[FIX] removeSub works for now but TODO inside.

app/Main.hs

@@ -13,24 +13,32 @@ import Data.Conduit (ConduitT)
 import Data.ByteString.Lazy (ByteString)
 import Data.ByteString.Char8 (pack)
 import Control.Monad.Catch (MonadThrow)
-import Text.Parsec hiding (runParser)
-import Text.Parsec.ByteString.Lazy
+
+import Control.Applicative
+import Data.Attoparsec.ByteString
+import Data.Attoparsec.ByteString.Char8 (anyChar)
+
 import qualified Data.ByteString.Lazy as LBS
+import qualified Data.ByteString      as DB
 import qualified Data.Text as T
 
 runParser :: Show res => (Cursor -> res) -> LBS.ByteString -> res
 runParser parser = parser . fromDocument . parseLBS_ def
 
+-- | TODO this parser need at least one subs at the end
+-- (use endOfInput)
 removeSub :: Parser ByteString
 removeSub = do
-  dt <- many1 anyChar <* (skipMany $ sub <|> asub)
-  return . LBS.fromStrict $ pack dt
+  dt <- many textWithBalise
+  pure $ LBS.fromStrict $ pack $ concat dt
   where
-    sub = string "<sub>"
+    textWithBalise = manyTill anyChar (try subs)
+    subs = sub <|> asub -- <|> isEndOfInput
+    sub  = string "<sub>"
     asub = string "</sub>"
 
 runSimpleFindPubmedAbstractRequest :: T.Text -> IO [PubMed]
-runSimpleFindPubmedAbstractRequest rq = do
+runSimpleFindPubmedAbstractRequest rq =  do
   manager' <- newManager tlsManagerSettings
   res <- runClientM
     (search (Just rq))
@@ -45,7 +53,8 @@ runSimpleFindPubmedAbstractRequest rq = do
       case res' of
         (Left err) -> return []
         (Right (BsXml abstracts)) -> do
-          case (parse removeSub "error" abstracts) of
+          -- TODO remove "</sub>" maybe there is a cleaner way with isEndOfInput
+          case (parseOnly removeSub $ LBS.toStrict abstracts <> "</sub>") of
             (Left _) -> return []
             (Right v) -> pubMedParser v
 

package.yaml

@@ -35,7 +35,7 @@ dependencies:
 - time
 - data-time-segment
 - protolude
-- parsec
+- attoparsec
 - either
 
 library: