From: "Joe Wölfel" <joe@...>
Date: 2008-12-02T07:41:31+09:00
Subject: Re: Regular Expressions


On 1 déc. 08, at 17:08, Rob Biedenharn wrote:

> On Dec 1, 2008, at 4:32 PM, Joe Wölfel wrote:
>> On 1 déc. 08, at 14:52, Kyle Schmitt wrote:
>>> On Mon, Dec 1, 2008 at 1:51 PM, Kyle Schmitt  
>>> <kyleaschmitt@gmail.com> wrote:
>>>> I just wanted to mention another way of combining regexes that may
>>>> help you stay sane: union.
>>>>
>>>> #You write each regex nice and simple like..
>>>> startswith=/~23430000/
>>>> codered=/CodeRed/
>>>>
>>>> #Then combine them to a complex one
>>>> combined_regex=Regexp.union(startswith,codered)
>>>>
>>>> When you've got to build up some large regular expressions, this  
>>>> can
>>>> be a godsend, especially when revisiting code you haven't looked  
>>>> at in
>>>> awhile.
>>>>
>>>> --Kyle
>>>
>>> Scratch that, not thinking clearly!  This is to match startswith OR
>>> codered, not necessarily both.
>>>
>>> Still, I maintain that this is a way of staying sane with complex  
>>> regexes :)
>>>
>>
>> Interesting that there is a union function but no intersection  
>> function.
>
>
> How would you even define a regexp (re) that matched only when both  
> of two other regexps (re1, re2) matched?
>
>     class Regexp
>       def self.intersection(re1,re2)
>         union(compile(/(?>#{re1}).*#{re2}/),
>               compile(/(?>#{re2}).*#{re1}/))
>       end
>     end
>
>     re = Regexp.intersection(re1,re2)
>
> What would you expect the value to be?  And while Regexp.union is  
> well-behaved for multiple arguments, the expansion for more  
> arguments in the intersection gets ugly fast.
>
> -Rob
>
> Rob Biedenharn		http://agileconsultingllc.com
> Rob@AgileConsultingLLC.com

Not sure I understand.  Are you arguing that an intersection cannot  
exist as a regular expression or merely that it is hard?