Files
jsmr/JSMR.Infrastructure/Common/Languages/LanguageIdentifier.cs
Brian Bicknell 06f8c2ec9e
All checks were successful
ci / build-test (push) Successful in 1m31s
ci / publish-image (push) Has been skipped
Addded language identifier tests.
2025-11-03 23:25:25 -05:00

47 lines
1.3 KiB
C#

using JSMR.Domain.Enums;
using NTextCat;
using System.Reflection;
namespace JSMR.Infrastructure.Common.Languages;
public class LanguageIdentifier : ILanguageIdentifier
{
private readonly string[] _languages =
[
"eng",
"jpn",
"kor",
"zho"
];
private readonly RankedLanguageIdentifier _identifier;
public LanguageIdentifier()
{
RankedLanguageIdentifierFactory factory = new();
using var stream = Assembly.GetExecutingAssembly().GetManifestResourceStream("JSMR.Infrastructure.Common.Languages.Language.xml");
_identifier = factory.Load(stream);
}
public Language GetLanguage(string text)
{
if (string.IsNullOrWhiteSpace(text))
return Language.Unknown;
var rankedLanguages = _identifier.Identify(text).Where(x => _languages.Contains(x.Item1.Iso639_3));
var identifiedLanguage = rankedLanguages.OrderBy(x => x.Item2).FirstOrDefault();
if (identifiedLanguage == null)
return Language.Unknown;
return identifiedLanguage.Item1.Iso639_3 switch
{
"jpn" => Language.Japanese,
"eng" => Language.English,
"kor" => Language.Korean,
"zho" => Language.ChineseTraditional,// Or ChineseSimplified?
_ => Language.Unknown,
};
}
}