]> jfr.im git - yt-dlp.git/blame - yt_dlp/socks.py
[ie/youtube] Extract upload timestamp if available (#9856)
[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
ac668111 11import struct
4350b745 12
ac668111 13from .compat import compat_ord
4350b745 14
71aff188 15__author__ = 'Timo Schmid <coding@timoschmid.de>'
4350b745 16
9e9cd724
YCH
17SOCKS4_VERSION = 4
18SOCKS4_REPLY_VERSION = 0x00
19# Excerpt from SOCKS4A protocol:
20# if the client cannot resolve the destination host's domain name to find its
21# IP address, it should set the first three bytes of DSTIP to NULL and the last
22# byte to a non-zero value.
ac668111 23SOCKS4_DEFAULT_DSTIP = struct.pack('!BBBB', 0, 0, 0, 0xFF)
9e9cd724
YCH
24
25SOCKS5_VERSION = 5
26SOCKS5_USER_AUTH_VERSION = 0x01
27SOCKS5_USER_AUTH_SUCCESS = 0x00
28
29
86e5f3ed 30class Socks4Command:
9e9cd724
YCH
31 CMD_CONNECT = 0x01
32 CMD_BIND = 0x02
33
34
35class Socks5Command(Socks4Command):
36 CMD_UDP_ASSOCIATE = 0x03
37
38
86e5f3ed 39class Socks5Auth:
9e9cd724
YCH
40 AUTH_NONE = 0x00
41 AUTH_GSSAPI = 0x01
42 AUTH_USER_PASS = 0x02
43 AUTH_NO_ACCEPTABLE = 0xFF # For server response
44
45
86e5f3ed 46class Socks5AddressType:
9e9cd724
YCH
47 ATYP_IPV4 = 0x01
48 ATYP_DOMAINNAME = 0x03
49 ATYP_IPV6 = 0x04
50
4350b745 51
f9fb3ce8 52class ProxyError(OSError):
9e9cd724
YCH
53 ERR_SUCCESS = 0x00
54
55 def __init__(self, code=None, msg=None):
56 if code is not None and msg is None:
83442966 57 msg = self.CODES.get(code) or 'unknown error'
86e5f3ed 58 super().__init__(code, msg)
9e9cd724
YCH
59
60
61class InvalidVersionError(ProxyError):
62 def __init__(self, expected_version, got_version):
86e5f3ed 63 msg = ('Invalid response version from server. Expected {:02x} got '
64 '{:02x}'.format(expected_version, got_version))
65 super().__init__(0, msg)
4350b745
YCH
66
67
4350b745 68class Socks4Error(ProxyError):
9e9cd724
YCH
69 ERR_SUCCESS = 90
70
4350b745 71 CODES = {
9e9cd724 72 91: 'request rejected or failed',
a2406fce 73 92: 'request rejected because SOCKS server cannot connect to identd on the client',
9e9cd724 74 93: 'request rejected because the client program and identd report different user-ids'
4350b745 75 }
71aff188 76
4350b745 77
9e9cd724
YCH
78class Socks5Error(ProxyError):
79 ERR_GENERAL_FAILURE = 0x01
71aff188 80
4350b745
YCH
81 CODES = {
82 0x01: 'general SOCKS server failure',
83 0x02: 'connection not allowed by ruleset',
84 0x03: 'Network unreachable',
85 0x04: 'Host unreachable',
86 0x05: 'Connection refused',
87 0x06: 'TTL expired',
88 0x07: 'Command not supported',
89 0x08: 'Address type not supported',
90 0xFE: 'unknown username or invalid password',
91 0xFF: 'all offered authentication methods were rejected'
92 }
93
71aff188 94
86e5f3ed 95class ProxyType:
71aff188 96 SOCKS4 = 0
4350b745 97 SOCKS4A = 1
71aff188 98 SOCKS5 = 2
4350b745 99
582be358 100
9e9cd724
YCH
101Proxy = collections.namedtuple('Proxy', (
102 'type', 'host', 'port', 'username', 'password', 'remote_dns'))
4350b745 103
4350b745 104
71aff188 105class sockssocket(socket.socket):
9e9cd724
YCH
106 def __init__(self, *args, **kwargs):
107 self._proxy = None
86e5f3ed 108 super().__init__(*args, **kwargs)
9e9cd724
YCH
109
