From: Nicholas Van Weerdenburg <vanweerd@...>
Date: 2005-01-07T00:45:47+09:00
Subject: Re: Ruby Data Structure Query Abstractions/Patterns

`On Thu, 6 Jan 2005 21:36:30 +0900, Robert Klemme <bob.news@gmx.net> wrote:
> 
> "Nicholas Van Weerdenburg" <vanweerd@gmail.com> schrieb im Newsbeitrag 
> news:632154f705010520196f99e845@mail.gmail.com...
> > On Thu, 6 Jan 2005 05:56:31 +0900, Robert Klemme <bob.news@gmx.net> wrote:
> >>
> >> "Nicholas Van Weerdenburg" <vanweerd@gmail.com> schrieb im Newsbeitrag
> >> news:632154f705010510264aea7a10@mail.gmail.com...
> >> > Hi all,
> >> >
> >> > I've written a small ruby application that uses YAML and the
> >> > associated hash/array Ruby structure.
> >> >
> >> > I am now looking to do some query-like actions on the basic ruby data
> >> > structure. I'm wondering if there are any packages of
> >> > query-abstractions that make sense for Ruby in a generalized sort-of
> >> > way. E.g. XPath and XQuery for XML, SQL for RDMS, etc.
> >> >
> >> > Currently I've written my structure-parsing code with a few of my
> >> > cases, and I'm starting to refactor with blocks and lambdas in order
> >> > to keep duplication low and offer an easily configurable higher-level
> >> > of abstraction. Now the code is starting to pressure me to create a
> >> > simple query language. Maybe I should go to a ruby db or sqlite, but I
> >> > expect I ultimately only need the most basic functionality.
> >> >
> >> > Any suggestions or recommendations?
> >>
> >> Another thought: maybe you can even use Ruby itself.  I mean, one can
> >> produce quite readable select expressions in Ruby and we have
> >> Enumerable#all?, #any? etc.  You might want to add some functionality for
> >> recursive traversal though.
> >>
> >> Kind regards
> >>
> >>     robert
> >>
> >>
> >
> > I like the idea of using Ruby, and have been extending my code in that
> > manner.  However, XML becomes interesting soon  because of it's DOM
> > traversal capabilities.
> >
> > I'm also wondering if I'm missing some  common Ruby/Perl/Python idiom
> > for handling in-memory data-structure traversal in a more abstract
> > way, but so far it doesn't look like it.
> >
> > Does it make sense to have a general data-structure traversal/query
> > api in a language like Ruby?
> 
> I'd say yes, but maybe only for certain types (DOM tree nodes) - not
> necessarily for general Hash, Array and the like.  OTOH you can cook up
> something quite easily though:
> 
> class Object
>   def traverse() yield self end
> end
> 
> module Enumerable
>   def traverse(&b)
>     each {|o| o.traverse(&b)}
>   end
> end
> 
> class Hash
>   def traverse(&b)
>     each {|k,v| v.traverse(&b)}
>   end
> end
> 
> Note that this does not take care of looks in the object graph.  To do that,
> you need to do a bit more:
> 
> class Object
>   def traverse(context = {}) context[self.id] ||= (yield self; true) end
> end
> 
> module Enumerable
>   def traverse(context = {}, &b)
>     unless context[self.id]
>       context[self.id] = true
>       each {|o| o.traverse(context, &b)}
>     end
>   end
> end
> 
> class Hash
>   def traverse(context = {}, &b)
>     unless context[self.id]
>       context[self.id] = true
>       each {|k,v| v.traverse(context, &b)}
>     end
>   end
> end
> 
> Kind regards
> 
>     robert
> 
> 

That looks pretty good.

Thanks,
Nick

-- 
Nicholas Van Weerdenburg