]> jfr.im git - yt-dlp.git/blame - yt_dlp/socks.py
[toggo] Improve `_VALID_URL` (#3689)
[yt-dlp.git] / yt_dlp / socks.py
CommitLineData
71aff188
YCH
1# Public Domain SOCKS proxy protocol implementation
2# Adapted from https://gist.github.com/bluec0re/cafd3764412967417fd3
9e9cd724
YCH
3# References:
4# SOCKS4 protocol http://www.openssh.com/txt/socks4.protocol
5# SOCKS4A protocol http://www.openssh.com/txt/socks4a.protocol
6# SOCKS5 protocol https://tools.ietf.org/html/rfc1928
7# SOCKS5 username/password authentication https://tools.ietf.org/html/rfc1929
8
71aff188
YCH
9import collections
10import socket
4350b745 11
f8271158 12from .compat import compat_ord, compat_struct_pack, compat_struct_unpack
4350b745 13
71aff188 14__author__ = 'Timo Schmid <coding@timoschmid.de>'
4350b745 15
9e9cd724
YCH
16SOCKS4_VERSION = 4
17SOCKS4_REPLY_VERSION = 0x00
18# Excerpt from SOCKS4A protocol:
19# if the client cannot resolve the destination host's domain name to find its
20# IP address, it should set the first three bytes of DSTIP to NULL and the last
21# byte to a non-zero value.
edaa23f8 22SOCKS4_DEFAULT_DSTIP = compat_struct_pack('!BBBB', 0, 0, 0, 0xFF)
9e9cd724
YCH
23
24SOCKS5_VERSION = 5
25SOCKS5_USER_AUTH_VERSION = 0x01
26SOCKS5_USER_AUTH_SUCCESS = 0x00
27
28
86e5f3ed 29class Socks4Command:
9e9cd724
YCH
30 CMD_CONNECT = 0x01
31 CMD_BIND = 0x02
32
33
34class Socks5Command(Socks4Command):
35 CMD_UDP_ASSOCIATE = 0x03
36
37
86e5f3ed 38class Socks5Auth:
9e9cd724
YCH
39 AUTH_NONE = 0x00
40 AUTH_GSSAPI = 0x01
41 AUTH_USER_PASS = 0x02
42 AUTH_NO_ACCEPTABLE = 0xFF # For server response
43
44
86e5f3ed 45class Socks5AddressType:
9e9cd724
YCH
46 ATYP_IPV4 = 0x01
47 ATYP_DOMAINNAME = 0x03
48 ATYP_IPV6 = 0x04
49
4350b745 50
4bd7d9d4 51class ProxyError(socket.error):
9e9cd724
YCH
52 ERR_SUCCESS = 0x00
53
54 def __init__(self, code=None, msg=None):
55 if code is not None and msg is None:
83442966 56 msg = self.CODES.get(code) or 'unknown error'
86e5f3ed 57 super().__init__(code, msg)
9e9cd724
YCH
58
59
60class InvalidVersionError(ProxyError):
61 def __init__(self, expected_version, got_version):
86e5f3ed 62 msg = ('Invalid response version from server. Expected {:02x} got '
63 '{:02x}'.format(expected_version, got_version))
64 super().__init__(0, msg)
4350b745
YCH
65
66
4350b745 67class Socks4Error(ProxyError):
9e9cd724
YCH
68 ERR_SUCCESS = 90
69
4350b745 70 CODES = {
9e9cd724 71 91: 'request rejected or failed',
a2406fce 72 92: 'request rejected because SOCKS server cannot connect to identd on the client',
9e9cd724 73 93: 'request rejected because the client program and identd report different user-ids'
4350b745 74 }
71aff188 75
4350b745 76
9e9cd724
YCH
77class Socks5Error(ProxyError):
78 ERR_GENERAL_FAILURE = 0x01
71aff188 79
4350b745
YCH
80 CODES = {
81 0x01: 'general SOCKS server failure',
82 0x02: 'connection not allowed by ruleset',
83 0x03: 'Network unreachable',
84 0x04: 'Host unreachable',
85 0x05: 'Connection refused',
86 0x06: 'TTL expired',
87 0x07: 'Command not supported',
88 0x08: 'Address type not supported',
89 0xFE: 'unknown username or invalid password',
90 0xFF: 'all offered authentication methods were rejected'
91 }
92
71aff188 93
86e5f3ed 94class ProxyType:
71aff188 95 SOCKS4 = 0
4350b745 96 SOCKS4A = 1
