[hal] use conduit for hal parsing

This still isn't optimal because in our code we use lists, so this
will fetch everything (using conduit), coerce into a list and then
process.

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

@@ -18,6 +18,7 @@ New corpus means either:
 module Gargantext.API.Node.Corpus.New
       where
 
+
 import Control.Lens hiding (elements, Empty)
 import Data.Aeson
 import Data.Aeson.TH (deriveJSON)

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

@@ -39,7 +39,7 @@ get :: ExternalAPIs
     -> Maybe Limit
     -> IO [HyperdataDocument]
 get PubMed  _la q _l = PUBMED.get   q default_limit -- EN only by default
-get HAL      la q _l = HAL.get   la q default_limit
+get HAL      la q _l = HAL.getC  la q Nothing
 get IsTex    la q _l = ISTEX.get la q default_limit
 get Isidore  la q _l = ISIDORE.get la (fromIntegral <$> default_limit) (Just q) Nothing
 get _        _  _ _ = undefined

src/Gargantext/Core/Text/Corpus/API/Hal.hs

@@ -12,6 +12,8 @@ Portability : POSIX
 module Gargantext.Core.Text.Corpus.API.Hal
     where
 
+import Conduit
+import Data.Either
 import Data.Maybe
 import Data.Text (Text, pack, intercalate)
 
@@ -25,8 +27,15 @@ import qualified HAL.Doc.Corpus as HAL
 
 get :: Lang -> Text -> Maybe Integer -> IO [HyperdataDocument]
 get la q ml = do
-  docs <- HAL.getMetadataWith q (Just 0) (fromIntegral <$> ml)
-  either (panic . pack . show) (\d -> mapM (toDoc' la) $ HAL._docs d) docs
+  eDocs <- HAL.getMetadataWith q (Just 0) ml
+  either (panic . pack . show) (\d -> mapM (toDoc' la) $ HAL._docs d) eDocs
+
+getC :: Lang -> Text -> Maybe Integer -> IO [HyperdataDocument]
+getC la q ml = do
+  eDocs <- HAL.getMetadataRecursively q (Just 0) ml
+  case eDocs of
+    Left err -> panic $ pack $ show err
+    Right docsC -> runConduit $ docsC .| mapMC (toDoc' la) .| sinkList
 
 toDoc' :: Lang -> HAL.Corpus -> IO HyperdataDocument
 toDoc' la (HAL.Corpus i t ab d s aus affs struct_id) = do

stack.yaml

@@ -73,7 +73,7 @@ extra-deps:
 - git: https://gitlab.iscpif.fr/gargantext/crawlers/istex.git
   commit: daeae80365250c4bd539f0a65e271f9aa37f731f
 - git: https://gitlab.iscpif.fr/gargantext/crawlers/hal.git
-  commit: 020f5f9b308f5c23c925aedf5fb11f8b4728fb19
+  commit: a51cb46fdf8622ec48b29206a40a37bbce1380bc
 - git: https://gitlab.iscpif.fr/gargantext/crawlers/isidore.git
   commit: 3db385e767d2100d8abe900833c6e7de3ac55e1b