gistfile1.cs

/// <summary>
	/// </summary>
	/// <remarks> There is a lot of useless information 
	/// in <see cref="http://buscon.rae.es/dpdI/SrvltGUIBusDPD?lema=g%E9nero2">RAE.</see></remarks>
	public class GenderGuesser
	{
		private const string LastVocalPattern = @"(?<vocal>[aeiouáéíóú])[^aeiouáéíóú]*\z";

		private static readonly string[] femaleEndings = new[]
		                                                 	{
		                                                 		"ción", "sión", "triz",
		                                                 	};

		private static readonly string[] maleStart = new[]
		                                                 	{
		                                                 		"a", "ha"
		                                                 	};

		private static readonly Regex reg = new Regex(LastVocalPattern);

		public virtual Gender Guess(string word)
		{
			word = word.Split(' ').First();
			var hasFemaleEnding = HasFemaleEnding(word);
			if (!hasFemaleEnding && HasMaleStart(word) && !word.EndsWith("s", StringComparison.InvariantCultureIgnoreCase))
			{
				return Gender.Male;
			}
			if (hasFemaleEnding )
			{
				return Gender.Female;
			}
			if (GetLastVocal(word).Equals('a'))
			{
				return Gender.Female;
			}
			return Gender.Male;
		}

		private static bool HasMaleStart(string word)
		{
			return maleStart.Any(e => word.StartsWith(e, StringComparison.InvariantCultureIgnoreCase));
		}

		private static bool HasFemaleEnding(string word)
		{
			return femaleEndings.Any(e => word.EndsWith(e, StringComparison.InvariantCultureIgnoreCase));
		}

		private static char GetLastVocal(string w)
		{
			var value = reg.Match(w).Groups["vocal"].Value;

			if (value.Length > 0) return value[0];
			throw new InvalidOperationException("Palabra sin vocales...mmmmmmm.");
		}
	}


	public enum Gender
	{
		Male,
		Female,
		Both,
		Semantic
	}