71aff188 97 SOCKS5 = 2
4350b745 98
582be358 99
9e9cd724
YCH
100Proxy = collections.namedtuple('Proxy', (
101 'type', 'host', 'port', 'username', 'password', 'remote_dns'))
4350b745 102
4350b745 103
71aff188 104class sockssocket(socket.socket):
9e9cd724
YCH
105 def __init__(self, *args, **kwargs):
106 self._proxy = None
86e5f3ed 107 super().__init__(*args, **kwargs)
9e9cd724
YCH
108
109 def setproxy(self, proxytype, addr, port, rdns=True, username=None, password=None):
110 assert proxytype in (ProxyType.SOCKS4, ProxyType.SOCKS4A, ProxyType.SOCKS5)
111
112 self._proxy = Proxy(proxytype, addr, port, username, password, rdns)
4350b745
YCH
113
114 def recvall(self, cnt):
115 data = b''
116 while len(data) < cnt:
117 cur = self.recv(cnt - len(data))
118 if not cur:
86e5f3ed 119 raise EOFError(f'{cnt - len(data)} bytes missing')
4350b745
YCH
120 data += cur
121 return data
122
9e9cd724
YCH
123 def _recv_bytes(self, cnt):
124 data = self.recvall(cnt)
86e5f3ed 125 return compat_struct_unpack(f'!{cnt}B', data)
9e9cd724
YCH
126
127 @staticmethod
128 def _len_and_data(data):
edaa23f8 129 return compat_struct_pack('!B', len(data)) + data
9e9cd724
YCH
130
131 def _check_response_version(self, expected_version, got_version):
132 if got_version != expected_version:
133 self.close()
134 raise InvalidVersionError(expected_version, got_version)
4350b745 135
9e9cd724 136 def _resolve_address(self, destaddr, default, use_remote_dns):
4350b745 137 try:
9e9cd724 138 return socket.inet_aton(destaddr)
86e5f3ed 139 except OSError:
9e9cd724
YCH
140 if use_remote_dns and self._proxy.remote_dns:
141 return default
4350b745 142 else:
9e9cd724
YCH
143 return socket.inet_aton(socket.gethostbyname(destaddr))
144
145 def _setup_socks4(self, address, is_4a=False):
146 destaddr, port = address
147
148 ipaddr = self._resolve_address(destaddr, SOCKS4_DEFAULT_DSTIP, use_remote_dns=is_4a)
149
edaa23f8 150 packet = compat_struct_pack('!BBH', SOCKS4_VERSION, Socks4Command.CMD_CONNECT, port) + ipaddr
9e9cd724
YCH
151
152 username = (self._proxy.username or '').encode('utf-8')
153 packet += username + b'\x00'
4350b745
YCH
154
155 if is_4a and self._proxy.remote_dns:
9e9cd724 156 packet += destaddr.encode('utf-8') + b'\x00'
4350b745
YCH
157
158 self.sendall(packet)
159
edaa23f8 160 version, resp_code, dstport, dsthost = compat_struct_unpack('!BBHI', self.recvall(8))
4350b745 161
9e9cd724 162 self._check_response_version(SOCKS4_REPLY_VERSION, version)
4350b745 163
9e9cd724 164 if resp_code != Socks4Error.ERR_SUCCESS:
4350b745
YCH
165 self.close()
166 raise Socks4Error(resp_code)
167
71aff188
YCH
168 return (dsthost, dstport)
169
9e9cd724
YCH
170 def _setup_socks4a(self, address):
171 self._setup_socks4(address, is_4a=True)
4350b745 172
9e9cd724 173 def _socks5_auth(self):
edaa23f8 174 packet = compat_struct_pack('!B', SOCKS5_VERSION)
4350b745 175
9e9cd724 176 auth_methods = [Socks5Auth.AUTH_NONE]
4350b745 177 if self._proxy.username and self._proxy.password:
9e9cd724
YCH
178 auth_methods.append(Socks5Auth.AUTH_USER_PASS)
179
edaa23f8 180 packet += compat_struct_pack('!B', len(auth_methods))
86e5f3ed 181 packet += compat_struct_pack(f'!{len(auth_methods)}B', *auth_methods)
4350b745
YCH
182
183 self.sendall(packet)
184
