gistfile1.rb

COMMON_WORDS = []

desc 'Move buisnesses to one sub category to another'
task :move_businesses, [:from, :to, :heuristic_file] => :environment do |t, args|
  puts args
  if valid_arguments?(args)
    read_heuristic_file(args[:heuristic_file])
    move_businesses(args[:from], args[:to])
  else
    print_usage
    print_whats_wrong_with_args(args)
  end
end

def move_businesses(from, to)
  @from_sub_cat          = GCS::SubCategory.first_by_title(from)
  @to_sub_cat            = GCS::SubCategory.first_by_title(to)
  @to_sub_cat_parent     = @to_sub_cat.categories.first
  @businesses_to_migrate = @from_sub_cat.businesses

  result = move_and_collect_failed(@businesses_to_migrate)

  print_results(result)
end

def move_and_collect_failed(businesses_to_migrate)
  success_count = 0;
  failed_to_save = []
  businesses_to_migrate.each do |biz|
    if heuristic_says_move?(biz)
      move_to_sub_category(biz)
      move_to_category(biz)
      if biz.save
        success_count += 1
      else
        failed_to_save << biz 
      end
    end
  end
  {failed:failed_to_save, success:success_count}
end

def heuristic_says_move?(biz)
  COMMON_WORDS.detect{|w| biz.name =~ /#{w}/xi}
end

def move_to_sub_category(biz)
  biz.sub_categories.delete_if{|cat| cat.title == @from_sub_cat.title}
  biz.sub_categories << @to_sub_cat
  biz.primary_sub_cat = @to_sub_cat
end

def move_to_category(biz)
  biz.categories << @to_sub_cat_parent
  biz.primary_cat = @to_sub_cat_parent
end

def valid_arguments?(args)
  return false unless args[:from] and args[:to] and args[:heuristic_file] and File.exists?(args[:heuristic_file]) and !File.zero?(args[:heuristic_file])

  from_sub_cat = GCS::SubCategory.first_by_title(args[:from])
  to_sub_cat = GCS::SubCategory.first_by_title(args[:to])
  return false unless from_sub_cat and to_sub_cat

  true
end

def print_whats_wrong_with_args(args)
  puts "You did not specify the sub category to move busnineses from." if !args[:from]
  puts "You did not specify the sub category to move busnineses to." if !args[:to]

  from_sub_cat = GCS::SubCategory.first_by_title(args[:from])
  to_sub_cat = GCS::SubCategory.first_by_title(args[:to])
  puts "The sub category '#{args[:from]}' doesn't exist in the database." if !from_sub_cat
  puts "The category '#{args[:to]}' doesn't exist in the database." if !to_sub_cat

  puts "The file #{args[:heuristic_file]} wasn't found." if !File.exists?(args[:heuristic_file]||'')
  puts "The file #{args[:heuristic_file]} empty." if File.zero?(args[:heuristic_file]||'')
end

def print_usage
  puts "Usage: rake move_businesses[<from_sub_cat>,<to_sub_cat>,<filename>]"
  puts "Make sure the both sub categories and the heuristic file exists and is not empty."
end

def print_results(results)
  failed_to_save = results[:failed]
  success        = results[:success]
  puts "Moved #{success} businesses out of #{@businesses_to_migrate.count}"
  failed_to_save.each do |failed|
    puts "Failed to save #{failed.name}"
    failed.errors.each{|e| puts e}
  end
end

def read_heuristic_file(filename)
  File.open(filename, 'r') do |f|
    f.each_line do |l| 
      COMMON_WORDS << l.split[0]
    end
  end
end