helloworld/vendor/gems/chronic-0.2.2/lib/numerizer/numerizer.rb

   1 require 'strscan'
   2
   3 class Numerizer
   4
   5   DIRECT_NUMS = [
   6                   ['eleven', '11'],
   7                   ['twelve', '12'],
   8                   ['thirteen', '13'],
   9                   ['fourteen', '14'],
  10                   ['fifteen', '15'],
  11                   ['sixteen', '16'],
  12                   ['seventeen', '17'],
  13                   ['eighteen', '18'],
  14                   ['nineteen', '19'],
  15                   ['ninteen', '19'], # Common mis-spelling
  16                   ['zero', '0'],
  17                   ['one', '1'],
  18                   ['two', '2'],
  19                   ['three', '3'],
  20                   ['four(\W|$)', '4\1'],  # The weird regex is so that it matches four but not fourty
  21                   ['five', '5'],
  22                   ['six(\W|$)', '6\1'],
  23                   ['seven(\W|$)', '7\1'],
  24                   ['eight(\W|$)', '8\1'],
  25                   ['nine(\W|$)', '9\1'],
  26                   ['ten', '10'],
  27                   ['\ba[\b^$]', '1'] # doesn't make sense for an 'a' at the end to be a 1
  28                 ]
  29
  30   TEN_PREFIXES = [ ['twenty', 20],
  31                     ['thirty', 30],
  32                     ['fourty', 40],
  33                     ['fifty', 50],
  34                     ['sixty', 60],
  35                     ['seventy', 70],
  36                     ['eighty', 80],
  37                     ['ninety', 90]
  38                   ]
  39
  40   BIG_PREFIXES = [ ['hundred', 100],
  41                     ['thousand', 1000],
  42                     ['million', 1_000_000],
  43                     ['billion', 1_000_000_000],
  44                     ['trillion', 1_000_000_000_000],
  45                   ]
  46
  47 class << self
  48   def numerize(string)
  49     string = string.dup
  50
  51     # preprocess
  52     string.gsub!(/ +|([^\d])-([^d])/, '\1 \2') # will mutilate hyphenated-words but shouldn't matter for date extraction
  53     string.gsub!(/a half/, 'haAlf') # take the 'a' out so it doesn't turn into a 1, save the half for the end
  54
  55     # easy/direct replacements
  56
  57     DIRECT_NUMS.each do |dn|
  58       string.gsub!(/#{dn[0]}/i, dn[1])
  59     end
  60
  61     # ten, twenty, etc.
  62
  63     TEN_PREFIXES.each do |tp|
  64       string.gsub!(/(?:#{tp[0]})( *\d(?=[^\d]|$))*/i) { (tp[1] + $1.to_i).to_s }
  65     end
  66
  67     # hundreds, thousands, millions, etc.
  68
  69     BIG_PREFIXES.each do |bp|
  70       string.gsub!(/(\d*) *#{bp[0]}/i) { (bp[1] * $1.to_i).to_s}
  71       andition(string)
  72       #combine_numbers(string) # Should to be more efficient way to do this
  73     end
  74
  75     # fractional addition
  76     # I'm not combining this with the previous block as using float addition complicates the strings
  77     # (with extraneous .0's and such )
  78     string.gsub!(/(\d+)(?: | and |-)*haAlf/i) { ($1.to_f + 0.5).to_s }
  79
  80     string
  81   end
  82
  83 private
  84   def andition(string)
  85     sc = StringScanner.new(string)
  86     while(sc.scan_until(/(\d+)( | and )(\d+)(?=[^\w]|$)/i))
  87       if sc[2] =~ /and/ || sc[1].size > sc[3].size
  88         string[(sc.pos - sc.matched_size)..(sc.pos-1)] = (sc[1].to_i + sc[3].to_i).to_s
  89         sc.reset
  90       end
  91     end
  92   end
  93
  94 #  def combine_numbers(string)
  95 #    sc = StringScanner.new(string)
  96 #    while(sc.scan_until(/(\d+)(?: | and |-)(\d+)(?=[^\w]|$)/i))
  97 #      string[(sc.pos - sc.matched_size)..(sc.pos-1)] = (sc[1].to_i + sc[2].to_i).to_s
  98 #      sc.reset
  99 #    end
 100 #  end
 101
 102 end
 103 end