9e9cd724 185 version, method = self._recv_bytes(2)
4350b745 186
9e9cd724 187 self._check_response_version(SOCKS5_VERSION, version)
4350b745 188
54f54fcc
YCH
189 if method == Socks5Auth.AUTH_NO_ACCEPTABLE or (
190 method == Socks5Auth.AUTH_USER_PASS and (not self._proxy.username or not self._proxy.password)):
4350b745 191 self.close()
54f54fcc 192 raise Socks5Error(Socks5Auth.AUTH_NO_ACCEPTABLE)
4350b745 193
9e9cd724
YCH
194 if method == Socks5Auth.AUTH_USER_PASS:
195 username = self._proxy.username.encode('utf-8')
196 password = self._proxy.password.encode('utf-8')
edaa23f8 197 packet = compat_struct_pack('!B', SOCKS5_USER_AUTH_VERSION)
9e9cd724 198 packet += self._len_and_data(username) + self._len_and_data(password)
4350b745
YCH
199 self.sendall(packet)
200
9e9cd724 201 version, status = self._recv_bytes(2)
4350b745 202
9e9cd724 203 self._check_response_version(SOCKS5_USER_AUTH_VERSION, version)
4350b745 204
9e9cd724 205 if status != SOCKS5_USER_AUTH_SUCCESS:
4350b745 206 self.close()
9e9cd724 207 raise Socks5Error(Socks5Error.ERR_GENERAL_FAILURE)
4350b745 208
9e9cd724
YCH
209 def _setup_socks5(self, address):
210 destaddr, port = address
211
212 ipaddr = self._resolve_address(destaddr, None, use_remote_dns=True)
213
214 self._socks5_auth()
215
216 reserved = 0
edaa23f8 217 packet = compat_struct_pack('!BBB', SOCKS5_VERSION, Socks5Command.CMD_CONNECT, reserved)
4350b745 218 if ipaddr is None:
9e9cd724 219 destaddr = destaddr.encode('utf-8')
edaa23f8 220 packet += compat_struct_pack('!B', Socks5AddressType.ATYP_DOMAINNAME)
9e9cd724 221 packet += self._len_and_data(destaddr)
4350b745 222 else:
edaa23f8
YCH
223 packet += compat_struct_pack('!B', Socks5AddressType.ATYP_IPV4) + ipaddr
224 packet += compat_struct_pack('!H', port)
4350b745
YCH
225
226 self.sendall(packet)
227
9e9cd724 228 version, status, reserved, atype = self._recv_bytes(4)
4350b745 229
9e9cd724 230 self._check_response_version(SOCKS5_VERSION, version)
4350b745 231
9e9cd724 232 if status != Socks5Error.ERR_SUCCESS:
4350b745
YCH
233 self.close()
234 raise Socks5Error(status)
235
9e9cd724 236 if atype == Socks5AddressType.ATYP_IPV4:
4350b745 237 destaddr = self.recvall(4)
9e9cd724
YCH
238 elif atype == Socks5AddressType.ATYP_DOMAINNAME:
239 alen = compat_ord(self.recv(1))
4350b745 240 destaddr = self.recvall(alen)
9e9cd724 241 elif atype == Socks5AddressType.ATYP_IPV6:
4350b745 242 destaddr = self.recvall(16)
edaa23f8 243 destport = compat_struct_unpack('!H', self.recvall(2))[0]
71aff188
YCH
244
245 return (destaddr, destport)
4350b745
YCH
246
247 def _make_proxy(self, connect_func, address):
9e9cd724 248 if not self._proxy:
4350b745
YCH
249 return connect_func(self, address)
250
9e9cd724
YCH
251 result = connect_func(self, (self._proxy.host, self._proxy.port))
252 if result != 0 and result is not None:
253 return result
254 setup_funcs = {
255 ProxyType.SOCKS4: self._setup_socks4,
256 ProxyType.SOCKS4A: self._setup_socks4a,
257 ProxyType.SOCKS5: self._setup_socks5,
258 }
259 setup_funcs[self._proxy.type](address)
260 return result
261
4350b745 262 def connect(self, address):
71aff188 263 self._make_proxy(socket.socket.connect, address)
4350b745
YCH
264
265 def connect_ex(self, address):
71aff188 266 return self._make_proxy(socket.socket.connect_ex, address)