strategy for splitting hit file efficiently

hit_load.rb

require 'set'
require 'csv'

data_path = '/Users/zach/Code/metrics_loader/data/'
$hits_path = "#{data_path}/analytic_recording_tags_20121216.csv"
$recordings_path = "#{data_path}/recordings_20121216.csv"

$raw_ids = Set.new
IO.foreach($hits_path).each do |line|
  next if(ids = line.slice!(0, 13)) == 'record_ident|'
  $raw_ids.add ids
end

$hit_ids = Set.new

$raw_ids.each do |id|
  $hit_ids.add({
    recording_id: Integer(id.slice(/^\d+(?=\|)/)),
    tag_id: Integer(id.slice(/(?<=\|)\d+/))
  })
end
$raw_ids.clear


$recordings = []
class Recording
  attr_accessor :id, :recording_time, :inbound, :tags
end

CSV.foreach($recordings_path, headers: true, col_sep: '|') do |line|
  recording = Recording.new
  recording.id = line.fetch('ident')
  recording.recording_time = line.fetch('recording_time')
  $recordings << recording
end