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