Merge pull request #32 from hmlendea/arabic

Improved the `Arabic` transliteration
hmlendea · May 24, 2023 · 96c5fd9 · 96c5fd9
2 parents c738abf + 3322954
commit 96c5fd9
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 1 deletion.
diff --git a/TransliterationAPI.UnitTests/Service/Transliterators/ArabicTransliteratorTests.cs b/TransliterationAPI.UnitTests/Service/Transliterators/ArabicTransliteratorTests.cs
@@ -25,7 +25,9 @@ public void SetUp()
         [TestCase("روما", "Rūmā")]
         [TestCase("عَسْقَلَان", "ʿAsqalān")]
         [TestCase("فِلَسْطِينَ", "Filasṭīn")]
+        [TestCase("فهرج", "Fahraǧ")]
         [TestCase("هرتسليا", "Hirtsiliyā")]
+        [TestCase("يوانينا", "Īānīnā")]
         public void GivenATextInArabicScript_WhenTransliteratingIntoLatin_ThenTheCorrectTextIsReturned(
             string arabicText,
             string expectedTransliteratedText)

diff --git a/TransliterationAPI/Service/TransliterationService.cs b/TransliterationAPI/Service/TransliterationService.cs
@@ -95,6 +95,8 @@ async Task<string> GetTransliteratedText(string text, string language)
                 case "ady": // Adyghe
                     return await translitterationDotComTransliterator.Transliterate(text, "ady", "iso-9");
                 case "ar": // Arabic
+                case "ary": // Maghrebi Arabic
+                case "arz": // Egyptian Arabic
                     return arabicTransliterator.Transliterate(text);
                 case "ba": // Bashkir
                     return await translitterationDotComTransliterator.Transliterate(text, "bak", "iso-9");

diff --git a/TransliterationAPI/Service/Transliterators/ArabicTransliterator.cs b/TransliterationAPI/Service/Transliterators/ArabicTransliterator.cs
@@ -140,10 +140,12 @@ string ApplyFixes(string text)
             fixedText = fixedText.Replace("-r", "-R");
 
             fixedText = Regex.Replace(fixedText, "([Ǧǧ])b", "$1ib");
+            fixedText = Regex.Replace(fixedText, "([Īī])wā", "$1ā");
             fixedText = Regex.Replace(fixedText, "([Nn])m", "$1im");
             fixedText = Regex.Replace(fixedText, "([Qq])f", "$1if");
             fixedText = Regex.Replace(fixedText, "Bā", "Ba");
             fixedText = Regex.Replace(fixedText, "Dm", "Dim");
+            fixedText = Regex.Replace(fixedText, "Fh", "Fah");
             fixedText = Regex.Replace(fixedText, "Hr", "Hir");
             fixedText = Regex.Replace(fixedText, "Is", "ʾIs");
             fixedText = Regex.Replace(fixedText, "R([hm])", "Ra$1");
@@ -170,7 +172,7 @@ string ApplyFixes(string text)
             fixedText = Regex.Replace(fixedText, "lnd", "land");
             fixedText = Regex.Replace(fixedText, "lẗ", "laḧ");
             fixedText = Regex.Replace(fixedText, "mš", "maš");
-            fixedText = Regex.Replace(fixedText, "rzn", "rzin");
+            fixedText = Regex.Replace(fixedText, "rǧ", "raǧ");
             fixedText = Regex.Replace(fixedText, "tsl", "tsil");
 
             fixedText = Regex.Replace(fixedText, "rzn", "rzin");