[pubmed] some api_key pubmed work

gargantext.ini_toModify

@@ -22,6 +22,7 @@ DATA_FILEPATH = FILEPATH_TO_CHANGE
 # Data path to local files (do not use quotes)
 REPO_FILEPATH = FILEPATH_TO_CHANGE
 
+PUBMED_API_KEY = ENTER_PUBMED_API_KEY
 
 
 # [external]

src/Gargantext/API/Admin/Orchestrator/Types.hs

@@ -6,6 +6,7 @@ module Gargantext.API.Admin.Orchestrator.Types
   where
 
 import Control.Lens hiding (elements)
+import Control.Monad.Reader (MonadReader)
 import Data.Aeson
 import Data.Morpheus.Types
   ( GQLType
@@ -23,7 +24,9 @@ import Test.QuickCheck.Arbitrary
 
 import qualified Gargantext.API.GraphQL.Utils as GQLU
 import Gargantext.Core.Types (TODO(..))
+import Gargantext.Database.Prelude (HasConfig(..))
 import Gargantext.Prelude
+import Gargantext.Prelude.Config (gc_pubmed_api_key)
 
 ------------------------------------------------------------------------
 instance Arbitrary a => Arbitrary (JobStatus 'Safe a) where
@@ -35,26 +38,41 @@ instance Arbitrary a => Arbitrary (JobOutput a) where
 -- | Main Types
 -- TODO IsidoreAuth
 data ExternalAPIs = All
-                  | PubMed
+                  | PubMed { mAPIKey :: Maybe Text }
                   | Arxiv
                   | HAL
                   | IsTex
                   | Isidore
-  deriving (Show, Eq, Enum, Bounded, Generic)
+  deriving (Show, Eq, Generic)
 
 
 -- | Main Instances
 instance FromJSON ExternalAPIs
 instance ToJSON ExternalAPIs
 
-externalAPIs :: [ExternalAPIs]
-externalAPIs = [minBound..maxBound]
+externalAPIs :: ( MonadReader env m
+                , HasConfig env) => m [ExternalAPIs]
+externalAPIs = do
+  pubmed_api_key <- view $ hasConfig . gc_pubmed_api_key
+
+  pure [ All
+       , PubMed { mAPIKey = Just pubmed_api_key }
+       , Arxiv
+       , HAL
+       , IsTex
+       , Isidore ]
 
 instance Arbitrary ExternalAPIs
   where
-    arbitrary = elements externalAPIs
-
-instance ToSchema ExternalAPIs
+    arbitrary = elements [ All
+                         , PubMed { mAPIKey = Nothing }
+                         , Arxiv
+                         , HAL
+                         , IsTex
+                         , Isidore ]
+
+instance ToSchema ExternalAPIs where
+  declareNamedSchema = genericDeclareNamedSchemaUnrestricted defaultSchemaOptions
 
 instance ToSchema URL where
   declareNamedSchema _ = declareNamedSchema (Proxy :: Proxy TODO)

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

@@ -130,7 +130,10 @@ deriveJSON (unPrefix "") 'ApiInfo
 instance ToSchema ApiInfo
 
 info :: FlowCmdM env err m => UserId -> m ApiInfo
-info _u = pure $ ApiInfo API.externalAPIs
+info _u = do
+  ext <- API.externalAPIs
+
+  pure $ ApiInfo ext
 
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
@@ -217,7 +220,8 @@ addToCorpusWithQuery user cid (WithQuery { _wq_query = q
       --      if cid is corpus -> add to corpus
       --      if cid is root   -> create corpus in Private
       printDebug "[G.A.N.C.New] getDataText with query" q
-      eTxts <- mapM (\db -> getDataText db (Multi l) q maybeLimit) [database2origin dbs]
+      databaseOrigin <- database2origin dbs
+      eTxts <- mapM (\db -> getDataText db (Multi l) q maybeLimit) [databaseOrigin]
 
