Redirecting FORTRAN (called via F2PY) output in Python

Question

I'm trying to figure out how to redirect output from some FORTRAN code for which I've generated a Python interface by using F2PY. I've tried:

from fortran_code import fortran_function
stdout_holder = sys.stdout
stderr_holder = sys.stderr
sys.stdout = file("/dev/null","w")
fortran_function()
sys.stdout.close()
sys.stderr.close()
sys.stdout = stdout_holder
sys.stderr = stderr_holder

This is the de facto method of redirecting output in Python, but it doesn't seem to work in this case (i.e., the output is displayed anyway).

I did find a mailing list post from 2002 saying that "It is possible to read messages from pts devices, e.g. ttysnoop does this". Information on ttysnoop seems to be pretty difficult to find online (I don't think it's been updated in quite a few years; for example, the first result on Google for "ttysnoop" has only dead links to tarballs, RPMs, and .deb's), and this request for a port to OS X received the response "No luck, it requires some linux specific utmp functions which I can't create."

I'm open to any suggestions on how to redirect the output (it doesn't have to use ttysnoop).

Thanks!

Solution 1

The stdin and stdout fds are being inherited by the C shared library.

from fortran_code import fortran_function
import os

print "will run fortran function!"

# open 2 fds
null_fds = [os.open(os.devnull, os.O_RDWR) for x in xrange(2)]
# save the current file descriptors to a tuple
save = os.dup(1), os.dup(2)
# put /dev/null fds on 1 and 2
os.dup2(null_fds[0], 1)
os.dup2(null_fds[1], 2)

# *** run the function ***
fortran_function()

# restore file descriptors so I can print the results
os.dup2(save[0], 1)
os.dup2(save[1], 2)
# close the temporary fds
os.close(null_fds[0])
os.close(null_fds[1])

print "done!"

Solution 2

Here's a context manager that I recently wrote and found useful, because I was having a similar problem with distutils.ccompiler.CCompiler.has_function while working on pymssql. I also used the file descriptor approach but I used a context manager. Here's what I came up with:

import contextlib


@contextlib.contextmanager
def stdchannel_redirected(stdchannel, dest_filename):
    """
    A context manager to temporarily redirect stdout or stderr

    e.g.:


    with stdchannel_redirected(sys.stderr, os.devnull):
        if compiler.has_function('clock_gettime', libraries=['rt']):
            libraries.append('rt')
    """

    try:
        oldstdchannel = os.dup(stdchannel.fileno())
        dest_file = open(dest_filename, 'w')
        os.dup2(dest_file.fileno(), stdchannel.fileno())

        yield
    finally:
        if oldstdchannel is not None:
            os.dup2(oldstdchannel, stdchannel.fileno())
        if dest_file is not None:
            dest_file.close()

The context for why I created this is at this blog post. Similar to yours I think.

I use it like this in a setup.py:

with stdchannel_redirected(sys.stderr, os.devnull):
    if compiler.has_function('clock_gettime', libraries=['rt']):
        libraries.append('rt')

Comments

[habanoz]

Are you sure the fortran output isn't going to stderr instead of stdout?

[Samir Unni]

Will this suppress stderr as well? If not, how can that be accomplished?

[nosklo]

@aberration: dunno, have you tested it with any fortran programs that write to stderr?

[josliber]

Thanks for this very helpful answer. When I added this code to a script that needed to redirect output many times as part of a loop, I noticed that I was leaking file descriptors. I believe you also need os.close(save[0]) and os.close(save[1]) at the end to prevent the leak.

[Sibbs Gambling]

Works like charm, but what is FDS short for?

[nosklo]

@SibbsGambling File Descriptors