ruby - Understanding negative look aheads in regular expressions -

i want match urls not contain string 'localhost' using ruby regex

based on answers , comments here, put 2 solutions, both of seem work:

solution a:

(?!.*localhost)^.*$  

example: http://rubular.com/r/tqtbwacl3g

solution b:

^((?!localhost).)*$  

example: http://rubular.com/r/2kknqzumwf

the problem don't understand they're doing. example, according docs, ^ can used in various ways:

[^abc]  single character except: a, b, or c   ^ start of line   

but don't how it's being applied here.

can breakdown these expressions me, , how differ 1 another?

in both of cases, ^ start of line (since it's not used inside character class). since both ^ , lookahead zero-width assertions, can switch them around in first case - think makes bit easier explain:

^(?!.*localhost).*$  

the ^ anchors expression beginning of string. lookahead starts position , tries find localhost anywhere string (the "anywhere" taken care of .* in front of localhost). if localhost can found, subexpression of lookahead matches , therefore negative lookahead causes pattern fail. since lookahead bound start @ beginning of string adjacent ^ means, pattern overall cannot match. if, .*localhost not match (and hence localhost not occur in string), lookahead succeeds, , .*$ takes care of matching rest of string.

now other one

^((?!localhost).)*$ 

this time lookahead checks @ current position (there no .* inside it). but lookahead repeated every single character. way check every single position again. here happens: ^ makes sure we're starting @ beginning of string again. lookahead checks whether word localhost found @ position. if not, well, , . consumes 1 character. * repeats both of steps. 1 character further in string, , lookahead checks whether second character starts word localhost - again, if not, well, , . consumes character. done every single character in string, until reach end.

in particular case both methods equivalent, , select 1 based on performance (if matters) or readability (if not; first one). however, in other cases second variant preferable, because allows repetition fixed part of string, whereas first variant check entire string.