quay/data/registry_model/registry_pre_oci_model.py

# pylint: disable=protected-access
import logging

from collections import defaultdict
from contextlib import contextmanager

from peewee import IntegrityError

from data import database
from data import model
from data.database import db_transaction
from data.registry_model.interface import RegistryDataInterface
from data.registry_model.datatypes import (Tag, Manifest, LegacyImage, Label,
                                           SecurityScanStatus, ManifestLayer, Blob, DerivedImage)
from data.registry_model.shared import SharedModel
from data.registry_model.label_handlers import apply_label_to_manifest
from image.docker.schema1 import (DockerSchema1ManifestBuilder, ManifestException,
                                  DockerSchema1Manifest)
from util.validation import is_json


logger = logging.getLogger(__name__)


class PreOCIModel(SharedModel, RegistryDataInterface):
  """
  PreOCIModel implements the data model for the registry API using a database schema
  before it was changed to support the OCI specification.
  """

  def find_matching_tag(self, repository_ref, tag_names):
    """ Finds an alive tag in the repository matching one of the given tag names and returns it
        or None if none.
    """
    found_tag = model.tag.find_matching_tag(repository_ref._db_id, tag_names)
    assert found_tag is None or not found_tag.hidden
    return Tag.for_repository_tag(found_tag)

  def get_most_recent_tag(self, repository_ref):
    """ Returns the most recently pushed alive tag in the repository, if any. If none, returns
        None.
    """
    found_tag = model.tag.get_most_recent_tag(repository_ref._db_id)
    assert found_tag is None or not found_tag.hidden
    return Tag.for_repository_tag(found_tag)

  def get_manifest_for_tag(self, tag, backfill_if_necessary=False):
    """ Returns the manifest associated with the given tag. """
    try:
      tag_manifest = database.TagManifest.get(tag_id=tag._db_id)
    except database.TagManifest.DoesNotExist:
      if backfill_if_necessary:
        return self.backfill_manifest_for_tag(tag)

      return

    return Manifest.for_tag_manifest(tag_manifest)

  def lookup_manifest_by_digest(self, repository_ref, manifest_digest, allow_dead=False,
                                include_legacy_image=False):
    """ Looks up the manifest with the given digest under the given repository and returns it
        or None if none. """
    repo = model.repository.lookup_repository(repository_ref._db_id)
    if repo is None:
      return None

    try:
      tag_manifest = model.tag.load_manifest_by_digest(repo.namespace_user.username,
                                                       repo.name,
                                                       manifest_digest,
                                                       allow_dead=allow_dead)
    except model.tag.InvalidManifestException:
      return None

    legacy_image = None
    if include_legacy_image:
      legacy_image = self.get_legacy_image(repository_ref, tag_manifest.tag.image.docker_image_id,
                                           include_parents=True)

    return Manifest.for_tag_manifest(tag_manifest, legacy_image)

  def create_manifest_and_retarget_tag(self, repository_ref, manifest_interface_instance, tag_name):
    """ Creates a manifest in a repository, adding all of the necessary data in the model.

        The `manifest_interface_instance` parameter must be an instance of the manifest
        interface as returned by the image/docker package.

        Note that all blobs referenced by the manifest must exist under the repository or this
        method will fail and return None.

        Returns a reference to the (created manifest, tag) or (None, None) on error.
    """
    # NOTE: Only Schema1 is supported by the pre_oci_model.
    assert isinstance(manifest_interface_instance, DockerSchema1Manifest)
    if not manifest_interface_instance.layers:
      return None, None

    # Ensure all the blobs in the manifest exist.
    digests = manifest_interface_instance.checksums
    query = model.storage.lookup_repo_storages_by_content_checksum(repository_ref._db_id, digests)
    blob_map = {s.content_checksum: s for s in query}
    for layer in manifest_interface_instance.layers:
      digest_str = str(layer.digest)
      if digest_str not in blob_map:
        return None, None

