bash sed/awk: replace string from a file searching in another file

Question

i have a question about search and replacement on Debian. I have two files. One with:

a:b
c:d
e:f

and the other with:

 e
 c
 a`

In the second file I want to replace a with b, c with d, e with f. How can I do this?

Answer 1

$ awk -F: 'FNR==NR{a[$1]=$2;next} {for (i in a)sub(i, a[i]);print}' file1 file2
 f
 d
 b

How it works

• -F:

  This tells awk to split fields on the colon.

• FNR==NR{a[$1]=$2;next}

  While reading the first file, this tells awk to create a dictionary a of the translations that we want to do.

• for (i in a)sub(i, a[i])

  While reading the second file file, this tells awk to substitute in for every entry that we have stored in our dictionary a.

• print

  After we have made the substitutions, this tells awk to print the line.

Replacing file2

To replace file2 with the new version:

awk -F: 'FNR==NR{a[$1]=$2;next} {for (i in a)sub(i, a[i]);print}' file1 file2 >tmp && mv tmp file2

With very recent versions of awk, there is a shortcut option for this: -i inplace. Beneath the surface, though, what this option does is exactly what the command above does.

Answer 2

A bit more complicated:

sed -f <(sed 's!\(.*\):\(.*\)!s/\1/\2/!' file1) file2

• sed 's!\(.*\):\(.*\)!s/\1/\2/!' file1 reads the first file and outputs:

  s/a/b/
  s/c/d/
  s/e/f/
  

• <(the_above) runs the above command with output to a temporary file.
• sed -f <(…) file2 runs sed on file2 using that temporary file as an input (script) file.

To send the output back into file2, add a -i option:

sed -i -f <(sed 's!\(.*\):\(.*\)!s/\1/\2/!' file1) file2