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   --printDebug "[Istex.get] calling getMetadataScrollProgress for la" la
  33   --printDebug "[Istex.get] calling getMetadataScrollProgress for q" q
  34   --printDebug "[Istex.get] calling getMetadataScrollProgress for ml" ml
  35   -- The "scroll" expects "d/h/m/s/ms" time interval. Let's set it to "1 month"
  36   --eDocs <- ISTEX.getMetadataScroll q ((\_n -> pack $ "1m") <$> ml) Nothing 0  --(fromIntegral <$> ml)
  37
  38   -- TODO check if abstract is in query already if not add like below
  39   -- eDocs <- ISTEX.getMetadataScroll (q <> " abstract:*")  "1m" Nothing 0  --(fromIntegral <$> ml)
  40   -- eDocs <- ISTEX.getMetadataScroll q "1m" Nothing 0  --(fromIntegral <$> ml)
  41   eDocs <- ISTEX.getMetadataWith q (Just 5000)
  42   printDebug "[Istex.get] will print length" (0 :: Int)
  43   case eDocs of
  44     Left _ -> pure ()
  45     Right (ISTEX.Documents { _documents_hits }) -> printDebug "[Istex.get] length docs" $ length _documents_hits
  46   --ISTEX.getMetadataScrollProgress q ((\_ -> pack $ "1m") <$> ml) Nothing progress errorHandler
  47   case eDocs of
  48     Left err -> panic . pack . show $ err
  49     Right docs -> toDoc' la docs
  50   --pure $ either (panic . pack . show) (toDoc' la) eDocs
  51 --  where
  52 --    progress (ISTEX.ScrollResponse { _scroll_documents = ISTEX.Documents { _documents_hits }}) =
  53 --      printDebug "[Istex.get] got docs: " $ length _documents_hits
  54 --    errorHandler err = printDebug "[Istex.get] error" $ show err
  55
  56 toDoc' :: Lang -> ISTEX.Documents -> IO [HyperdataDocument]
  57 toDoc' la docs' =  mapM (toDoc la) (ISTEX._documents_hits docs')
  58   --printDebug "ISTEX" (ISTEX._documents_total docs')
  59
  60 -- | TODO remove dateSplit here
  61 -- TODO current year as default
  62 toDoc :: Lang -> ISTEX.Document -> IO HyperdataDocument
  63 toDoc la (ISTEX.Document i t a ab d s) = do
  64   --printDebug "ISTEX date" d
  65   (utctime, (pub_year, pub_month, pub_day)) <-
  66         Date.dateSplit la (maybe (Just $ pack $ show Defaults.year) (Just . pack . show) d)
  67   --printDebug "toDoc Istex" (utctime, (pub_year, pub_month, pub_day))
  68   pure $ HyperdataDocument { _hd_bdd       = Just "Istex"
  69                            , _hd_doi       = Just i
  70                            , _hd_url       = Nothing
  71                            , _hd_uniqId    = Nothing
  72                            , _hd_uniqIdBdd = Nothing
  73                            , _hd_page      = Nothing
  74                            , _hd_title     = t
  75                            , _hd_authors = Just $ foldl (\x y -> x <> ", " <> y) "" (map ISTEX._author_name a)
  76                            , _hd_institutes = Just $ foldl (\x y -> x <> ", " <> y) "" (concat $ (map ISTEX._author_affiliations) a)
  77                            , _hd_source = Just $ foldl (\x y -> x <> ", " <> y) "" (catMaybes $ map ISTEX._source_title s)
  78                            , _hd_abstract = ab
  79                            , _hd_publication_date = fmap (pack . show) utctime
  80                            , _hd_publication_year = pub_year
  81                            , _hd_publication_month = pub_month
  82                            , _hd_publication_day = pub_day
  83                            , _hd_publication_hour = Nothing
  84                            , _hd_publication_minute = Nothing
  85                            , _hd_publication_second = Nothing
  86                            , _hd_language_iso2 = Just $ (pack . show) la
  87                            }