Get licencing information of packages and their dependencies from pip

licensing_info.py

from email import message_from_string
from pkg_resources import get_distribution

def get_pkg_info(package, extras=False):
    """Get information about an installed package.

    This function will also attempt to get the licence text from a bundled
    licence file, if available.

    Parameters
    ----------
    package : str
        the package to get information about
    extras : bool, optional
        whether to consider 'extras' required packages (default=False)

    Returns
    -------
    metadata : dict
        package metadata
    required : list
        required packages

    """

    d = get_distribution(package)
    m = message_from_string(d.get_metadata(d.PKG_INFO))

    if "Requires-Dist" in m.keys():
        deps = m.get_all("Requires-Dist")
        if extras:
            required = [x.split()[0] for x in deps]
        else:
            required = [x.split()[0] for x in deps if not "extra == " in x]
    else:
        required = []

    metadata = {}
    for key in m.keys():
        if m.keys().count(key) == 1:
            metadata[key] = m.get(key)
        else:
            metadata[key] = m.get_all(key)
    metadata["License-text"] = ""
    for l in ["LICENSE*", "LICENCE*", "COPYING*", "NOTICE*"]:
        for licence_file in glob.glob(os.path.join(d.egg_info, l)):
            with open(licence_file) as f:
                metadata["License-text"] += f.read()

    return metadata, required

def get_licensing_info(packages, extras=False):
    """Get licensing information of a list of package and their dependencies.

    Parameters
    ----------
    packages : list
        the packages to get licensing information from
    extras : bool, optional
        whether to consider 'extras' required packages (default=False)

    Returns
    -------
    licensing_info : list
        the licensing information for each package

    """

    licensing_info = []
    for package in packages:
        try:
            metadata, required = get_pkg_info(package, extras)
            packages.extend([x for x in required if x not in packages])
        except:
            continue

    for package in packages:
        try:
            metadata, required = get_pkg_info(package, extras)
            if not metadata in licensing_info:
                licensing_info.append(metadata)
        except:
            continue

    return sorted(licensing_info, key=lambda x: x["Name"])