110 def setproxy(self, proxytype, addr, port, rdns=True, username=None, password=None):
111 assert proxytype in (ProxyType.SOCKS4, ProxyType.SOCKS4A, ProxyType.SOCKS5)
112
113 self._proxy = Proxy(proxytype, addr, port, username, password, rdns)
4350b745
YCH
114
115 def recvall(self, cnt):
116 data = b''
117 while len(data) < cnt:
118 cur = self.recv(cnt - len(data))
119 if not cur:
86e5f3ed 120 raise EOFError(f'{cnt - len(data)} bytes missing')
4350b745
YCH
121 data += cur
122 return data
123
9e9cd724
YCH
124 def _recv_bytes(self, cnt):
125 data = self.recvall(cnt)
ac668111 126 return struct.unpack(f'!{cnt}B', data)
9e9cd724
YCH
127
128 @staticmethod
129 def _len_and_data(data):
ac668111 130 return struct.pack('!B', len(data)) + data
9e9cd724
YCH
131
132 def _check_response_version(self, expected_version, got_version):
133 if got_version != expected_version:
134 self.close()
135 raise InvalidVersionError(expected_version, got_version)
4350b745 136
20fbbd92 137 def _resolve_address(self, destaddr, default, use_remote_dns, family=None):
138 for f in (family,) if family else (socket.AF_INET, socket.AF_INET6):
139 try:
140 return f, socket.inet_pton(f, destaddr)
141 except OSError:
142 continue
143
144 if use_remote_dns and self._proxy.remote_dns:
145 return 0, default
146 else:
147 res = socket.getaddrinfo(destaddr, None, family=family or 0)
148 f, _, _, _, ipaddr = res[0]
149 return f, socket.inet_pton(f, ipaddr[0])
9e9cd724
YCH
150
151 def _setup_socks4(self, address, is_4a=False):
152 destaddr, port = address
153
20fbbd92 154 _, ipaddr = self._resolve_address(destaddr, SOCKS4_DEFAULT_DSTIP, use_remote_dns=is_4a, family=socket.AF_INET)
9e9cd724 155
ac668111 156 packet = struct.pack('!BBH', SOCKS4_VERSION, Socks4Command.CMD_CONNECT, port) + ipaddr
9e9cd724 157
0f06bcd7 158 username = (self._proxy.username or '').encode()
9e9cd724 159 packet += username + b'\x00'
4350b745 160
20fbbd92 161 if is_4a and self._proxy.remote_dns and ipaddr == SOCKS4_DEFAULT_DSTIP:
0f06bcd7 162 packet += destaddr.encode() + b'\x00'
4350b745
YCH
163
164 self.sendall(packet)
165
ac668111 166 version, resp_code, dstport, dsthost = struct.unpack('!BBHI', self.recvall(8))
4350b745 167
9e9cd724 168 self._check_response_version(SOCKS4_REPLY_VERSION, version)
4350b745 169
9e9cd724 170 if resp_code != Socks4Error.ERR_SUCCESS:
4350b745
YCH
171 self.close()
172 raise Socks4Error(resp_code)
173
71aff188
YCH
174 return (dsthost, dstport)
175
9e9cd724
YCH
176 def _setup_socks4a(self, address):
177 self._setup_socks4(address, is_4a=True)
4350b745 178
9e9cd724 179 def _socks5_auth(self):
ac668111 180 packet = struct.pack('!B', SOCKS5_VERSION)
4350b745 181
9e9cd724 182 auth_methods = [Socks5Auth.AUTH_NONE]
4350b745 183 if self._proxy.username and self._proxy.password:
9e9cd724
YCH
184 auth_methods.append(Socks5Auth.AUTH_USER_PASS)
185
ac668111 186 packet += struct.pack('!B', len(auth_methods))
187 packet += struct.pack(f'!{len(auth_methods)}B', *auth_methods)
4350b745
YCH
188
189 self.sendall(packet)
190
9e9cd724 191 version, method = self._recv_bytes(2)
4350b745 192
9e9cd724 193 self._check_response_version(SOCKS5_VERSION, version)
4350b745 194
54f54fcc
YCH
195 if method == Socks5Auth.AUTH_NO_ACCEPTABLE or (
