using sed to replace multiple strings

castor0troy · 04-28-2013, 02:56 AM

Hi Guys
I know this has been covered in the past but there is a small twist/

a.txt is
redcat
redhat
bluebat

b.txt is
cat mat
hat tat
.
.thousand keywords

i want to use sed to replace cat with mat[example] from b.txt

output should be
redmat
redtat
bluebat

iv tried the regular sed command but how do i do it for thousand of keywords?

thanks

jdkaye · 04-28-2013, 03:06 AM

Quote:

Originally Posted by castor0troy

iv tried the regular sed command but how do i do it for thousand of keywords?

thanks

Hi Castor,
Can you show us what you've go so far. It will make it easier to help you.
Ciao,
jdk

castor0troy · 04-28-2013, 03:09 AM

Hi
ive got the results using sed.
$sed 's/string1/string2/' file1 > file1

the questions is how do i incorprate file names in sed.
string1 and string 2 are in b.txt
example
string1=cat
string2=mat

allend · 04-28-2013, 03:10 AM

Use sed to make a sed interpreter script from your b.txt. http://www.grymoire.com/Unix/Sed.html#uh-20

castor0troy · 04-28-2013, 04:46 AM

thanks.
this link helped.

shivaa · 04-28-2013, 05:16 AM

Even you can replace the word cat with mat in vi editor, as:

Code:

~$ vi b.txt
:%s/cat/mat/g

theNbomr · 04-28-2013, 12:36 PM

allend's suggestion of creating a sed interpreter script is probably the definitive solution. Another possibility is to loop over all line entries in b.txt:

Code:

while read from to; do
   echo sed -i s/$from/$to/g a.txt
done < b.txt

This only works if the substitution text contains no whitespace. It is left as an exercise to the reader to make the script actually work on real data files. Test on backup data first.
--- rod.

David the H. · 04-29-2013, 11:42 AM

A trick I would probably use would be to run one sed command inside a process substitution, to reformat the file into a sed script that can be used directly in a second instance.

Code:

sed -f <( sed -r 's|(\w+) (\w+)|s/\1/\2/|' b.txt )  a.txt