Pattern: how subtract matched character in character class?

Question

Is it possible to subtract a matched character in a character class?

Java docs are having examples about character classes with subtraction:

[a-z&&[^bc]]    - a through z, except for b and c: [ad-z] (subtraction)
[a-z&&[^m-p]]   - a through z, and not m through p: [a-lq-z](subtraction)

I want to write pattern, which matches two pairs of word characters, when pairs are not the same:

1) "aaaa123" - should NOT match
2) "aabb123" - should match "aabb" part
3) "aa--123" - should NOT match

I am close to success with following pattern:

([\w])\1([\w])\2

but of course it does not work in case 1, so I need to subtract the match of first group. But when I try to do this:

Pattern p = Pattern.compile("([\\w])\\1([\\w&&[^\\1]])\\2");

I am getting an exception:

Exception in thread "main" java.util.regex.PatternSyntaxException: Illegal/unsupported escape sequence near index 17
([\w])\1([\w&&[^\1]])\2
                 ^
    at java.util.regex.Pattern.error(Pattern.java:1713)

So seems it does not work with groups, but just with listing specific characters. Following pattern compiles with no problems:

Pattern p = Pattern.compile("([\\w])\\1([\\w&&[^a]])\\2");

Is there any other way to write such pattern?

Answer 1

Use

Pattern p = Pattern.compile("((\\w)\\2(?!\\2))((\\w)\\4)");

Your characters will be in groups 1 and 3.

This works by using a negative lookahead, to make sure the character following the second character in the first character group is a different character.