Suppose you have a text file containing the following:

strings=hello world

and you wanted to use the value of strings, which is “hello world” in your script

You can parse it with the following:

grep '^strings=' test | sed 's/strings=//g'

So the generic format is:

grep '^variablename=' filename | sed 's/variablename=//g'

The grep searches for a string beginning with ‘strings=’, in the file ‘test’, and outputs the following:

$grep '^strings=' test
strings=hello world

Next we have to remove the initial assignment part upto and including the ‘=’ sign, this is done by piping the output of grep, to sed.

Now the sed line substitutes the string “strings=” with an empty string. The substitution is done globally. Finally sed outputs just the value of the assigned variable.

Practical example:

I wanted to use the main User directory on the server, to make my backups more generic. I know that the file /etc/imscp/imscp.conf contains the following line:

USER_HOME_DIR = /var/www/virtual

So what I want is to get /var/www/virtual from this file, removing the assignment and spaces.

The solution is:

grep -i 'USER_HOME_DIR' imscp/imscp.conf | sed 's/USER_HOME_DIR[ ]=[ ]//'

which outputs:

/var/www/virtual

without the initial spaces. A space within a square brackets matches any number of spaces in the expression.

However, if you want to use these values in variables, you need to make minor adjustments to take care of variable escaping:

apache_home_var="USER_HOME_DIR"
apache_conf_file=/etc/imscp/imscp.conf
grep -i "$apache_home_var" "$apache_conf_file" | sed 's/'$apache_home_var'[ ]=[ ]//'