196 method == Socks5Auth.AUTH_USER_PASS and (not self._proxy.username or not self._proxy.password)):
4350b745 197 self.close()
54f54fcc 198 raise Socks5Error(Socks5Auth.AUTH_NO_ACCEPTABLE)
4350b745 199
9e9cd724 200 if method == Socks5Auth.AUTH_USER_PASS:
0f06bcd7 201 username = self._proxy.username.encode()
202 password = self._proxy.password.encode()
ac668111 203 packet = struct.pack('!B', SOCKS5_USER_AUTH_VERSION)
9e9cd724 204 packet += self._len_and_data(username) + self._len_and_data(password)
4350b745
YCH
205 self.sendall(packet)
206
9e9cd724 207 version, status = self._recv_bytes(2)
4350b745 208
9e9cd724 209 self._check_response_version(SOCKS5_USER_AUTH_VERSION, version)
4350b745 210
9e9cd724 211 if status != SOCKS5_USER_AUTH_SUCCESS:
4350b745 212 self.close()
9e9cd724 213 raise Socks5Error(Socks5Error.ERR_GENERAL_FAILURE)
4350b745 214
9e9cd724
YCH
215 def _setup_socks5(self, address):
216 destaddr, port = address
217
20fbbd92 218 family, ipaddr = self._resolve_address(destaddr, None, use_remote_dns=True)
9e9cd724
YCH
219
220 self._socks5_auth()
221
222 reserved = 0
ac668111 223 packet = struct.pack('!BBB', SOCKS5_VERSION, Socks5Command.CMD_CONNECT, reserved)
4350b745 224 if ipaddr is None:
0f06bcd7 225 destaddr = destaddr.encode()
ac668111 226 packet += struct.pack('!B', Socks5AddressType.ATYP_DOMAINNAME)
9e9cd724 227 packet += self._len_and_data(destaddr)
20fbbd92 228 elif family == socket.AF_INET:
ac668111 229 packet += struct.pack('!B', Socks5AddressType.ATYP_IPV4) + ipaddr
20fbbd92 230 elif family == socket.AF_INET6:
231 packet += struct.pack('!B', Socks5AddressType.ATYP_IPV6) + ipaddr
ac668111 232 packet += struct.pack('!H', port)
4350b745
YCH
233
234 self.sendall(packet)
235
9e9cd724 236 version, status, reserved, atype = self._recv_bytes(4)
4350b745 237
9e9cd724 238 self._check_response_version(SOCKS5_VERSION, version)
4350b745 239
9e9cd724 240 if status != Socks5Error.ERR_SUCCESS:
4350b745
YCH
241 self.close()
242 raise Socks5Error(status)
243
9e9cd724 244 if atype == Socks5AddressType.ATYP_IPV4:
4350b745 245 destaddr = self.recvall(4)
9e9cd724
YCH
246 elif atype == Socks5AddressType.ATYP_DOMAINNAME:
247 alen = compat_ord(self.recv(1))
4350b745 248 destaddr = self.recvall(alen)
9e9cd724 249 elif atype == Socks5AddressType.ATYP_IPV6:
4350b745 250 destaddr = self.recvall(16)
ac668111 251 destport = struct.unpack('!H', self.recvall(2))[0]
71aff188
YCH
252
253 return (destaddr, destport)
4350b745
YCH
254
255 def _make_proxy(self, connect_func, address):
9e9cd724 256 if not self._proxy:
4350b745
YCH
257 return connect_func(self, address)
258
9e9cd724
YCH
259 result = connect_func(self, (self._proxy.host, self._proxy.port))
260 if result != 0 and result is not None:
261 return result
262 setup_funcs = {
263 ProxyType.SOCKS4: self._setup_socks4,
264 ProxyType.SOCKS4A: self._setup_socks4a,
265 ProxyType.SOCKS5: self._setup_socks5,
266 }
267 setup_funcs[self._proxy.type](address)
268 return result
269
4350b745 270 def connect(self, address):
71aff188 271 self._make_proxy(socket.socket.connect, address)
4350b745
YCH
272
273 def connect_ex(self, address):
71aff188 274 return self._make_proxy(socket.socket.connect_ex, address)