Doc/lib/libsocket.tex

   1 \section{\module{socket} ---
   2          Low-level networking interface}
   3
   4 \declaremodule{builtin}{socket}
   5 \modulesynopsis{Low-level networking interface.}
   6
   7
   8 This module provides access to the BSD \emph{socket} interface.
   9 It is available on all modern \UNIX{} systems, Windows, MacOS, BeOS,
  10 OS/2, and probably additional platforms.
  11
  12 For an introduction to socket programming (in C), see the following
  13 papers: \citetitle{An Introductory 4.3BSD Interprocess Communication
  14 Tutorial}, by Stuart Sechrest and \citetitle{An Advanced 4.3BSD
  15 Interprocess Communication Tutorial}, by Samuel J.  Leffler et al,
  16 both in the \citetitle{\UNIX{} Programmer's Manual, Supplementary Documents 1}
  17 (sections PS1:7 and PS1:8).  The platform-specific reference material
  18 for the various socket-related system calls are also a valuable source
  19 of information on the details of socket semantics.  For \UNIX, refer
  20 to the manual pages; for Windows, see the WinSock (or Winsock 2)
  21 specification.
  22
  23 The Python interface is a straightforward transliteration of the
  24 \UNIX{} system call and library interface for sockets to Python's
  25 object-oriented style: the \function{socket()} function returns a
  26 \dfn{socket object}\obindex{socket} whose methods implement the
  27 various socket system calls.  Parameter types are somewhat
  28 higher-level than in the C interface: as with \method{read()} and
  29 \method{write()} operations on Python files, buffer allocation on
  30 receive operations is automatic, and buffer length is implicit on send
  31 operations.
  32
  33 Socket addresses are represented as a single string for the
  34 \constant{AF_UNIX} address family and as a pair
  35 \code{(\var{host}, \var{port})} for the \constant{AF_INET} address
  36 family, where \var{host} is a string representing
  37 either a hostname in Internet domain notation like
  38 \code{'daring.cwi.nl'} or an IP address like \code{'100.50.200.5'},
  39 and \var{port} is an integral port number.  Other address families are
  40 currently not supported.  The address format required by a particular
  41 socket object is automatically selected based on the address family
  42 specified when the socket object was created.
  43
  44 For IP addresses, two special forms are accepted instead of a host
  45 address: the empty string represents \constant{INADDR_ANY}, and the string
  46 \code{'<broadcast>'} represents \constant{INADDR_BROADCAST}.
  47
  48 All errors raise exceptions.  The normal exceptions for invalid
  49 argument types and out-of-memory conditions can be raised; errors
  50 related to socket or address semantics raise the error
  51 \exception{socket.error}.
  52
  53 Non-blocking mode is supported through the
  54 \method{setblocking()} method.
  55
  56 The module \module{socket} exports the following constants and functions:
  57
  58
  59 \begin{excdesc}{error}
  60 This exception is raised for socket- or address-related errors.
  61 The accompanying value is either a string telling what went wrong or a
  62 pair \code{(\var{errno}, \var{string})}
  63 representing an error returned by a system
  64 call, similar to the value accompanying \exception{os.error}.
  65 See the module \refmodule{errno}\refbimodindex{errno}, which contains
  66 names for the error codes defined by the underlying operating system.
  67 \end{excdesc}
  68
  69 \begin{datadesc}{AF_UNIX}
  70 \dataline{AF_INET}
  71 These constants represent the address (and protocol) families,
  72 used for the first argument to \function{socket()}.  If the
  73 \constant{AF_UNIX} constant is not defined then this protocol is
  74 unsupported.
  75 \end{datadesc}
  76
  77 \begin{datadesc}{SOCK_STREAM}
  78 \dataline{SOCK_DGRAM}
  79 \dataline{SOCK_RAW}
  80 \dataline{SOCK_RDM}
  81 \dataline{SOCK_SEQPACKET}
  82 These constants represent the socket types,
  83 used for the second argument to \function{socket()}.
  84 (Only \constant{SOCK_STREAM} and
  85 \constant{SOCK_DGRAM} appear to be generally useful.)
  86 \end{datadesc}
  87
  88 \begin{datadesc}{SO_*}
  89 \dataline{SOMAXCONN}
  90 \dataline{MSG_*}
  91 \dataline{SOL_*}
  92 \dataline{IPPROTO_*}
  93 \dataline{IPPORT_*}
  94 \dataline{INADDR_*}
  95 \dataline{IP_*}
  96 Many constants of these forms, documented in the \UNIX{} documentation on
  97 sockets and/or the IP protocol, are also defined in the socket module.
