bootstrap_upstream.py

import os, io, zipfile, shutil, requests

GITHUB_SOURCES = [
    ("RTLucassen", "B-line_detection"),
    ("ljhowell", "LUS-Segmentation-RT"),
    ("GuoJiaqi-1020", "MeDiVLAD"),
]

def _dl_and_extract(owner, repo, ref):
    url = f"https://github.com/{owner}/{repo}/archive/refs/heads/{ref}.zip"
    r = requests.get(url, timeout=60)
    r.raise_for_status()
    z = zipfile.ZipFile(io.BytesIO(r.content))

    tmp = f"_tmp_{repo}"
    os.makedirs(tmp, exist_ok=True)
    z.extractall(tmp)

    top = [os.path.join(tmp, d) for d in os.listdir(tmp) if os.path.isdir(os.path.join(tmp, d))]
    if not top:
        raise RuntimeError(f"Could not find extracted folder for {owner}/{repo}")

    src = top[0]
    dst = os.path.join("external", repo)
    os.makedirs("external", exist_ok=True)
    if os.path.exists(dst):
        shutil.rmtree(dst)
    shutil.move(src, dst)
    shutil.rmtree(tmp, ignore_errors=True)
    return dst

def fetch_upstream():
    os.makedirs("external", exist_ok=True)
    results = {}
    for owner, repo in GITHUB_SOURCES:
        dst = os.path.join("external", repo)
        if os.path.isdir(dst):
            results[repo] = "exists"
            continue
        # try 'main' then 'master'
        try:
            _dl_and_extract(owner, repo, "main")
            results[repo] = "downloaded-main"
        except Exception:
            _dl_and_extract(owner, repo, "master")
            results[repo] = "downloaded-master"
    return results