Merge remote-tracking branch 'origin/dev-openalex' into dev-merge

src/Gargantext/Core.hs

@@ -80,10 +80,24 @@ instance FromHttpApiData Lang
 instance ToHttpApiData Lang where
   toUrlPiece = pack . show
 instance Hashable Lang
-
 instance Arbitrary Lang where
   arbitrary = arbitraryBoundedEnum
 
+-- | https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes
+toISO639Lang :: Lang -> Maybe Text
+toISO639Lang All = Nothing
+toISO639Lang DE = Just "de"
+toISO639Lang EL = Just "el"
+toISO639Lang EN = Just "en"
+toISO639Lang ES = Just "es"
+toISO639Lang FR = Just "fr"
+toISO639Lang IT = Just "it"
+toISO639Lang PL = Just "pl"
+toISO639Lang PT = Just "pt"
+toISO639Lang RU = Just "ru"
+toISO639Lang UK = Just "uk"
+toISO639Lang ZH = Just "zh"
+
 allLangs :: [Lang]
 allLangs = [minBound .. maxBound]
 

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

@@ -57,7 +57,7 @@ get externalAPI la q mPubmedAPIKey limit = do
     Left err -> pure $ Left $ InvalidInputQuery q (T.pack err)
     Right corpusQuery -> case externalAPI of
       OpenAlex -> first ExternalAPIError <$>
-                    OpenAlex.get (fromMaybe "" Nothing {- email -}) q limit
+                    OpenAlex.get (fromMaybe "" Nothing {- email -}) q la limit
       PubMed -> first ExternalAPIError <$>
                   PUBMED.get (fromMaybe "" mPubmedAPIKey) corpusQuery limit
       --docs <- PUBMED.get   q default_limit -- EN only by default

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

@@ -11,6 +11,7 @@ module Gargantext.Core.Text.Corpus.API.OpenAlex where
 
 import Conduit
 import qualified Data.Text as T
+import Gargantext.Core (Lang, toISO639Lang)
 import Gargantext.Core.Text.Corpus.Query as Corpus
 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
 import Protolude
@@ -21,11 +22,14 @@ import Servant.Client (ClientError)
 
 get :: Text
     -> Corpus.RawQuery
+    -> Lang
     -> Maybe Limit
     -> IO (Either ClientError (Maybe Integer, ConduitT () HyperdataDocument IO ()))
-get _email q _l = do
-  eRes <- OA.fetchWorksC Nothing Nothing $ Just $ Corpus.getRawQuery q
-  pure $ (\(len, docsC) -> (len, docsC .| takeC 1000 .| mapC toDoc)) <$> eRes
+get _email q lang mLimit = do
+  let limit = getLimit $ fromMaybe 10000 mLimit
+  let mFilter = (\l -> "language:" <> l) <$> toISO639Lang lang
+  eRes <- OA.fetchWorksC Nothing mFilter $ Just $ Corpus.getRawQuery q
+  pure $ (\(len, docsC) -> (len, docsC .| takeC limit .| mapC toDoc)) <$> eRes
 
 toDoc :: OA.Work -> HyperdataDocument
 toDoc (OA.Work { .. } ) =