Multiple process output

[Paulo J. Matos]

Hi all,

I am have a function which executes a command in the shell. The stdout 
and stderr of the command should be multipled to two strings for stdout 
and stderr respectively and stdout and stderr of the current process 
respectively.

I have done like this:
from subprocess import Popen, PIPE, STDOUT
from select import select
from os import read
from sys import stdout, stderr

def communicate(p):
     """ 
 

     Multiplex the subprocess stdout/stderr to the process stdout/stderr 
 

     and a tuple of strings 
 

     """
     output = []
     errput = []

     while True:
         (ready_to_read, _, _) = select([p.stdout, p.stderr], [], [])
         dataout = ""
         dataerr = ""

         if p.stdout in ready_to_read:
             dataout = read(p.stdout.fileno(), 1024)
             stdout.write(dataout)
             output.append(dataout)

         if p.stderr in ready_to_read:
             dataerr = read(p.stderr.fileno(), 1024)
             stderr.write(dataerr)
             errput.append(dataerr)

         if dataout == "" and dataerr == "":
             p.stdout.close()
             p.stderr.close()
             break

     return (''.join(output), ''.join(errput))

def exe(s, cwd=None, output_command=True):
     if output_command:
         print s
     p = Popen(s, stdin=None, stdout=PIPE, stderr=PIPE, shell=True, cwd=cwd)
     (output, err) = communicate(p)
     rc = p.wait()
     return (rc, output, err)


Unfortunately, the program is _sometimes_ but not always getting stuck 
on the call to select. I don't really understand when this happens. Any 
suggestions to the above code so select doesn't block the function?

Cheers,

-- 
PMatos