    # 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.
    docker_image_ids = list(manifest_interface_instance.legacy_image_ids)
    images_query = model.image.lookup_repository_images(repository_ref._db_id, docker_image_ids)
    image_storage_map = {i.docker_image_id: i.storage for i in images_query}

    # Rewrite any v1 image IDs that do not match the checksum in the database.
    try:
      rewritten_images = manifest_interface_instance.rewrite_invalid_image_ids(image_storage_map)
      rewritten_images = list(rewritten_images)
      parent_image_map = {}

      for rewritten_image in rewritten_images:
        if not rewritten_image.image_id in image_storage_map:
          parent_image = None
          if rewritten_image.parent_image_id:
            parent_image = parent_image_map.get(rewritten_image.parent_image_id)
            if parent_image is None:
              parent_image = model.image.get_image(repository_ref._db_id,
                                                   rewritten_image.parent_image_id)
              if parent_image is None:
                return None, None

          synthesized = model.image.synthesize_v1_image(
            repository_ref._db_id,
            blob_map[rewritten_image.content_checksum].id,
            blob_map[rewritten_image.content_checksum].image_size,
            rewritten_image.image_id,
            rewritten_image.created,
            rewritten_image.comment,
            rewritten_image.command,
            rewritten_image.compat_json,
            parent_image,
          )

          parent_image_map[rewritten_image.image_id] = synthesized
    except ManifestException:
      logger.exception("exception when rewriting v1 metadata")
      return None, None

    # Store the manifest pointing to the tag.
    leaf_layer_id = rewritten_images[-1].image_id
    tag_manifest, newly_created = model.tag.store_tag_manifest_for_repo(repository_ref._db_id,
                                                                        tag_name,
                                                                        manifest_interface_instance,
                                                                        leaf_layer_id,
                                                                        blob_map)

    manifest = Manifest.for_tag_manifest(tag_manifest)

    # Save the labels on the manifest.
    if newly_created:
      with self.batch_create_manifest_labels(manifest) as add_label:
        for key, value in manifest_interface_instance.layers[-1].v1_metadata.labels.iteritems():
          media_type = 'application/json' if is_json(value) else 'text/plain'
          add_label(key, value, 'manifest', media_type)

    return manifest, Tag.for_repository_tag(tag_manifest.tag)

  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 create_manifest_label(self, manifest, key, value, source_type_name, media_type_name=None):
    """ Creates a label on the manifest with the given key and value. """
    try:
      tag_manifest = database.TagManifest.get(id=manifest._db_id)
    except database.TagManifest.DoesNotExist:
      return None

    label_data = dict(key=key, value=value, source_type_name=source_type_name,
                      media_type_name=media_type_name)

    with db_transaction():
      # Create the label itself.
      label = model.label.create_manifest_label(tag_manifest, key, value, source_type_name,
                                                media_type_name)

      # Apply any changes to the manifest that the label prescribes.
      apply_label_to_manifest(label_data, manifest, self)

    return Label.for_label(label)

  @contextmanager
  def batch_create_manifest_labels(self, manifest):
    """ Returns a context manager for batch creation of labels on a manifest.

        Can raise InvalidLabelKeyException or InvalidMediaTypeException depending
        on the validation errors.
    """
    try:
      tag_manifest = database.TagManifest.get(id=manifest._db_id)
    except database.TagManifest.DoesNotExist:
      yield None
      return

    labels_to_add = []
    def add_label(key, value, source_type_name, media_type_name=None):
      labels_to_add.append(dict(key=key, value=value, source_type_name=source_type_name,
                                media_type_name=media_type_name))

    yield add_label

    # TODO: make this truly batch once we've fully transitioned to V2_2 and no longer need
    # the mapping tables.
    for label in labels_to_add:
      with db_transaction():
        # Create the label itself.
        model.label.create_manifest_label(tag_manifest, **label)

