fix(cleanTextForNLP): removes transformation for hyphaneted words

src/Gargantext/Core/Text/Terms/Multi.hs

@@ -11,10 +11,10 @@ Multi-terms are ngrams where n > 1.
 
 -}
 
-module Gargantext.Core.Text.Terms.Multi (multiterms, multiterms_rake, tokenTagsWith, tokenTags, cleanTextForNLP)
+module Gargantext.Core.Text.Terms.Multi (multiterms, Terms(..), tokenTag2terms, multiterms_rake, tokenTagsWith, tokenTags, cleanTextForNLP)
   where
 
-import Data.Attoparsec.Text as DAT ( digit, space, notChar, string )
+import Data.Attoparsec.Text as DAT (space, notChar, string )
 import Gargantext.Core (Lang(..), NLPServerConfig(..), PosTagAlgo(..))
 import Gargantext.Core.Text.Terms.Multi.Lang.En qualified as En
 import Gargantext.Core.Text.Terms.Multi.Lang.Fr qualified as Fr
@@ -82,12 +82,10 @@ groupTokens _  = Fr.groupTokens
 
 -- TODO: make tests here
 cleanTextForNLP :: Text -> Text
-cleanTextForNLP = unifySpaces . removeDigitsWith "-" . removeUrls
+cleanTextForNLP = unifySpaces . removeUrls
   where
     remove x = RAT.streamEdit x (const "")
 
     unifySpaces         = RAT.streamEdit (many DAT.space) (const " ")
-    removeDigitsWith x  = remove (many DAT.digit *> DAT.string x <* many DAT.digit)
-
     removeUrls          = removeUrlsWith "http" . removeUrlsWith "www"
     removeUrlsWith w    = remove (DAT.string w *> many (DAT.notChar ' ') <* many DAT.space)