[WIP] epo: some work

cabal.project

@@ -6,6 +6,8 @@ with-compiler: ghc-8.10.7
 
 packages:
     ./
+    ../epo-api-client
+    ../openalex
 
 source-repository-package
     type: git
@@ -91,6 +93,11 @@ source-repository-package
     location: https://gitlab.iscpif.fr/gargantext/crawlers/isidore.git
     tag: 3db385e767d2100d8abe900833c6e7de3ac55e1b
 
+source-repository-package
+    type: git
+    location: https://gitlab.iscpif.fr/gargantext/iso639.git
+    tag: 1a9e23e210a02da3b846d7cdc666541e2148334d
+
 source-repository-package
     type: git
     location: https://gitlab.iscpif.fr/gargantext/crawlers/istex.git

gargantext.cabal

@@ -431,6 +431,7 @@ library
     , duckling ^>= 0.2.0.0
     , ekg-core ^>= 0.1.1.7
     , ekg-json ^>= 0.1.0.7
+    , epo-api-client
     , exceptions ^>= 0.10.4
     , extra ^>= 1.7.9
     , fast-logger ^>= 3.0.5
@@ -461,7 +462,7 @@ library
     , ihaskell ^>= 0.10.2.2
     , ini ^>= 0.4.1
     , insert-ordered-containers ^>= 0.2.5.1
-    , iso639 ^>= 0.1.0.3
+    , iso639
     , jose ^>= 0.8.4
     , json-stream ^>= 0.4.2.4
     , lens ^>= 4.19.2
@@ -494,7 +495,7 @@ library
     , process ^>= 1.6.13.2
     , product-profunctors ^>= 0.11.0.3
     , profunctors ^>= 5.6.2
-    , protolude ^>= 0.3.0
+    , protolude ^>= 0.3.3
     , pureMD5 ^>= 2.1.4
     , quickcheck-instances ^>= 0.3.25.2
     , rake ^>= 0.0.1
@@ -667,7 +668,7 @@ executable gargantext-cli
     , gargantext-prelude
     , ini ^>= 0.4.1
     , optparse-generic ^>= 1.4.7
-    , protolude ^>= 0.3.0
+    , protolude ^>= 0.3.3
     , split ^>= 0.2.3.4
     , text ^>= 1.2.4.1
     , unordered-containers ^>= 0.2.16.0

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

@@ -36,6 +36,7 @@ data ExternalAPIs = OpenAlex
                   | HAL
                   | IsTex
                   | Isidore
+                  | EPO
   deriving (Show, Eq, Generic, Enum, Bounded)
 
 

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

@@ -32,6 +32,7 @@ data Database = Empty
               | HAL
               | IsTex
               | Isidore
+              | EPO
   deriving (Eq, Show, Generic, Enum, Bounded)
 
 instance Arbitrary Database where
@@ -49,6 +50,7 @@ database2origin Arxiv    = ExternalOrigin Types.Arxiv
 database2origin HAL      = ExternalOrigin Types.HAL
 database2origin IsTex    = ExternalOrigin Types.IsTex
 database2origin Isidore  = ExternalOrigin Types.Isidore
+database2origin EPO      = ExternalOrigin Types.EPO
 
 ------------------------------------------------------------------------
 data Datafield = Gargantext

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

@@ -21,9 +21,11 @@ module Gargantext.Core.Text.Corpus.API
 import Conduit
 import Control.Monad.Except
 import Data.Text qualified as T
+import EPO.API.Client.Types qualified as EPO
 import Gargantext.API.Admin.Orchestrator.Types (ExternalAPIs(..), externalAPIs)
 import Gargantext.Core (Lang(..), toISO639)
 import Gargantext.Core.Text.Corpus.API.Arxiv qualified as Arxiv
+import Gargantext.Core.Text.Corpus.API.EPO qualified as EPO
 import Gargantext.Core.Text.Corpus.API.Hal qualified as HAL
 import Gargantext.Core.Text.Corpus.API.Isidore qualified as ISIDORE
 import Gargantext.Core.Text.Corpus.API.Istex qualified as ISTEX
@@ -47,6 +49,7 @@ get :: ExternalAPIs
     -> Lang
     -> Corpus.RawQuery
     -> Maybe PUBMED.APIKey
+    -> Maybe EPO.AuthKey
     -> Maybe Corpus.Limit
     -- -> IO [HyperdataDocument]
     -> IO (Either GetCorpusError (Maybe Integer, ConduitT () HyperdataDocument IO ()))
@@ -69,5 +72,7 @@ get externalAPI la q mPubmedAPIKey limit = do
       Isidore  -> do
         docs <- ISIDORE.get la (Corpus.getLimit <$> limit) (Just $ Corpus.getRawQuery q) Nothing
         pure $ Right (Just $ fromIntegral $ length docs, yieldMany docs)
+      EPO -> do
+        first ExternalAPIError <$> EPO.get (fromMaybe "" Nothing {- email -}) q (toISO639 la) limit
   where
     parse_query = first (InvalidInputQuery q . T.pack) $ Corpus.parseQuery q

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

@@ -45,10 +45,10 @@ toDoc (OA.Work { .. } ) =
                     , _hd_institutes = institutes authorships
                     , _hd_source = source
                     , _hd_abstract = Just abstract_reconstructed
-                    , _hd_publication_date = Just $ show publication_date
+                    , _hd_publication_date = Just $ OA.showDate publication_date
                     , _hd_publication_year = Just $ publication_year
-                    , _hd_publication_month = Nothing  -- TODO
-                    , _hd_publication_day = Nothing  -- TODO
+                    , _hd_publication_month = Just $ OA.dateMonth publication_date
+                    , _hd_publication_day = Just $ OA.dateDay publication_date
                     , _hd_publication_hour = Nothing  -- TODO
                     , _hd_publication_minute = Nothing  -- TODO
                     , _hd_publication_second = Nothing  -- TODO