Question

In a text file I need to remove non-escaped characters like in the following example

"\\s" -> "\s"
"\s" -> ""

I need to use sed for this excercise. I made this command:

s/\([^\]\)\\s/\1/g
s/\\\\/\\/g

The problem I now encounter is the fact that \s at the beginning of a line is not beïng removed. What sould be added to the regex to acomplish this. I also tried the * and ? but the problem with these is that they also remove the \\s strings.

So I really need a way to match ‘not a backslash or a line start’.

Answer 1

This might work for you:

echo '\s xxxxx \\s yyyyy \s zzzzz \\s \\s end' |
sed 's/\\s/\n/g;s/\\\n/\\s/g;s/\n//g'
xxxxx \s yyyyy  zzzzz \s \s end

a three step process:

1. Change \s to \n
2. Change \\n to \s
3. Delete \n‘s i.e. the original \s‘s

N.B. I use \n because it cannot appear in the original line as sed uses it as a line delimiter.