Bash substring with regular expression

Question

In a bash script, I´d like to extract a variable string from a given string. I mean, i´d like to extract the string file.txt from the string:

This is the file.txt from my folder.

I tried:

var=$(echo "This is the file.txt from my folder.")
var=echo ${var##'This'}
...

but I´d like to make it in a cleaner way, using the expr, sed or awk commands.

Thanks

Edited:

I found another way (nevertheless, the answer with the sed command is the best one for me):

var=$(echo 'This is the file.txt from my folder.')
front=$(echo 'This is the ')
back=$(echo ' from my folder.')
var=${var##$front}
var=${var%$back} 
echo $var

Daniel S. · Accepted Answer · 2013-10-14 09:11:08Z

18

The following solution uses sed with s/ (substitution) to remove the leading and trailing parts:

echo "This is the file.txt from my folder." | sed "s/^This is the \(.*\) from my folder.$/\1/"

Output:

file.txt

The $ and $ enclose the part which we want to keep. This is called a group. Because it's the first (and only) group which we use in this expression, it's group 1. We later reference this group inside of the replacement string with \1.

The ^ and $ signs make sure that the complete string is matched. This is only necessary for the special case that the filename contains either "from my folder." or "This is the".

edited Oct 14, 2013 at 9:11

answered Oct 14, 2013 at 8:59

Daniel S.

6,7004 gold badges42 silver badges85 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

EverythingRightPlace · Accepted Answer · 2013-10-14 09:00:42Z

1

You could try grep:

var=$(egrep -o file.txt)

answered Oct 14, 2013 at 9:00

EverythingRightPlace

1,1971 gold badge12 silver badges37 bronze badges

Comments

Dániel · Accepted Answer · 2013-10-14 09:01:16Z

1

If 'file.txt' is a fixed string, and won't change, then you can do it like this:

var="This is the file.txt from my folder"

Notice that you don't need to echo the string to the variable, you just type it on the right side of the binary '=' operator.

echo $var |sed -e 's/^.*$file\.txt$.*$/\1/'

Depending on your sed(1) version, you can loose the escaping of the parenthesis if you have the -r (extended regexp) option in sed(1).

If 'file.txt' changes, than you can create a pattern on a best effort basis, like:

echo $var |sed -e 's/^.* $[^ ]\+\.[^ ]\+$ .*$/\1/'

answered Oct 14, 2013 at 9:01

Dániel

585 bronze badges

Comments

erwin · Accepted Answer · 2023-01-19 07:13:41Z

0

No need to use sed or awk. Since 2004, bash has built in regex matching with the =~ operator.

input="This is the file.txt from my folder."
[[ $input =~ ([[:alnum:]]+\.[[:alnum:]]+) ]]
echo ${BASH_REMATCH[0]}

Output:

file.txt

If you're not comfortable writing Regular Expressions, it's easier to do interactively with regex101. For bash, use their default PCRE (perl compatible regular expressions) flavor.

answered Jan 19, 2023 at 7:13

erwin

8081 gold badge13 silver badges17 bronze badges

Comments

RARE Kpop Manifesto · Accepted Answer · 2023-01-19 07:50:42Z

0

using gawk :

gawk '_<($_ = RT)' RS='[^ /\0]+[.][^\0/\n ]+'

file.txt

"_" serves 2 different (implicit) purposes here :

left of <, it's an empty string ""
right of <, it's used as numeric zero, yielding $0

answered Jan 19, 2023 at 7:50

RARE Kpop Manifesto

3,0356 silver badges15 bronze badges

Collectives™ on Stack Overflow

Bash substring with regular expression

5 Answers 5

Comments

Comments

Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

5 Answers 5

Comments

Comments

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related