metaphone file

metaphone.rb

def self.lookup(str, pos, length, last)
      case str[pos, 1]
        when /^A|E|I|O|U|Y$/
          if 0 == pos
            return 'A', 'A', 1
          else
            return nil, nil, 1
          end
        when 'B'
          return 'P', 'P', ('B' == str[pos + 1, 1] ? 2 : 1)
        when 'Ç'
          return 'S', 'S', 1
          end
        when 'C'
          if pos > 1 &&
            !vowel?(str[pos - 2, 1]) &&
            'ACH' == str[pos - 1, 3] &&
            str[pos + 2, 1] != 'I' && (
              str[pos + 2, 1] != 'E' ||
              str[pos - 2, 6] =~ /^(B|M)ACHER$/
            ) then
            return 'K', 'K', 2
          elsif 0 == pos && 'CAESAR' == str[pos, 6]
            return 'S', 'S', 2
          elsif 'CHIA' == str[pos, 4]
            return 'K', 'K', 2
          elsif 'CH' == str[pos, 2]
            if 0 == pos && 'CHAE' == str[pos, 4]
              return 'K', 'X', 2
            elsif 0 == pos && (
                ['HARAC', 'HARIS'].include?(str[pos + 1, 5]) ||
                ['HOR', 'HYM', 'HIA', 'HEM'].include?(str[pos + 1, 3])
              ) && str[0, 5] != 'CHORE' then
              return 'K', 'K', 2
            elsif ['VAN ','VON '].include?(str[0, 4]) ||
                  'SCH' == str[0, 3] ||
                  ['ORCHES','ARCHIT','ORCHID'].include?(str[pos - 2, 6]) ||
                  ['T','S'].include?(str[pos + 2, 1]) || (
                    ((0 == pos) || ['A','O','U','E'].include?(str[pos - 1, 1])) &&
                    ['L','R','N','M','B','H','F','V','W',' '].include?(str[pos + 2, 1])
                  ) then
              return 'K', 'K', 2
            elsif pos > 0
              return ('MC' == str[0, 2] ? 'K' : 'X'), 'K', 2
            else
              return 'X', 'X', 2
            end
          elsif 'CZ' == str[pos, 2] && 'WICZ' != str[pos - 2, 4]
            return 'S', 'X', 2
          elsif 'CIA' == str[pos + 1, 3]
            return 'X', 'X', 3
          elsif 'CC' == str[pos, 2] && !(1 == pos && 'M' == str[0, 1])
            if /^I|E|H$/ =~ str[pos + 2, 1] && 'HU' != str[pos + 2, 2]
              if (1 == pos && 'A' == str[pos - 1, 1]) ||
                /^UCCE(E|S)$/ =~ str[pos - 1, 5] then
                return 'KS', 'KS', 3
              else
                return 'X', 'X', 3
              end
            else
              return 'K', 'K', 2
            end
          elsif /^C(K|G|Q)$/ =~ str[pos, 2]
            return 'K', 'K', 2
          elsif /^C(I|E|Y)$/ =~ str[pos, 2]
            return 'S', (/^CI(O|E|A)$/ =~ str[pos, 3] ? 'X' : 'S'), 2
          else
            if /^ (C|Q|G)$/ =~ str[pos + 1, 2]
              return 'K', 'K', 3
            else
              return 'K', 'K', (/^C|K|Q$/ =~ str[pos + 1, 1] && !(['CE','CI'].include?(str[pos + 1, 2])) ? 2 : 1)
            end
          end