Merge pull request #34 from hmlendea/arabic

Improved the `Arabic` transliteration
hmlendea · May 25, 2023 · d7c07b2 · d7c07b2
2 parents 4ae056b + 97a66cf
commit d7c07b2
Show file tree

Hide file tree

Showing 4 changed files with 219 additions and 125 deletions.
diff --git a/TransliterationAPI.UnitTests/Service/Transliterators/ArabicTransliteratorTests.cs b/TransliterationAPI.UnitTests/Service/Transliterators/ArabicTransliteratorTests.cs
@@ -15,22 +15,41 @@ public void SetUp()
         }
 
         [Test]
+        [TestCase("أريحا", "ʾArīḥā")]
+        [TestCase("ٱلنَّقَب", "an-Naqab")]
         [TestCase("إسدود", "ʾIsdūd")]
+        [TestCase("إِسْرَائِيل", "ʾIsrāʾīl")]
         [TestCase("اسلو", "Āslū")]
-        [TestCase("الرملة", "āl-Ramlaḧ")] // Or ar-Ramleh
+        [TestCase("الأردن", "Al-ʾUrdunn")]
+        [TestCase("البيرة", "al-Bīra")]
+        [TestCase("الخليل", "al-Khalīl")]
+        [TestCase("السامرة", "as-Sāmirah")]
+        [TestCase("العمارنة", "al-ʿAmārnah")]
+        [TestCase("الفرات", "al-Furāt")]
+        [TestCase("الكويت", "al-Kuwayt")]
+        [TestCase("المندائيّة", "al-Mandāʾiyya")]
+        [TestCase("بئر السبع", "Biʾr as-Sabʿ")]
         [TestCase("بابل", "Babil")]
         [TestCase("تشيشيناو", "Tšīšīnāū")]
+        [TestCase("حَدِيثَةٌ", "Ḥadīthah")]
+        [TestCase("حركة المقاومة الإسلامية", "Ḥarakah al-Muqāwamah al-ʾIslāmiyyah")]
+        [TestCase("حماس", "Ḥamās")]
         [TestCase("حَيْفَا", "Ḥayfā")]
         [TestCase("دمشق", "Dimašq")]
+        [TestCase("رام الله", "Rām Allāh")]
         [TestCase("روما", "Rūmā")]
+        [TestCase("سُورِيَا", "Sūriyā")]
         [TestCase("عَسْقَلَان", "ʿAsqalān")]
+        [TestCase("غَزَّة", "Ġazzah")]
+        [TestCase("فتح", "Fatḥ")]
         [TestCase("فِلَسْطِينَ", "Filasṭīn")]
         [TestCase("فهرج", "Fahraǧ")]
+        [TestCase("مَنْف", "Manf")]
         [TestCase("هرتسليا", "Hirtsiliyā")]
         [TestCase("يوانينا", "Īānīnā")]
         public void GivenATextInArabicScript_WhenTransliteratingIntoLatin_ThenTheCorrectTextIsReturned(
             string arabicText,
             string expectedTransliteratedText)
-            => Assert.That(transliterator.Transliterate(arabicText), Is.EqualTo(expectedTransliteratedText));
+            => Assert.That(transliterator.Transliterate(arabicText, "ar"), Is.EqualTo(expectedTransliteratedText));
     }
 }
diff --git a/TransliterationAPI/Service/TransliterationService.cs b/TransliterationAPI/Service/TransliterationService.cs
@@ -86,9 +86,9 @@ async Task<string> TryGetTransliteratedText(string text, string language)
             }
         }
 
-        async Task<string> GetTransliteratedText(string text, string language)
+        async Task<string> GetTransliteratedText(string text, string languageCode)
         {
-            switch (language)
+            switch (languageCode)
             {
                 case "ab": // Abkhaz
                     return await translitterationDotComTransliterator.Transliterate(text, "abk", "iso-9");
@@ -97,7 +97,7 @@ async Task<string> GetTransliteratedText(string text, string language)
                 case "ar": // Arabic
                 case "ary": // Maghrebi Arabic
                 case "arz": // Egyptian Arabic
-                    return arabicTransliterator.Transliterate(text);
+                    return arabicTransliterator.Transliterate(text, languageCode);
                 case "ba": // Bashkir
                     return await translitterationDotComTransliterator.Transliterate(text, "bak", "iso-9");
                 case "be": // Belarussian
@@ -107,7 +107,7 @@ async Task<string> GetTransliteratedText(string text, string language)
                 case "bn": // Bengali
                     return await ushuaiaTransliterator.Transliterate(text, "bengali_iso_transliterate");
                 case "cu": // Old Church Slavonic
-                    return await podolakTransliterator.Transliterate(text, language);
+                    return await podolakTransliterator.Transliterate(text, languageCode);
                 case "cv": // Chuvash
                     return await translitterationDotComTransliterator.Transliterate(text, "chv", "ala-lc");
                 case "el": // Greek
@@ -172,7 +172,7 @@ async Task<string> GetTransliteratedText(string text, string language)
                 case "zh-hans": // Simplified Chinese
                     return pinyinTransliterator.Transliterate(text);
                 default:
-                    throw new ArgumentException($"The \"{language}\" language is not supported");
+                    throw new ArgumentException($"The \"{languageCode}\" language is not supported");
             }
         }