Fix bug in GargPorter implementation

src/Gargantext/Core/Text/Terms/Mono/Stem/Internal/GargPorter.hs

@@ -122,9 +122,14 @@ afterStep1b word = fromMaybe word result
 step1b :: [Char] -> [Char]
 step1b = either identity afterStep1b . beforeStep1b
 
+-- Issue #415: According to the Porter stemming rules, we need to replace `y` with `i` only if there
+-- are no other vocals at the end.
 step1c :: [Char] -> [Char]
-step1c word = fromMaybe word result
-    where result = replaceEnd containsVowel word "y" "i"
+step1c word
+  | length word > 2 && List.last word == 'y' && isConsonant word (List.length word - 2)
+  = fromMaybe word $ replaceEnd containsVowel word "y" "i"
+  | otherwise
+  = word
 
 step1 :: [Char] -> [Char]
 step1 = step1c . step1b . step1a

test/Test/API/Private/Table.hs

@@ -52,6 +52,7 @@ tests = sequential $ aroundAll withTestDBAndPort $ do
         let corpusId = _sctx_data
         withApplication _sctx_app $ do
           withValidLogin _sctx_port "alice" (GargPassword "alice") $ \clientEnv token -> do
+
             liftIO $ do
               (HashedResponse _ tr1)
                 <- checkEither $ runClientM (get_table token
@@ -63,8 +64,8 @@ tests = sequential $ aroundAll withTestDBAndPort $ do
                                                        (Just $ RawQuery "soy")
                                                        Nothing
                                             ) clientEnv
-              length (tr_docs tr1) `shouldBe` 4
 
+              length (tr_docs tr1) `shouldBe` 3
 
 createSoySauceCorpus :: SpecContext () -> IO (SpecContext CorpusId)
 createSoySauceCorpus ctx@SpecContext{..} = do

test/Test/Database/Operations/DocumentSearch.hs

@@ -135,6 +135,8 @@ stemmingTest :: TestEnv -> Assertion
 stemmingTest _env = do
   stem EN GargPorterAlgorithm "Ajeje"        `shouldBe` "Ajeje"
   stem EN GargPorterAlgorithm "PyPlasm:"     `shouldBe` "PyPlasm:"
+  stem EN GargPorterAlgorithm "soy"          `shouldBe` "soy"
+  stem EN GargPorterAlgorithm "cry"          `shouldBe` "cri"
   -- This test outlines the main differences between Porter and Lancaster.
   stem EN GargPorterAlgorithm "dancer"       `shouldBe` "dancer"
   stem EN LancasterAlgorithm  "dancer"       `shouldBe` "dant"