task.rb

require 'fuzzystringmatch'
desc "This task eliminates books without ISBN, finds the duplicated books and merges them"
task find_duplicated_books: :environment do

  jarow = FuzzyStringMatch::JaroWinkler.create(:native)
  
  #Destroy books without isbn that have no shelf_books and progresses
  Book.where(isbn: nil).where(shelf_books_count: 0, progresses_count: 0).destroy_all
  
  books = Book.all # These are 30K books.
  books.each do |book|
    books.each do |otherbook|  # <- this double loop is the one that concerns me the most. 30,000 * 30,000 = 900,000,000 loops.
      next if same_book?(book, otherbook)
      if books_are_duplicate(jarow, book, otherbook)
        merge_books(book, otherbook)
      end
    end
  end
end

def merge_books(book, otherbook)
  sleep 2 # omitted for brevity. traversing several AR associations, some db calls, etc.
end


def same_book?(book_1, book_2)
  book_1.id == book_2.id
end

def books_are_duplicate(jarow, book, otherbook)
  title = book.title.parameterize(" ")
  other_title = otherbook.title.parameterize(" ")
  jarow.getDistance(title, other_title) > 0.99 # this is a really fast implementation of the Jaro-Winkler distance for strings.
end