Merge pull request #70 from hmlendea/normalisation

Improved normalisation of arabic names
hmlendea · Dec 4, 2023 · a03e1db · a03e1db
2 parents 1ba5ae0 + e1cc7d2
commit a03e1db
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 5 deletions.
diff --git a/MoreCulturalNamesBuilder.UnitTests/Service/NameNormaliserTests.cs b/MoreCulturalNamesBuilder.UnitTests/Service/NameNormaliserTests.cs
@@ -79,6 +79,7 @@ public void SetUp()
         [TestCase("Ǧuzur al-Īǧẗ", "Ğuzur al-Īğah")]
         [TestCase("Ḥadīṯẗ", "Hadīthah")]
         [TestCase("Ȟaȟáwakpa", "Ĥaĥáwakpa")]
+        [TestCase("Ḥamāẗ", "Hamāh")]
         [TestCase("H̱rūnīnġn", "Khrūnīnġn")]
         [TestCase("Ins Br̥k", "Ins Bruk")]
         [TestCase("Iṉspruk", "Iņspruk")]
@@ -93,6 +94,7 @@ public void SetUp()
         [TestCase("Kȁzahstān", "Kàzahstān")]
         [TestCase("Khar‌gōn", "Khargōn")]
         [TestCase("K‍ragujevak", "Kragujevak")]
+        [TestCase("Lablaẗ", "Lablah")]
         [TestCase("Lāip‌ॎsiśa", "Lāipsiśa")]
         [TestCase("Lėnkėjė", "Lėnkėjė")]
         [TestCase("Likṟṟaṉ‌sṟṟaiṉ", "Likrransrrain")]
@@ -150,11 +152,11 @@ public void SetUp()
         [TestCase("Užhorod", "Užhorod")]
         [TestCase("Vialikaja Poĺšča", "Vialikaja Poĺšča")]
         [TestCase("Vюrцby’rg", "Viurcby’rg")]
-        [TestCase("پwyrṭūrīkū", "Bwyrțūrīkū")]
         [TestCase("Ẇel‌ś‌", "Ẃelś")]
         [TestCase("Вуллонгонг", "Vullongong")]
         [TestCase("Эstoniья", "Estoni'ia")]
         [TestCase("Юli’h", "Iuli’h")]
+        [TestCase("پwyrṭūrīkū", "Bwyrțūrīkū")]
         public void WhenNormalisingForCK3_ReturnsTheExpectedNormalisedName(
             string name,
             string expectedResult)
@@ -218,6 +220,7 @@ public void WhenNormalisingForCK3_ReturnsTheExpectedNormalisedName(
         [TestCase("G‍roseṭō", "Groseţō")]
         [TestCase("Ḥadīṯẗ", "Hadīthah")]
         [TestCase("Ȟaȟáwakpa", "Ĥaĥáwakpa")]
+        [TestCase("Ḥamāẗ", "Hamāh")]
         [TestCase("H̱rūnīnġn", "Khrūnīnġn")]
         [TestCase("Ins Br̥k", "Ins Bruk")]
         [TestCase("Iṉspruk", "Iņspruk")]
@@ -232,6 +235,7 @@ public void WhenNormalisingForCK3_ReturnsTheExpectedNormalisedName(
         [TestCase("Kȁzahstān", "Kàzahstān")]
         [TestCase("Khar‌gōn", "Khargōn")]
         [TestCase("K‍ragujevak", "Kragujevak")]
+        [TestCase("Lablaẗ", "Lablah")]
         [TestCase("Lāip‌ॎsiśa", "Lāipsiśa")]
         [TestCase("Lėnkėjė", "Lėnkėjė")]
         [TestCase("Likṟṟaṉ‌sṟṟaiṉ", "Likrransrrain")]
@@ -355,6 +359,7 @@ public void WhenNormalisingForHOI4City_ReturnsTheExpectedNormalisedName(
         [TestCase("G‍roseṭō", "Grosetõ")]
         [TestCase("Ḥadīṯẗ", "Hadïthah")]
         [TestCase("Ȟaȟáwakpa", "Haháwakpa")]
+        [TestCase("Ḥamāẗ", "Hamãh")]
         [TestCase("H̱rūnīnġn", "Khrünïnghn")]
         [TestCase("Ins Br̥k", "Ins Bruk")]
         [TestCase("Iṉspruk", "Inspruk")]
@@ -369,6 +374,7 @@ public void WhenNormalisingForHOI4City_ReturnsTheExpectedNormalisedName(
         [TestCase("Kȁzahstān", "Kàzahstãn")]
         [TestCase("Khar‌gōn", "Khargõn")]
         [TestCase("K‍ragujevak", "Kragujevak")]
+        [TestCase("Lablaẗ", "Lablah")]
         [TestCase("Lāip‌ॎsiśa", "Lãipsisa")]
         [TestCase("Lėnkėjė", "Lénkéjé")]
         [TestCase("Likṟṟaṉ‌sṟṟaiṉ", "Likrransrrain")]
@@ -493,6 +499,7 @@ public void WhenNormalisingForWindow1252_ReturnsTheExpectedNormalisedName(
         [TestCase("G‍roseṭō", "Grosetō")]
         [TestCase("Ḥadīṯẗ", "Hadīthah")]
         [TestCase("Ȟaȟáwakpa", "Haháwakpa")]
