Downloads the first blob for a given image and tag from the public Docker Hub

download_blob.py

#!/usr/bin/env python3
"""
Downloads the first blob for a given
image and tag from the public Docker Hub
"""
import argparse
import urllib.parse

import requests

DOCKERHUB_BASE_URL = 'https://registry.hub.docker.com/'


def get_auth_token(name):
    """Get token to make pull requests to Docker API"""
    url = f'https://auth.docker.io/token?service=registry.docker.io&scope=repository:{name}:pull'

    response = requests.get(url)
    response.raise_for_status()

    return response.json()['token']


def get_manifest(name, tag, token):
    """Get manifest for image 'name:tag'"""
    url = _build_url(f'v2/{name}/manifests/{tag}')

    response = requests.get(
        url,
        headers={'Authorization': f'Bearer {token}'},
    )
    response.raise_for_status()

    return response.json()


def get_blob(name, digest, token):
    """Download a blob with specified digest"""
    url = _build_url(f'/v2/{name}/blobs/{digest}')

    response = requests.get(
        url,
        headers={'Authorization': f'Bearer {token}'},
    )
    response.raise_for_status()

    return response.content


def _build_url(path):
    return urllib.parse.urljoin(DOCKERHUB_BASE_URL, path)


if __name__ == '__main__':
    default_name = 'elementsinteractive/flask-graphql-neo4j-app.dockerapp'
    default_tag = '0.1.0'
    parser = argparse.ArgumentParser(description='Downloads blob for a docker image')
    parser.add_argument('--name', dest='name', default=default_name)
    parser.add_argument('--tag', dest='tag', default=default_tag)
    parser.add_argument('--output', dest='output', default='blob.tar.gz')
    args = parser.parse_args()

    print(f'Getting content for {args.name}:{args.tag}')

    token = get_auth_token(args.name)
    manifests = get_manifest(name=args.name, tag=args.tag, token=token)
    digest = manifests['fsLayers'][0]['blobSum']

    blob = get_blob(name=args.name, digest=digest, token=token)
    with open(args.output, 'wb') as f:
        f.write(blob)

    print(f'File downloaded: {args.output}')