def parse_url(self):
""" Parse the proxy url into its component pieces
"""
# NOTE: If this changes, update tests/regex/proxy.py
#
# proxy=[protocol://][username[:password]@]host[:port][path][?query][#fragment]
# groups (both named and numbered)
# 1 = protocol
# 2 = username
# 3 = password
# 4 = host
# 5 = port
# 6 = path
# 7 = query
# 8 = fragment
m = URL_PARSE.match(self.url)
if not m:
raise ProxyStringError(_("malformed URL, cannot parse it."))
# If no protocol was given default to http.
self.protocol = m.group("protocol") or "http://"
if m.group("username"):
self.username = ensure_str(unquote(m.group("username")))
if m.group("password"):
self.password = ensure_str(unquote(m.group("password")))
if m.group("host"):
self.host = m.group("host")
if m.group("port"):
self.port = m.group("port")
else:
raise ProxyStringError(_("URL has no host component"))
self.parse_components()
def parse_url(self):
""" Parse the proxy url into its component pieces
"""
# NOTE: If this changes, update tests/regex/proxy.py
#
# proxy=[protocol://][username[:password]@]host[:port][path][?query][#fragment]
# groups (both named and numbered)
# 1 = protocol
# 2 = username
# 3 = password
# 4 = host
# 5 = port
# 6 = path
# 7 = query
# 8 = fragment
m = URL_PARSE.match(self.url)
if not m:
raise ProxyStringError(_("malformed URL, cannot parse it."))
# If no protocol was given default to http.
self.protocol = m.group("protocol") or "http://"
if m.group("username"):
self.username = ensure_str(unquote(m.group("username")))
if m.group("password"):
self.password = ensure_str(unquote(m.group("password")))
if m.group("host"):
self.host = m.group("host")
if m.group("port"):
self.port = m.group("port")
else:
raise ProxyStringError(_("URL has no host component"))
self.parse_components()
def url_regex_test(self):
"""
Run a list of possible URL values through the regex and check for
correct results.
tests are in the form of: (URL string, match.groups() tuple)
"""
tests = [
("proxy.host", (None, None, None, 'proxy.host', None, None, None,
None)),
("proxy.host:3128", (None, None, None, 'proxy.host', '3128', None,
None, None)),
("user:[email protected]",
(None, 'user', 'password', 'proxy.host', None, None, None, None)),
("*****@*****.**", (None, 'user', None, 'proxy.host', None, None,
None, None)),
("user:[email protected]:3128",
(None, 'user', 'password', 'proxy.host', '3128', None, None,
None)),
("[email protected]:3128", (None, 'user', None, 'proxy.host', '3128',
None, None, None)),
("proxy.host/blah/blah", (None, None, None, 'proxy.host', None,
'/blah/blah', None, None)),
("proxy.host:3128/blah/blah", (None, None, None, 'proxy.host',
'3128', '/blah/blah', None, None)),
("user:[email protected]/blah/blah",
(None, 'user', 'password', 'proxy.host', None, '/blah/blah', None,
None)),
("[email protected]/blah/blah", (None, 'user', None, 'proxy.host',
None, '/blah/blah', None, None)),
("user:[email protected]:3128/blah/blah",
(None, 'user', 'password', 'proxy.host', '3128', "/blah/blah",
None, None)),
("[email protected]:3128/blah/blah",
(None, 'user', None, 'proxy.host', '3128', "/blah/blah", None,
None)),
("http://proxy.host", ('http://', None, None, 'proxy.host', None,
None, None, None)),
("http://proxy.host:3128", ('http://', None, None, 'proxy.host',
'3128', None, None, None)),
("http://*****:*****@proxy.host",
('http://', 'user', 'password', 'proxy.host', None, None, None,
None)),
("http://[email protected]", ('http://', 'user', None, 'proxy.host',
None, None, None, None)),
("http://*****:*****@proxy.host:3128",
('http://', 'user', 'password', 'proxy.host', '3128', None, None,
None)),
("http://[email protected]:3128",
('http://', 'user', None, 'proxy.host', '3128', None, None,
None)),
("http://proxy.host/blah/blah",
