Bugfix

Author: abhik1505040

normalizer/normalize.py

@@ -49,7 +49,7 @@ def normalize(
 
     # apply unicode replacements    
     text = const.UNICODE_REPLACEMENTS_REGEX.sub(
-        lambda match: const.UNICODE_REPLACEMENTS[match.group(0)], 
+        lambda match: const.UNICODE_REPLACEMENTS.get(match.group(0), f"{match.group(1)}\u09cc"), 
         text
     )
 

tests/test.py

@@ -65,12 +65,16 @@ def test_unicode_replacements():
         [
             "\u00a0",
             " "
+        ],
+        [
+            'েশৗ',
+            'শৌ'
         ]
     ]
 
     for d in data:
         normalized_text = normalize(d[0])
-        assert normalized_text == d[1]
+        assert normalized_text == d[1], f"{normalized_text} != {d[1]}"