Added more special characters for unaccent
authorNeil Smith <neil.git@njae.me.uk>
Fri, 21 Oct 2016 12:33:30 +0000 (13:33 +0100)
committerNeil Smith <neil.git@njae.me.uk>
Fri, 21 Oct 2016 12:33:30 +0000 (13:33 +0100)
language_models.py

index bf00875c43e134fd2b46327e80c56c8468c60e58..644338eda837e845095353ed882b7daf048cf1d0 100644 (file)
@@ -11,7 +11,7 @@ import itertools
 from math import log10
 import os 
 
-unaccent_specials = ''.maketrans({"’": "'"})
+unaccent_specials = ''.maketrans({"’": "'", '“': '"', '”': '"'})
 
 def letters(text):
     """Remove all non-alphabetic characters from a text