class Numerizer

Constants

BIG_PREFIXES
DIRECT_NUMS
TEN_PREFIXES

Public Class Methods

numerize(string) click to toggle source
# File lib/numerizer/numerizer.rb, line 47
def self.numerize(string)
  string = string.dup

  # preprocess
  string.gsub!(/ +|([^\d])-([^\d])/, '\1 \2') # will mutilate hyphenated-words but shouldn't matter for date extraction
  string.gsub!(/a half/, 'haAlf') # take the 'a' out so it doesn't turn into a 1, save the half for the end
  
  # easy/direct replacements
  
  DIRECT_NUMS.each do |dn|
    string.gsub!(/#{dn[0]}/, '<num>' + dn[1])
  end
  
  # ten, twenty, etc.
  
  TEN_PREFIXES.each do |tp|
    string.gsub!(/(?:#{tp[0]}) *<num>(\d(?=[^\d]|$))*/) { '<num>' + (tp[1] + $1.to_i).to_s }
  end
  
  TEN_PREFIXES.each do |tp|
    string.gsub!(/#{tp[0]}/) { '<num>' + tp[1].to_s }
  end
  
  # hundreds, thousands, millions, etc.
  
  BIG_PREFIXES.each do |bp|
    string.gsub!(/(?:<num>)?(\d*) *#{bp[0]}/) { '<num>' + (bp[1] * $1.to_i).to_s}
    andition(string)
  end
  
  # fractional addition
  # I'm not combining this with the previous block as using float addition complicates the strings
  # (with extraneous .0's and such )
  string.gsub!(/(\d+)(?: | and |-)*haAlf/) { ($1.to_f + 0.5).to_s }
  
  string.gsub(/<num>/, '')
end