('http://', None, None, 'proxy.host', None, '/blah/blah', None,
None)),
("http://proxy.host:3128/blah/blah",
('http://', None, None, 'proxy.host', '3128', '/blah/blah', None,
None)),
("http://*****:*****@proxy.host/blah/blah",
("http://", 'user', 'password', 'proxy.host', None, '/blah/blah',
None, None)),
("http://%75ser:[email protected]/blah/blah",
("http://", '%75ser', 'password', 'proxy.host', None,
'/blah/blah', None, None)),
("http://*****:*****@proxy.host/blah/blah",
("http://", 'user', '%70assword', 'proxy.host', None,
'/blah/blah', None, None)),
("http://[email protected]/blah/blah",
("http://", 'user', None, 'proxy.host', None, '/blah/blah', None,
None)),
("http://[email protected]/blah/bla%68",
("http://", 'user', None, 'proxy.host', None, '/blah/bla%68',
None, None)),
("http://*****:*****@proxy.host:3128/blah/blah",
("http://", 'user', 'password', 'proxy.host', '3128',
'/blah/blah', None, None)),
("http://[email protected]:3128/blah/blah",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
None, None)),
("http://[email protected]:3128/blah/blah?query",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
"query", None)),
("http://[email protected]:3128/blah/blah?query?",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
"query?", None)),
("http://[email protected]:3128/blah/blah?query=whatever",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
"query=whatever", None)),
("http://[email protected]:3128/blah/blah?query=whate%76er",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
"query=whate%76er", None)),
("http://[email protected]:3128/blah/blah?",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah', "",
None)),
("http://[email protected]:3128/blah/blah#fragment",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
None, "fragment")),
("http://[email protected]:3128/blah/blah#",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
None, "")),
("http://[email protected]:3128/blah/blah#fragm%65nt",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
None, "fragm%65nt")),
("http://[email protected]:3128/blah/blah?query=whatever#fragment",
("http://", 'user', None, 'proxy.host', '3128', '/blah/blah',
"query=whatever", "fragment")),
# Same, but with IPv4 literals
("1.2.3.4", (None, None, None, '1.2.3.4', None, None, None, None)),
("1.2.3.4:3128", (None, None, None, '1.2.3.4', '3128', None, None,
None)),
("user:[email protected]", (None, 'user', 'password', '1.2.3.4',
None, None, None, None)),
("[email protected]", (None, 'user', None, '1.2.3.4', None, None, None,
None)),
("user:[email protected]:3128",
(None, 'user', 'password', '1.2.3.4', '3128', None, None, None)),
("[email protected]:3128", (None, 'user', None, '1.2.3.4', '3128', None,
None, None)),
("1.2.3.4/blah/blah", (None, None, None, '1.2.3.4', None,
'/blah/blah', None, None)),
("1.2.3.4:3128/blah/blah", (None, None, None, '1.2.3.4', '3128',
'/blah/blah', None, None)),
("user:[email protected]/blah/blah",
(None, 'user', 'password', '1.2.3.4', None, '/blah/blah', None,
None)),
("[email protected]/blah/blah", (None, 'user', None, '1.2.3.4', None,
'/blah/blah', None, None)),
("user:[email protected]:3128/blah/blah",
(None, 'user', 'password', '1.2.3.4', '3128', "/blah/blah", None,
None)),
("[email protected]:3128/blah/blah",
(None, 'user', None, '1.2.3.4', '3128', "/blah/blah", None,
None)),
("http://1.2.3.4", ('http://', None, None, '1.2.3.4', None, None,
None, None)),
("http://1.2.3.4:3128", ('http://', None, None, '1.2.3.4', '3128',
None, None, None)),
("http://*****:*****@1.2.3.4",
('http://', 'user', 'password', '1.2.3.4', None, None, None,
None)),
("http://[email protected]", ('http://', 'user', None, '1.2.3.4', None,
None, None, None)),
("http://*****:*****@1.2.3.4:3128",
('http://', 'user', 'password', '1.2.3.4', '3128', None, None,
None)),
("http://[email protected]:3128", ('http://', 'user', None, '1.2.3.4',
'3128', None, None, None)),
("http://1.2.3.4/blah/blah", ('http://', None, None, '1.2.3.4',
None, '/blah/blah', None, None)),
("http://1.2.3.4:3128/blah/blah",
('http://', None, None, '1.2.3.4', '3128', '/blah/blah', None,
None)),
("http://*****:*****@1.2.3.4/blah/blah",
("http://", 'user', 'password', '1.2.3.4', None, '/blah/blah',
None, None)),
("http://%75ser:[email protected]/blah/blah",
("http://", '%75ser', 'password', '1.2.3.4', None, '/blah/blah',
None, None)),
("http://*****:*****@1.2.3.4/blah/blah",
("http://", 'user', '%70assword', '1.2.3.4', None, '/blah/blah',
None, None)),
("http://[email protected]/blah/blah",
("http://", 'user', None, '1.2.3.4', None, '/blah/blah', None,
None)),
("http://[email protected]/blah/bla%68",
("http://", 'user', None, '1.2.3.4', None, '/blah/bla%68', None,
None)),
("http://*****:*****@1.2.3.4:3128/blah/blah",
("http://", 'user', 'password', '1.2.3.4', '3128', '/blah/blah',
None, None)),
("http://[email protected]:3128/blah/blah",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah', None,
None)),
("http://[email protected]:3128/blah/blah?query",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah',
"query", None)),
("http://[email protected]:3128/blah/blah?query?",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah',
"query?", None)),
("http://[email protected]:3128/blah/blah?query=whatever",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah',
"query=whatever", None)),
("http://[email protected]:3128/blah/blah?query=whate%76er",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah',
"query=whate%76er", None)),
("http://[email protected]:3128/blah/blah?",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah', "",
None)),
("http://[email protected]:3128/blah/blah#fragment",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah', None,
"fragment")),
("http://[email protected]:3128/blah/blah#",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah', None,
"")),
("http://[email protected]:3128/blah/blah#fragm%65nt",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah', None,
"fragm%65nt")),
("http://[email protected]:3128/blah/blah?query=whatever#fragment",
("http://", 'user', None, '1.2.3.4', '3128', '/blah/blah',
"query=whatever", "fragment")),
# An again, but with IPv6 literals
("[dead::beef]", (None, None, None, '[dead::beef]', None, None,
None, None)),
("[dead::beef]:3128", (None, None, None, '[dead::beef]', '3128',
None, None, None)),
("user:password@[dead::beef]",
(None, 'user', 'password', '[dead::beef]', None, None, None,
None)),
("user@[dead::beef]", (None, 'user', None, '[dead::beef]', None,
None, None, None)),
("user:password@[dead::beef]:3128",
(None, 'user', 'password', '[dead::beef]', '3128', None, None,
None)),
("user@[dead::beef]:3128", (None, 'user', None, '[dead::beef]',
'3128', None, None, None)),
("[dead::beef]/blah/blah", (None, None, None, '[dead::beef]', None,
'/blah/blah', None, None)),
("[dead::beef]:3128/blah/blah",
(None, None, None, '[dead::beef]', '3128', '/blah/blah', None,
None)),
("user:password@[dead::beef]/blah/blah",
(None, 'user', 'password', '[dead::beef]', None, '/blah/blah',
None, None)),
("user@[dead::beef]/blah/blah",
(None, 'user', None, '[dead::beef]', None, '/blah/blah', None,
None)),
("user:password@[dead::beef]:3128/blah/blah",
(None, 'user', 'password', '[dead::beef]', '3128', "/blah/blah",
None, None)),
("user@[dead::beef]:3128/blah/blah",
(None, 'user', None, '[dead::beef]', '3128', "/blah/blah", None,
None)),
("http://[dead::beef]", ('http://', None, None, '[dead::beef]',
None, None, None, None)),
("http://[dead::beef]:3128",
('http://', None, None, '[dead::beef]', '3128', None, None,
None)),
