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 qualified Data.Text as Text
  22 import qualified Data.List as List
  23 import Gargantext.Core (Lang(..))
  24 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
  25 import qualified Gargantext.Defaults as Defaults
  26 import Gargantext.Prelude
  27 import qualified Gargantext.Core.Text.Corpus.Parsers.Date as Date
  28 import qualified ISTEX        as ISTEX
  29 import qualified ISTEX.Client as ISTEX
  30
  31 type Query = Text
  32 type MaxResults = Maybe Integer
  33
  34 get :: Lang -> Query -> MaxResults -> IO [HyperdataDocument]
  35 get la query' maxResults = do
  36   --printDebug "[Istex.get] calling getMetadataScrollProgress for la" la
  37   --printDebug "[Istex.get] calling getMetadataScrollProgress for q" q
  38   --printDebug "[Istex.get] calling getMetadataScrollProgress for ml" ml
  39   -- The "scroll" expects "d/h/m/s/ms" time interval. Let's set it to "1 month"
  40   --eDocs <- ISTEX.getMetadataScroll q ((\_n -> pack $ "1m") <$> ml) Nothing 0  --(fromIntegral <$> ml)
  41
  42   -- TODO check if abstract is in query already if not add like below
  43   -- eDocs <- ISTEX.getMetadataScroll (q <> " abstract:*")  "1m" Nothing 0  --(fromIntegral <$> ml)
  44   -- eDocs <- ISTEX.getMetadataScroll q "1m" Nothing 0  --(fromIntegral <$> ml)
  45
  46   let query = case (List.length $ Text.splitOn ":" query') == 1 of
  47         -- True case means users is entering default search of IsTex
  48         -- In that case we need to enrich his query with 2 parameters
  49         -- First expected language: user has to define it in GTXT
  50         -- Second : query in abstract
  51         True  -> ("language:"<> lang la) <> " AND abstract:"<>query'
  52             where
  53               lang FR = "fre"
  54               lang _  = "eng"
  55
  56         False -> query'
  57         -- Complex queries of IsTex needs parameters using ":" so we leave the query as it is
  58         -- in that case we suppose user is knowing what s.he is doing
  59
  60   eDocs <- ISTEX.getMetadataWith query (fromIntegral <$> maxResults)
  61   -- printDebug "[Istex.get] will print length" (0 :: Int)
  62   case eDocs of
  63     Left _ -> pure ()
  64     Right (ISTEX.Documents { _documents_hits }) -> printDebug "[Istex.get] length docs" $ length _documents_hits
  65   --ISTEX.getMetadataScrollProgress q ((\_ -> pack $ "1m") <$> ml) Nothing progress errorHandler
  66   case eDocs of
  67     Left err -> panic . Text.pack . show $ err
  68     Right docs -> toDoc' la docs
  69   --pure $ either (panic . pack . show) (toDoc' la) eDocs
  70 --  where
  71 --    progress (ISTEX.ScrollResponse { _scroll_documents = ISTEX.Documents { _documents_hits }}) =
  72 --      printDebug "[Istex.get] got docs: " $ length _documents_hits
  73 --    errorHandler err = printDebug "[Istex.get] error" $ show err
  74
  75 toDoc' :: Lang -> ISTEX.Documents -> IO [HyperdataDocument]
  76 toDoc' la docs' =  mapM (toDoc la) (ISTEX._documents_hits docs')
  77   --printDebug "ISTEX" (ISTEX._documents_total docs')
  78
  79 -- | TODO remove dateSplit here
  80 -- TODO current year as default
  81 toDoc :: Lang -> ISTEX.Document -> IO HyperdataDocument
  82 toDoc la (ISTEX.Document i t a ab d s) = do
  83   --printDebug "ISTEX date" d
  84   (utctime, (pub_year, pub_month, pub_day)) <-
  85         Date.dateSplit la (maybe (Just $ pack $ show Defaults.year) (Just . pack . show) d)
  86   --printDebug "toDoc Istex" (utctime, (pub_year, pub_month, pub_day))
  87   pure $ HyperdataDocument { _hd_bdd       = Just "Istex"
  88                            , _hd_doi       = Just i
  89                            , _hd_url       = Nothing
  90                            , _hd_uniqId    = Nothing
  91                            , _hd_uniqIdBdd = Nothing
  92                            , _hd_page      = Nothing
  93                            , _hd_title     = t
  94                            , _hd_authors = Just $ foldl (\x y -> x <> ", " <> y) "" (map ISTEX._author_name a)
  95                            , _hd_institutes = Just $ foldl (\x y -> x <> ", " <> y) "" (concat $ (map ISTEX._author_affiliations) a)
  96                            , _hd_source = Just $ foldl (\x y -> x <> ", " <> y) "" (catMaybes $ map ISTEX._source_title s)
  97                            , _hd_abstract = ab
  98                            , _hd_publication_date = fmap (pack . show) utctime
  99                            , _hd_publication_year = pub_year
 100                            , _hd_publication_month = pub_month
 101                            , _hd_publication_day = pub_day
 102                            , _hd_publication_hour = Nothing
 103                            , _hd_publication_minute = Nothing
 104                            , _hd_publication_second = Nothing
 105                            , _hd_language_iso2 = Just $ (pack . show) la
 106                            }