This repository has been archived on 2020-03-24. You can view files and clone it, but cannot push or open issues or pull requests.
quay/data/interfaces/v1.py
Joseph Schorr b5bb76cdea Optimize repository search by changing our lookup strategy
Previous to this change, repositories were looked up unfiltered in six different queries, and then filtered using the permissions model, which issued a query per repository found, making search incredibly slow. Instead, we now lookup a chunk of repositories unfiltered and then filter them via a single query to the database. By layering the filtering on top of the lookup, each as queries, we can minimize the number of queries necessary, without (at the same time) using a super expensive join.

Other changes:
- Remove the 5 page pre-lookup on V1 search and simply return that there is one more page available, until there isn't. While technically not correct, it is much more efficient, and no one should be using pagination with V1 search anyway.
- Remove the lookup for repos without entries in the RAC table. Instead, we now add a new RAC entry when the repository is created for *the day before*, with count 0, so that it is immediately searchable
- Remove lookup of results with a matching namespace; these aren't very relevant anyway, and it overly complicates sorting
2017-03-09 19:47:55 -05:00

402 lines
14 KiB
Python

from abc import ABCMeta, abstractmethod
from collections import namedtuple
from six import add_metaclass
from app import app, storage as store
from data import model
from data.model import db_transaction
from util.morecollections import AttrDict
class Repository(namedtuple('Repository', ['id', 'name', 'namespace_name', 'description',
'is_public'])):
"""
Repository represents a namespaced collection of tags.
"""
@add_metaclass(ABCMeta)
class DockerRegistryV1DataInterface(object):
"""
Interface that represents all data store interactions required by a Docker Registry v1.
"""
@abstractmethod
def placement_locations_and_path_docker_v1(self, namespace_name, repo_name, image_id):
"""
Returns all the placements for the image with the given V1 Docker ID, found under the given
repository or None if no image was found.
"""
pass
@abstractmethod
def docker_v1_metadata(self, namespace_name, repo_name, image_id):
"""
Returns various pieces of metadata associated with an image with the given V1 Docker ID,
including the checksum and its V1 JSON metadata.
"""
pass
@abstractmethod
def update_docker_v1_metadata(self, namespace_name, repo_name, image_id, created_date_str,
comment, command, compat_json, parent_image_id=None):
"""
Updates various pieces of V1 metadata associated with a particular image.
"""
pass
@abstractmethod
def storage_exists(self, namespace_name, repo_name, image_id):
"""
Returns whether storage already exists for the image with the V1 Docker ID under the given
repository.
"""
pass
@abstractmethod
def store_docker_v1_checksums(self, namespace_name, repo_name, image_id, checksum,
content_checksum):
"""
Stores the various V1 checksums for the image with the V1 Docker ID.
"""
pass
@abstractmethod
def is_image_uploading(self, namespace_name, repo_name, image_id):
"""
Returns whether the image with the V1 Docker ID is currently marked as uploading.
"""
pass
@abstractmethod
def update_image_uploading(self, namespace_name, repo_name, image_id, is_uploading):
"""
Marks the image with the V1 Docker ID with the given uploading status.
"""
pass
@abstractmethod
def update_image_sizes(self, namespace_name, repo_name, image_id, size, uncompressed_size):
"""
Updates the sizing information for the image with the given V1 Docker ID.
"""
pass
@abstractmethod
def get_image_size(self, namespace_name, repo_name, image_id):
"""
Returns the wire size of the image with the given Docker V1 ID.
"""
pass
@abstractmethod
def create_bittorrent_pieces(self, namespace_name, repo_name, image_id, pieces_bytes):
"""
Saves the BitTorrent piece hashes for the image with the given Docker V1 ID.
"""
pass
@abstractmethod
def image_ancestry(self, namespace_name, repo_name, image_id):
"""
Returns a list containing the full ancestry of Docker V1 IDs, in order, for the image with the
given Docker V1 ID.
"""
pass
@abstractmethod
def repository_exists(self, namespace_name, repo_name):
"""
Returns whether the repository with the given name and namespace exists.
"""
pass
@abstractmethod
def create_or_link_image(self, username, namespace_name, repo_name, image_id, storage_location):
"""
Adds the given image to the given repository, by either linking to an existing image visible to
the user with the given username, or creating a new one if no existing image matches.
