Python Language More flexibility with Popen


Example

Using subprocess.Popen give more fine-grained control over launched processes than subprocess.call.

Launching a subprocess

process = subprocess.Popen([r'C:\path\to\app.exe', 'arg1', '--flag', 'arg'])

The signature for Popen is very similar to the call function; however, Popen will return immediately instead of waiting for the subprocess to complete like call does.

Waiting on a subprocess to complete

process = subprocess.Popen([r'C:\path\to\app.exe', 'arg1', '--flag', 'arg'])
process.wait()

Reading output from a subprocess

process = subprocess.Popen([r'C:\path\to\app.exe'], stdout=subprocess.PIPE, stderr=subprocess.PIPE)

# This will block until process completes
stdout, stderr = process.communicate()
print stdout
print stderr

Interactive access to running subprocesses

You can read and write on stdin and stdout even while the subprocess hasn't completed. This could be useful when automating functionality in another program.

Writing to a subprocess

process = subprocess.Popen([r'C:\path\to\app.exe'], stdout = subprocess.PIPE, stdin = subprocess.PIPE)


process.stdin.write('line of input\n') # Write input

line  = process.stdout.readline() # Read a line from stdout

# Do logic on line read.

However, if you only need one set of input and output, rather than dynamic interaction, you should use communicate() rather than directly accessing stdin and stdout.

Reading a stream from a subprocess

In case you want to see the output of a subprocess line by line, you can use the following snippet:

process = subprocess.Popen(<your_command>, stdout=subprocess.PIPE)
while process.poll() is None:
    output_line = process.stdout.readline()

in the case the subcommand output do not have EOL character, the above snippet does not work. You can then read the output character by character as follows:

process = subprocess.Popen(<your_command>, stdout=subprocess.PIPE)
while process.poll() is None:
    output_line = process.stdout.read(1)

The 1 specified as argument to the read method tells read to read 1 character at time. You can specify to read as many characters you want using a different number. Negative number or 0 tells to read to read as a single string until the EOF is encountered (see here).

In both the above snippets, the process.poll() is None until the subprocess finishes. This is used to exit the loop once there is no more output to read.

The same procedure could be applied to the stderr of the subprocess.