From: Bob Showalter <bob_showalter@...>
Date: 2005-12-22T06:48:30+09:00
Subject: Re: Odd behavior of String#scan

Warren Brown wrote:
>     First off, the problem I am trying to solve can be simplified down
> to:
> 
> 'abcSTARTdef,ghi,jkl,ENDmno'.scan(/START([^,]*,)*END/)
> 
>     What I want is [["def,"], ["ghi,"], ["jkl,"]] (or the same thing
> without the commas), and I still need a way to achieve this.  I can
> accomplish it with:
> 
> 'abcSTARTdef,ghi,jkl,ENDmno'.scan(/START(.*)END/)[0][0].split(/,/)
> 
>     But this does two operations where it seems like one should suffice.
> Does someone know of a way to do this in a single operation?

I would suggest:

   'abcSTARTdef,ghi,jkl,ENDmno'.match(/START(.*)END/)[1].split(',')

I don't know of a way to accomplish it in one step. String#scan attempts 
to match the whole regex at multiple places within the string; but you 
need the START and END to delimit the substring over which scan operates.

> 
> 
>     Back to the odd behavior, the first expression actually returns
> [["jkl,"]].  I can't figure out how that is the correct answer by any
> reasonable definition of "scan".  However, the equivalent String#match
> does the same kind of thing, so I must be missing something.  Can
> someone please explain this behavior?

Because you have this:

   ([^,]*,)*

The final * allows the group to match multiple times. The MatchData will 
  hold only the last match however, which is "jkl,".