quay/data/registry_model/shared.py

# pylint: disable=protected-access
import logging

from collections import defaultdict

from data import database
from data import model
from data.cache import cache_key
from data.registry_model.datatype import FromDictionaryException
from data.registry_model.datatypes import (RepositoryReference, Blob, TorrentInfo, BlobUpload,
                                           LegacyImage, ManifestLayer, DerivedImage)
from image.docker.schema1 import ManifestException, DockerSchema1ManifestBuilder

logger = logging.getLogger(__name__)


class SharedModel:
  """
  SharedModel implements those data model operations for the registry API that are unchanged
  between the old and new data models.
  """
  def lookup_repository(self, namespace_name, repo_name, kind_filter=None):
    """ Looks up and returns a reference to the repository with the given namespace and name,
        or None if none. """
    repo = model.repository.get_repository(namespace_name, repo_name, kind_filter=kind_filter)
    return RepositoryReference.for_repo_obj(repo)

  def is_existing_disabled_namespace(self, namespace_name):
    """ Returns whether the given namespace exists and is disabled. """
    namespace = model.user.get_namespace_user(namespace_name)
    return namespace is not None and not namespace.enabled

  def is_namespace_enabled(self, namespace_name):
    """ Returns whether the given namespace exists and is enabled. """
    namespace = model.user.get_namespace_user(namespace_name)
    return namespace is not None and namespace.enabled

  def get_derived_image_signature(self, derived_image, signer_name):
    """
    Returns the signature associated with the derived image and a specific signer or None if none.
    """
    try:
      derived_storage = database.DerivedStorageForImage.get(id=derived_image._db_id)
    except database.DerivedStorageForImage.DoesNotExist:
      return None

    storage = derived_storage.derivative
    signature_entry = model.storage.lookup_storage_signature(storage, signer_name)
    if signature_entry is None:
      return None

    return signature_entry.signature

  def set_derived_image_signature(self, derived_image, signer_name, signature):
    """
    Sets the calculated signature for the given derived image and signer to that specified.
    """
    try:
      derived_storage = database.DerivedStorageForImage.get(id=derived_image._db_id)
    except database.DerivedStorageForImage.DoesNotExist:
      return None

    storage = derived_storage.derivative
    signature_entry = model.storage.find_or_create_storage_signature(storage, signer_name)
    signature_entry.signature = signature
    signature_entry.uploading = False
    signature_entry.save()

  def delete_derived_image(self, derived_image):
    """
    Deletes a derived image and all of its storage.
    """
    try:
      derived_storage = database.DerivedStorageForImage.get(id=derived_image._db_id)
    except database.DerivedStorageForImage.DoesNotExist:
      return None

    model.image.delete_derived_storage(derived_storage)

  def set_derived_image_size(self, derived_image, compressed_size):
    """
    Sets the compressed size on the given derived image.
    """
    try:
      derived_storage = database.DerivedStorageForImage.get(id=derived_image._db_id)
    except database.DerivedStorageForImage.DoesNotExist:
      return None

    storage_entry = derived_storage.derivative
    storage_entry.image_size = compressed_size
    storage_entry.uploading = False
    storage_entry.save()

  def get_torrent_info(self, blob):
    """
    Returns the torrent information associated with the given blob or None if none.
    """
    try:
      image_storage = database.ImageStorage.get(id=blob._db_id)
    except database.ImageStorage.DoesNotExist:
      return None

    try:
      torrent_info = model.storage.get_torrent_info(image_storage)
    except model.TorrentInfoDoesNotExist:
      return None

    return TorrentInfo.for_torrent_info(torrent_info)

  def set_torrent_info(self, blob, piece_length, pieces):
    """
    Sets the torrent infomation associated with the given blob to that specified.
    """
    try:
      image_storage = database.ImageStorage.get(id=blob._db_id)
    except database.ImageStorage.DoesNotExist:
      return None

    torrent_info = model.storage.save_torrent_info(image_storage, piece_length, pieces)
    return TorrentInfo.for_torrent_info(torrent_info)

  def get_cached_repo_blob(self, model_cache, namespace_name, repo_name, blob_digest):
    """
    Returns the blob in the repository with the given digest if any or None if none.
    Caches the result in the caching system.
    """
    def load_blob():
      repository_ref = self.lookup_repository(namespace_name, repo_name)
      if repository_ref is None:
        return None

