Splitting on "," but not "\,"

Question

I'm looking for a regular expression to match , but ignore \, in Java's regex engine. This comes close:

[^\\],

However, it matches the previous character (in addition to the comma), which won't work.

Perhaps the regular expression approach is the wrong one altogether. I was intending to use String.split() to parse a simple CSV file (can't use an external library) with escaped commas.

Ah, definitely a duplicate. Sorry, I missed it when searching for solutions. — David Chouinard, Oct 06 '13 at 19:34

Rohit Jain · Answer 1 · 2013-10-06T19:54:01.687

7

You need a negative look-behind assertion here:

String[] arr = str.split("(?<![^\\\\]\\\\),");

Note that you need 4 backslashes there. First escape the backslash for Java string literal. And then again escape both the backslashes for regex.

edited Oct 06 '13 at 19:54

answered Oct 06 '13 at 19:30

Rohit Jain

203,151
43
392
509

Great, thanks! I was reading up on lookaheads before I gave up and asked this question. – David Chouinard Oct 06 '13 at 19:34
@DavidChouinard You're welcome :) – Rohit Jain Oct 06 '13 at 19:36
This fails on a string like `Split\\,this`. – Tim Pietzcker Oct 06 '13 at 19:51
@TimPietzcker. oops. missed the edge case. Will edit the regex. Thanks :) – Rohit Jain Oct 06 '13 at 19:53
OK, so what about `Don't split\\\,this`? – Tim Pietzcker Oct 06 '13 at 19:58
@TimPietzcker. Ah! if only Java regex supported variable length look-behinds. Still thinking of a way. – Rohit Jain Oct 06 '13 at 20:03

Splitting on "," but not "\,"

1 Answers1