[FEAT] Iramuteq parser (WIP)

src/Gargantext/API/Node/Corpus/New.hs

@@ -263,6 +263,7 @@ addToCorpusWithForm user cid (NewWithForm ft ff d l _n sel) jobHandle = do
       CSV       -> Parser.parseFormatC Parser.CsvGargV3
       WOS       -> Parser.parseFormatC Parser.WOS
       PresseRIS -> Parser.parseFormatC Parser.RisPresse
+      Iramuteq  -> Parser.parseFormatC Parser.Iramuteq
 
   -- TODO granularity of the logStatus
   let data' = case ff of

src/Gargantext/API/Node/Corpus/New/Types.hs

@@ -14,6 +14,7 @@ data FileType = CSV
               | CSV_HAL
               | PresseRIS
               | WOS
+              | Iramuteq
   deriving (Eq, Show, Generic)
 instance ToSchema FileType
 instance Arbitrary FileType where arbitrary = elements [CSV, PresseRIS]
@@ -26,7 +27,8 @@ instance FromHttpApiData FileType where
   parseUrlPiece "CSV_HAL"   = pure CSV_HAL
   parseUrlPiece "PresseRis" = pure PresseRIS
   parseUrlPiece "WOS"       = pure WOS
-  parseUrlPiece _           = pure CSV -- TODO error here
+  parseUrlPiece "Iramuteq"  = pure Iramuteq
+  parseUrlPiece _           = panic "[G.A.A.Node.Corpus.New] File Type not implemented (yet)"
 instance ToHttpApiData FileType where
   toUrlPiece = pack . show
 

src/Gargantext/Core/Text/Corpus/Parsers.hs

@@ -118,6 +118,16 @@ parseFormatC WOS Plain bs = do
               .| mapC (map $ first WOS.keys)
               .| mapC (map $ both decodeUtf8)
               .| mapMC (toDoc WOS)) ) <$> eDocs
+
+parseFormatC Iramuteq Plain bs = do
+  let eDocs = runParser' Iramuteq bs
+  pure $ (\docs ->
+            ( Just $ fromIntegral $ length docs
+            , yieldMany docs
+              .| mapC (map $ first Iramuteq.keys)
+              .| mapC (map $ both decodeUtf8)
+              .| mapMC ((toDoc Iramuteq) . (map (second (Text.replace "_" " ")))) ))<$> eDocs
+
 parseFormatC ft ZIP bs = do
   path <- liftBase $ emptySystemTempFile "parsed-zip"
   liftBase $ DB.writeFile path bs

src/Gargantext/Core/Text/Corpus/Parsers/RIS.hs

@@ -25,8 +25,8 @@ import Data.Attoparsec.ByteString.Char8 (isEndOfLine)
 import Data.ByteString (ByteString, intercalate)
 import Gargantext.Prelude hiding (takeWhile, take)
 import qualified Data.List as DL
--------------------------------------------------------------
 
+-------------------------------------------------------------
 parser :: Parser [[(ByteString, ByteString)]]
 parser = do
     n  <- notice "TY  -"