#!/usr/bin/env python3
import argparse
from pathlib import Path
import re
import subprocess
import sys
import urllib.parse

parser = argparse.ArgumentParser(
    description="""
    noRMS for APK-based systems (Alpine).
    Checks to see which installed packages are attached to RMS (i.e. the FSF and GNU).
    """
)
parser.add_argument('--verbose', '-v', help='Print more stuff', action='count', default=0)
args = parser.parse_args()

verbose = args.verbose

PACKAGE = re.compile(r'^(?P<name>\S+)-[\d\w\._]+-r\d+ \S+ \{(?P<origin>[^}]+)\} \([^)]+\) \[installed\]$')

# from https://directory.fsf.org/wiki/GNU:
# `new Set([].map.call(document.querySelectorAll('table tr td:nth-child(3)'), x => new URL(x.innerText).host))`
GNU_DOMAINS = set([
    "www.gnu.org",
    "puszcza.gnu.org.ua",
    "gcc.gnu.org",
    "gcompris.net",
    "gmplib.org",
    # "www.gnome.org", # the GNOME foundation signed https://rms-open-letter.github.io/ so they're OK
    "www.gnucash.org",
    "gnunet.org",
    "www.gnuhealth.org",
    "www.list.org",
    "www.midnight-commander.org",
    "mediagoblin.org",
    "www.multiprecision.org",
    "www.mpfr.org",
    "www.nano-editor.org",
    "jemarch.net",
    "www.r-project.org",
    "savannah.gnu.org",
    "www.texmacs.org",
    "unifoundry.com"
])

origins = dict()

all_packages = subprocess.check_output(['apk', 'list', '--installed'], text=True)
for package in all_packages.splitlines():
    match = PACKAGE.match(package)
    if match is None:
        print('oops!', package)
    name, origin = match['name'], match['origin']
    if origin not in origins:
        origins[origin] = []
    origins[origin].append(name)

info_output = subprocess.check_output(['apk', 'info', '-wv', *origins.keys()], text=True)
package_webpage_map = dict(x.split(': ', 1) for x in info_output.splitlines())

rms_count = 0

for origin in origins:
    webpage = package_webpage_map[origin]
    parsed_url = urllib.parse.urlparse(webpage)
    host = parsed_url.netloc
    if verbose >= 1:
        print('origin:', origin, 'host:', host, 'webpage:', webpage)
    if host in GNU_DOMAINS:
        subpackages = [x for x in origins[origin] if x != origin]
        if len(subpackages) == 0:
            package_text = ''
        else:
            package_text = ' (& ' + ', '.join(sorted(subpackages)) + ')'
        print(f'error: {origin}{package_text} is GNU software (website is {webpage})')
        rms_count += 1

print()
print('Total RMS-affiliated packages found:', rms_count)

if rms_count == 0:
    print('Hell yeah!')
    sys.exit(0)
else:
    sys.exit(1)