        # Apply any changes to the manifest that the label prescribes.
        apply_label_to_manifest(label, manifest, self)

  def list_manifest_labels(self, manifest, key_prefix=None):
    """ Returns all labels found on the manifest. If specified, the key_prefix will filter the
        labels returned to those keys that start with the given prefix.
    """
    labels = model.label.list_manifest_labels(manifest._db_id, prefix_filter=key_prefix)
    return [Label.for_label(l) for l in labels]

  def get_manifest_label(self, manifest, label_uuid):
    """ Returns the label with the specified UUID on the manifest or None if none. """
    return Label.for_label(model.label.get_manifest_label(label_uuid, manifest._db_id))

  def delete_manifest_label(self, manifest, label_uuid):
    """ Delete the label with the specified UUID on the manifest. Returns the label deleted
        or None if none.
    """
    return Label.for_label(model.label.delete_manifest_label(label_uuid, manifest._db_id))

  def list_repository_tags(self, repository_ref, include_legacy_images=False,
                           start_pagination_id=None,
                           limit=None):
    """
    Returns a list of all the active tags in the repository. Note that this can be a *heavy*
    operation on repositories with a lot of tags, and should be avoided for more targetted
    operations wherever possible.
    """
    # NOTE: include_legacy_images isn't used here because `list_active_repo_tags` includes the
    # information already, so we might as well just use it. However, the new model classes will
    # *not* include it by default, so we make it a parameter now.
    tags = model.tag.list_active_repo_tags(repository_ref._db_id, start_pagination_id, limit)
    return [Tag.for_repository_tag(tag,
                                   legacy_image=LegacyImage.for_image(tag.image),
                                   manifest_digest=(tag.tagmanifest.digest
                                                    if hasattr(tag, 'tagmanifest')
                                                    else None))
            for tag in tags]

  def list_repository_tag_history(self, repository_ref, page=1, size=100, specific_tag_name=None,
                                  active_tags_only=False):
    """
    Returns the history of all tags in the repository (unless filtered). This includes tags that
    have been made in-active due to newer versions of those tags coming into service.
    """
    tags, manifest_map, has_more = model.tag.list_repository_tag_history(repository_ref._db_id,
                                                                         page, size,
                                                                         specific_tag_name,
                                                                         active_tags_only)
    return [Tag.for_repository_tag(tag, manifest_map.get(tag.id),
                                   legacy_image=LegacyImage.for_image(tag.image))
            for tag in tags], has_more

  def has_expired_tag(self, repository_ref, tag_name):
    """
    Returns true if and only if the repository contains a tag with the given name that is expired.
    """
    try:
      model.tag.get_expired_tag_in_repo(repository_ref._db_id, tag_name)
      return True
    except database.RepositoryTag.DoesNotExist:
      return False

  def get_repo_tag(self, repository_ref, tag_name, include_legacy_image=False):
    """
    Returns the latest, *active* tag found in the repository, with the matching name
    or None if none.
    """
    assert isinstance(tag_name, basestring)
    tag = model.tag.get_active_tag_for_repo(repository_ref._db_id, tag_name)
    if tag is None:
      return None

    legacy_image = LegacyImage.for_image(tag.image) if include_legacy_image else None
    tag_manifest = model.tag.get_tag_manifest(tag)
    manifest_digest = tag_manifest.digest if tag_manifest else None
    return Tag.for_repository_tag(tag, legacy_image=legacy_image, manifest_digest=manifest_digest)

  def retarget_tag(self, repository_ref, tag_name, manifest_or_legacy_image,
                   is_reversion=False):
    """
    Creates, updates or moves a tag to a new entry in history, pointing to the manifest or
    legacy image specified. If is_reversion is set to True, this operation is considered a
    reversion over a previous tag move operation. Returns the updated Tag or None on error.
    """
    # TODO: unify this.
    if not is_reversion:
      if isinstance(manifest_or_legacy_image, Manifest):
        raise NotImplementedError('Not yet implemented')
      else:
        model.tag.create_or_update_tag_for_repo(repository_ref._db_id, tag_name,
                                                manifest_or_legacy_image.docker_image_id)
    else:
      if isinstance(manifest_or_legacy_image, Manifest):
        model.tag.restore_tag_to_manifest(repository_ref._db_id, tag_name,
                                          manifest_or_legacy_image.digest)
      else:
        model.tag.restore_tag_to_image(repository_ref._db_id, tag_name,
                                       manifest_or_legacy_image.docker_image_id)

