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

   1 {-|
   2 Module      : Gargantext.Core.Text.Corpus.API.Pubmed
   3 Description : Pubmed API connection
   4 Copyright   : (c) CNRS, 2017
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 -}
  11
  12
  13 module Gargantext.Core.Text.Corpus.API.Pubmed
  14     ( get )
  15     where
  16
  17 import Conduit
  18 import Control.Monad.Reader (runReaderT)
  19 import Data.Either (Either)
  20 import Data.Maybe
  21 import Data.Text (Text)
  22 import qualified Data.Text as Text
  23 import Servant.Client (ClientError)
  24
  25 import Gargantext.Prelude
  26 import Gargantext.Core (Lang(..))
  27 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
  28
  29 import qualified PUBMED as PubMed
  30 import qualified PUBMED.Parser as PubMedDoc
  31 import PUBMED.Types (Config(..))
  32
  33
  34 type Query = Text
  35 type Limit = Int
  36
  37
  38 -- | TODO put default pubmed query in gargantext.ini
  39 -- by default: 10K docs
  40 get :: Text
  41     -> Query
  42     -> Maybe Limit
  43     -> IO (Either ClientError (Maybe Integer, ConduitT () HyperdataDocument IO ()))
  44 get apiKey q l = do
  45   eRes <- runReaderT PubMed.getMetadataWithC (Config { apiKey  = Just apiKey
  46                                                      , query   = q
  47                                                      , perPage = Just 200
  48                                                      , mWebEnv = Nothing })
  49   let takeLimit = case l of
  50         Nothing -> mapC identity
  51         Just l' -> takeC $ fromIntegral l'
  52   pure $ (\(len, docsC) -> (len, docsC .| takeLimit .| mapC (toDoc EN))) <$> eRes
  53   --either (\e -> panic $ "CRAWL: PubMed" <> e) (map (toDoc EN))
  54   --      <$> PubMed.getMetadataWithC q l
  55
  56 toDoc :: Lang -> PubMedDoc.PubMed -> HyperdataDocument
  57 toDoc l (PubMedDoc.PubMed { pubmed_id
  58                           , pubmed_article = PubMedDoc.PubMedArticle t j as aus
  59                           , pubmed_date = PubMedDoc.PubMedDate a y m d }
  60           ) = HyperdataDocument { _hd_bdd = Just "PubMed"
  61                                 , _hd_doi = Nothing
  62                                 , _hd_url = Nothing
  63                                 , _hd_uniqId = Just $ Text.pack $ show pubmed_id
  64                                 , _hd_uniqIdBdd = Nothing
  65                                 , _hd_page = Nothing
  66                                 , _hd_title = t
  67                                 , _hd_authors = authors aus
  68                                 , _hd_institutes = institutes aus
  69                                 , _hd_source = j
  70                                 , _hd_abstract = abstract as
  71                                 , _hd_publication_date = Just $ Text.pack $ show a
  72                                 , _hd_publication_year = Just $ fromIntegral y
  73                                 , _hd_publication_month = Just m
  74                                 , _hd_publication_day = Just d
  75                                 , _hd_publication_hour = Nothing
  76                                 , _hd_publication_minute = Nothing
  77                                 , _hd_publication_second = Nothing
  78                                 , _hd_language_iso2 = Just $ (Text.pack . show) l }
  79       where
  80         authors :: [PubMedDoc.Author] -> Maybe Text
  81         authors [] = Nothing
  82         authors au = Just $ (Text.intercalate ", ")
  83                           $ catMaybes
  84                           $ map (\n -> PubMedDoc.foreName n <> Just " " <> PubMedDoc.lastName n) au
  85
  86         institutes :: [PubMedDoc.Author] -> Maybe Text
  87         institutes [] = Nothing
  88         institutes au = Just $ (Text.intercalate ", ")
  89                              $ (map (Text.replace ", " " - "))
  90                              $ catMaybes
  91                              $ map PubMedDoc.affiliation au
  92
  93
  94         abstract :: [Text] -> Maybe Text
  95         abstract [] = Nothing
  96         abstract as' = Just $ Text.intercalate ", " as'