Addded language identifier tests.
This commit is contained in:
@@ -26,6 +26,9 @@ public class LanguageIdentifier : ILanguageIdentifier
|
||||
|
||||
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();
|
||||
|
||||
|
||||
21
JSMR.Tests/Unit/LanguageIdentifierTests.cs
Normal file
21
JSMR.Tests/Unit/LanguageIdentifierTests.cs
Normal file
@@ -0,0 +1,21 @@
|
||||
using JSMR.Domain.Enums;
|
||||
using JSMR.Infrastructure.Common.Languages;
|
||||
using Shouldly;
|
||||
|
||||
namespace JSMR.Tests.Unit;
|
||||
|
||||
public class LanguageIdentifierTests
|
||||
{
|
||||
private readonly LanguageIdentifier _languageIdentifier = new();
|
||||
|
||||
[Theory]
|
||||
[InlineData("これは日本語のサンプル文章です。今日はいい天気ですね。", Language.Japanese)]
|
||||
[InlineData("This is an English sample sentence used for language detection.", Language.English)]
|
||||
[InlineData("這是一段中文(繁體)的範例文字,用於語言辨識。", Language.ChineseTraditional)]
|
||||
[InlineData("이것은 한국어 예문으로, 언어 식별을 위한 문장입니다。", Language.Korean)]
|
||||
[InlineData("", Language.Unknown)]
|
||||
public void Identify_Languages(string text, Language expectedLanguage)
|
||||
{
|
||||
_languageIdentifier.GetLanguage(text).ShouldBe(expectedLanguage);
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user