    # Generate a manifest for the tag, if necessary.
    tag = self.get_repo_tag(repository_ref, tag_name, include_legacy_image=True)
    if tag is None:
      return None

    self.backfill_manifest_for_tag(tag)
    return tag

  def delete_tag(self, repository_ref, tag_name):
    """
    Deletes the latest, *active* tag with the given name in the repository.
    """
    repo = model.repository.lookup_repository(repository_ref._db_id)
    if repo is None:
      return None

    deleted_tag = model.tag.delete_tag(repo.namespace_user.username, repo.name, tag_name)
    return Tag.for_repository_tag(deleted_tag)

  def delete_tags_for_manifest(self, manifest):
    """
    Deletes all tags pointing to the given manifest, making the manifest inaccessible for pulling.
    Returns the tags deleted, if any. Returns None on error.
    """
    try:
      tagmanifest = database.TagManifest.get(id=manifest._db_id)
    except database.TagManifest.DoesNotExist:
      return None

    namespace_name = tagmanifest.tag.repository.namespace_user.username
    repo_name = tagmanifest.tag.repository.name
    tags = model.tag.delete_manifest_by_digest(namespace_name, repo_name, manifest.digest)
    return [Tag.for_repository_tag(tag) for tag in tags]

  def change_repository_tag_expiration(self, tag, expiration_date):
    """ Sets the expiration date of the tag under the matching repository to that given. If the
        expiration date is None, then the tag will not expire. Returns a tuple of the previous
        expiration timestamp in seconds (if any), and whether the operation succeeded.
    """
    try:
      tag_obj = database.RepositoryTag.get(id=tag._db_id)
    except database.RepositoryTag.DoesNotExist:
      return (None, False)

    return model.tag.change_tag_expiration(tag_obj, expiration_date)

  def get_legacy_images_owned_by_tag(self, tag):
    """ Returns all legacy images *solely owned and used* by the given tag. """
    try:
      tag_obj = database.RepositoryTag.get(id=tag._db_id)
    except database.RepositoryTag.DoesNotExist:
      return None

    # Collect the IDs of all images that the tag uses.
    tag_image_ids = set()
    tag_image_ids.add(tag_obj.image.id)
    tag_image_ids.update(tag_obj.image.ancestor_id_list())

    # Remove any images shared by other tags.
    for current_tag in model.tag.list_active_repo_tags(tag_obj.repository_id):
      if current_tag == tag_obj:
        continue

      tag_image_ids.discard(current_tag.image.id)
      tag_image_ids = tag_image_ids.difference(current_tag.image.ancestor_id_list())
      if not tag_image_ids:
        return []

    if not tag_image_ids:
      return []

    # Load the images we need to return.
    images = database.Image.select().where(database.Image.id << list(tag_image_ids))
    all_image_ids = set()
    for image in images:
      all_image_ids.add(image.id)
      all_image_ids.update(image.ancestor_id_list())

    # Build a map of all the images and their parents.
    images_map = {}
    all_images = database.Image.select().where(database.Image.id << list(all_image_ids))
    for image in all_images:
      images_map[image.id] = image

    return [LegacyImage.for_image(image, images_map=images_map) for image in images]

  def get_security_status(self, manifest_or_legacy_image):
    """ Returns the security status for the given manifest or legacy image or None if none. """
    image = None

    if isinstance(manifest_or_legacy_image, Manifest):
      try:
        tag_manifest = database.TagManifest.get(id=manifest_or_legacy_image._db_id)
        image = tag_manifest.tag.image
      except database.TagManifest.DoesNotExist:
        return None
    else:
      try:
        image = database.Image.get(id=manifest_or_legacy_image._db_id)
      except database.Image.DoesNotExist:
        return None

    if image.security_indexed_engine is not None and image.security_indexed_engine >= 0:
      return SecurityScanStatus.SCANNED if image.security_indexed else SecurityScanStatus.FAILED

    return SecurityScanStatus.QUEUED

  def backfill_manifest_for_tag(self, tag):
    """ Backfills a manifest for the V1 tag specified.
        If a manifest already exists for the tag, returns that manifest.