  98 They are generally used in arguments to the \method{setsockopt()} and
  99 \method{getsockopt()} methods of socket objects.  In most cases, only
 100 those symbols that are defined in the \UNIX{} header files are defined;
 101 for a few symbols, default values are provided.
 102 \end{datadesc}
 103
 104 \begin{funcdesc}{getfqdn}{\optional{name}}
 105 Return a fully qualified domain name for \var{name}.
 106 If \var{name} is omitted or empty, it is interpreted as the local
 107 host.  To find the fully qualified name, the hostname returned by
 108 \function{gethostbyaddr()} is checked, then aliases for the host, if
 109 available.  The first name which includes a period is selected.  In
 110 case no fully qualified domain name is available, the hostname is
 111 returned.
 112 \versionadded{2.0}
 113 \end{funcdesc}
 114
 115 \begin{funcdesc}{gethostbyname}{hostname}
 116 Translate a host name to IP address format.  The IP address is
 117 returned as a string, e.g.,  \code{'100.50.200.5'}.  If the host name
 118 is an IP address itself it is returned unchanged.  See
 119 \function{gethostbyname_ex()} for a more complete interface.
 120 \end{funcdesc}
 121
 122 \begin{funcdesc}{gethostbyname_ex}{hostname}
 123 Translate a host name to IP address format, extended interface.
 124 Return a triple \code{(hostname, aliaslist, ipaddrlist)} where
 125 \code{hostname} is the primary host name responding to the given
 126 \var{ip_address}, \code{aliaslist} is a (possibly empty) list of
 127 alternative host names for the same address, and \code{ipaddrlist} is
 128 a list of IP addresses for the same interface on the same
 129 host (often but not always a single address).
 130 \end{funcdesc}
 131
 132 \begin{funcdesc}{gethostname}{}
 133 Return a string containing the hostname of the machine where
 134 the Python interpreter is currently executing.  If you want to know the
 135 current machine's IP address, use \code{gethostbyname(gethostname())}.
 136 Note: \function{gethostname()} doesn't always return the fully qualified
 137 domain name; use \code{gethostbyaddr(gethostname())}
 138 (see below).
 139 \end{funcdesc}
 140
 141 \begin{funcdesc}{gethostbyaddr}{ip_address}
 142 Return a triple \code{(\var{hostname}, \var{aliaslist},
 143 \var{ipaddrlist})} where \var{hostname} is the primary host name
 144 responding to the given \var{ip_address}, \var{aliaslist} is a
 145 (possibly empty) list of alternative host names for the same address,
 146 and \var{ipaddrlist} is a list of IP addresses for the same interface
 147 on the same host (most likely containing only a single address).
 148 To find the fully qualified domain name, use the function
 149 \function{getfqdn()}.
 150 \end{funcdesc}
 151
 152 \begin{funcdesc}{getprotobyname}{protocolname}
 153 Translate an Internet protocol name (e.g.\ \code{'icmp'}) to a constant
 154 suitable for passing as the (optional) third argument to the
 155 \function{socket()} function.  This is usually only needed for sockets
 156 opened in ``raw'' mode (\constant{SOCK_RAW}); for the normal socket
 157 modes, the correct protocol is chosen automatically if the protocol is
 158 omitted or zero.
 159 \end{funcdesc}
 160
 161 \begin{funcdesc}{getservbyname}{servicename, protocolname}
 162 Translate an Internet service name and protocol name to a port number
 163 for that service.  The protocol name should be \code{'tcp'} or
 164 \code{'udp'}.
 165 \end{funcdesc}
 166
 167 \begin{funcdesc}{socket}{family, type\optional{, proto}}
 168 Create a new socket using the given address family, socket type and
 169 protocol number.  The address family should be \constant{AF_INET} or
 170 \constant{AF_UNIX}.  The socket type should be \constant{SOCK_STREAM},
 171 \constant{SOCK_DGRAM} or perhaps one of the other \samp{SOCK_} constants.
