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

   1 {-|
   2 Module      : Gargantext.Core.Text.Corpus.API.Istex
   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.Istex
  14     where
  15
  16 import Data.Either (Either(..))
  17 import Data.List (concat)
  18 import Data.Maybe
  19 import Data.Text (Text, pack)
  20
  21 import Gargantext.Core (Lang(..))
  22 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
  23 import qualified Gargantext.Defaults as Defaults
  24 import Gargantext.Prelude
  25 import qualified Gargantext.Core.Text.Corpus.Parsers.Date as Date
  26 import qualified ISTEX        as ISTEX
  27 import qualified ISTEX.Client as ISTEX
  28
  29
  30 get :: Lang -> Text -> Maybe Integer -> IO [HyperdataDocument]
  31 get la q _ml = do
  32   --docs <- ISTEX.getMetadataWith q (fromIntegral <$> ml)
  33   --printDebug "[Istex.get] calling getMetadataScrollProgress for la" la
  34   --printDebug "[Istex.get] calling getMetadataScrollProgress for q" q
  35   --printDebug "[Istex.get] calling getMetadataScrollProgress for ml" ml
  36   -- The "scroll" expects "d/h/m/s/ms" time interval. Let's set it to "1 month"
  37   --eDocs <- ISTEX.getMetadataScroll q ((\_n -> pack $ "1m") <$> ml) Nothing 0  --(fromIntegral <$> ml)
  38   --eDocs <- ISTEX.getMetadataScroll (q <> " abstract:*")  "1m" Nothing 0  --(fromIntegral <$> ml)
  39   eDocs <- ISTEX.getMetadataWith q (Just 5000)
  40   printDebug "[Istex.get] will print length" (0 :: Int)
  41   case eDocs of
  42     Left _ -> pure ()
  43     Right (ISTEX.Documents { _documents_hits }) -> printDebug "[Istex.get] length docs" $ length _documents_hits
  44   --ISTEX.getMetadataScrollProgress q ((\_ -> pack $ "1m") <$> ml) Nothing progress errorHandler
  45   case eDocs of
  46     Left err -> panic . pack . show $ err
  47     Right docs -> toDoc' la docs
  48   --pure $ either (panic . pack . show) (toDoc' la) eDocs
  49 --  where
  50 --    progress (ISTEX.ScrollResponse { _scroll_documents = ISTEX.Documents { _documents_hits }}) =
  51 --      printDebug "[Istex.get] got docs: " $ length _documents_hits
  52 --    errorHandler err = printDebug "[Istex.get] error" $ show err
  53
  54 toDoc' :: Lang -> ISTEX.Documents -> IO [HyperdataDocument]
  55 toDoc' la docs' =  mapM (toDoc la) (ISTEX._documents_hits docs')
  56   --printDebug "ISTEX" (ISTEX._documents_total docs')
  57
  58 -- | TODO remove dateSplit here
  59 -- TODO current year as default
  60 toDoc :: Lang -> ISTEX.Document -> IO HyperdataDocument
  61 toDoc la (ISTEX.Document i t a ab d s) = do
  62   --printDebug "ISTEX date" d
  63   (utctime, (pub_year, pub_month, pub_day)) <-
  64         Date.dateSplit la (maybe (Just $ pack $ show Defaults.year) (Just . pack . show) d)
  65   --printDebug "toDoc Istex" (utctime, (pub_year, pub_month, pub_day))
  66   pure $ HyperdataDocument { _hd_bdd       = Just "Istex"
  67                            , _hd_doi       = Just i
  68                            , _hd_url       = Nothing
  69                            , _hd_uniqId    = Nothing
  70                            , _hd_uniqIdBdd = Nothing
  71                            , _hd_page      = Nothing
  72                            , _hd_title     = t
  73                            , _hd_authors = Just $ foldl (\x y -> x <> ", " <> y) "" (map ISTEX._author_name a)
  74                            , _hd_institutes = Just $ foldl (\x y -> x <> ", " <> y) "" (concat $ (map ISTEX._author_affiliations) a)
  75                            , _hd_source = Just $ foldl (\x y -> x <> ", " <> y) "" (catMaybes $ map ISTEX._source_title s)
  76                            , _hd_abstract = ab
  77                            , _hd_publication_date = fmap (pack . show) utctime
  78                            , _hd_publication_year = pub_year
  79                            , _hd_publication_month = pub_month
  80                            , _hd_publication_day = pub_day
  81                            , _hd_publication_hour = Nothing
  82                            , _hd_publication_minute = Nothing
  83                            , _hd_publication_second = Nothing
  84                            , _hd_language_iso2 = Just $ (pack . show) la
  85                            }