Improve text2ngrams to strip spaces

src/Gargantext/Core/Text/Terms.hs

@@ -55,7 +55,7 @@ import Gargantext.Core.Text.Terms.Mono.Token.En (tokenize)
 import Gargantext.Core.Text.Terms.Multi (multiterms)
 import Gargantext.Core.Types
 import Gargantext.Database.Prelude (Cmd)
-import Gargantext.Database.Schema.Ngrams (Ngrams(..), NgramsType(..))
+import Gargantext.Database.Schema.Ngrams (Ngrams(..), NgramsType(..), ngramsTerms, text2ngrams)
 import Gargantext.Prelude
 
 
@@ -118,11 +118,11 @@ class ExtractNgramsT h
 
 filterNgramsT :: Int -> Map Ngrams (Map NgramsType Int)
                      -> Map Ngrams (Map NgramsType Int)
-filterNgramsT s ms = Map.fromList $ map (\a -> filter' s a) $ Map.toList ms
+filterNgramsT s ms = Map.fromList $ map filter' $ Map.toList ms
   where
-    filter' s' (ng@(Ngrams t n),y) = case (Text.length t) < s' of
-          True  -> (ng,y)
-          False -> (Ngrams (Text.take s' t) n , y)
+    filter' (ng,y)
+      | Text.length (ng ^. ngramsTerms) < s = (ng,y)
+      | otherwise                           = (text2ngrams (Text.take s (ng ^. ngramsTerms)), y)
 
 
 -- =======================================================

src/Gargantext/Database/Schema/Ngrams.hs

@@ -25,7 +25,7 @@ import Control.Monad (mzero)
 import Data.Aeson
 import Data.Aeson.Types (toJSONKeyText)
 import Data.Map (Map, fromList, lookup)
-import Data.Text (Text, splitOn, pack)
+import Data.Text (Text, splitOn, pack, strip)
 import Gargantext.Core.Types (TODO(..))
 import Gargantext.Prelude
 import Prelude (Functor)
@@ -140,16 +140,18 @@ fromNgramsTypeId id = lookup id
 
 ------------------------------------------------------------------------
 -- | TODO put it in Gargantext.Core.Text.Ngrams
-data Ngrams = Ngrams { _ngramsTerms :: Text
-                     , _ngramsSize  :: Int
-           } deriving (Generic, Show, Eq, Ord)
+data Ngrams = UnsafeNgrams { _ngramsTerms :: Text
+                           , _ngramsSize  :: Int
+                           } deriving (Generic, Show, Eq, Ord)
 
 makeLenses ''Ngrams
 instance PGS.ToRow Ngrams where
-  toRow (Ngrams t s) = [toField t, toField s]
+  toRow (UnsafeNgrams t s) = [toField t, toField s]
 
 text2ngrams :: Text -> Ngrams
-text2ngrams txt = Ngrams txt $ length $ splitOn " " txt
+text2ngrams txt = UnsafeNgrams txt' $ length $ splitOn " " txt'
+  where
+    txt' = strip txt
 
 -------------------------------------------------------------------------
 -- | TODO put it in Gargantext.Core.Text.Ngrams