+        [TestCase("Ḥamāẗ", "Hamāh")]
         [TestCase("H̱rūnīnġn", "Khrūnīnghn")]
         [TestCase("Ins Br̥k", "Ins Bruk")]
         [TestCase("Iṉspruk", "Inspruk")]
@@ -507,6 +514,7 @@ public void WhenNormalisingForWindow1252_ReturnsTheExpectedNormalisedName(
         [TestCase("Kȁzahstān", "Kàzahstān")]
         [TestCase("Khar‌gōn", "Khargōn")]
         [TestCase("K‍ragujevak", "Kragujevak")]
+        [TestCase("Lablaẗ", "Lablah")]
         [TestCase("Lāip‌ॎsiśa", "Lāipsisa")]
         [TestCase("Lėnkėjė", "Lénkéjé")]
         [TestCase("Likṟṟaṉ‌sṟṟaiṉ", "Likrransrrain")]

diff --git a/MoreCulturalNamesBuilder/Service/NameNormaliser.cs b/MoreCulturalNamesBuilder/Service/NameNormaliser.cs
@@ -87,7 +87,8 @@ public string ToCK3Charset(string name)
             processedName = Regex.Replace(processedName, "[ả]", "à");
             processedName = Regex.Replace(processedName, "[ǟ]", "ä");
             processedName = Regex.Replace(processedName, "[ậ]", "â");
-            processedName = Regex.Replace(processedName, "[ẗ]", "ah");
+            processedName = Regex.Replace(processedName, "ā[ẗ]", "āh");
+            processedName = Regex.Replace(processedName, "[a]*[ẗ]", "ah");
             processedName = Regex.Replace(processedName, "[ḃḅ]", "b");
             processedName = Regex.Replace(processedName, "[ḏḍɗɖḑ]", "d");
             processedName = Regex.Replace(processedName, "[ẹ]", "e");
@@ -204,7 +205,8 @@ public string ToHOI4CityCharset(string name)
             processedName = Regex.Replace(processedName, "[ả]", "à");
             processedName = Regex.Replace(processedName, "[ǟ]", "ä");
             processedName = Regex.Replace(processedName, "[ậ]", "â");
-            processedName = Regex.Replace(processedName, "[ẗ]", "ah");
+            processedName = Regex.Replace(processedName, "ā[ẗ]", "āh");
+            processedName = Regex.Replace(processedName, "[a]*[ẗ]", "ah");
             processedName = Regex.Replace(processedName, "[ḃḅ]", "b");
             processedName = Regex.Replace(processedName, "[ḏḍɗɖḑ]", "d");
             processedName = Regex.Replace(processedName, "[ẹ]", "e");
@@ -420,7 +422,8 @@ public string ToImperatorRomeCharset(string name)
             processedName = Regex.Replace(processedName, "[ǟ]", "ä");
             processedName = Regex.Replace(processedName, "[ậ]", "â");
             processedName = Regex.Replace(processedName, "[ả]", "à");
-            processedName = Regex.Replace(processedName, "[ẗ]", "ah");
+            processedName = Regex.Replace(processedName, "ā[ẗ]", "āh");
+            processedName = Regex.Replace(processedName, "[a]*[ẗ]", "ah");
             processedName = Regex.Replace(processedName, "[č]", "ch");
             processedName = Regex.Replace(processedName, "[ćĉċ]", "c");
             processedName = Regex.Replace(processedName, "[ď]", "d");
@@ -552,7 +555,8 @@ public string ToWindows1252(string name)
             processedName = Regex.Replace(processedName, "[Ż]", "Ž");
             processedName = Regex.Replace(processedName, "[ǣ]", "æ");
             processedName = Regex.Replace(processedName, "[ạəą]", "a");
-            processedName = Regex.Replace(processedName, "[ẗ]", "ah");
+            processedName = Regex.Replace(processedName, "ā[ẗ]", "āh");
+            processedName = Regex.Replace(processedName, "[a]*[ẗ]", "ah");
             processedName = Regex.Replace(processedName, "[ả]", "à");
             processedName = Regex.Replace(processedName, "[ậ]", "â");
             processedName = Regex.Replace(processedName, "[ăā]", "ã");