Splitting a string based on a number/char pattern - Java

Question

I want to split a string with the following patterns:

abc3xab     -> ('abc', '3xa', 'b')
abc3x3xab   -> ('abc', '3x3x', 'ab')
abc3xxab    -> ('abc', '3xx', 'ab')
abc34xxab   -> ('abc', '34xx', 'ab')
abc100ab    -> ('abc100ab')
abc10axb    -> ('abc10axb')

That's it, the pattern is '[0-9]+x[a-z]' (if there's a number followed by an 'x' and any char between a-z. I tried

String[] words = string.split("[0-9]+x")

but then I only got abc and ab. What would be the correct way to do it?

Please see numerous Q&A: https://www.google.com/search?q=java+split+keep+separator — Nayuki, Feb 23 '16 at 15:50
Use this regex and use groups instead of split : `(.*?)((?:[0-9]+x+)+)(.*)` — Goufalite, Feb 23 '16 at 16:00
I will reopen the question as the solution is not an evident one. — Wiktor Stribiżew, Feb 23 '16 at 16:19

score 2 · Answer 1 · edited Feb 23 '16 at 16:15

2

Try using a combination of positive lookahead and lookbehind to keep the delimiter:

String[] words = string.split("((?<=[0-9]+x)|(?=[0-9]+x))");

Sample IDEONE demo:

String[] words = "abc3xab".split("((?<=[0-9]+x)|(?=[0-9]+x))");
System.out.println(Arrays.toString(words));
String[] words1 = "abc3xxab".split("((?<=[0-9]+x)|(?=[0-9]+x))");
System.out.println(Arrays.toString(words1));
String[] words2 = "abc34xxab".split("((?<=[0-9]+x)|(?=[0-9]+x))");
System.out.println(Arrays.toString(words2));
String[] words3 = "abc100ab".split("((?<=[0-9]+x)|(?=[0-9]+x))");
System.out.println(Arrays.toString(words3));
String[] words4 = "abc10axb".split("((?<=[0-9]+x)|(?=[0-9]+x))");
System.out.println(Arrays.toString(words4));

Results:

[abc, 3x, ab]
[abc, 3x, xab]
[abc, 3, 4x, xab]
[abc100ab]
[abc10axb]

edited Feb 23 '16 at 16:15

Wiktor Stribiżew

561,645
34
376
476

answered Feb 23 '16 at 15:50

M A

69,673
13
131
165

Java does not support infinite width lookbehind. This does not work. Upvoters give you much credit though :) – Wiktor Stribiżew Feb 23 '16 at 15:51
@WiktorStribiżew I'm aware of this limitation but it's a bit weird as it worked fine on my machine (in Java 8). – M A Feb 23 '16 at 16:09
It works for some reason, but the output is also weird. I will re-edit. – Wiktor Stribiżew Feb 23 '16 at 16:11

Splitting a string based on a number/char pattern - Java

1 Answers1