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