 172 The protocol number is usually zero and may be omitted in that case.
 173 \end{funcdesc}
 174
 175 \begin{funcdesc}{fromfd}{fd, family, type\optional{, proto}}
 176 Build a socket object from an existing file descriptor (an integer as
 177 returned by a file object's \method{fileno()} method).  Address family,
 178 socket type and protocol number are as for the \function{socket()} function
 179 above.  The file descriptor should refer to a socket, but this is not
 180 checked --- subsequent operations on the object may fail if the file
 181 descriptor is invalid.  This function is rarely needed, but can be
 182 used to get or set socket options on a socket passed to a program as
 183 standard input or output (e.g.\ a server started by the \UNIX{} inet
 184 daemon).
 185 \end{funcdesc}
 186
 187 \begin{funcdesc}{ntohl}{x}
 188 Convert 32-bit integers from network to host byte order.  On machines
 189 where the host byte order is the same as network byte order, this is a
 190 no-op; otherwise, it performs a 4-byte swap operation.
 191 \end{funcdesc}
 192
 193 \begin{funcdesc}{ntohs}{x}
 194 Convert 16-bit integers from network to host byte order.  On machines
 195 where the host byte order is the same as network byte order, this is a
 196 no-op; otherwise, it performs a 2-byte swap operation.
 197 \end{funcdesc}
 198
 199 \begin{funcdesc}{htonl}{x}
 200 Convert 32-bit integers from host to network byte order.  On machines
 201 where the host byte order is the same as network byte order, this is a
 202 no-op; otherwise, it performs a 4-byte swap operation.
 203 \end{funcdesc}
 204
 205 \begin{funcdesc}{htons}{x}
 206 Convert 16-bit integers from host to network byte order.  On machines
 207 where the host byte order is the same as network byte order, this is a
 208 no-op; otherwise, it performs a 2-byte swap operation.
 209 \end{funcdesc}
 210
 211 \begin{funcdesc}{inet_aton}{ip_string}
 212 Convert an IP address from dotted-quad string format
 213 (e.g.\ '123.45.67.89') to 32-bit packed binary format, as a string four
 214 characters in length.
 215
 216 Useful when conversing with a program that uses the standard C library
 217 and needs objects of type \ctype{struct in_addr}, which is the C type
 218 for the 32-bit packed binary this function returns.
 219
 220 If the IP address string passed to this function is invalid,
 221 \exception{socket.error} will be raised. Note that exactly what is
 222 valid depends on the underlying C implementation of
 223 \cfunction{inet_aton()}.
 224 \end{funcdesc}
 225
 226 \begin{funcdesc}{inet_ntoa}{packed_ip}
 227 Convert a 32-bit packed IP address (a string four characters in
 228 length) to its standard dotted-quad string representation
 229 (e.g. '123.45.67.89').
 230
 231 Useful when conversing with a program that uses the standard C library
 232 and needs objects of type \ctype{struct in_addr}, which is the C type
 233 for the 32-bit packed binary this function takes as an argument.
 234
 235 If the string passed to this function is not exactly 4 bytes in
 236 length, \exception{socket.error} will be raised.
 237 \end{funcdesc}
 238
 239 \begin{datadesc}{SocketType}
 240 This is a Python type object that represents the socket object type.
 241 It is the same as \code{type(socket(...))}.
 242 \end{datadesc}
 243
 244
 245 \begin{seealso}
 246   \seemodule{SocketServer}{Classes that simplify writing network servers.}
 247 \end{seealso}
 248
 249
 250 \subsection{Socket Objects \label{socket-objects}}
 251
 252 Socket objects have the following methods.  Except for
 253 \method{makefile()} these correspond to \UNIX{} system calls
 254 applicable to sockets.
 255
 256 \begin{methoddesc}[socket]{accept}{}
 257 Accept a connection.
 258 The socket must be bound to an address and listening for connections.
 259 The return value is a pair \code{(\var{conn}, \var{address})}
 260 where \var{conn} is a \emph{new} socket object usable to send and
 261 receive data on the connection, and \var{address} is the address bound
 262 to the socket on the other end of the connection.