      blob_found = self.get_repo_blob_by_digest(repository_ref, blob_digest,
                                                include_placements=True)
      if blob_found is None:
        return None

      return blob_found.asdict()

    blob_cache_key = cache_key.for_repository_blob(namespace_name, repo_name, blob_digest, 2)
    blob_dict = model_cache.retrieve(blob_cache_key, load_blob)

    try:
      return Blob.from_dict(blob_dict) if blob_dict is not None else None
    except FromDictionaryException:
      # The data was stale in some way. Simply reload.
      repository_ref = self.lookup_repository(namespace_name, repo_name)
      if repository_ref is None:
        return None

      return self.get_repo_blob_by_digest(repository_ref, blob_digest, include_placements=True)

  def get_repo_blob_by_digest(self, repository_ref, blob_digest, include_placements=False):
    """
    Returns the blob in the repository with the given digest, if any or None if none. Note that
    there may be multiple records in the same repository for the same blob digest, so the return
    value of this function may change.
    """
    try:
      image_storage = model.blob.get_repository_blob_by_digest(repository_ref._db_id, blob_digest)
    except model.BlobDoesNotExist:
      return None

    assert image_storage.cas_path is not None

    placements = None
    if include_placements:
      placements = list(model.storage.get_storage_locations(image_storage.uuid))

    return Blob.for_image_storage(image_storage,
                                  storage_path=model.storage.get_layer_path(image_storage),
                                  placements=placements)

  def create_blob_upload(self, repository_ref, new_upload_id, location_name, storage_metadata):
    """ Creates a new blob upload and returns a reference. If the blob upload could not be
        created, returns None. """
    repo = model.repository.lookup_repository(repository_ref._db_id)
    if repo is None:
      return None

    try:
      upload_record = model.blob.initiate_upload(repo.namespace_user.username, repo.name,
                                                 new_upload_id, location_name, storage_metadata)
      return BlobUpload.for_upload(upload_record)
    except database.Repository.DoesNotExist:
      return None

  def lookup_blob_upload(self, repository_ref, blob_upload_id):
    """ Looks up the blob upload withn the given ID under the specified repository and returns it
        or None if none.
    """
    upload_record = model.blob.get_blob_upload_by_uuid(blob_upload_id)
    if upload_record is None:
      return None

    return BlobUpload.for_upload(upload_record)

  def update_blob_upload(self, blob_upload, uncompressed_byte_count, piece_hashes, piece_sha_state,
                         storage_metadata, byte_count, chunk_count, sha_state):
    """ Updates the fields of the blob upload to match those given. Returns the updated blob upload
        or None if the record does not exists.
    """
    upload_record = model.blob.get_blob_upload_by_uuid(blob_upload.upload_id)
    if upload_record is None:
      return None

    upload_record.uncompressed_byte_count = uncompressed_byte_count
    upload_record.piece_hashes = piece_hashes
    upload_record.piece_sha_state = piece_sha_state
    upload_record.storage_metadata = storage_metadata
    upload_record.byte_count = byte_count
    upload_record.chunk_count = chunk_count
    upload_record.sha_state = sha_state
    upload_record.save()
    return BlobUpload.for_upload(upload_record)

  def delete_blob_upload(self, blob_upload):
    """ Deletes a blob upload record. """
    upload_record = model.blob.get_blob_upload_by_uuid(blob_upload.upload_id)
    if upload_record is not None:
      upload_record.delete_instance()

  def commit_blob_upload(self, blob_upload, blob_digest_str, blob_expiration_seconds):
    """ Commits the blob upload into a blob and sets an expiration before that blob will be GCed.
    """
    upload_record = model.blob.get_blob_upload_by_uuid(blob_upload.upload_id)
    if upload_record is None:
      return None

    repository = upload_record.repository
    namespace_name = repository.namespace_user.username
    repo_name = repository.name

    # Create the blob and temporarily tag it.
    location_obj = model.storage.get_image_location_for_name(blob_upload.location_name)
    blob_record = model.blob.store_blob_record_and_temp_link(
      namespace_name, repo_name, blob_digest_str, location_obj.id, blob_upload.byte_count,
      blob_expiration_seconds, blob_upload.uncompressed_byte_count)

