regex to convert find instances a single \

Question

I am looking to replace \n with \\n but so far my regex attempts are not working (Really it is any \ by itself, \n just happens to be the use case I have in the data).

What I need is something along the lines of:

any-non-\ followed by \ followed by any-non-\

Ultimately I'll be passing the regex to java.lang.String.replaceAll so a regex formatted for that would be great, but I can probably translate another style regex into what I need.

For example I after this program to print out "true"...

public class Main
{
    public static void main(String[] args)
    {
        final String original;
        final String altered;
        final String expected;

        original = "hello\nworld";
        expected = "hello\\nworld";
        altered  = original.replaceAll("([^\\\\])\\\\([^\\\\])", "$1\\\\$2");
        System.out.println(altered.equals(expected));
   }
}

using this does work:

    altered  = original.replaceAll("\\n", "\\\\n");

Answer 1

The string should be

"[^\\\\]\\\\[^\\\\]"

You have to quadruple backslashes in a String constant that's meant for a regex; if you only doubled them, they would be escaped for the String but not for the regex.

So the actual code would be

myString = myString.replaceAll("([^\\\\])\\\\([^\\\\])", "$1\\\\$2");

Note that in the replacement, a quadruple backslash is now interpreted as two backslashes rather than one, since the regex engine is not parsing it. Edit: Actually, the regex engine does parse it since it has to check for the backreferences.

Edit: The above was assuming that there was a literal \n in the input string, which is represented in a string literal as "\\n". Since it apparently has a newline instead (represented as "\n"), the correct substitution would be

myString = myString.replaceAll("\\n", "\\\\n");

This must be repeated for any other special characters (\t, \r, \0, \\, etc.). As above, the replacement string looks exactly like the regex string but isn't.

Answer 2

I don't know exactly what you need it for, but you could have a look at StringEscapeUtils from Commons Lang. They have plenty of methods doing things like that, and if you don't find exactly what you're searching for, you could have a look at the source to find inspiration :)

Answer 3

So whenever there is 1 backslash, you want 2, but if there is 2, 3 or 4... in a row, leave them alone?

you want to replace

(?<=[^\\])\\(?!\\+)([^\\])

with

\\$1

That changes the string

hello\nworld and hello\\nworld and hello\\\nworld

into

hello\\nworld and hello\\nworld and hello\\\nworld

Answer 4

Whats wrong with using altered = original.replaceAll("\\n", "\\\\n"); ? That's exactly what i would have done.