gistfile1.rb

#!/usr/local/bin/ruby -w
require 'csv'
require 'iconv' unless String.method_defined?(:encode)



## THE MAIN FILE TO IMPORT FROM
CSV_FILE_PATH     = File.join(File.dirname(__FILE__), "eloqua_20140130.csv")

## THE BASE NAME AND LOCATION OF EXPORT FILES
OUTPUT_FILE_NAME  = File.join(File.dirname(__FILE__), "export_#{Time.now}")

## COMPANY NAMES TO EXCLUDE
EXCLUDES          = ["eloqua","oracle","gmail","hotmail"]

## THE COLUMN WE WILL USE TO EXCLUDE
EXCLUDE_HEADER    = "company_unique"

## IN THE EVENT OF MISSING COMPANY NAMES
## THIS WILL STORE AS A DICTIONARY OF ALL COMPANY_UNIQUES => COMPANY NAMES
## IN CASE SOMEONE ELSE FROM THE SAME COMPANY IN THE LIST ENTERED
## THEIR COMPANY NAME
COMPANY_NAME_LOOKUP = Hash.new

## THE HEADERS TO EXPORT
HEADERS = [:first_name, :last_name, :company, :email, :industry, :region, :company_unique]


##
## OUTPUT THE HEADERS TO THE CSV
##
def output_headers (csv)
  tmp = []
  HEADERS.each { |sym| tmp << sym.to_s }
  csv << tmp
end



##
## OUTPUT A ROW TO THE PASSED CSV
##
def output_row (row, csv)
  tmp = []
  HEADERS.each { |sym| tmp << row[sym] }
  csv << tmp
end



##
## OUTS PUT THE QUEUE AND EXCLUDED ARRAYS TO CSV FILES
##
def output (queue, excludeds)

  count = 1
  queue.each do |group|
    CSV.open("#{OUTPUT_FILE_NAME}_#{count}.csv", "w") do |csv|
      output_headers csv
      group.each do |row|
        output_row row, csv
      end
    end
    count +=1
  end

  if excludeds
    CSV.open("excludeds_#{Time.now}.csv", "w") do |csv|
      output_headers csv
      excludeds.each do |row|
        output_row row, csv
      end
    end
  end
end



##
## BUILDS THE COMPANY LOOKUP DICIONARY AS ANNOTATED ABOVE
##
def build_company_name_lookup
  CSV.open(CSV_FILE_PATH, {:headers => true, :header_converters => :symbol}).each do |row|
    t = row[:company_unique]
    if !row[:company].nil? && !COMPANY_NAME_LOOKUP[t.to_sym]
      COMPANY_NAME_LOOKUP[t.to_sym] = row[:company]
    end
  end
end



##
## ADDS ITEMS TO THE SPECIFIED QUEUE
##
def add_to_queue(queue, index, row)
  unless queue[index]
    queue[index] = []
  end
  queue[index] << row
end


##
## OPENS THE SOURCE FILE AND ENQUEUES THE ROWS
##
def enqeue

  # HOLDS THE QUEUES TO BE OUTPUT
  queue       = []

  # HOLD THE QUEUE OF EXCLUDED ROWS TO BE OUTPUT
  excludeds   = []

  # CHECKS TO SEE IF ANOTHER PERSON FROM THE SAME COMPANY EXISTS
  # IN THE CURRENT QUEUE. IF SO, MOVES THEM TO THE NEXT QUEUE
  lookup      = Hash.new


  CSV.open(CSV_FILE_PATH, {:headers => true, :header_converters => :symbol}).each do |row|
    if EXCLUDES.include? row[EXCLUDE_HEADER.to_sym]
      excludeds << row
    else
      t = row[:company_unique]
      search = lookup[t.to_sym]

      # ADD THE COMPANY NAME IF THE CURRENT RECORD IS MISSING
      # ONE AND WE ARE ABLE TO FIND THE COMPANY NAME IN THE DICTIONARY
      if row[:company].nil? && COMPANY_NAME_LOOKUP[t.to_sym]
        row[:company] = COMPANY_NAME_LOOKUP[t.to_sym]
      end

      if search
        add_to_queue queue, (search + 1), row
        lookup[row[:company_unique].to_sym] += 1
      else
        lookup[row[:company_unique].to_sym] = 0
        add_to_queue queue, 0, row
      end
    end
  end
  output queue, excludeds
end


##
## INITIALIZE
build_company_name_lookup
enqeue
##

You'd probably have fewer hoops to jump through if you dealt exclusively with Strings, rather than converting some things to symbols.