    # Delete the blob upload.
    upload_record.delete_instance()
    return Blob.for_image_storage(blob_record,
                                  storage_path=model.storage.get_layer_path(blob_record))

  def mount_blob_into_repository(self, blob, target_repository_ref, expiration_sec):
    """
    Mounts the blob from another repository into the specified target repository, and adds an
    expiration before that blob is automatically GCed. This function is useful during push
    operations if an existing blob from another repository is being pushed. Returns False if
    the mounting fails.
    """
    repo = model.repository.lookup_repository(target_repository_ref._db_id)
    if repo is None:
      return False

    namespace_name = repo.namespace_user.username
    repo_name = repo.name

    storage = model.blob.temp_link_blob(namespace_name, repo_name, blob.digest,
                                        expiration_sec)
    return bool(storage)

  def get_legacy_images(self, repository_ref):
    """
    Returns an iterator of all the LegacyImage's defined in the matching repository.
    """
    repo = model.repository.lookup_repository(repository_ref._db_id)
    if repo is None:
      return None

    all_images = model.image.get_repository_images_without_placements(repo)
    all_images_map = {image.id: image for image in all_images}

    all_tags = model.tag.list_repository_tags(repo.namespace_user.username, repo.name)
    tags_by_image_id = defaultdict(list)
    for tag in all_tags:
      tags_by_image_id[tag.image_id].append(tag)

    return [LegacyImage.for_image(image, images_map=all_images_map, tags_map=tags_by_image_id)
            for image in all_images]

  def get_legacy_image(self, repository_ref, docker_image_id, include_parents=False,
                       include_blob=False):
    """
    Returns the matching LegacyImages under the matching repository, if any. If none,
    returns None.
    """
    repo = model.repository.lookup_repository(repository_ref._db_id)
    if repo is None:
      return None

    image = model.image.get_image(repository_ref._db_id, docker_image_id)
    if image is None:
      return None

    parent_images_map = None
    if include_parents:
      parent_images = model.image.get_parent_images(repo.namespace_user.username, repo.name, image)
      parent_images_map = {image.id: image for image in parent_images}

    blob = None
    if include_blob:
      placements = list(model.storage.get_storage_locations(image.storage.uuid))
      blob = Blob.for_image_storage(image.storage,
                                    storage_path=model.storage.get_layer_path(image.storage),
                                    placements=placements)

    return LegacyImage.for_image(image, images_map=parent_images_map, blob=blob)

  def _list_manifest_layers(self, manifest, repo_id, include_placements=False):
    """ Returns an *ordered list* of the layers found in the manifest, starting at the base and
        working towards the leaf, including the associated Blob and its placements (if specified).
        Returns None if the manifest could not be parsed and validated.
    """
    try:
      parsed = manifest.get_parsed_manifest()
    except ManifestException:
      logger.exception('Could not parse and validate manifest `%s`', manifest._db_id)
      return None

    blob_query = model.storage.lookup_repo_storages_by_content_checksum(repo_id, parsed.checksums)
    storage_map = {blob.content_checksum: blob for blob in blob_query}

    manifest_layers = []
    for layer in parsed.layers:
      digest_str = str(layer.digest)
      if digest_str not in storage_map:
        logger.error('Missing digest `%s` for manifest `%s`', layer.digest, manifest._db_id)
        return None

      image_storage = storage_map[digest_str]
      assert image_storage.cas_path is not None

      placements = None
      if include_placements:
        placements = list(model.storage.get_storage_locations(image_storage.uuid))

      blob = Blob.for_image_storage(image_storage,
                                    storage_path=model.storage.get_layer_path(image_storage),
                                    placements=placements)
      manifest_layers.append(ManifestLayer(layer, blob))

    return manifest_layers

  def _build_derived(self, derived, verb, varying_metadata, include_placements):
    if derived is None:
      return None

    derived_storage = derived.derivative
    placements = None
    if include_placements:
      placements = list(model.storage.get_storage_locations(derived_storage.uuid))

    blob = Blob.for_image_storage(derived_storage,
                                  storage_path=model.storage.get_layer_path(derived_storage),
                                  placements=placements)

    return DerivedImage.for_derived_storage(derived, verb, varying_metadata, blob)

  def _build_manifest_for_legacy_image(self, tag_name, legacy_image_row):
    import features

    from app import app, docker_v2_signing_key

    repo = legacy_image_row.repository
    namespace_name = repo.namespace_user.username
    repo_name = repo.name

    # Find the v1 metadata for this image and its parents.
    parents = model.image.get_parent_images(namespace_name, repo_name, legacy_image_row)

    # 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 = DockerSchema1ManifestBuilder(manifest_namespace, repo_name, tag_name)

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

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

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