quay/endpoints/v2/manifest.py

import logging

from functools import wraps

from flask import request, url_for, Response

import features

from app import app, metric_queue, storage
from auth.registry_jwt_auth import process_registry_jwt_auth
from digest import digest_tools
from data.registry_model import registry_model
from endpoints.decorators import anon_protect, parse_repository_name
from endpoints.v2 import v2_bp, require_repo_read, require_repo_write
from endpoints.v2.errors import (ManifestInvalid, ManifestUnknown, NameInvalid, TagExpired,
                                 NameUnknown)
from image.docker import ManifestException
from image.docker.schema1 import DOCKER_SCHEMA1_MANIFEST_CONTENT_TYPE
from image.docker.schema2 import DOCKER_SCHEMA2_CONTENT_TYPES, OCI_CONTENT_TYPES
from image.docker.schemas import parse_manifest_from_bytes
from notifications import spawn_notification
from util.audit import track_and_log
from util.names import VALID_TAG_PATTERN
from util.registry.replication import queue_replication_batch


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)


@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):
  repository_ref = registry_model.lookup_repository(namespace_name, repo_name)
  if repository_ref is None:
    raise NameUnknown()

  tag = registry_model.get_repo_tag(repository_ref, manifest_ref)
  if tag is None:
    if registry_model.has_expired_tag(repository_ref, manifest_ref):
      logger.debug('Found expired tag %s for repository %s/%s', manifest_ref, namespace_name,
                   repo_name)
      msg = 'Tag %s was deleted or has expired. To pull, revive via time machine' % manifest_ref
      raise TagExpired(msg)

    raise ManifestUnknown()

  manifest = registry_model.get_manifest_for_tag(tag, backfill_if_necessary=True)
  if manifest is None:
    # Something went wrong.
    raise ManifestInvalid()

  try:
    parsed = manifest.get_parsed_manifest()
  except ManifestException:
    logger.exception('Got exception when trying to parse manifest `%s`', manifest_ref)
    raise ManifestInvalid()

  supported = _rewrite_schema_if_necessary(namespace_name, repo_name, manifest_ref, manifest,
                                           parsed)
  if supported is None:
    raise ManifestUnknown()

  track_and_log('pull_repo', repository_ref, analytics_name='pull_repo_100x', analytics_sample=0.01,
                tag=manifest_ref)
  metric_queue.repository_pull.Inc(labelvalues=[namespace_name, repo_name, 'v2', True])

  return Response(
    supported.bytes,
    status=200,
    headers={
      'Content-Type': supported.media_type,
      'Docker-Content-Digest': supported.digest,
    },
  )


@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):
  repository_ref = registry_model.lookup_repository(namespace_name, repo_name)
  if repository_ref is None:
    raise NameUnknown()

  manifest = registry_model.lookup_manifest_by_digest(repository_ref, manifest_ref)
  if manifest is None:
    raise ManifestUnknown()

  try:
    parsed = manifest.get_parsed_manifest()
  except ManifestException:
    logger.exception('Got exception when trying to parse manifest `%s`', manifest_ref)
    raise ManifestInvalid()

  supported = _rewrite_schema_if_necessary(namespace_name, repo_name, '$digest', manifest,
                                           parsed)
  if supported is None:
    raise ManifestUnknown()

  track_and_log('pull_repo', repository_ref, manifest_digest=manifest_ref)
  metric_queue.repository_pull.Inc(labelvalues=[namespace_name, repo_name, 'v2', True])

  return Response(supported.bytes, status=200, headers={
    'Content-Type': supported.media_type,
    'Docker-Content-Digest': supported.digest,
  })


def _rewrite_schema_if_necessary(namespace_name, repo_name, tag_name, manifest, parsed):
  # As per the Docker protocol, if the manifest is not schema version 1 and the manifest's
  # media type is not in the Accept header, we return a schema 1 version of the manifest for
  # the amd64+linux platform, if any, or None if none.
  # See: https://docs.docker.com/registry/spec/manifest-v2-2
  mimetypes = [mimetype for mimetype, _ in request.accept_mimetypes]
  if parsed.media_type in mimetypes:
    return parsed

  converted = registry_model.convert_manifest(manifest, namespace_name, repo_name, tag_name,
                                              mimetypes, storage)
  if converted is not None:
    return converted

  # For back-compat, we always default to schema 1 if the manifest could not be converted.
  return registry_model.get_schema1_parsed_manifest(manifest, namespace_name, repo_name, tag_name,
                                                    storage)


def _reject_manifest2_schema2(func):
  @wraps(func)
  def wrapped(*args, **kwargs):
    namespace_name = kwargs['namespace_name']
    if registry_model.supports_schema2(namespace_name):
      return func(*args, **kwargs)

    if _doesnt_accept_schema_v1() or \
      request.content_type in DOCKER_SCHEMA2_CONTENT_TYPES | OCI_CONTENT_TYPES:
      raise ManifestInvalid(detail={'message': 'manifest schema version not supported'},
                            http_status_code=415)
    return func(*args, **kwargs)

  return wrapped


def _doesnt_accept_schema_v1():
  # If the client doesn't specify anything, still give them Schema v1.
  return len(request.accept_mimetypes) != 0 and \
         DOCKER_SCHEMA1_MANIFEST_CONTENT_TYPE not in request.accept_mimetypes


