage=True)
  parents = model.image.get_parent_images(namespace_name, repo_name, image)

  # If the manifest is being generated under the library namespace, then we make its namespace
  # empty.
  manifest_namespace = namespace_name
  if features.LIBRARY_SUPPORT and namespace_name == app.config['LIBRARY_NAMESPACE']:
    manifest_namespace = ''

  # Create and populate the manifest builder
  builder = SignedManifestBuilder(manifest_namespace, repo_name, tag_name)

  # Add the leaf layer
  builder.add_layer(image.storage.content_checksum, image.v1_json_metadata)

  for parent in parents:
    builder.add_layer(parent.storage.content_checksum, parent.v1_json_metadata)

  # Sign the manifest with our signing key.
  manifest = builder.build(docker_v2_signing_key)

  # Write the manifest to the DB. If an existing manifest already exists, return the
  # one found.
  try:
    return model.tag.associate_generated_tag_manifest(namespace_name, repo_name, tag_name,
                                                      manifest.digest, manifest.bytes)
  except IntegrityError as ie:
    logger.debug('Got integrity error: %s', ie)
    try:
      return model.tag.load_tag_manifest(namespace_name, repo_name, tag_name)
    except model.InvalidManifestException:
      logger.exception('Exception when generating manifest')
      raise model.DataModelException('Could not load or generate manifest')