urlparse

#!/usr/bin/env python3

import os, sys, pyaml, urllib.parse as up

def dissect_url(url):
	url = up.urlparse(url)._asdict()
	q = url.pop('query', None)
	if q: q = sorted(up.parse_qsl(q))
	url = dict(sorted((k,v) for k,v in url.items() if v))
	if q: url['query'] = q # should always be last
	return url

def decode_url(url):
	for k, v in url.items():
		if isinstance(v, str): url[k] = up.unquote_plus(v)
	for k, v in url.get('query') or dict():
		if isinstance(v, str): url['query'][k] = up.unquote_plus(v)

def main():
	import argparse
	parser = argparse.ArgumentParser(
		usage='%(prog)s [options] URL1 [URL2] [ -- <diff options> ]',
		description='Process and dump URLs in a human-readable format.')
	parser.add_argument('url', help='URL or string to process.')
	parser.add_argument('url2', nargs='?', help='URL to show diff against.')
	parser.add_argument('-d', '--decode', action='store_true',
		help='Decode (aka "unquote") URL-encoded (aka "percent-encoded") URL components.')
	parser.add_argument('--decode-string', action='store_true',
		help='Decode and print passed URL as one string, without breaking it down into components.')

	argv, diff_opts = sys.argv[1:], list()
	try: idx = argv.index('--')
	except ValueError: pass
	else: argv, diff_opts = argv[:idx], argv[idx+1:]
	opts = parser.parse_args(argv)

	if opts.decode_string:
		return print(up.unquote_plus(opts.url))

	url = dissect_url(opts.url)
	if opts.decode: decode_url(url)
	if not opts.url2: return pyaml.dump(url, sys.stdout)

	import tempfile, subprocess as sp
	with \
			tempfile.NamedTemporaryFile(prefix='url1.') as tmp1,\
			tempfile.NamedTemporaryFile(prefix='url2.') as tmp2:
		pyaml.dump(url, tmp1)
		url2 = dissect_url(opts.url2)
		if opts.decode: decode_url(url2)
		pyaml.dump(url2, tmp2)
		tmp1.flush(), tmp2.flush()
		diff_cmd = 'colordiff' if os.path.exists('/usr/bin/colordiff') else 'diff'
		diff_cmd = [diff_cmd, '-uw'] + diff_opts + [tmp1.name, tmp2.name]
		return sp.run(diff_cmd).returncode

if __name__ == '__main__': sys.exit(main())