"""
pass
@abstractmethod
def create_temp_hidden_tag(self, namespace_name, repo_name, image_id, expiration):
"""
Creates a hidden tag under the matching namespace pointing to the image with the given V1 Docker
ID.
"""
pass
@abstractmethod
def list_tags(self, namespace_name, repo_name):
"""
Returns all the tags defined in the repository with the given namespace and name.
"""
pass
@abstractmethod
def create_or_update_tag(self, namespace_name, repo_name, image_id, tag_name):
"""
Creates or updates a tag under the matching repository to point to the image with the given
Docker V1 ID.
"""
pass
@abstractmethod
def find_image_id_by_tag(self, namespace_name, repo_name, tag_name):
"""
Returns the Docker V1 image ID for the HEAD image for the tag with the given name under the
matching repository, or None if none.
"""
pass
@abstractmethod
def delete_tag(self, namespace_name, repo_name, tag_name):
"""
Deletes the given tag from the given repository.
"""
pass
@abstractmethod
def load_token(self, token):
"""
Loads the data associated with the given (deprecated) access token, and, if
found returns True.
"""
pass
@abstractmethod
def verify_robot(self, username, token):
"""
Returns True if the given robot username and token match an existing robot
account.
"""
pass
@abstractmethod
def change_user_password(self, user, new_password):
"""
Changes the password associated with the given user.
"""
pass
@abstractmethod
def get_repository(self, namespace_name, repo_name):
"""
Returns the repository with the given name under the given namespace or None
if none.
"""
pass
@abstractmethod
def create_repository(self, namespace_name, repo_name, user=None):
"""
Creates a new repository under the given namespace with the given name, for
the given user.
"""
pass
@abstractmethod
def repository_is_public(self, namespace_name, repo_name):
"""
Returns whether the repository with the given name under the given namespace
is public. If no matching repository was found, returns False.
"""
pass
@abstractmethod
def validate_oauth_token(self, token):
""" Returns whether the given OAuth token validates. """
pass
@abstractmethod
def get_sorted_matching_repositories(self, search_term, filter_username=None, offset=0, limit=25):
"""
Returns a sorted list of repositories matching the given search term.
"""
pass
class PreOCIModel(DockerRegistryV1DataInterface):
"""
PreOCIModel implements the data model for the v1 Docker Registry protocol using a database schema
before it was changed to support the OCI specification.
"""
def placement_locations_and_path_docker_v1(self, namespace_name, repo_name, image_id):
repo_image = model.image.get_repo_image_extended(namespace_name, repo_name, image_id)
if not repo_image or repo_image.storage is None:
return None, None
return repo_image.storage.locations, model.storage.get_layer_path(repo_image.storage)
def docker_v1_metadata(self, namespace_name, repo_name, image_id):
repo_image = model.image.get_repo_image(namespace_name, repo_name, image_id)
if repo_image is None:
return None
return AttrDict({
'namespace_name': namespace_name,
'repo_name': repo_name,
'image_id': image_id,
'checksum': repo_image.v1_checksum,
'compat_json': repo_image.v1_json_metadata,
})
def update_docker_v1_metadata(self, namespace_name, repo_name, image_id, created_date_str,
comment, command, compat_json, parent_image_id=None):
parent_image = None
if parent_image_id is not None:
parent_image = model.image.get_repo_image(namespace_name, repo_name, parent_image_id)
model.image.set_image_metadata(image_id, namespace_name, repo_name, created_date_str, comment,
command, compat_json, parent=parent_image)
def storage_exists(self, namespace_name, repo_name, image_id):
repo_image = model.image.get_repo_image_and_storage(namespace_name, repo_name, image_id)
if repo_image is None or repo_image.storage is None:
return False
if repo_image.storage.uploading:
return False
layer_path = model.storage.get_layer_path(repo_image.storage)
return store.exists(repo_image.storage.locations, layer_path)
def store_docker_v1_checksums(self, namespace_name, repo_name, image_id, checksum,
content_checksum):
repo_image = model.image.get_repo_image_and_storage(namespace_name, repo_name, image_id)
if repo_image is None or repo_image.storage is None:
return
with db_transaction():
repo_image.storage.content_checksum = content_checksum
repo_image.v1_checksum = checksum
repo_image.storage.save()
repo_image.save()
def is_image_uploading(self, namespace_name, repo_name, image_id):
