Develop Reference

Can't assign bash variable in python subprocess

I am trying to assign to a variable the fingerprint of a pgp key in a bash subprocess of a python script.
Here's a snippet:
import subprocess
subprocess.run(
'''
export KEYFINGERPRINT="$(gpg --with-colons --fingerprint --list-secret-keys | sed -n 's/^fpr:::::::::\([[:alnum:]]\+\):/\1/p')"
echo "KEY FINGERPRINT IS: ${KEYFINGERPRINT}"
''',
shell=True, check=True,
executable='/bin/bash')
The code runs but echo shows an empty variable:
KEY FINGERPRINT IS:
and if I try to use that variable for other commands I get the following error:
gpg: key "" not found: Not found
HOWEVER, if I run the same exact two lines of bash code in a bash script, everything works perfectly, and the variable is correctly assigned.
What is my python script missing?
Thank you all in advance.

The problem is the backslashes in your sed command. When you paste those into a Python string, python is escaping the backslashes. To fix this, simply add an r in front of your string to make it a raw string:
import subprocess
subprocess.run(
r'''
export KEYFINGERPRINT="$(gpg --with-colons --fingerprint --list-secret-keys | sed -n 's/^fpr:::::::::\([[:alnum:]]\+\):/\1/p')"
echo "KEY FINGERPRINT IS: ${KEYFINGERPRINT}"
''',
shell=True, check=True,
executable='/bin/bash')

in order to run 2 commands in subprocess you need to run them one after each other or use ;
import subprocess
ret = subprocess.run('export KEYFINGERPRINT="$(gpg --with-colons --fingerprint --list-secret-keys | sed -n 's/^fpr:::::::::\([[:alnum:]]\+\):/\1/p')"; echo "KEY FINGERPRINT IS: ${KEYFINGERPRINT}"', capture_output=True, shell=True)
print(ret.stdout.decode())
you can use popen:
commands = '''
export KEYFINGERPRINT="$(gpg --with-colons --fingerprint --list-secret-keys | sed -n 's/^fpr:::::::::\([[:alnum:]]\+\):/\1/p')"
echo "KEY FINGERPRINT IS: ${KEYFINGERPRINT}"
'''
process = subprocess.Popen('/bin/bash', stdin=subprocess.PIPE, stdout=subprocess.PIPE)
out, err = process.communicate(commands)
print out

How can I use run instead of communicate when providing text on stdin?

trying to figure out how to do this:
command = f"adb -s {i} shell"
proc = Popen(command, stdin=PIPE, stdout=PIPE)
out, err = proc.communicate(f'dumpsys package {app_name} | grep version'.encode('utf-8'))
but in this:
command = f"adb -s {i} shell"
proc = run(command, stdin=PIPE, stdout=PIPE, shell=True)
out, err = run(f'dumpsys package {app_name} | grep version', shell=True, text=True, stdin=proc.stdout )
The idea is to make a command which require input of some kind( for example(entering shell)) and afterwards inserting another command to shell.
I've found a way online with communicate, But I wonder how to do it with run() func.
Thanks!

You only need to call run once -- pass the remote command in the input argument (and don't use shell=True in places where you don't need it).
import subprocess, shlex
proc = subprocess.run(['adb', '-s', i, 'shell'],
capture_output=True,
input=f'dumpsys package {shlex.quote(app_name)} | grep version')
shlex.quote prevents an app name that contains $(...), ;, etc from running unwanted commands on your device.

Sed with space in python

I was trying perform replace using sed in VMkernel. I used the following command,
sed s/myname/sample name/g txt.txt
I got an error saying sed: unmatched '/'.
I replaced space with \. It worked.
When I tried the same using python,
def executeCommand(cmd):
process = subprocess.Popen(cmd.split(), stdout=subprocess.PIPE)
output, error = process.communicate()
print (output.decode("utf-8"))
executeCommand('sed s/myname/sample\ name/g txt.txt')
I am getting the error sed: unmatched '/' again. I used \s instead of space I am getting the name replaced with samplesname.
How can I replace a string with space?

The simplest thing would be to not be smart about splitting the command:
executeCommand(['sed', 's/myname/sample name/g', 'txt.txt'])
Otherwise you are opening a can of worms, effectively playing a shell parser role.
Alternatively you may run the command in a shell and let the shell parse and run the command:
import subprocess
def executeCommand(cmd):
process = subprocess.Popen(cmd, shell=True, stdout=subprocess.PIPE)
# Or:
# This will run the command in /bin/bash (instead of /bin/sh)
process = subprocess.Popen(['/bin/bash', '-c', cmd], stdout=subprocess.PIPE)
output, error = process.communicate()
print (output.decode("utf-8"))
executeCommand("sed 's/myname/sample name/g' txt.txt")

Escaping both types of quotes in subprocess.Popen call to awk

My subprocess call should be calling tabix 1kg.phase1.snp.bed.gz -B test.bed | awk '{FS="\t";OFS="\t"} $4 >= 10' but is giving me errors because it has both " and ' in it. I have tried using r for a raw string but I can't figure out the right combination to prevent errors. My current call looks like:
snp_tabix = subprocess.Popen(["tabix", tgp_snp, "-B", infile, "|", "awk", """'{FS="\t";OFS="\t"}""", "$4", ">=", maf_cut_off, r"'"], stdout=subprocess.PIPE)
Which gives the error TypeError: execv() arg 2 must contain only strings

r"'" is not the issue. Most likely you're passing maf_cut_off as an integer, which is incorrect. You should use str(maf_cut_off).

