src-test/Ngrams/NLP.hs

   1
   2 {-|
   3 Module      : Ngrams.NLP
   4 Description :
   5 Copyright   : (c) CNRS, 2017-Present
   6 License     : AGPL + CECILL v3
   7 Maintainer  : team@gargantext.org
   8 Stability   : experimental
   9 Portability : POSIX
  10
  11 -}
  12
  13
  14 module Ngrams.NLP where
  15
  16 import Data.Text (Text)
  17 import Test.Hspec
  18 import Gargantext.Prelude
  19 import Gargantext.Core.Text.Terms.Multi
  20
  21
  22 test :: Spec
  23 test = do
  24   describe "Text that should be cleaned before sending it to NLP tools as micro-services." $ do
  25     let text   = "This is a url http://cnrs.gargantext.org to be remove and another one www.gargantext.org and digits 343242-2332 to be remove and some to keep: 232 231 33." :: Text
  26     let result = "This is a url to be remove and another one and digits to be remove and some to keep: 232 231 33."
  27     it "NLP Clean Text before sending to micro services:" $ cleanTextForNLP text `shouldBe` result
  28