java负面lookbehind正则表达式bug？

Question

My java @windows is:

java version "1.6.0_29"
Java(TM) SE Runtime Environment (build 1.6.0_29-b11)
Java HotSpot(TM) Client VM (build 20.4-b02, mixed mode, sharing)

Code

Pattern.compile(".+(?<!(xxx|idea|perforator|pycharm|s).*)").matcher("xxx").matches() //returns TRUE while it should return FALSE
Pattern.compile(".+(?<!(xxx|idea|perforator|pycharm|s).*)").matcher("perforator").matches() //returns FALSE

Looks like negative lookbehind fails if string is smaller than 8 chars.

Is it bug or do I misunderstand something about regex?

Answer 1

Lookbehinds in Java cannot have variable-length things like .* , only things like alternations and finite repetitions.

More Information: http://www.regular-expressions.info/lookaround.html#limitbehind

Answer 2

If you need to match substrings that do not contain some words (as a part of a bigger expression) you can use:

(?s:(?!xxx|idea|perforator|pycharm|s).)*

If that is the only thing you are doing, just invert the result of the match:

xxx|idea|perforator|pycharm|s