debian-pypi-median.py

#!/usr/bin/env python3

import json
import statistics
import sys


def parse_list_from_stdin() -> list[str]:
    raw_packages = []
    
    raw_package = ""
    for line in sys.stdin:
        if line == "\n":
            raw_packages.append(raw_package)
            raw_package = ""
            continue
        raw_package += line
    return raw_packages


def match_and_get(prefix: str, data: str) -> str:
    if data.startswith(prefix):
        return data.removeprefix(prefix)
    return ""


def parse_raw_package(data: str) -> dict[str, str]:
    out = {}
    out["python-package"] = False

    for line in data.splitlines():
        line = line.strip()
        if (d := match_and_get("Package: ", line)) != "":
            out["name"] = d
        if (d := match_and_get("Depends: ", line)) != "":
            if "python3:any" in d:
                out["python-package"] = True

            deps = d.split(",")
            deps = list(map(lambda x: x.strip(), deps))
            out["depends"] = deps
    return out


def main() -> None:
    packages = []
    for raw_package in parse_list_from_stdin():
        package = parse_raw_package(raw_package)
        if not package["name"].startswith("python3-") or not package["python-package"]:
            continue

        package["python-deps"] = 0

        if "depends" in package:
            for dep in package["depends"]:
                if dep.startswith("python3-"):
                    package["python-deps"] += 1

        packages.append(package)
    print(packages)

    print(f"number of packages: {len(packages)}")

    dep_numbers = map(lambda x: x["python-deps"], packages)
    median = statistics.mean(dep_numbers)

    print(f"median of python dependencies: {median}")


if __name__ == "__main__":
    try:
        main()
    except KeyboardInterrupt:
        pass

Use this script like this:

$ curl "http://ftp.de.debian.org/debian/dists/bookworm/main/binary-amd64/Packages.gz" | zcat | ./debian-pypi-median.py