pyLyricFlow/LyricFlow.Core.Backend/LyricFlow.Core/Engine/PhoneticProcessor.cs

using System.Text.RegularExpressions;
using System.IO;
using System.Collections.Generic;
using System.Linq;

namespace LyricFlow.Core.Engine;

public class PhoneticProcessor
{
    private readonly Dictionary<string, List<List<string>>> _dictionary = new();
    private static readonly Regex WordCleanupRegex = new(@"[^a-z']", RegexOptions.Compiled | RegexOptions.IgnoreCase);

    public PhoneticProcessor(string? cmudictPath)
    {
        if (!string.IsNullOrWhiteSpace(cmudictPath))
        {
            LoadCmuDict(cmudictPath);
        }
    }

    private void LoadCmuDict(string path)
    {
        if (!File.Exists(path)) return;

        foreach (var line in File.ReadLines(path))
        {
            if (string.IsNullOrWhiteSpace(line) || line.StartsWith(";;;")) continue;

            var parts = line.Split(' ', StringSplitOptions.RemoveEmptyEntries);
            if (parts.Length < 3) continue;

            var word = parts[0].ToLower();
            // The format from NLTK has [WORD] [VARIATION_ID] [PH1] [PH2]...
            var phonemes = parts.Skip(2).ToList();

            if (!_dictionary.TryGetValue(word, out var variations))
            {
                variations = new List<List<string>>();
                _dictionary[word] = variations;
            }
            variations.Add(phonemes);
        }
    }

    public string NormalizeWord(string word)
    {
        word = word.ToLower().Trim();
        word = WordCleanupRegex.Replace(word, "");

        if (word.EndsWith("in'"))
        {
            word = word[..^1] + "g";
        }

        return word;
    }

    public List<List<string>> GetPhonemes(string word)
    {
        var normalized = NormalizeWord(word);
        if (_dictionary.TryGetValue(normalized, out var phones))
        {
            return phones;
        }
        return new List<List<string>>();
    }

    public Dictionary<string, List<List<string>>> Dictionary => _dictionary;
}