Downloads and extracts all .deb files of libc6-dev and its dependencies for given Ubuntu release

fetch-libc-lib-files.py

import re
import subprocess
import sys
import tempfile
from pathlib import Path

import requests


def get_package_info(package_name, ubuntu_version, arch):
    domain = "https://launchpad.net"
    print(f"Loading {package_name} info...")
    path = f"/ubuntu/{ubuntu_version}/{arch}/{package_name}"
    html = requests.get(f"{domain}{path}").text
    latest_version = re.search(rf'href="{path}/([^"]+)"', html).group(1)
    info_html = requests.get(f"{domain}{path}/{latest_version}").text
    deb_url = re.search(rf'href="([^"]+\.deb)"', info_html).group(1)
    depends = []
    if '<dl id="depends">' in info_html:
        depends_section = info_html.split('<dl id="depends">')[1].split("</dl>")[0]
        depends = re.findall(rf'{arch}/([^/"]+)"', depends_section)
    return latest_version, deb_url, depends


def get_transitive_dev_packages(root_package, ubuntu_version, arch="amd64"):
    def recurse(pkg):
        if pkg not in info and pkg.endswith("-dev"):
            version, deb_url, depends = get_package_info(pkg, ubuntu_version, arch)
            info[pkg] = version, deb_url
            for dep in depends:
                recurse(dep)

    info = {}
    recurse(root_package)
    return info


def download_dev_packages(output_dir, ubuntu_version, pkg_name, arch="amd64"):
    packages = get_transitive_dev_packages(pkg_name, ubuntu_version, arch)
    for package, (version, deb_url) in packages.items():
        print(f"Downloading {package} {version}...")
        deb = requests.get(deb_url).content
        with open(f"{output_dir}/{package}_{version}_{arch}.deb", "wb") as f:
            f.write(deb)


def extract_deb(deb_path, output_dir):
    return subprocess.check_call(["dpkg-deb", "-xv", deb_path, output_dir])


def main(argv):
    output_dir = Path(argv[1])
    ubuntu_version = argv[2]
    pkg = "libc6-dev"
    with tempfile.TemporaryDirectory() as temp_dir:
        temp_dir = Path(temp_dir)
        temp_dir.mkdir(exist_ok=True, parents=True)
        download_dev_packages(temp_dir, ubuntu_version, pkg)
        for deb in temp_dir.glob("*.deb"):
            extract_deb(deb, output_dir)


if __name__ == "__main__":
    main(sys.argv)

The list of symbols defined in the .a and .so files can then be dumped from the extracted files using:

find usr -name '*.a' -type f | xargs objdump -t | grep ' F .text' | awk 'NF>1{print $NF}' | sort -u > static.list
find usr -name '*.so' -type f | xargs objdump -T | grep ' DF .text' | awk 'NF>1{print $NF}' | sort -u > shared.list