("http://*****:*****@[dead::beef]",
('http://', 'user', 'password', '[dead::beef]', None, None, None,
None)),
("http://user@[dead::beef]",
('http://', 'user', None, '[dead::beef]', None, None, None,
None)),
("http://*****:*****@[dead::beef]:3128",
('http://', 'user', 'password', '[dead::beef]', '3128', None,
None, None)),
("http://user@[dead::beef]:3128",
('http://', 'user', None, '[dead::beef]', '3128', None, None,
None)),
("http://[dead::beef]/blah/blah",
('http://', None, None, '[dead::beef]', None, '/blah/blah', None,
None)),
("http://[dead::beef]:3128/blah/blah",
('http://', None, None, '[dead::beef]', '3128', '/blah/blah',
None, None)),
("http://*****:*****@[dead::beef]/blah/blah",
("http://", 'user', 'password', '[dead::beef]', None,
'/blah/blah', None, None)),
("http://%75ser:password@[dead::beef]/blah/blah",
("http://", '%75ser', 'password', '[dead::beef]', None,
'/blah/blah', None, None)),
("http://*****:*****@[dead::beef]/blah/blah",
("http://", 'user', '%70assword', '[dead::beef]', None,
'/blah/blah', None, None)),
("http://user@[dead::beef]/blah/blah",
("http://", 'user', None, '[dead::beef]', None, '/blah/blah',
None, None)),
("http://user@[dead::beef]/blah/bla%68",
("http://", 'user', None, '[dead::beef]', None, '/blah/bla%68',
None, None)),
("http://*****:*****@[dead::beef]:3128/blah/blah",
("http://", 'user', 'password', '[dead::beef]', '3128',
'/blah/blah', None, None)),
("http://user@[dead::beef]:3128/blah/blah",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
None, None)),
("http://user@[dead::beef]:3128/blah/blah?query",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
"query", None)),
("http://user@[dead::beef]:3128/blah/blah?query?",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
"query?", None)),
("http://user@[dead::beef]:3128/blah/blah?query=whatever",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
"query=whatever", None)),
("http://user@[dead::beef]:3128/blah/blah?query=whate%76er",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
"query=whate%76er", None)),
("http://user@[dead::beef]:3128/blah/blah?",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
"", None)),
("http://user@[dead::beef]:3128/blah/blah#fragment",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
None, "fragment")),
("http://user@[dead::beef]:3128/blah/blah#",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
None, "")),
("http://user@[dead::beef]:3128/blah/blah#fragm%65nt",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
None, "fragm%65nt")),
("http://user@[dead::beef]:3128/blah/blah?query=whatever#fragment",
("http://", 'user', None, '[dead::beef]', '3128', '/blah/blah',
"query=whatever", "fragment")),
# Invalid schemes
("0http://proxy.host/", None),
("h~ttp://proxy.host/", None),
# Invalid usernames and passwords
("http://%[email protected]/", None),
("http://*[email protected]/", None),
("http://*****:*****@proxy.host/", None),
("http://*****:*****@[email protected]/", None),
# Invalid paths
("http://*****:*****@proxy.host/%xxlah/blah", None),
("http://*****:*****@proxy.host/[]lah/blah", None),
# Invalid queries
("http://proxy.host/blah/blah?quer%xx", None),
("http://proxy.host/blah/blah?que[]y", None),
# Invalid fragments
("http://proxy.host/blah/blah#fragment#", None),
("http://proxy.host/blah/blah#%xxragment", None),
# Unbracketed IPv6
("fe80::1234:56:78", None),
("fe80::1234:56:78/blah/blah", None),
("http://fe80::1234:56:78/blah/blah", None)
]
got_error = False
for proxy, result in tests:
match = URL_PARSE.match(proxy)
if match:
match = match.groups()
else:
match = None
try:
self.assertEqual(match, result)
except AssertionError:
got_error = True
print("Proxy parse error: `%s' did not parse as `%s': %s" %
(proxy, result, match))
if got_error:
self.fail()