/*public Rhyme EvaluateRhyme(Phrase word) {
* var consecutive = 0;
* var count = 0;
* var index = word.Words.SelectMany(each => each.Syllables).Count() - 1;
* var isConsecutive = true;
*
* foreach (var syllable in word.Words.SelectMany(each => each.Syllables).Reverse()) {
* if (RhymeFinder.IsRhyme(syllable.Phonemes, Syllables[index].Phonemes)) {
* if (isConsecutive) {
* consecutive += 1;
* }
*
* count += 1;
* }
* else {
* isConsecutive = false;
* }
*
* index -= 1;
* }
*
* return new Rhyme {
* Phrase = word,
* Consecutive = consecutive,
* Count = count
* };
* }*/
private IEnumerable <Phrase> InternalGetRhymes(bool strict = true)
{
var index = 1;
var results = new Dictionary <int, List <Word> >();
var ngrams = new Dictionary <string, NGram>();
foreach (var syllable in Syllables)
{
var rhymes = RhymeFinder.GetRhymes(syllable, index).ToList();
var validWords = new List <Word>();
foreach (var rhyme in rhymes.Where(each => index - each.Syllables.Count + 1 == 1).ToList())
{
validWords.Add(rhyme);
if (strict)
{
foreach (var ngram in rhyme.NGrams.Where(each => each.Index == 1))
{
if (!ngrams.ContainsKey(ngram.NGram.Text))
{
ngrams.Add(ngram.NGram.Text, ngram.NGram);
}
}
}
}
foreach (var rhyme in rhymes.Where(each => index - each.Syllables.Count + 1 != 1).ToList())
{
if (strict)
{
var requiredSyllables = index - rhyme.Syllables.Count;
foreach (var ngram in rhyme.NGrams.OrderByDescending(each => each.NGram.Frequency))
{
if (ngrams.ContainsKey(ngram.NGram.Text))
{
var priorSyllables = ngram.NGram.Words.Where(each => each.Index < ngram.Index).Sum(each => each.Word.Syllables.Count);
if (priorSyllables == requiredSyllables)
{
validWords.Add(rhyme);
// Todo: Should filter all valid NGrams for all permutations
break;
}
}
}
}
else
{
validWords.Add(rhyme);
}
}
var bestResults = validWords.OrderByDescending(each => each.NGrams.Sum(ngram => ngram.NGram.Frequency)) /*.Take(100)*/.ToList();
results.Add(index, bestResults);
index += 1;
}
var final = FindMultiSyllables(results, Syllables.Count);
if (strict)
{
final = final.Where(each => each.GetNGramScore() > 0);
}
return(final);
}
public IEnumerable <Phrase> GetRhymes(bool strict = true)
{
if (Syllables.Count > 10)
{
throw new Exception("Too many syllables.");
}
var rhymes = InternalGetRhymes(strict).Where(rhyme => {
var index = 0;
foreach (var word in Words)
{
if (word.Text == rhyme.Words[index].Text && RhymeFinder.IsRhyme(word, rhyme.Words[index]))
{
return(false);
}
if (rhyme.Words.Count > index + 1)
{
index += 1;
}
}
return(true);
})
.Select(rhyme => {
var consecutive = 0;
var count = 0;
var index = rhyme.Words.SelectMany(each => each.Syllables).Count() - 1;
var isConsecutive = true;
foreach (var syllable in rhyme.Words.SelectMany(each => each.Syllables).Reverse())
{
if (RhymeFinder.IsRhyme(syllable, Syllables[index]))
{
if (isConsecutive)
{
consecutive += 1;
}
count += 1;
}
else
{
isConsecutive = false;
}
index -= 1;
}
return(new {
Rhyme = rhyme,
Consecutive = consecutive,
Count = count
});
})
//.Where(each => each.Consecutive == each.Rhyme.Words.Sum(w => w.Syllables.Count))
.OrderByDescending(each => each.Count)
.ThenByDescending(each => each.Consecutive)
.ThenByDescending(each => each.Rhyme.GetNGramScore())
.Take(500)
.Select(each => each.Rhyme)
.ToList();
return(rhymes);
}
static English()
{
Vocabulary = new Dictionary <string, Word>();
//var pronounce = EnglishResources.pronounce.Replace("\t", " ").ToLower();
//File.WriteAllText(@"C:\Users\Josh\Downloads\pronounce.txt", pronounce);
var words = EnglishResources.pronounce.Replace("\t", " ").ToLower().ReadLines();
/*var punctuated = new List<string>();
*
* foreach (var word in words) {
* if (word.Contains("'") && !word.Contains("'s"))) {
* punctuated.Add(word);
* }
* }*/
var tokens = IllegalTokens;
var results = words.ForEachParallel(line => {
if (tokens.Any(line.Contains))
{
return(null);
}
var segments = line.Replace("'", "").Replace("-", "").Split(new[] { " " }, StringSplitOptions.RemoveEmptyEntries);
var word = segments.First().Trim();
var hasNumbers = word.HasNumbers();
if (!word[0].IsAlphabetical() || (hasNumbers && !word.Contains("(")) || !word.HasVowel())
{
return(null);
}
if (hasNumbers)
{
word = word.Split('(')[0];
}
var phonemes = segments.Skip(1).Select(each => new Phoneme(each)).ToList();
var resolver = new PhonemeResolver(word);
resolver.BindPhonemes(phonemes.ToArray());
return(new Word(word, phonemes));
});
foreach (var word in results)
{
if (!Vocabulary.ContainsKey(word.Text))
{
Vocabulary.Add(word.Text, word);
}
else
{
Vocabulary[word.Text].Homographs.Add(word);
}
}
RhymeFinder.BuildRhymes();
}