@v2_bp.route(MANIFEST_TAGNAME_ROUTE, methods=['PUT'])
@parse_repository_name()
@_reject_manifest2_schema2
@process_registry_jwt_auth(scopes=['pull', 'push'])
@require_repo_write
@anon_protect
def write_manifest_by_tagname(namespace_name, repo_name, manifest_ref):
  parsed = _parse_manifest()
  return _write_manifest_and_log(namespace_name, repo_name, manifest_ref, parsed)


@v2_bp.route(MANIFEST_DIGEST_ROUTE, methods=['PUT'])
@parse_repository_name()
@_reject_manifest2_schema2
@process_registry_jwt_auth(scopes=['pull', 'push'])
@require_repo_write
@anon_protect
def write_manifest_by_digest(namespace_name, repo_name, manifest_ref):
  parsed = _parse_manifest()
  if parsed.digest != manifest_ref:
    raise ManifestInvalid(detail={'message': 'manifest digest mismatch'})

  if parsed.schema_version != 2:
    return _write_manifest_and_log(namespace_name, repo_name, parsed.tag, parsed)

  # If the manifest is schema version 2, then this cannot be a normal tag-based push, as the
  # manifest does not contain the tag and this call was not given a tag name. Instead, we write the
  # manifest with a temporary tag, as it is being pushed as part of a call for a manifest list.
  repository_ref = registry_model.lookup_repository(namespace_name, repo_name)
  if repository_ref is None:
    raise NameUnknown()

  expiration_sec = app.config['PUSH_TEMP_TAG_EXPIRATION_SEC']
  manifest = registry_model.create_manifest_with_temp_tag(repository_ref, parsed, expiration_sec,
                                                          storage)
  if manifest is None:
    raise ManifestInvalid()

  return Response(
    'OK',
    status=202,
    headers={
      'Docker-Content-Digest': manifest.digest,
      'Location':
        url_for('v2.fetch_manifest_by_digest',
                repository='%s/%s' % (namespace_name, repo_name),
                manifest_ref=manifest.digest),
    },
  )


def _parse_manifest():
  content_type = request.content_type or DOCKER_SCHEMA1_MANIFEST_CONTENT_TYPE
  if content_type == 'application/json':
    # For back-compat.
    content_type = DOCKER_SCHEMA1_MANIFEST_CONTENT_TYPE

  try:
    return parse_manifest_from_bytes(request.data, content_type)
  except ManifestException as me:
    logger.exception("failed to parse manifest when writing by tagname")
    raise ManifestInvalid(detail={'message': 'failed to parse manifest: %s' % me.message})


@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.
  """
  repository_ref = registry_model.lookup_repository(namespace_name, repo_name)
  if repository_ref is None:
    raise NameUnknown()

  manifest = registry_model.lookup_manifest_by_digest(repository_ref, manifest_ref)
  if manifest is None:
    raise ManifestUnknown()

  tags = registry_model.delete_tags_for_manifest(manifest)
  if not tags:
    raise ManifestUnknown()

  for tag in tags:
    track_and_log('delete_tag', repository_ref, tag=tag.name, digest=manifest_ref)

  return Response(status=202)


def _write_manifest_and_log(namespace_name, repo_name, tag_name, manifest_impl):
  repository_ref, manifest, tag = _write_manifest(namespace_name, repo_name, tag_name,
                                                  manifest_impl)

  # Queue all blob manifests for replication.
  if features.STORAGE_REPLICATION:
    blobs = registry_model.get_manifest_local_blobs(manifest)
    if blobs is None:
      logger.error('Could not lookup blobs for manifest `%s`', manifest.digest)
    else:
      with queue_replication_batch(namespace_name) as queue_storage_replication:
        for blob_digest in blobs:
          queue_storage_replication(blob_digest)

  track_and_log('push_repo', repository_ref, tag=tag_name)
  spawn_notification(repository_ref, 'repo_push', {'updated_tags': [tag_name]})
  metric_queue.repository_push.Inc(labelvalues=[namespace_name, repo_name, 'v2', True])

  return Response(
    'OK',
    status=202,
    headers={
      'Docker-Content-Digest': manifest.digest,
      'Location':
        url_for('v2.fetch_manifest_by_digest',
                repository='%s/%s' % (namespace_name, repo_name),
                manifest_ref=manifest.digest),
    },
  )


def _write_manifest(namespace_name, repo_name, tag_name, manifest_impl):
  # NOTE: These extra checks are needed for schema version 1 because the manifests
  # contain the repo namespace, name and tag name.
  if manifest_impl.schema_version == 1:
    if (manifest_impl.namespace == '' and features.LIBRARY_SUPPORT and
        namespace_name == app.config['LIBRARY_NAMESPACE']):
      pass
    elif manifest_impl.namespace != namespace_name:
      raise NameInvalid()

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

    if not manifest_impl.layers:
      raise ManifestInvalid(detail={'message': 'manifest does not reference any layers'})

  # Ensure that the repository exists.
  repository_ref = registry_model.lookup_repository(namespace_name, repo_name)
  if repository_ref is None:
    raise NameUnknown()

  # Create the manifest(s) and retarget the tag to point to it.
  manifest, tag = registry_model.create_manifest_and_retarget_tag(repository_ref, manifest_impl,
                                                                  tag_name, storage)
  if manifest is None:
    raise ManifestInvalid()

  return repository_ref, manifest, tag