Question

I have a regex defined follow to match http urls, anyone can help to explain in English?

^/foo/.*(?<!\.css|\.js|\.jpg)$

Answer 1

Any file or directory in directory /foo or below that is not css, js or jpg file.

^        start of string anchor
/foo/    literal "/foo/"
.*       any number of characters
(?<!...) match from here backwards must fail:
  \.     dot
  css    literal "css"
  |      or
$        end of string anchor

So, start of string, /foo/, maybe some other characters, then string ends – but just before, can’t be .css, .js or .jpg.

EDIT: Apologies for being stubborn. It is indeed an invalid regular expression for most engines, including Perl. The reason is, negative lookbehind must have fixed width; and this lookbehind can be either four characters (in case of .jpg or .css) or three characters (.js). A fix is to insert an additional “match anything” into the lookbehind so that the width is always four:

^/foo/.*(?<!\.css|.\.js|\.jpg)$

With that, it works:

perl -e 'print "/foo/bar" =~ m[^/foo/.*(?<!\.css|.\.js|\.jpg)$];'
=> 1

OP: Your problem with regexpal.com is the fact that they test for JavaScript regular expressions, which do not implement negative lookbehind at all. Regexp dialects differ in details.