From: Mathieu Bouchard <matju@...>
Date: 2001-03-18T02:55:34+09:00
Subject: [ruby-talk:12780] Re: email address regexp

On Sun, 18 Mar 2001, Cameron Laird wrote:
> In article <m18zm531s9.fsf@halfdome.holdit.com>,
> Randal L. Schwartz <merlyn@stonehenge.com> wrote:
> >which I include here to show that it's not a simple regex:
> There are several positive results to emphasize here:
> 1.  Regexps are great for some things, impossible
>     for others, and ... clumsy for still others.
>     e-mail addresses nicely demonstrate that the
>     third category is non-empty.

If you look at Merlyn's post, you'll notice that the big regexp validates
most kinds of URL. You can then extract the 5 lines that apply
to email addresses (starting at "(?:mailto:")

I'll assume Merlyn's post's purpose is to show off Abigail's ability to
write code that looks like the output of a compiler. Besides, Larry
thinks LISP looks like nail clippings in oatmeal. I don't know what he
thinks about that regexp, but I think I prefer LISP.

A big improvement to the regexp (the email part or the whole) would be to
rewrite it as a bunch of functions that each generate parts of the regexp.
Each part would be given a name and redundancy would be eliminated. If
properly done it could even look like a grammar description.

matju