From: Robert Klemme <bob.news@...>
Date: 2006-01-08T07:38:01+09:00
Subject: Re: One-liner for regex subsitition (with upcasing)!

Donkey Agony <root@[127.0.0.1]> wrote:
> Robert Klemme wrote:
>>>  def methodB(s)  # this doesn't work, "will" stays lowercase
>>>    s.sub(/w.ll/, '\0'.upcase)
>>>  end
>
>> You want the block form of sub:
>>
>>>> "I will drill for a well in walla walla
>>>> washington.".sub(/w.ll/){|m| m.upcase}
>> => "I WILL drill for a well in walla walla washington."
>
> Ahhhhhhh.

I hope nobody's hurt. :-)

>>>> "I will drill for a well in walla walla
>>>> washington.".sub(/w.ll/){|m| m.upcase!}
>> => "I WILL drill for a well in walla walla washington."
>
> What's the difference between `upcase` and `upcase!` in these
> instances (i.e., inside the block)?  In irb, those both seem to yield
> the same result.  In a program, I had to use `sub!` if modifying the
> variable directly, i.e.:

upcase! modifies in place - but it's not a problem here as the match string 
is a new instance anyway.  In this case it boils down to upcase! being more 
efficint.

>  string1 = "I will drill for a well in walla walla washington."
>  string1.sub!(/w.ll/) { |m| m.upcase }
>  puts string1
>
> (also same result whether using `upcase` or `upcase!` in the block...)

Same story.

>> It's the only reasonable thing to do here because you won't know the
>> matched text before the RX actually matched and that is the point in
>> time where you have to calculate the replacement.  The non block form
>> calculates the replacement *before* the invocation - of course you
>> can use \\0, \\1 etc, to access parts - but you can't modify them.
>
> Bingo!  It's so obvious once you see it.  :)
>
> Thank you, Robert!

You're welcome!

    robert