[FIX] IsTex basic working for simple queries without quotes

gargantext.cabal

@@ -5,7 +5,7 @@ cabal-version: 1.12
 -- see: https://github.com/sol/hpack
 
 name:           gargantext
-version: 0.0.5.8.9.5
+version:        0.0.5.8.9.5
 synopsis:       Search, map, share
 description:    Please see README.md
 category:       Data

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

@@ -26,6 +26,7 @@ import qualified Gargantext.Core.Text.Corpus.Parsers.Date as Date
 import qualified ISTEX        as ISTEX
 import qualified ISTEX.Client as ISTEX
 
+
 get :: Lang -> Text -> Maybe Integer -> IO [HyperdataDocument]
 get la q _ml = do
   --docs <- ISTEX.getMetadataWith q (fromIntegral <$> ml)
@@ -34,7 +35,7 @@ get la q _ml = do
   --printDebug "[Istex.get] calling getMetadataScrollProgress for ml" ml
   -- The "scroll" expects "d/h/m/s/ms" time interval. Let's set it to "1 month"
   --eDocs <- ISTEX.getMetadataScroll q ((\_n -> pack $ "1m") <$> ml) Nothing 0  --(fromIntegral <$> ml)
-  eDocs <- ISTEX.getMetadataScroll q "1m" Nothing 0  --(fromIntegral <$> ml)
+  eDocs <- ISTEX.getMetadataScroll (q <> " abstract:*")  "1m" Nothing 0  --(fromIntegral <$> ml)
   printDebug "[Istex.get] will print length" (0 :: Int)
   case eDocs of
     Left _ -> pure ()
@@ -57,15 +58,17 @@ toDoc' la docs' =  mapM (toDoc la) (ISTEX._documents_hits docs')
 -- TODO current year as default
 toDoc :: Lang -> ISTEX.Document -> IO HyperdataDocument
 toDoc la (ISTEX.Document i t a ab d s) = do
+  --printDebug "ISTEX date" d
   (utctime, (pub_year, pub_month, pub_day)) <-
         Date.dateSplit la (maybe (Just $ pack $ show Defaults.year) (Just . pack . show) d)
-  pure $ HyperdataDocument { _hd_bdd = Just "Istex"
-                           , _hd_doi = Just i
-                           , _hd_url = Nothing
-                           , _hd_uniqId = Nothing
+  --printDebug "toDoc Istex" (utctime, (pub_year, pub_month, pub_day))
+  pure $ HyperdataDocument { _hd_bdd       = Just "Istex"
+                           , _hd_doi       = Just i
+                           , _hd_url       = Nothing
+                           , _hd_uniqId    = Nothing
                            , _hd_uniqIdBdd = Nothing
-                           , _hd_page = Nothing
-                           , _hd_title = t
+                           , _hd_page      = Nothing
+                           , _hd_title     = t
                            , _hd_authors = Just $ foldl (\x y -> x <> ", " <> y) "" (map ISTEX._author_name a)
                            , _hd_institutes = Just $ foldl (\x y -> x <> ", " <> y) "" (concat $ (map ISTEX._author_affiliations) a)
                            , _hd_source = Just $ foldl (\x y -> x <> ", " <> y) "" (catMaybes $ map ISTEX._source_title s)
@@ -77,5 +80,5 @@ toDoc la (ISTEX.Document i t a ab d s) = do
                            , _hd_publication_hour = Nothing
                            , _hd_publication_minute = Nothing
                            , _hd_publication_second = Nothing
-                           , _hd_language_iso2 = Just $ (pack . show) la }
-                         
+                           , _hd_language_iso2 = Just $ (pack . show) la
+                           }

src/Gargantext/Core/Text/Corpus/Parsers/Date.hs

@@ -67,14 +67,15 @@ type Day   = Int
 
 -- | Date Parser
 -- Parses dates mentions in full text given the language.
--- >>> parse FR (pack "10 avril 1900 à 19H")
--- 1900-04-10 19:00:00 UTC
--- >>> parse EN (pack "April 10 1900")
--- 1900-04-10 00:00:00 UTC
+-- >>> parse FR (pack "1 avril 1900 à 19H")
+-- 1900-04-01 19:00:00 UTC
+-- >>> parse EN (pack "April 1 1900")
+-- 1900-04-01 00:00:00 UTC
 parse :: Lang -> Text -> IO UTCTime
 parse lang s = do
-  printDebug "Date: " s
+  --printDebug "Date: " s
   dateStr' <- pure $ dateFlow (DucklingFailure s) -- parseRawSafe lang s
+  --printDebug "Date': " dateStr'
   case dateFlow dateStr' of
     DateFlowSuccess ok -> pure ok
     _                  -> withDebugMode (DebugMode True)
@@ -93,7 +94,7 @@ data DateFlow = DucklingSuccess { ds_result  :: Text }
               | DucklingFailure { df_result  :: Text }
               | ReadFailure1    { rf1_result :: Text }
               | ReadFailure2    { rf2_result :: Text }
-              | DateFlowSuccess  { success :: UTCTime }
+              | DateFlowSuccess { success :: UTCTime }
               | DateFlowFailure
   deriving Show
 
@@ -126,7 +127,7 @@ readDate txt = do
 -- | To get Homogeinity of the languages
 --   TODO : put this in a more generic place in the source code
 parserLang :: Lang -> DC.Lang
-parserLang FR = DC.FR
+parserLang FR    = DC.FR
 parserLang EN    = DC.EN
 parserLang lang  = panic $ "[G.C.T.C.P.Date] Lang not implemented" <> (cs $ show lang)
 

stack.yaml

@@ -73,7 +73,7 @@ extra-deps:
 - git: https://gitlab.iscpif.fr/gargantext/crawlers/pubmed.git
   commit: 02e03d9b856bd35d391f43da8525330f9d184615
 - git: https://gitlab.iscpif.fr/gargantext/crawlers/istex.git
-  commit: 6821f41655e298fa1fac5021c5776a6aed2d18fd
+  commit: a34bb341236d82cf3d488210bc1d8448a98f5808
 - git: https://gitlab.iscpif.fr/gargantext/crawlers/hal.git
   commit: 9a43470241690a19c1c381c42a62c5dd4e28dff2
 - git: https://gitlab.iscpif.fr/gargantext/crawlers/isidore.git