Python socket recv function

Question

In the python socket module recv method socket.recv(bufsize[, flags]) docs here , it states:

Receive data from the socket. The return value is a bytes object representing the data received.
The maximum amount of data to be received at once is specified by bufsize.

I'm aware that bufsize represents the MAX amount of data received at once, and that if the amount of data received is LESS than bufsize, that means that the number of bytes sent by socket on the other end is less than bufsize

Is it possible that the data returned from the 1st call to socket.recv(bufsize) is < bufsize but there is still data left in the network buffer?

Eg.

data = socket.recv(10)
print(len(data))       # outputs 5
data = socket.recv(10) # calling `socket.recv(10)` returns more data without the
                       # socket on the other side doing `socket.send(data)`

Can a scenario in the example ever occur and does this apply for unix domain sockets as well as regular TCP/IP sockets?

Don’t know about unix sockets, but for TCP sockets the answer is yes - you have to assume that what you receive may be more or less than any one send() call content. That’s why you often need to add a protocol on top of TCP to get delimited messages. Or use a protocol like Websocket whic runs on top of tcp and does ensure message integrity. — DisappointedByUnaccountableMod
– DisappointedByUnaccountableMod, Commented Oct 6, 2020 at 9:44

Serge Ballesta · Accepted Answer · 2020-10-06 10:07:24Z

2

The real problem in network communication is that the receiver cannot control when and how the network delivers the data.

If the size of the data returned by recv is less than the requested size, that means that at the moment of the recv call, no more data was available in the local network buffer. So if you can make sure that:

the sender has stopped sending data
the network could deliver all the data

then a new revc call will block.

The problem is that in real world cases, you can never make sure of the two above assumptions. TCP is a stream protocol, which only guarantees that all sent bytes will go to the receiver and in correct order. But if offers no guarantee on the timing, and sent packets can be fragmented or re-assembled by the network (starting from the network TCP stack on the sender and ending at the reciever TCP stack)

answered Oct 6, 2020 at 10:07

Serge Ballesta

150k13 gold badges137 silver badges267 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

applecider Over a year ago

so the network buffer is empty at the moment of the the recv call. For a TCP socket, there is no guarantee on timing, what about unix domain sockets? Considering just 1 send() call and 1 recv() call, can the above scenario occur?

user207421 Over a year ago

@ShideFoo Same. There is nothing in the specification of send() or recv() that entitles you to make that assumption.

applecider · Accepted Answer · 2020-10-15 04:07:09Z

1

Found a similar post that follows up on this: How can I reliably read exactly n bytes from a TCP socket?

Basically use socket.makefile() to return a file object and call read(num_bytes) to return exactly the amount requested, else block.

fh = socket.makefile(mode='b')
data = fh.read(LENGTH_BYTES)

answered Oct 15, 2020 at 4:07

applecider

3031 gold badge3 silver badges9 bronze badges

Collectives™ on Stack Overflow

Python socket recv function

2 Answers 2

2 Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related