repo_image = model.image.get_repo_image_and_storage(namespace_name, repo_name, image_id)
if repo_image is None or repo_image.storage is None:
return False
return repo_image.storage.uploading
def update_image_uploading(self, namespace_name, repo_name, image_id, is_uploading):
repo_image = model.image.get_repo_image_and_storage(namespace_name, repo_name, image_id)
if repo_image is None or repo_image.storage is None:
return
repo_image.storage.uploading = is_uploading
repo_image.storage.save()
return repo_image.storage
def update_image_sizes(self, namespace_name, repo_name, image_id, size, uncompressed_size):
model.storage.set_image_storage_metadata(image_id, namespace_name, repo_name, size,
uncompressed_size)
def get_image_size(self, namespace_name, repo_name, image_id):
repo_image = model.image.get_repo_image_and_storage(namespace_name, repo_name, image_id)
if repo_image is None or repo_image.storage is None:
return None
return repo_image.storage.image_size
def create_bittorrent_pieces(self, namespace_name, repo_name, image_id, pieces_bytes):
repo_image = model.image.get_repo_image_and_storage(namespace_name, repo_name, image_id)
if repo_image is None or repo_image.storage is None:
return
model.storage.save_torrent_info(repo_image.storage, app.config['BITTORRENT_PIECE_SIZE'],
pieces_bytes)
def image_ancestry(self, namespace_name, repo_name, image_id):
try:
image = model.image.get_image_by_id(namespace_name, repo_name, image_id)
except model.InvalidImageException:
return None
parents = model.image.get_parent_images(namespace_name, repo_name, image)
ancestry_docker_ids = [image.docker_image_id]
ancestry_docker_ids.extend([parent.docker_image_id for parent in parents])
return ancestry_docker_ids
def repository_exists(self, namespace_name, repo_name):
repo = model.repository.get_repository(namespace_name, repo_name)
return repo is not None
def create_or_link_image(self, username, namespace_name, repo_name, image_id, storage_location):
repo = model.repository.get_repository(namespace_name, repo_name)
model.image.find_create_or_link_image(image_id, repo, username, {}, storage_location)
def create_temp_hidden_tag(self, namespace_name, repo_name, image_id, expiration):
repo_image = model.image.get_repo_image(namespace_name, repo_name, image_id)
if repo_image is None:
return
repo = repo_image.repository
model.tag.create_temporary_hidden_tag(repo, repo_image, expiration)
def list_tags(self, namespace_name, repo_name):
return model.tag.list_repository_tags(namespace_name, repo_name)
def create_or_update_tag(self, namespace_name, repo_name, image_id, tag_name):
model.tag.create_or_update_tag(namespace_name, repo_name, tag_name, image_id)
def find_image_id_by_tag(self, namespace_name, repo_name, tag_name):
try:
tag_image = model.tag.get_tag_image(namespace_name, repo_name, tag_name)
except model.DataModelException:
return None
return tag_image.docker_image_id
def delete_tag(self, namespace_name, repo_name, tag_name):
model.tag.delete_tag(namespace_name, repo_name, tag_name)
def load_token(self, token):
try:
model.token.load_token_data(token)
return True
except model.InvalidTokenException:
return False
def verify_robot(self, username, token):
try:
return bool(model.user.verify_robot(username, token))
except model.InvalidRobotException:
return False
def change_user_password(self, user, new_password):
model.user.change_password(user, new_password)
def get_repository(self, namespace_name, repo_name):
repo = model.repository.get_repository(namespace_name, repo_name)
if repo is None:
return None
return _repository_for_repo(repo)
def create_repository(self, namespace_name, repo_name, user=None):
model.repository.create_repository(namespace_name, repo_name, user)
def repository_is_public(self, namespace_name, repo_name):
return model.repository.repository_is_public(namespace_name, repo_name)
def validate_oauth_token(self, token):
return bool(model.oauth.validate_access_token(token))
def get_sorted_matching_repositories(self, search_term, filter_username=None, offset=0, limit=25):
repos = model.repository.get_filtered_matching_repositories(search_term, filter_username,
offset, limit)
return [_repository_for_repo(repo) for repo in repos]
def _repository_for_repo(repo):
""" Returns a Repository object representing the Pre-OCI data model instance of a repository. """
return Repository(
id=repo.id,
name=repo.name,
namespace_name=repo.namespace_user.username,
description=repo.description,
is_public=model.repository.is_repository_public(repo)
)
pre_oci_model = PreOCIModel()