       let lTxts = lefts eTxts
       printDebug "[G.A.N.C.New] lTxts" lTxts
@@ -242,7 +246,7 @@ addToCorpusWithQuery user cid (WithQuery { _wq_query = q
                       , _scst_remaining = Just 0
                       , _scst_events    = Just []
                       }
-        
+
         (err:_) -> do
           printDebug "Error: " err
           let jl = addEvent "ERROR" (T.pack $ show err) $
@@ -283,7 +287,7 @@ addToCorpusWithForm user cid (NewWithForm ft ff d l _n) logStatus jobLog = do
       CSV       -> Parser.parseFormatC Parser.CsvGargV3
       WOS       -> Parser.parseFormatC Parser.WOS
       PresseRIS -> Parser.parseFormatC Parser.RisPresse
-  
+
   -- TODO granularity of the logStatus
   let data' = case ff of
         Plain -> cs d
@@ -422,4 +426,3 @@ addToCorpusWithFile user cid nwf@(NewWithFile _d _l fName) logStatus = do
                 , _scst_remaining = Just 0
                 , _scst_events    = Just []
                 }
-

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

@@ -4,6 +4,7 @@ module Gargantext.API.Node.Corpus.Types where
 
 import Control.Lens hiding (elements, Empty)
 import Control.Monad.Fail (fail)
+import Control.Monad.Reader (MonadReader)
 import Data.Aeson
 import Data.Aeson.TH (deriveJSON)
 import Data.Monoid (mempty)
@@ -19,6 +20,8 @@ import Gargantext.Prelude
 import qualified Gargantext.API.Admin.Orchestrator.Types as T
 import Gargantext.Core.Utils.Prefix (unPrefix)
 import Gargantext.Database.Action.Flow (DataOrigin(..))
+import Gargantext.Database.Prelude (HasConfig(..))
+import Gargantext.Prelude.Config (gc_pubmed_api_key)
 
 data Database = Empty
               | PubMed
@@ -31,13 +34,17 @@ data Database = Empty
 deriveJSON (unPrefix "") ''Database
 instance ToSchema Database
 
-database2origin :: Database -> DataOrigin
-database2origin Empty   = InternalOrigin T.IsTex
-database2origin PubMed  = ExternalOrigin T.PubMed
-database2origin Arxiv   = ExternalOrigin T.Arxiv
-database2origin HAL     = ExternalOrigin T.HAL
-database2origin IsTex   = ExternalOrigin T.IsTex
-database2origin Isidore = ExternalOrigin T.Isidore
+database2origin :: ( MonadReader env m
+                   , HasConfig env ) => Database -> m DataOrigin
+database2origin Empty   = pure $ InternalOrigin T.IsTex
+database2origin PubMed  = do
+  pubmed_api_key <- view $ hasConfig . gc_pubmed_api_key
+
+  pure $ ExternalOrigin $ T.PubMed { mAPIKey = Just pubmed_api_key }
+database2origin Arxiv   = pure $ ExternalOrigin T.Arxiv
+database2origin HAL     = pure $ ExternalOrigin T.HAL
+database2origin IsTex   = pure $ ExternalOrigin T.IsTex
+database2origin Isidore = pure $ ExternalOrigin T.Isidore
 
 ------------------------------------------------------------------------
 data Datafield = Gargantext
@@ -66,4 +73,3 @@ instance ToSchema Datafield where
   declareNamedSchema _ = do
     return $ NamedSchema (Just "Datafield") $ mempty
       & type_ ?~ SwaggerObject
-        

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

@@ -38,7 +38,7 @@ get :: ExternalAPIs
     -> Maybe Limit
     -- -> IO [HyperdataDocument]
     -> IO (Either ClientError (Maybe Integer, ConduitT () HyperdataDocument IO ()))
-get PubMed  _la q limit = PUBMED.get q limit
+get PubMed { mAPIKey = mAPIKey }  _la q limit = PUBMED.get mAPIKey q limit
   --docs <- PUBMED.get   q default_limit -- EN only by default
   --pure (Just $ fromIntegral $ length docs, yieldMany docs)
 get Arxiv    la q limit = Arxiv.get la q (fromIntegral <$> limit)

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

@@ -26,6 +26,7 @@ import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
 
 import qualified PUBMED as PubMed
 import qualified PUBMED.Parser as PubMedDoc
+import PUBMED.Types (Config(..))
 
