From: Gavin Kistner <gavin@...>
Date: 2005-09-28T14:00:37+09:00
Subject: Re: Splitting a string with escapable separator?

On Sep 27, 2005, at 5:46 PM, Michael Schuerig wrote:
> I'm trying to come up with an *elegant* way to split a string into an
> array at a separator with the additional feature that the separators
> can be escaped. It should work like this
>
>   "Hello\, World,Hi".split_escapable(',' '\')
>   # => ["Hello, World", "Hi"]
>
> Through a number of permutations with regexps, scan and the rest of  
> the
> family, I was unable to find a solution.

Your above example is missing a couple of \, but I assume I know what  
you meant.

Is the following elegant or not?

class String
     def split_escapable( separator, escape_char=nil )
         results = []
         re = /(.+?)(?:#{escape_char ? "([^\\#{escape_char}])" : ''}# 
{separator}|$)/
         self.scan( re ){ |str,last_char|
             results << str + last_char.to_s
         }
         results
     end
end

p "Hello\\, World,Hi".split_escapable( ',', '\\' )
#=> ["Hello\\, World", "Hi"]

Note that the above does not account for the case of:
Hello \\,World
(where an escaped backslash is intended to end the first entry)
but if that was important, that's just a matter of a bit of odd/even  
backslash counting.

Something like (untested):
re = /(.+?)(?:#{escape_char ? "([^\\#{escape_char}](\\#{escape_char}\ 
\#{escape_char})*)" : ''}#{separator}|$)/