SAVE THE WORLD REMIX!!!

gistfile1.rb

namespace :gts do
  desc "get source of lead for client"

#target output
#client id 1003, from date 2001-01-01
#   coupon
#      www.google.com 13
#      www.yahoo.com   7
#   quotes
#      www.google.com 7
#      www.yahoo.com  4
#

  task(:source_for_client_lead => :environment) do
    start_date = ENV['start_date'] || Date.parse('2011-01-01')
    client_id = ENV['client_id'] || 1003

    lead_types = LeadType.all(:select=>'id, name')
    lead_types_by_id = {}
    lead_types.each{|lead| lead_types_by_id[lead.id] = lead_types_by_id.name } 

    #this makes an assumption that a lead type is 1:1 to a session
    leads = Lead.all(:select=>'DISTINCT(session_id), lead_type_id', :conditions=>["client_id=? AND session_id IS NOT NULL AND DATE(created_at) >= ?", client_id, start_date])

    lead_sources = {}
    leads.each do |lead|
      first_source_in_session = G5PageView::PageView.collection.find({:session_id => lead.session_id}, :fields => ["source"]).sort([['created_at', -1]]).first["source"]
      lead_type_name = lead_types_by_id[lead.lead_type_id]
      sources_for_lead_type = lead_sources[lead_type_name] || {}
      num_leads = sources_for_lead_type.fetch(first_source_in_session, 0)
      num_leads += 1      
      sources_for_lead_type[first_source_in_session] = num_leads
      lead_sources[lead_type_name] = sources_for_lead_type
    end

    puts "Client id #{client_id}, from date #{start_date}"
    lead_sources.each { |lead_type_name, sources|
       puts "\t#{lead_type_name}"
          sources.each {|source, num|
            puts "\t\t#{source}\tnum"
          } 
    }
  end
end