ssurt.py

#!/usr/bin/python3
# coding=utf-8

import re

SSURT_RE = r"""
\A
(?P<scheme> [a-zA-Z] [a-zA-Z0-9+.-]* : )?
(?P<authority>
  (?P<slashes> /* )
  (?P<userinfo> [^/?#\[\]@]* @ )?
  (?P<host>
      (?P<ip6> \[ [^\]]* \] )
    | (?P<ip4> [0-9]+ \. [0-9]+ \. [0-9]+ \. [0-9]+ )
    | (?P<domain> [^:/?#\[\]@]+ )
  )
  (?P<port> : [0-9]+ )?
)?
(?P<path> .*? )
\Z
"""

def ssurt(input):
    m = re.match(SSURT_RE, input, re.VERBOSE | re.DOTALL)
    assert m

    if m.group('domain'):
        host = revdomain(m.group('domain'))
    else:
        host = m.group('host')

    return ((host or '')
            + (m.group('port') or '')
            + (m.group('slashes') or '')
            + ":" +
            (m.group('scheme') or '')
            + (m.group('userinfo') or '')
            + m.group("path"))

def revdomain(domain):
    return '(' +  ','.join(reversed(domain.split('.'))) + ',' + ')'

UNSSURT_RE = r"""
\A
(?P<authority>
    (?P<host>
      (?P<ip6> \[ [^\]]* \] )
    | (?P<ip4> [0-9]+ \. [0-9]+ \. [0-9]+ \. [0-9]+ )
    | (?P<revdomain> \( [^:/?#\[\]@]+ , \) )
    )
    (?P<port> : [0-9]+ )?
    (?P<slashes> /* )
)?
:
(?P<scheme> [a-zA-Z] [a-zA-Z0-9+.-]* :)?
(?P<userinfo> [^/?#\[\]@]* @ )?
(?P<path> .*? )
\Z
"""

def unssurt(input):
    m = re.match(UNSSURT_RE, input, re.VERBOSE | re.DOTALL)
    assert m
    if m.group('revdomain'):
        host = unrevdomain(m.group('revdomain'))
    else:
        host = m.group('host')

    return ((m.group('scheme') or '')
            + (m.group('slashes') or '')
            + (m.group('userinfo') or '')
            + (host or '')
            + (m.group('port') or '')
            + m.group('path'))

def unrevdomain(revdomain):
    return '.'.join(reversed(revdomain[1:-2].split(',')))


urls = ["https://user:pass@www.example.org:1443/path",
        "random!!garbage",
        "screenshot+http://user:pass@www.example.org:1443/path",
        "http:/日本.jp:80//.././[FÜNKY]",
        "https://[2001:db8::1:0:0:1]:8080/",
        "ssh://git@github.com/smola/galimatias.git",
        "dns:example.com",
        ":",
        "::",
        "x:[/]/foo",
        "//noscheme.tld/foo.html?query#frag"]


for url in urls:
    print(url, ssurt(url))
    assert unssurt(ssurt(url)) == url

# https://user:pass@www.example.org:1443/path            (org,example,www,):1443//:https:user:pass@/path
# random!!garbage                                        (random!!garbage,):
# screenshot+http://user:pass@www.example.org:1443/path  (org,example,www,):1443//:screenshot+http:user:pass@/path
# http:/日本.jp:80//.././[FÜNKY]                         (jp,日本,):80/:http://.././[FÜNKY]
# https://[2001:db8::1:0:0:1]:8080/                      [2001:db8::1:0:0:1]:8080//:https:/
# ssh://git@github.com/smola/galimatias.git              (com,github,)//:ssh:git@/smola/galimatias.git
# dns:example.com                                        (com,example,):dns:
# :                                                      ::
# ::                                                     :::
# x:[/]/foo                                              [/]:x:/foo
# //noscheme.tld/foo.html?query#frag                     (tld,noscheme,)//:/foo.html?query#frag