        NOTE: This method will only be necessary until we've completed the backfill, at which point
        it should be removed.
    """
    import features

    from app import app, docker_v2_signing_key

    # Ensure that there isn't already a manifest for the tag.
    tag_manifest = model.tag.get_tag_manifest(tag._db_id)
    if tag_manifest is not None:
      return Manifest.for_tag_manifest(tag_manifest)

    # Create the manifest.
    try:
      tag_obj = database.RepositoryTag.get(id=tag._db_id)
    except database.RepositoryTag.DoesNotExist:
      return None

    assert not tag_obj.hidden

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

    # Find the v1 metadata for this image and its parents.
    repo_image = tag_obj.image
    parents = model.image.get_parent_images(namespace_name, repo_name, repo_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 = DockerSchema1ManifestBuilder(manifest_namespace, repo_name, tag.name)

    # Add the leaf layer
    builder.add_layer(repo_image.storage.content_checksum, repo_image.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.
    manifest = builder.build(docker_v2_signing_key)

    # Write the manifest to the DB.
    blob_query = model.storage.lookup_repo_storages_by_content_checksum(repo,
                                                                        manifest.checksums)

    storage_map = {blob.content_checksum: blob.id for blob in blob_query}
    try:
      tag_manifest, _ = model.tag.associate_generated_tag_manifest(namespace_name, repo_name,
                                                                   tag.name, manifest, storage_map)
    except IntegrityError:
      tag_manifest = model.tag.get_tag_manifest(tag_obj)

    return Manifest.for_tag_manifest(tag_manifest)

  def list_manifest_layers(self, manifest, 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

    try:
      tag_manifest = database.TagManifest.get(id=manifest._db_id)
    except database.TagManifest.DoesNotExist:
      logger.exception('Could not find tag manifest for manifest `%s`', manifest._db_id)
      return None

    repo = tag_manifest.tag.repository
    blob_query = model.storage.lookup_repo_storages_by_content_checksum(repo, 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 lookup_derived_image(self, manifest, verb, varying_metadata=None, include_placements=False):
    """
    Looks up the derived image for the given manifest, verb and optional varying metadata and
    returns it or None if none.
    """
    try:
      tag_manifest = database.TagManifest.get(id=manifest._db_id)
    except database.TagManifest.DoesNotExist:
      logger.exception('Could not find tag manifest for manifest `%s`', manifest._db_id)
      return None

    repo_image = tag_manifest.tag.image
    derived = model.image.find_derived_storage_for_image(repo_image, verb, varying_metadata)
    return self._build_derived(derived, verb, varying_metadata, include_placements)

  def lookup_or_create_derived_image(self, manifest, verb, storage_location, varying_metadata=None,
                                     include_placements=False):
    """
    Looks up the derived image for the given maniest, verb and optional varying metadata
    and returns it. If none exists, a new derived image is created.
    """
    try:
      tag_manifest = database.TagManifest.get(id=manifest._db_id)
    except database.TagManifest.DoesNotExist:
      logger.exception('Could not find tag manifest for manifest `%s`', manifest._db_id)
      return None

    repo_image = tag_manifest.tag.image
    derived = model.image.find_or_create_derived_storage(repo_image, verb, storage_location,
                                                         varying_metadata)
    return self._build_derived(derived, verb, varying_metadata, include_placements)

  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 set_tags_expiration_for_manifest(self, manifest, expiration_sec):
    """
    Sets the expiration on all tags that point to the given manifest to that specified.
    """
    try:
      tag_manifest = database.TagManifest.get(id=manifest._db_id)
    except database.TagManifest.DoesNotExist:
      return

    model.tag.set_tag_expiration_for_manifest(tag_manifest, expiration_sec)

pre_oci_model = PreOCIModel()