// Count the number of syllables in a CMU pronunciation string.
countSyllables[str] :=
{
   length[str =~ %r/([\d])/g]
}

// Returns true if one word is an anagram of the other.
isAnagram[word1, word2] :=
{
   a2 = chars[word2.word]
   for c1 = char[word1.word]
      if (a2.removeValue[c1] == false)
         return false

   return true
}

weakHash[word] :=
{
   v = 0
   for c = chars[word]
      v = v + c

   return v
}

class Anagram
{
   var word
   var hash
   var length
   var syllables

   new[w, p] :=
   {
      word = w
      syllables = countSyllables[p]
      hash = weakHash[w]
      length = length[word]
   }
}


anagramList = []
// The wordlist files are part of the Moby wordlist project, available at:
//   http://icon.shef.ac.uk/Moby/
for line lines["file:/home/eliasen/prog/mobydict/mpron/cmupronunciation.txt"]
{
   if line =~ %r/^#/
      next;
   
   if [word, pron] = line =~ %r/^([\w\-]+)[^\s]*\s+(.*)/
   {
      a = new Anagram[word, pron]
      anagramList.push[a]
   }
}

var sd
for a1 = anagramList
   for a2 = anagramList
   {
      if (a1.word > a2.word)
         next
      
      //sd = abs[a1.syllables-a2.syllables]
      if (a1.length == a2.length) && (a1.hash == a2.hash) && (a1.word != a2.word) && ((sd = abs[a1.syllables-a2.syllables]) > 2) && isAnagram[a1, a2]
         println[a1.word + "\t" + a2.word + "\t" + sd]
   }