What would you guys think about adding the Thrush combinator to Object?

Hello,

I’d like to get some peoples’ opinion on this. I recently needed to
make a change to an enumerator that changed the number of items in the
enumerator. I wanted to be able to chain this function in like map and
inject, but it needed to operate on the enumerator, not just the items
in it. Adding functions like this to the Object or Enumerator class
seemed like poor design (since there could be many of them). I was
surprised to find out that there isn’t a Thrush combinator function on
the Object class. I ended up copying the code Reganwald
(https://github.com/raganwald/homoiconic/blob/master/2008-10-30/thrush.markdown#readme)
and wanted to see if anyone else thought it would be a good addition
to the Ruby language. Here’s the code:

class Object
def into(expr = nil)
expr.nil? ? yield(self) : expr.to_proc.call(self)
end
end

Thoughts?

Here’s my situation in more detail for those who are interested:

I had some data like this:

orig_data = [{:type => :products, :ids => [1, 2, 3, 4, 5]}, {:type =>
:stores, :ids => [6, 7, 8, 9]}]

that I needed to split into smaller chunks, like this:

[{:type => :products, :ids => [1, 2]},
{:type => :products, :ids => [3, 4]},
{:type => :products, :ids => [5]},
{:type => :stores, :ids => [6, 7]},
{:type => :stores, :ids => [8, 9]}]

My function for splitting them was pretty simple, but had to take in
an enumerator:

def slice_by_ids(hashes)
Enumerator.new do |yielder|
hashes.each do |hash|
hash[:ids].each_slice(2) do |slice|
yielder.yield hash.merge(:ids => slice)
end
end
end
end

This function seemed way too specific to put on the Enumerator class,
but if we had the Thrush combinator we could do something like this:

orig_data.into(&method(:slice_by_ids))

This is also still easily chainable and maintains laziness:

orig_data.
lazy_map(&method(:do_stuff)).
into(&method(:slice_by_ids)).
lazy_map(&method(:do_other_stuff))

On 03.06.2011 15:30, Brian Maddy wrote:

(homoiconic/2008-10-30/thrush.markdown at master · raganwald-deprecated/homoiconic · GitHub)

[{:type => :products, :ids => [1, 2]},
hashes.each do |hash|
orig_data.into(&method(:slice_by_ids))

This is also still easily chainable and maintains laziness:

orig_data.
lazy_map(&method(:do_stuff)).
into(&method(:slice_by_ids)).
lazy_map(&method(:do_other_stuff))

Hm, that looks quite complex to me. Why not just do this?

require ‘pp’

sliced = []

orig_data.each do |h|
h[:ids].each_slice 2 do |sl|
sliced << h.merge(:ids => sl)
end
end

pp orig_data, sliced

Kind regards

robert

Brian Maddy wrote in post #1002943:

orig_data.into(&method(:slice_by_ids))

Presumably you mean:

orig_data.into(:slice_by_ids)

(so that the .to_proc does something useful). But this is pretty ugly to
me. You defined slice_by_ids at the top level anyway, in which case I’d
simply write:

slice_by_ids(orig_data)

If you don’t want it at the top level, put it into a module.