 263 \end{methoddesc}
 264
 265 \begin{methoddesc}[socket]{bind}{address}
 266 Bind the socket to \var{address}.  The socket must not already be bound.
 267 (The format of \var{address} depends on the address family --- see
 268 above.)  \strong{Note:}  This method has historically accepted a pair
 269 of parameters for \constant{AF_INET} addresses instead of only a
 270 tuple.  This was never intentional and is no longer be available in
 271 Python 2.0.
 272 \end{methoddesc}
 273
 274 \begin{methoddesc}[socket]{close}{}
 275 Close the socket.  All future operations on the socket object will fail.
 276 The remote end will receive no more data (after queued data is flushed).
 277 Sockets are automatically closed when they are garbage-collected.
 278 \end{methoddesc}
 279
 280 \begin{methoddesc}[socket]{connect}{address}
 281 Connect to a remote socket at \var{address}.
 282 (The format of \var{address} depends on the address family --- see
 283 above.)  \strong{Note:}  This method has historically accepted a pair
 284 of parameters for \constant{AF_INET} addresses instead of only a
 285 tuple.  This was never intentional and is no longer available in
 286 Python 2.0 and later.
 287 \end{methoddesc}
 288
 289 \begin{methoddesc}[socket]{connect_ex}{address}
 290 Like \code{connect(\var{address})}, but return an error indicator
 291 instead of raising an exception for errors returned by the C-level
 292 \cfunction{connect()} call (other problems, such as ``host not found,''
 293 can still raise exceptions).  The error indicator is \code{0} if the
 294 operation succeeded, otherwise the value of the \cdata{errno}
 295 variable.  This is useful, e.g., for asynchronous connects.
 296 \strong{Note:}  This method has historically accepted a pair of
 297 parameters for \constant{AF_INET} addresses instead of only a tuple.
 298 This was never intentional and is no longer be available in Python
 299 2.0 and later.
 300 \end{methoddesc}
 301
 302 \begin{methoddesc}[socket]{fileno}{}
 303 Return the socket's file descriptor (a small integer).  This is useful
 304 with \function{select.select()}.
 305 \end{methoddesc}
 306
 307 \begin{methoddesc}[socket]{getpeername}{}
 308 Return the remote address to which the socket is connected.  This is
 309 useful to find out the port number of a remote IP socket, for instance.
 310 (The format of the address returned depends on the address family ---
 311 see above.)  On some systems this function is not supported.
 312 \end{methoddesc}
 313
 314 \begin{methoddesc}[socket]{getsockname}{}
 315 Return the socket's own address.  This is useful to find out the port
 316 number of an IP socket, for instance.
 317 (The format of the address returned depends on the address family ---
 318 see above.)
 319 \end{methoddesc}
 320
 321 \begin{methoddesc}[socket]{getsockopt}{level, optname\optional{, buflen}}
 322 Return the value of the given socket option (see the \UNIX{} man page
 323 \manpage{getsockopt}{2}).  The needed symbolic constants
 324 (\constant{SO_*} etc.) are defined in this module.  If \var{buflen}
 325 is absent, an integer option is assumed and its integer value
 326 is returned by the function.  If \var{buflen} is present, it specifies
 327 the maximum length of the buffer used to receive the option in, and
 328 this buffer is returned as a string.  It is up to the caller to decode
 329 the contents of the buffer (see the optional built-in module
 330 \refmodule{struct} for a way to decode C structures encoded as strings).
 331 \end{methoddesc}
 332
 333 \begin{methoddesc}[socket]{listen}{backlog}
 334 Listen for connections made to the socket.  The \var{backlog} argument
 335 specifies the maximum number of queued connections and should be at
 336 least 1; the maximum value is system-dependent (usually 5).
 337 \end{methoddesc}
 338
 339 \begin{methoddesc}[socket]{makefile}{\optional{mode\optional{, bufsize}}}
 340 Return a \dfn{file object} associated with the socket.  (File objects
 341 are described in \ref{bltin-file-objects}, ``File Objects.'')
 342 The file object references a \cfunction{dup()}ped version of the
 343 socket file descriptor, so the file object and socket object may be
 344 closed or garbage-collected independently.
 345 \index{I/O control!buffering}The optional \var{mode}
 346 and \var{bufsize} arguments are interpreted the same way as by the
 347 built-in \function{open()} function.
 348 \end{methoddesc}
 349
 350 \begin{methoddesc}[socket]{recv}{bufsize\optional{, flags}}
 351 Receive data from the socket.  The return value is a string representing
 352 the data received.  The maximum amount of data to be received
 353 at once is specified by \var{bufsize}.  See the \UNIX{} manual page
 354 \manpage{recv}{2} for the meaning of the optional argument
 355 \var{flags}; it defaults to zero.
