Socket Address

After defining the type of connection you need to specify the addresses involved in the connection. As sockets can be used to connect to a wide range of different types of network the address format could vary a lot and not just be a simple URL or an IP address. For this reason sockets are used with their own internal address representation and you have to use the getaddrinfo method:

socket.getaddrinfo(host, port, af=0, type=0, proto=0, flags=0)

The first two parameters are familiar and simple. The host has to be specified using a URL or an IP address. The second is the numeric port to use with the socket. The next three parameters are the same as used in creating the socket. In general you use the same parameter values for an address intended to be used with a socket as used to create the socket. If you don’t specify a parameter than getaddrinfo will return a list of tuples for each possible address type. Each tuple has the following format:

(af, type, proto, canonname, sockaddr)

where af, type and proto are as before and canonname is the cannonical name for the address type and sockaddr is the actual address, which may also be a tuple. In the case of an IP address, the sockaddr is a tuple with the format:

(IP, port)

At the time of writing the Pico only supports IPv4 and it only returns a list with a single tuple. The only really useful part of this tuple is the final element which is an IP tuple which most of the socket methods actually use. You could short circuit everything by simply creating the IP tuple and not use getaddrinfo, however, you should use it to allow for compatibility with the full sockets module and for future development.

For example:

ai = socket.getaddrinfo("www.example.com", 
                   80,socket.AF_INET,socket.SOCK_DGRAM)

returns the list:

[(2, 1, 0, '', ('93.184.216.34', 80))]

and is it the final tuple

('93.184.216.34', 80)

which we actually use in calls to socket methods. To extract this final tuple we use:

addr=ai[0][-1]

While on the subject of addresses there are two functions which can be used to convert an IP address in “dotted” form to a byte value and vice versa:

socket.inet_ntop(socket.AF_INET,bytes)

gives the dotted text form and:

socket.inet_pton(socket.AF_INET,string)

gives the binary form of the IP address.

Client Sockets

Once we have a socket object and the ability to specify an address we have to connect it to another socket to make a two-way communication channel. Data can be written and read at both sockets that form the connection. Exactly how sockets are connected depends on which is the client and which the server. In this section we look at how to connect to a server. By default all socket methods are blocking. See later for non-blocking operation.

If you want to connect a socket to a server socket than all you have to do is use:

socket.connect(address)

where address is the address of the sever socket you are trying to connect to and has to be specified as an IP tuple:

(ip,port)

For example:

socket.connect(('93.184.216.34',80))

While you can use an IP tuple, it is more usual to use socket.getaddrinfo to generate it.

You should always close a socket when you are finished using it with a call to the close method.

Once you have a connected socket you can send and receive data using a range of methods.

The methods used to send data are:

send(bytes)

sendall(bytes)

write(buf)

The receive methods are:

recv(len)

recvfrom(len)

read(len)

readinto(buf, len)

readline()

There are also:

sendto(bytes, address)

makefile(mode='rb', 
          buffering=0)