From: dblack@...
Date: 2002-09-16T12:08:15+09:00
Subject: Re: camelCaseTo_ruby_case.rb ??

Hi --

On Mon, 16 Sep 2002, Joel VanderWerf wrote:

>
> Two comments:
>
> 1. #gsub! has big teeth and will bite you:
>
>    "FooBar".decamel   # ==> nil
>
> It seems ok with #gsub instead of the first occurrence of #gsub!.

How strange.  When I do this in irb, I get nil, as you say.  But when
I run my little script:

  #!/usr/bin/ruby -wp

  BEGIN {
    class String
      def decamel
	gsub!(/\b[^\WA-Z]+[A-Z].*?\b/) do |x|
	  x.gsub!(/[A-Z][^\WA-Z]+/) {|s| "_" + s.downcase }
	  x.gsub!(/[A-Z][A-Z]+/)    {|s| "_" + s }
	  x
	end
      end
    end
  }

  $_.decamel


I get this (where odd-numbered lines are typed in and even ones are the
output):

  someMethodOrOther
  some_method_or_other
  SomeClass
  SomeClass

which is what I wanted.  I can't quite puzzle through why there would
be that difference.


> 2. How will you handle quotation of various kinds?
>
>    "decamel replaces 'camelVar' with 'camel_var'".decamel
>
>    # ==> "decamel replaces 'camel_var' with 'camel_var'"
>
> The fact that %{ } can be nested means you'll need something more than
> regexes...

Yes, a full-featured decameler would have to get over that hump (ha
ha).  Mine is very mini and quick-n-dirty, probably to the point of
pointlessness.  I imagine there's literally no way to do it perfectly,
without harnessing an actual Ruby parser.

Actually a decameler was one of things considered for the Austin Ruby
codefest.  Since we opted for scanf, the decamler project is still up
for grabs....  (Or has it been done?  I never did research the
original question :-)


David

-- 
David Alan Black                      | Register for RubyConf 2002!
home: dblack@candle.superlink.net     | November 1-3
work: blackdav@shu.edu                | Seattle, WA, USA
Web:  http://pirate.shu.edu/~blackdav | http://www.rubyconf.com