Dupfinder.rb

#!/usr/bin/env ruby

require 'nokogiri'
require 'active_record'
require "sqljdbc4.jar"

ActiveRecord::Base.establish_connection(
  :adapter => "jdbcmssql",
  :driver => "com.microsoft.sqlserver.jdbc.SQLServerDriver",
  :url => "jdbc:sqlserver://192.168.0.18;databaseName=oms_db_cod_12",
  :username => "secret",
  :password => "secret"
)

class Lpu < ActiveRecord::Base
  set_table_name "oms_LPU"
  set_primary_key "LPUID"

  has_many :doctors
end

class Doctor < ActiveRecord::Base
  set_table_name "oms_DOCTOR"
  set_primary_key "DOCTORID"

  belongs_to :lpu, :foreign_key => "rf_LPUID"
end

@filepath = ARGV[0]
@doc = Nokogiri::XML(File.open(@filepath))

def check_mcodes
  @mcodes = @doc.xpath('//PHARMACYRECIPE/MCOD/text()').map(&:to_s).uniq
  @lpu_counts = @mcodes.select { |mcode| Lpu.where(:MCOD => mcode).count > 1 }
  unless @lpu_counts.empty?
    puts "DUPLICATED MCODES: #{@lpu_counts.join(', ')}"
  end
end

def check_pcodes
  @recepies = @doc.xpath("//PHARMACYRECIPE")
  @codes = @recepies.reduce([]) do |memo, r|
    memo << [r.xpath("V_C_OGRN/text()").to_s, r.xpath("PCOD/text()").to_s]
  end.uniq
  @code_counts = @codes.reduce([]) do |memo, codes|
    ogrn = codes[0]
    pcod = codes[1]
    memo << [codes, Doctor.joins(:lpu).where("oms_DOCTOR.PCOD = ? and oms_LPU.C_OGRN = ?", pcod, ogrn).count]
  end.select do |pairs|
    pairs[1] > 1
  end
  @code_counts.each do |pair|
    puts "#{pair[1]} for OGRN #{pair[0][0]} and PCOD #{pair[0][1]}"
  end
end

check_mcodes
check_pcodes