 
 type Query = Text
@@ -34,9 +35,12 @@ type Limit = PubMed.Limit
 
 -- | TODO put default pubmed query in gargantext.ini
 -- by default: 10K docs
-get :: Query -> Maybe Limit -> IO (Either ClientError (Maybe Integer, ConduitT () HyperdataDocument IO ()))
-get q l = do
-  eRes <- PubMed.getMetadataWithC q l
+get :: Maybe Text
+    -> Query
+    -> Maybe Limit
+    -> IO (Either ClientError (Maybe Integer, ConduitT () HyperdataDocument IO ()))
+get mAPIKey q l = do
+  eRes <- PubMed.getMetadataWithC (Config { mAPIKey = mAPIKey }) q l
   pure $ (\(len, docsC) -> (len, docsC .| mapC (toDoc EN))) <$> eRes
   --either (\e -> panic $ "CRAWL: PubMed" <> e) (map (toDoc EN))
   --      <$> PubMed.getMetadataWithC q l

src/Gargantext/Database/Action/Flow.hs

@@ -49,6 +49,7 @@ module Gargantext.Database.Action.Flow -- (flowDatabase, ngrams2list)
 
 import Conduit
 import Control.Lens ((^.), view, _Just, makeLenses)
+import Control.Monad.Reader (MonadReader)
 import Data.Aeson.TH (deriveJSON)
 import Data.Conduit.Internal (zipSources)
 import Data.Either
@@ -128,9 +129,13 @@ deriveJSON (unPrefix "_do_") ''DataOrigin
 instance ToSchema DataOrigin where
   declareNamedSchema = genericDeclareNamedSchema (unPrefixSwagger "_do_")
 
-allDataOrigins :: [DataOrigin]
-allDataOrigins = map InternalOrigin API.externalAPIs
-              <> map ExternalOrigin API.externalAPIs
+allDataOrigins :: ( MonadReader env m
+                  , HasConfig env) => m [DataOrigin]
+allDataOrigins = do
+  ext <- API.externalAPIs
+
+  pure $ map InternalOrigin ext
+      <> map ExternalOrigin ext
 
 ---------------
 data DataText = DataOld ![NodeId]
@@ -284,7 +289,7 @@ flow c u cn la mfslw (mLength, docsC) logStatus = do
                            , _scst_events    = Just []
                            }
       pure $ Prelude.head id
-      
+
 
 
 ------------------------------------------------------------------------
@@ -465,7 +470,7 @@ documentIdWithNgrams f = traverse toDocumentIdWithNgrams
 mapNodeIdNgrams :: (Ord b, Hashable b)
                 => [DocumentIdWithNgrams a b]
                 -> HashMap b
-                       (Map NgramsType 
+                       (Map NgramsType
                             (Map NodeId Int)
                        )
 mapNodeIdNgrams = HashMap.unionsWith (Map.unionWith (Map.unionWith (+))) . fmap f

src/Gargantext/Database/GargDB.hs

@@ -115,7 +115,7 @@ toFilePath fp1 fp2 = fp1 <> "/" <> fp2
 
 -- | Disk operations
 -- | For example, this write file with a random filepath
--- better use a hash of json of Type used to parameter as input 
+-- better use a hash of json of Type used to parameter as input
 -- the functions
 writeFile :: ( MonadReader env m
              , HasConfig   env
@@ -139,7 +139,7 @@ writeFile a = do
 
 ---
 
--- | Example to read a file with Type 
+-- | Example to read a file with Type
 readGargFile :: ( MonadReader  env m
             , HasConfig    env
             , MonadBase IO     m
@@ -205,4 +205,3 @@ onDisk_2 action fp1 fp2 = do
         | isDoesNotExistError e = return ()
         | otherwise = throwIO e
 ------------------------------------------------------------------------
-

stack.yaml

@@ -32,8 +32,10 @@ ghc-options:
 #  "$everything": -haddock
 
 extra-deps:
-- git: https://gitlab.iscpif.fr/gargantext/haskell-gargantext-prelude.git
-  commit: 03c3c381ba9df6da02a7a3c8d7b78cde9a380d04
+#- git: https://gitlab.iscpif.fr/gargantext/haskell-gargantext-prelude.git
+#  commit: 03c3c381ba9df6da02a7a3c8d7b78cde9a380d04
+- git: https://gitlab.iscpif.fr/cgenie/haskell-gargantext-prelude
+  commit: 7b9656ce5180d2723301561b0715f06c7b905182
 - git: https://gitlab.iscpif.fr/gargantext/gargantext-graph.git
   commit: 588e104fe7593210956610cab0041fd16584a4ce
   # Data Mining Libs