Merge pull request #13 from hmlendea/improvements

Improved greek transliteration
hmlendea · Jan 26, 2022 · d80aa25 · d80aa25
2 parents bbbcbef + 5a81e4a
commit d80aa25
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 2 deletions.
diff --git a/Service/Transliterators/TransliterateDotComTransliterator.cs b/Service/Transliterators/TransliterateDotComTransliterator.cs
@@ -41,23 +41,33 @@ private string ApplyFixes(string text, string language)
                 fixedText = Regex.Replace(fixedText, "ō̂", "ô");
 
                 fixedText = Regex.Replace(fixedText, "Ch", "Kh");
+                fixedText = Regex.Replace(fixedText, "R([^\\s])", "Rh$1");
 
                 fixedText = Regex.Replace(fixedText, "ch", "kh");
-                fixedText = Regex.Replace(fixedText, "rh", "r");
+                fixedText = Regex.Replace(fixedText, "([^\\s])rh", "$1r");
 
                 fixedText = Regex.Replace(fixedText, "Mp", "V");
                 fixedText = Regex.Replace(fixedText, "Nk", "G");
                 fixedText = Regex.Replace(fixedText, "Nt", "D");
                 fixedText = Regex.Replace(fixedText, "([nrs])mp", "$1v");
                 fixedText = Regex.Replace(fixedText, "mp([aáioó])", "v$1");
                 fixedText = Regex.Replace(fixedText, "nknt", "gd");
+                fixedText = Regex.Replace(fixedText, "nnt", "ng");
                 fixedText = Regex.Replace(fixedText, "ntm", "dm");
                 fixedText = Regex.Replace(fixedText, "rnk", "rk");
                 fixedText = Regex.Replace(fixedText, "rnt", "rd");
                 fixedText = Regex.Replace(fixedText, "snt", "sht");
                 fixedText = Regex.Replace(fixedText, "([A-Za-z])'([A-Za-z])", "$1$2");
                 fixedText = Regex.Replace(fixedText, "([Rr])(a|ṓ)", "$1h$2");
                 fixedText = Regex.Replace(fixedText, "([Rr])h([a])", "$1$2");
+
+                fixedText = Regex.Replace(fixedText, "Agí(a|os)", "Hagí$1");
+                fixedText = Regex.Replace(fixedText, "polē$", "polis");
+                fixedText = Regex.Replace(fixedText, "polē ", "polis");
+
+                fixedText = Regex.Replace(fixedText, "ín$", "ínon");
+                fixedText = Regex.Replace(fixedText, "o$", "on");
+
             }
             else if (language == "he")
             {

diff --git a/Service/Transliterators/TranslitterationDotComTransliterator.cs b/Service/Transliterators/TranslitterationDotComTransliterator.cs
@@ -58,15 +58,24 @@ private string ApplyLanguageSpecificFixes(string text, string language)
                 fixedText = Regex.Replace(fixedText, "ήr([iíί])$", "éri");
 
                 fixedText = Regex.Replace(fixedText, "([a-zA-Z])H", "$1h");
+
+
+                fixedText = Regex.Replace(fixedText, "[Ά]", "Á");
+                fixedText = Regex.Replace(fixedText, "[Έ]", "É");
+                fixedText = Regex.Replace(fixedText, "[Ό]", "Ó");
+
                 fixedText = Regex.Replace(fixedText, "ᾶ", "a");
                 fixedText = Regex.Replace(fixedText, "ά", "á");
                 fixedText = Regex.Replace(fixedText, "[έ]", "é");
                 fixedText = Regex.Replace(fixedText, "[ίή]", "í");
-                fixedText = Regex.Replace(fixedText, "ώ", "ó");
+                fixedText = Regex.Replace(fixedText, "[ϊ]", "ï");
+                fixedText = Regex.Replace(fixedText, "[όώ]", "ó");
                 fixedText = Regex.Replace(fixedText, "ς", "s");
                 fixedText = Regex.Replace(fixedText, "ύ", "ú");
 
                 fixedText = fixedText.Replace("Mp", "B");
+                fixedText = fixedText.Replace("Nt", "D");
+
                 fixedText = Regex.Replace(fixedText, "([r])nt", "$1d");
                 fixedText = Regex.Replace(fixedText, "([nrs])mp", "$1b");
             }