There are several issues. You are trying to execute a shell command (there is a pipe | in the command). So it won't work even if you convert all variables to strings.
You could execute it using shell:
from pipes import quote
from subprocess import check_output
cmd = r"""tabix %s -B %s | awk '{FS="\t";OFS="\t"} $4 >= %d'""" % (
quote(tgp_snp), quote(infile), maf_cut_off)
output = check_output(cmd, shell=True)
Or you could use the pipe recipe from subprocess' docs:
from subprocess import Popen, PIPE
tabix = Popen(["tabix", tgp_snp, "-B", infile], stdout=PIPE)
awk = Popen(["awk", r'{FS="\t";OFS="\t"} $4 >= %d' % maf_cut_off],
stdin=tabix.stdout, stdout=PIPE)
tabix.stdout.close() # allow tabix to receive a SIGPIPE if awk exits
output = awk.communicate()[0]
tabix.wait()
Or you could use plumbum that provides some syntax sugar for shell commands:
from plumbum.cmd import tabix, awk
cmd = tabix[tgp_snp, '-B', infile]
cmd |= awk[r'{FS="\t";OFS="\t"} $4 >= %d' % maf_cut_off]
output = cmd() # run it and get output
Another option is to reproduce the awk command in pure Python. To get all lines that have 4th field larger than or equal to maf_cut_off numerically (as an integer):
from subprocess import Popen, PIPE
tabix = Popen(["tabix", tgp_snp, "-B", infile], stdout=PIPE)
lines = []
for line in tabix.stdout:
columns = line.split(b'\t', 4)
if len(columns) > 3 and int(columns[3]) >= maf_cut_off:
lines.append(line)
output = b''.join(lines)
tabix.communicate() # close streams, wait for the subprocess to exit
tgp_snp, infile should be strings and maf_cut_off should be an integer.
You could use bufsize=-1 (Popen()'s parameter) to improve time performance.

Python exec grep

I'm trying to grep a list of file from the "*.nasl" of "Openvas" which contains a certain port's number.
I can make it directly in the terminal with the command :
egrep --only-match '111' /home/gwvm/Openvas/var/lib/openvas/plugins/*.nasl |cut -d ":" -f1
This command return all the name of the nasl file which contains 111.
like :
/home/gwvm/Openvas/var/lib/openvas/plugins/SolarWinds_TFTP.nasl:111
/home/gwvm/Openvas/var/lib/openvas/plugins/trojan_horses.nasl:111
and after the cut :
/home/gwvm/Openvas/var/lib/openvas/plugins/SolarWinds_TFTP.nasl
/home/gwvm/Openvas/var/lib/openvas/plugins/trojan_horses.nasl
But when I'm in python(3.1.3) the output give me an error :
egrep:/home/gwvm/Openvas/var/lib/openvas/plugins/*.nasl: No such file or directory
i was thinking about a problem because of the "*.nasl" but when I'm trying with an existing file, same result.
Here is the part of code :
command = ("egrep --only-match '"+ str(port[0]) +"' "+ openvas_directory["locate"]["nasl"] + '*.nasl' + ' |cut -d ":" -f1 ')
process=sp.Popen(command,shell=True, stdout= sp.PIPE)
or
exec(command)
I was thinking too of a bad construction but wen I'm printing the command it gives me what i want :
egrep --only-match '111' /home/gwvm/Openvas/var/lib/openvas/plugins/*.nasl |cut -d ":" -f1
If there are any idea!

from subprocess import PIPE, Popen
x = Popen('egrep --only-match \'111\' /home/gwvm/Openvas/var/lib/openvas/plugins/*.nasl', stdout=PIPE, stderr=PIPE, shell=True)
y = Popen('cut -d ":" -f1', stdin=x.stdout, stdout=PIPE, stderr=PIPE, shell=True)
for row in y.stdout.readline():
print row
Or just use check_output()
And this is btw how you | in Popen ;)
Guidelines:
When using Popen, if you supply a command as a string, use shell=True.
If you however supply Popen with a list ['ls, '-l'] then use shell=False, that's just how it works.
If you're piping data, execute two different Popen's and use the output from the first command as stdin for the second command, this is equivilant to doing | in Linux.