syllables.rb

require "rubygems"
require "rspec"

class String
  def syllable
    # bl|qu|ng|ch|rt|
    consonants = "bcdfghjklmnpqrstvwxzh"
    vowels = "aeiouy"
    word = self.downcase
    parts = word.split(/(?=ly)/)
    parts = parts.first.split(/(?=ed)/) + parts[1..-1]
    parts = parts.first.split(/(?=[^i]al)/) + parts[1..-1]
    parts = parts.first.split(/(?=[#{consonants}])/) + parts[1..-1]
    ##### process 1
    parts = parts.inject([]) do |result, part|
      # [pa r ti al] => par, tial
      if result.size > 0 and part.size == 1 and part =~ /[#{consonants}]/
        result[-1] = "#{result.last}#{part}"
      else
        result.push part
      end
      result
    end
    ##### process 2
    # ["syl", "lab", "le"] => ["syl", "la", "ble"]
    parts = parts.inject([]) do |result, part|
      # |qu|ng|ch|rt
      if result.size > 0 and "#{result.last[-1]}#{part[0]}" =~ /bl/
        part = "#{result.last[-1]}#{part}"
        result[-1] = result.last[0..-2]
      end
      result.push part
      result
    end
    ##### process 3
    # Concate the first one to seconds
    if parts.first !~ /[#{consonants}]+[#{vowels}]+/
      ["#{parts[0]}#{parts[1]}", parts[2..-1]].flatten
    else
      parts
    end
  end
end


describe "String Syllables" do
  it "should be correct" do
    {
      surreptitious: %w{sur rep ti tious},
      successfully: %w{suc cess ful ly},
      processed: %w{pro cess ed},
      consonants: %w{con so nants},
      syllable: %w{syl la ble},
      china: %w{chi na},
      great: %w{great},
      threat: %w{threat},
      translationally: %w{trans la tion al ly},
      partial: %w{par tial},
      level: %w{lev el},
      usual: %w{us u al}
    }.each do |k, v|
      puts k.to_s.syllable.inspect
      # k.to_s.syllable.should == v
    end
  end
end