[^ pL]奇怪的正则表达式行为

Question

I have a snippet of PHP that replaces all characters from the left not being Unicode letters. It works fine, with one exception, and I can't figure out why. Can anyone help?

<?php
$B=$A;
do{
    $A=$B;
    $B=preg_replace('/^[^\pL\s]/','',$B);
    }
while($B!=$A);
echo $B;
?>

If I feed it with a string like "\\*^&\\\@@\816.80831téstmé" it nicely spits out "téstmé".
$A="*^&\\\@@\816.80831[+" gives an empty string, also correct.
But, when I enter "\\*^&\\\@@\816.80831", I end up with "831", when in fact it should be an empty string.

"^&\\\@@\8016.8048.31" gives "48.31"
"^&\\\@@\8016.8148.31" gives an empty string correctly
"^&\\\@@\8016.8148067" gives "16.8148067"
"^&\\\@@\8116.8148167" the again is empty
It seems to have somethinh to do with the zero and the dot, but I can't find a pattern nor a solution. I tried adding strval, but still the same result. Maybe someone has an answer? Thnx.

Answer 1

I honestly can not find out why this is going wrong. It has to be some sort of bug. However there is a simple solution.

<?php
$B=preg_replace('/^[^\pL\s]*/','',$A);

This way it has the same functionality, except it works and has a lot less overhead.

Update: i did some testing in Java, regex coach and regexpal.com and they all do it correctly. So this has to be a bug in preg_replace.