s/\\n/\n/g;
[download]

You've broken the solution by creating it: I might have had a plaintext \n in the text before, that now becomes a newline. Oops.

If you escape any characters, you need to escape the escape character as well:

s/\\/\\s/g;
s/\n/\\n/g;
s/\r/\\r/g;
[download]

so that a plaintext \n becomes \sn, which won't be matched by the regex that unescapes newlines. Then you can do the same thing backwards:

s/\\n/\n/g;
s/\\r/\r/g;
s/\\s/\\/g;
[download]

Just don't forget to leave the last unescaping last. If the escape code for the character is itself (ie s/\\/\\\\/g;), then unescaping becomes much more complicated, because you could match a \n that's actually part of a sequence like \\n. So you'll have to unwrap the string very carefully:

my %xlat = ('\\' => "\\", 'n' => "\n", 'r' => "\r");

s/\G(?:\\(.)|(.))/defined($1) ? $xlat{$1} : $2/eg;
[download]

And I haven't tested this, so I'm not even sure I got it right on the first try.

Quoting and escaping are hair-raisingly complex issues - be careful.

I have discovered that there are two types of command interfaces in the world of computing:
good interfaces and user interfaces.

— Dan J. Bernstein, The qmail security guarantee

Makeshifts last the longest.