From: Gavin Kistner <gavin@...>
Date: 2005-05-19T23:46:04+09:00
Subject: Re: Idiomatic way for collecting elements using REXML?

--Apple-Mail-1-767805386
Content-Transfer-Encoding: 7bit
Content-Type: text/plain;
	charset=US-ASCII;
	delsp=yes;
	format=flowed

On May 19, 2005, at 8:15 AM, John Lam wrote:
> item.details = []
> doc.elements.each('ItemLookupResponse/Items/Item') do |item|
>   # construct an item_detail
>   item_details << item_detail
> end
>
> This doesn't *feel* right. I wanted to use collect() but it's not
> implemented in REXML. Is there a more Ruby-esque approach to gathering
> some objects into an array while parsing XML?

You seem to have left a detail out of your code (where does  
'item_detail' come from?) but that's generally what I do in similar  
circumstances.

Alternatively, you could write your own #collect method:

class REXML::Elements
     def collect( xpath=nil )
         vals = []
         self.each(xpath){ |el| vals << yield el }
         vals
     end
end

Or, perhaps you could modify the existing each method to return an  
array with no block supplied:

class REXML::Elements
     alias_method :__each :each
     def each( xpath=nil, &block )
         if block_given?
             __each( xpath, &block )
         else
             els = []
             __each( xpath ){ |el| els << el }
             els
         end
     end
end

Then you could simply:
item_details = doc.elements.each( '...' ).collect{ ... }


But then this is really exactly what Xpath.match does:
item_details = XPath.match( doc, 'ItemLookupResponse/Items/ 
Item' ).collect{ ... }

(Aside: What possessed the author of the REXML library to put all the  
XPath stuff in its own class with class methods? Why not simply  
"doc.match( my_xpath )"? Whenever I see a class method that takes a  
single object for an argument to set scope, it screams to me that it  
should be a method on that object.)

Hope that helps,


--
(-, /\ \/ / /\/


--Apple-Mail-1-767805386--