 356 \end{methoddesc}
 357
 358 \begin{methoddesc}[socket]{recvfrom}{bufsize\optional{, flags}}
 359 Receive data from the socket.  The return value is a pair
 360 \code{(\var{string}, \var{address})} where \var{string} is a string
 361 representing the data received and \var{address} is the address of the
 362 socket sending the data.  The optional \var{flags} argument has the
 363 same meaning as for \method{recv()} above.
 364 (The format of \var{address} depends on the address family --- see above.)
 365 \end{methoddesc}
 366
 367 \begin{methoddesc}[socket]{send}{string\optional{, flags}}
 368 Send data to the socket.  The socket must be connected to a remote
 369 socket.  The optional \var{flags} argument has the same meaning as for
 370 \method{recv()} above.  Returns the number of bytes sent.
 371 \end{methoddesc}
 372
 373 \begin{methoddesc}[socket]{sendto}{string\optional{, flags}, address}
 374 Send data to the socket.  The socket should not be connected to a
 375 remote socket, since the destination socket is specified by
 376 \var{address}.  The optional \var{flags} argument has the same
 377 meaning as for \method{recv()} above.  Return the number of bytes sent.
 378 (The format of \var{address} depends on the address family --- see above.)
 379 \end{methoddesc}
 380
 381 \begin{methoddesc}[socket]{setblocking}{flag}
 382 Set blocking or non-blocking mode of the socket: if \var{flag} is 0,
 383 the socket is set to non-blocking, else to blocking mode.  Initially
 384 all sockets are in blocking mode.  In non-blocking mode, if a
 385 \method{recv()} call doesn't find any data, or if a
 386 \method{send()} call can't immediately dispose of the data, a
 387 \exception{error} exception is raised; in blocking mode, the calls
 388 block until they can proceed.
 389 \end{methoddesc}
 390
 391 \begin{methoddesc}[socket]{setsockopt}{level, optname, value}
 392 Set the value of the given socket option (see the \UNIX{} manual page
 393 \manpage{setsockopt}{2}).  The needed symbolic constants are defined in
 394 the \module{socket} module (\code{SO_*} etc.).  The value can be an
 395 integer or a string representing a buffer.  In the latter case it is
 396 up to the caller to ensure that the string contains the proper bits
 397 (see the optional built-in module
 398 \refmodule{struct}\refbimodindex{struct} for a way to encode C
 399 structures as strings).
 400 \end{methoddesc}
 401
 402 \begin{methoddesc}[socket]{shutdown}{how}
 403 Shut down one or both halves of the connection.  If \var{how} is
 404 \code{0}, further receives are disallowed.  If \var{how} is \code{1},
 405 further sends are disallowed.  If \var{how} is \code{2}, further sends
 406 and receives are disallowed.
 407 \end{methoddesc}
 408
 409 Note that there are no methods \method{read()} or \method{write()};
 410 use \method{recv()} and \method{send()} without \var{flags} argument
 411 instead.
 412
 413
 414 \subsection{Example \label{socket-example}}
 415
 416 Here are two minimal example programs using the TCP/IP protocol:\ a
 417 server that echoes all data that it receives back (servicing only one
 418 client), and a client using it.  Note that a server must perform the
 419 sequence \function{socket()}, \method{bind()}, \method{listen()},
 420 \method{accept()} (possibly repeating the \method{accept()} to service
 421 more than one client), while a client only needs the sequence
 422 \function{socket()}, \method{connect()}.  Also note that the server
 423 does not \method{send()}/\method{recv()} on the
 424 socket it is listening on but on the new socket returned by
 425 \method{accept()}.
 426
 427 \begin{verbatim}
 428 # Echo server program
 429 import socket
 430
 431 HOST = ''                 # Symbolic name meaning the local host
 432 PORT = 50007              # Arbitrary non-privileged port
 433 s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
 434 s.bind((HOST, PORT))
 435 s.listen(1)
 436 conn, addr = s.accept()
 437 print 'Connected by', addr
 438 while 1:
 439     data = conn.recv(1024)
 440     if not data: break
 441     conn.send(data)
 442 conn.close()
 443 \end{verbatim}
 444
 445 \begin{verbatim}
 446 # Echo client program
 447 import socket
 448
 449 HOST = 'daring.cwi.nl'    # The remote host
 450 PORT = 50007              # The same port as used by the server
 451 s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
 452 s.connect((HOST, PORT))
 453 s.send('Hello, world')
 454 data = s.recv(1024)
 455 s.close()
 456 print 'Received', `data`
 457 \end{verbatim}