tumbledry

gistfile1.rb

class TumbleDRYer
  def initialize filename
    @hash = Hash.new 0
    @links = Hash.new 0
    previous = nil
    open(filename).read.scan /\w+|\W+/ do |word|
      unless previous.nil?
        @links[[previous,word]] += 1
      end
      @hash[word] += 1
      previous = word
    end
  end

  attr_reader :hash, :links
end

def test
  TumbleDRYer.new('sample.sql')
end

gistfile2.rb

=> {["id", "`),\n  "]=>1, ["nickname", "` "]=>1, [") ", "NOT"]=>15, ["NOT", " "]=>18, [" '',\n  `", "nickname"]=>1, [" ", "auto_increment"]=>3, ["(", "70"]=>1, ["authors", "` (\n  `"]=>1, ["3", " ;\n  \n"]=>2, ["default", " '',\n  `"]=>7, ["` ", "int"]=>6, [" ;\n  \n", "CREATE"]=>3, [" ", "TABLE"]=>4, ["id", "` "]=>3, ["(", "20"]=>1, [" '',\n  `", "contact"]=>1, [" ", "NOT"]=>3, ["`),\n  ", "KEY"]=>1, ["00", "',\n  `"]=>1, ["document_id", "` "]=>1, ["50", ") "]=>7, ["firstname", "` "]=>1, ["=", "14"]=>1, ["  (`", "id"]=>3, ["password", "` "]=>1, ["',\n  `", "filename"]=>1, ["date", " "]=>1, [",\n  `", "author_id"]=>1, ["0", "',\n) "]=>1, ["(", "11"]=>6, ["(", "50"]=>7, [" ", "default"]=>13, ["KEY", " `"]=>1, ["0000", "-"]=>1, [" '", "0000"]=>1, ["',\n  `", "document_id"]=>1, ["0", "',\n  `"]=>2, ["contact", "` "]=>1, ["=", "3"]=>2, ["AUTO_INCREMENT", "="]=>3, ["` ", "text"]=>2, ["11", ") "]=>6, ["document", "` (`"]=>1, [" ", "NULL"]=>18, ["TYPE", "="]=>4, [" `", "authors"]=>1, ["20", ") "]=>1, ["PRIMARY", " "]=>3, ["` (`", "title"]=>1, [",\n  `", "title"]=>1, ["TABLE", " `"]=>4, ["NULL", ",\n  "]=>1, ["00", "-"]=>1, ["NULL", ",\n  `"]=>1, ["KEY", "  (`"]=>3, ["MyISAM", " "]=>3, [" ", "KEY"]=>3, ["default", " '',\n  "]=>2, [" '", "0"]=>3, [" '',\n  ", "PRIMARY"]=>2, ["id", "`)\n) "]=>2, [",\n  ", "PRIMARY"]=>1, ["14", " ;\n"]=>1, ["date", "` "]=>1, ["text", " "]=>2, ["`)\n) ", "TYPE"]=>3, ["NULL", " "]=>16, ["description", "` "]=>3, [" `", "document"]=>1, ["',\n) ", "TYPE"]=>1, ["default", " '"]=>4, ["int", "("]=>6, [",\n  `", "firstname"]=>1, ["varchar", "("]=>9, ["categories_documents", "` (\n  `"]=>1, ["` (\n  `", "id"]=>3, ["author_id", "` "]=>1, ["` (\n  `", "category_id"]=>1, [" `", "categories_documents"]=>1, ["filename", "` "]=>1, ["title", "` "]=>1, [" `", "documents"]=>1, ["name", "` "]=>2, [" '',\n  `", "description"]=>3, ["categories", "` (\n  `"]=>1, ["',\n  `", "date"]=>1, ["documents", "` (\n  `"]=>1, ["MyISAM", " ;\n  \n"]=>1, ["=", "MyISAM"]=>4, ["70", ") "]=>1, [" `", "categories"]=>1, ["category_id", "` "]=>1, ["` ", "varchar"]=>9, [" '',\n  `", "name"]=>1, [" '',\n  `", "password"]=>1, ["CREATE", " "]=>4, ["title", "`)\n) "]=>1, ["-", "00"]=>2, ["` ", "date"]=>1, ["auto_increment", ",\n  `"]=>3, [",\n  `", "name"]=>1, [" ", "AUTO_INCREMENT"]=>3}