Another rendition of the previous expression, without the lazy quantification. Here, the expression will match any character except for the underscore character, so if there is an underscore, it will stop and evaluate the lookahead.

I wonder which one is faster