From: Tim Pease <tim.pease@...>
Date: 2007-08-17T23:53:35+09:00
Subject: Re: Regexp riddle; escaping escapes

On 8/17/07, Phlip <phlip2005@gmail.com> wrote:
> Rubies:
>
> Someone didn't escape their & in their HTML correctly. Let's fix it.
>
> This regexp correctly does not escape &dude, because we only want to escape
> raw & markers:
>
>   p "yo &dude".gsub(/&([^a-z])/i, '&amp;\1')
>
> That passed "yo &dude" thru unchanged. (I am aware "dude" has no ; on the
> end; we are leaving that optional, for whatever reason...)
>
> Now escape & followed by a non-alphabetic character:
>
>   p "yo & dude".gsub(/&([^a-z])/i, '&amp;\1')
>
> That correctly provides: "yo &amp; dude"
>
> Now how to escape "yo && dude"? Note that the ([^a-z]) consumes the second
> &, leading to this incorrect output:
>
>   "yo &amp;& dude"
>
> The only workaround I can think of is to run the Regexp twice:
>
>   x = "yo && dude"
>   2.times{ x.gsub!(/&([^a-z])/i, '&amp;\1') }
>   p x
>
> Can someone help my feeb Regexp skills and get a "yo &amp;&amp; dude" in one
> line?
>

str = "yo && dude"
str.gsub!( %r/&(?=[^a-z])/i, '&amp;')
p str
=> "yo &amp;&amp; dude"


The regular expression trick here is the (?=re)  That's called the
"zero-width positive lookahead". It matches, but it does not consume
the string; so the gsub! will only replace the characters that are NOT
inside (?=re).

Blessings,
TwP