I’ve run into situations where it was necessary to read the values of bash environment variables defined in a file into a python script.

Something like that is achieved in bash simply by doing:

source env.vars

or

. env.vars

Where the content of env.vars could be something like this:

BASE=/home/user/scripts
BIN_DIR=$BASE/bin
VAR1=xxxxx
VAR2=yyyy

The solutions described here achieve that effect:

• http://stackoverflow.com/questions/17435056/read-bash-variables-into-a-python-script

But they require that every variable be exported in the source file. Some times this may not be possible due to file permissions or other restrictions.

The following solution takes care of that issue:

import re
import subprocess
from _collections import defaultdict

def load_vars(file_name):
    variables = defaultdict(str)

    command = ['bash', '-xc', 'source %s' % file_name]

    proc = subprocess.Popen(command, stderr = subprocess.PIPE)

    reg = re.compile('^\+\+ (?P<name>\w+)\=(?P<value>.+)')
    for line in proc.stderr:
        m = reg.match(line)
        if m:
            name = m.group('name')
            value = ''
            if m.group('value'):
                value = m.group('value')

            variables[name] = value

    proc.communicate()

    return variables

This also works better than just reading and parsing the source file since bash will perform variable substitution before outputting anything, so cases like

VAR2=$VAR1/abcd

are possible.

For the example file above, the function would return:

defaultdict(<type 'str'>, {'VAR1': 'xxxxx', 'BASE': '/home/user/scripts', 'BIN_DIR': '/home/user/scripts/bin', 'VAR2': 'yyyy'})