Splitting a string into two with its first, digit-only part and the rest?

Question

A string is comprised of the following:

An optional sequence of ASCII digits.
A sequence of ASCII lowercase letters.

I'm trying to do the split in one single regex that I could use like this:

String string = "123abc";
var array = string.split(...);
System.out.println(java.util.Arrays.toString(array));
// prints [123, abc]

The closest regex I've come to is the following:

(?<=\d+)

Example:

String string = "123abc";
var array = string.split("(?<=\\d+)");
System.out.println(java.util.Arrays.toString(array));
// prints [1, 2, 3, abc]

Technically, I could do this without any regex, but here, it's important to be done with regex.

A solution to prove I can do it normally:

String string = "123abc";
int i = 0;
for(; i < string.length() && Character.isDigit(string.charAt(i)); i++)
  ;
String[] array = {
    string.substring(0, i), string.substring(i)
  };
System.out.println(java.util.Arrays.toString(array));
// prints [123, abc]

Another way of doing it would be:

String string = "123abc";
String[] array = {
    string.replaceAll("\\D", ""),
    string.replaceAll("\\d", "")
  };
System.out.println(java.util.Arrays.toString(array));
// prints [123, abc]

Matching examples:

In:                 Out:
123abc              [ "123", "abc" ]
0a                  [ "0", "a" ]
a                   [ "", "a" ]

@WiktorStribiżew Thanks but it doesn't work for `"abc".split(...)` which returns `[ "abc" ]`, not `[ "", "abc" ]`. — Olivier Grégoire, Dec 20 '18 at 19:36
Then match with `"(\\d*)(\\D+)"` or `"(\\d*)(\\D*)"` and get Group 1 and 2 — Wiktor Stribiżew, Dec 20 '18 at 19:38
Is it also important to use `.split()`? If I wanted to use regex for this, I'd use matching groups instead: `string.matches("(\d+)([a-z]+)")` — Geoffrey Wiseman, Dec 20 '18 at 19:39
@GeoffreyWiseman Yes, it's important: I'm actually code-golfing something and split is usually the best way to win huge amounts of bytes. I've already golfed everything else but that part, and even with a 20-characters regex, I'd still gain bytes in the end. So yes, there should only be a `split`. As shown in the solutions that work without regex or without split, I have no issue to make it work without split. I'm only interested in a `split` solution. — Olivier Grégoire, Dec 20 '18 at 19:43
Try `split("(?<=\\d|^)(?=[a-z])")` or `split("(?<!\\D)(?=[a-z])")` — Wiktor Stribiżew, Dec 20 '18 at 19:45
[Pattern.split()](https://docs.oracle.com/javase/10/docs/api/java/util/regex/Pattern.html#split(java.lang.CharSequence,int)) documentation says `When there is a positive-width match at the beginning of the input sequence then an empty leading substring is included at the beginning of the resulting array. A zero-width match at the beginning however never produces such empty leading substring.`. — Venkata Raju, Dec 20 '18 at 21:50
@VenkataRaju You can answer that: saying "what you want is not possible" is a valid answer! ;) — Olivier Grégoire, Dec 20 '18 at 21:52

score 1 · Accepted Answer · answered Dec 20 '18 at 21:55

Pattern.split() documentation says:

When there is a positive-width match at the beginning of the input sequence then an empty leading substring is included at the beginning of the resulting array. A zero-width match at the beginning however never produces such empty leading substring.

So what you are trying to achieve may not be possible with Regular Expression.

Splitting a string into two with its first, digit-only part and the rest?

1 Answers1