Bash. Replace with sed. End of line problem

Question

I am trying to search and replace in file using sed. But sed seems to hate end of line special char.

text=$(shuf text.txt)
echo "This is text:
$text"
sed -i "s~%text%~$text~g" t.txt

Here is what I get:

This is text:
 line 2
 line 1
 line 3
 line 4
sed: -e expression #1, char 15: unterminated `s' command

I've tried to replace \n with \r but result is not satisfying at all. There is an option to do tr '\n' , '$' and then do it backwards afterwards but it doesn't seems right.

Help?

Answer 1

Using sed and bash

Since you are using bash, try:

sed -i "s~%text%~${text//$'\n'/\\n}~g" t.txt

${text//$'\n'/\\n} is an example of bash's pattern substitution. In this case, it replaces all newline characters with with backslash followed by n which sed will interpret as a newline.

Example

Consider this text variable:

$ echo "$text"
line 2
 line 1
 line 3
 line 4

And this input file:

$ cat t.txt
start
%text%
end

Now, run our command:

$ sed "s~%text%~${text//$'\n'/\\n}~g" t.txt
start
line 2
 line 1
 line 3
 line 4
end

To change the file in-place, of course, add the -i option back.

Using awk

With the same text variable and t.txt file as above:

$ awk -v new="$text" '{gsub(/%text%/, new)} 1' t.txt
start
line 2
 line 1
 line 3
 line 4
end