From: Jay Levitt <jay+news@...>
Date: 2007-10-01T07:50:03+09:00
Subject: Re: Detecting number ranges

On Fri, 28 Sep 2007 17:30:46 +0900, Pe�a, Botp wrote:

> i was impressed by James and Raf's solutions, so i combined the two, together using 1.9's group_by, so

That's beautiful!  I *love* ranger.  And there are a lot of other techiques
I learned from here.  Again, I've oversimplified it below for future
reference.

I've never seen "set.divide" before, and it's the key to the whole thing.
Question: Isn't it possible to give set.divide a block that'll give
inconsistent or crazy results?  I can't quite think of how, but it's
assuming that the relationships are commutative..

It's also really interesting that I was initially using the group_by from
version 3726 of Rails:

   def group_by 
     inject([]) do |groups, element| 
       value = yield(element) 
       if (last_group = groups.last) && last_group.first == value 
 	last_group.last << element 
       else 
 	groups << [value, [element]] 
       end 
       groups 
     end 
   end 

which resulted in two separate "server 7" groups - yet something later in
the chain healed that rift.


Finally, the function I broke out as "text_range" felt like an Escher
drawing at first...

def text_range(ary)
  "#{ary.first}#{" to #{ary.last}" if ary.size>1}"
end

But I finally realized that the braces must take precedence over the
quotes, so the middle quotes are "inner" quotes, as if it read:

def text_range(ary)
  s = "#{ary.first}"
  s << " to #{ary.last}" if ary.size > 1
  s
end

# botp.rb
require 'set'

# Using server 7 as our example

module Enumerable

  def ranger
    s = to_set
    #=> #<Set: {33, 30, 8, 31, 20, 9, 32, 10}>

    consecutives = s.divide {|i,j| (i - j).abs == 1}
    #=> #<Set: {#<Set: {20}>, #<Set: {8, 9, 10}>, 
                #<Set: {33, 30, 31, 32}>}>

    inner_sorted = consecutives.map{|group| group.to_a.sort}
    # each group is sorted internally now
    #=> [[30, 31, 32, 33], [20], [8, 9, 10]]

    inner_sorted.sort_by{|s| s.first}
    # now groups are in sorted order as well
    #=> [[8, 9, 10], [20], [30, 31, 32, 33]]
 
  end

  def group_by
    inject({}) do |groups, element|
      (groups[yield(element)] ||= []) << element
      groups
    end
  end if RUBY_VERSION < '1.9'

end

def text_range(ary)
  # Note the quoted text - and another #{} - *inside* 
  # the #{... if ary.size>1} construct.
  # Who knew you could do such a thing?

  "#{ary.first}#{" to #{ary.last}" if ary.size>1}"
end

db_server_pairs = DATA.readlines.map{|s| s.scan(/\d+/).map{|s| s.to_i}}
#=> [[8, 7], [10, 7], [9, 7], ... ]

pairs_by_server = db_server_pairs.group_by{|pair| pair.last}
#=> [[7, [[8, 7], [10, 7], [9, 7], ... ]], 3, ... ]

puts "Server\tDatabases"
pairs_by_server.sort.each do |server, pairs|
  # server => 7
  # pairs => [[8, 7], [10, 7], [9, 7], ...]

  my_dbs = pairs.map {|pair| pair.first}
  #=> [8, 10, 9, ... ]

  my_ranges = my_dbs.ranger
  #=> [[8, 9, 10], [20], [30, 31, 32, 33]]

  range_text =  my_ranges.map{|r| text_range(r)}.join(", ")
  puts "#{server}\t#{range_text}"
end

__END__
database 8 is on server 7
database 10 is on server 7
database 9 is on server 7
database 5 is on server 9
database 1 is on server 3
database 2 is on server 3
database 133 is on server 3
database 4 is on server 144
database 3 is on server 144
database 20 is on server 7
database 30 is on server 7
database 31 is on server 7
database 32 is on server 7
database 33 is on server 7


-- 
Jay Levitt                |
Boston, MA                | My character doesn't like it when they
Faster: jay at jay dot fm | cry or shout or hit.
http://www.jay.fm         | - Kristoffer