find_npm_licenses.py

import sys
import json
import requests
import re

def get_license(package_name):
    """Find the license of an NPM package, come hell or high water."""
    url = "https://registry.npmjs.org/{}".format(package_name)
    resp = requests.get(url).json()
    def read_license(license_):
        if isinstance(license_, basestring):
            return license_
        elif isinstance(license_, list):
            return read_license(license_[0])
        else:
            return license_["type"]
    if "license" in resp:
        return read_license(resp["license"])
    elif "licenses" in resp:
        return read_license(resp["licenses"][0])
    else:
        for version, info in resp["versions"].items():
            if "license" in info:
                return read_license(info["license"])
            elif "licenses" in info:
                return read_license(info["licenses"][0])
    if "repository" in resp:
        repo_url = resp["repository"]["url"]
        if "github.com" in repo_url:
            if repo_url.endswith(".git"):
                repo_url = repo_url[:-4]
            _url = repo_url\
                           .replace("git+ssh://", "https://")\
                           .replace("git://", "https://")\
                           .replace("git@", "")\
                           .replace("github.com", "raw.githubusercontent.com") \
                           .replace(".com:", ".com/")\
                           + "/master/package.json"
            if not _url.startswith("http"):
                _url = "https://" + _url
            pkg_json = json.loads(requests.get(_url).content)
            if "license" in pkg_json:
                return read_license(pkg_json["license"])
            elif "licenses" in pkg_json:
                return read_license(pkg_json["licenses"])