From: Rick DeNatale <rick.denatale@...>
Date: 2008-05-09T08:32:34+09:00
Subject: Re: Doing an AND in regexp char class

On Thu, May 8, 2008 at 5:40 PM, Todd Benson <caduceass@gmail.com> wrote:
> This question arises out of a couple of recent threads and may or may
> not be a Ruby-specific question.
>
> I can check with a character class if one of the characters in the
> class exists or does not exist, but can I use a regexp to check if a
> string absolutely contains all of the characters in the class?
>
> Using a set perspective, I can do it like this in irb...
>
> s1 = "hello there"
> s2 = "ohi"
> (s2.unpack('c*') & s1.unpack('c*')).size == s2.size
>
> => false
>
> I use unpack to avoid creating a bunch of String objects, one for each
> element in the array, which would happen if I used #split.  What I'm
> wondering is if there is a way to do this with a simple regexp.

REs can do this, but may not be the best way.  The way that comes to
mind is to see if the string matches the characters in any order,
i.e. for "ohi" either ohi, oih, hio, hoi, iho, or ioh
so something like

/(o([^h]*h[^i]i|[^i]*i[^h]*h)|(h([^i]*i[^o]*o|[^o]*[^i]*i)|o([^h]*h[^o]*o|[^o]*o[^h]*h)/

meaning

   o followed by either
      zero or more non-h's folllowed by an h followed by zero or more
non-i's folllowed by an i
      or
      zero or more non-i's followed by an i followed by zero or more
non-h's followed by an h
   or
   h followed by either
      ...
  ...

I would be possible to generate such an RE from the string.

But maybe someone cleverer with REs has a better approach.

-- 
Rick DeNatale

My blog on Ruby
http://talklikeaduck.denhaven2.com/