count_docker_pulls.py

import requests
import typing as t

repos = {
    "alpine": None,
    "debian": None,
    "ubuntu": None,
    "centos": None,
    "fedora": None,
    "archlinux": ["archlinux"],
    "opensuse": ["leap"],
    "rockylinux": ["rockylinux"],
    "almalinux": ["almalinux"]
}

def request(namespace: str, image: str):
    urltpl = f"https://hub.docker.com/v2/repositories/{namespace}/{image}"
    r = requests.get(urltpl)
    if r.status_code == 200:
        return r.json()
    raise Exception(f"{r.status_code}: {namespace}/{image}")

def get_count(namespace: str, image: str):
    return request(namespace, image)["pull_count"]

def count_pulls(os: str, tags: t.Union[t.List[str], None]) -> t.Tuple[int,int]:
    # get the count of the Official library namespace images
    library = get_count("library", os)
    nonlibrary = 0
    if tags:
        # if tags is not None, look inside the {os}/{t for t in tags} to count, too
        for t in tags:
            nonlibrary += get_count(os, t)
    return library, nonlibrary, library+nonlibrary
    
totals = {}
for os, tags in repos.items():
    totals[os] = count_pulls(os, tags)

for os, (library, nonlibrary, total) in sorted(totals.items(), reverse=True, key=lambda k: k[1][2]):
    print(f"{os : >20} {library:>20,} {nonlibrary:>20,} {total:>20,}")