using structIsChildOf_fs instead of instStructIsChildOf_fs for better grouping

src/HAL.hs

@@ -232,7 +232,7 @@ baseFields = [ "docid"
              , "authFullName_s"
              , "authOrganism_s"
              , "structIdName_fs"
-             , "instStructIsChildOf_fs" ]
+             , "structIsChildOf_fs" ]
 
 structFields :: Text
 structFields = "docid,label_s,parentDocid_i"

src/HAL/Doc/Document.hs

@@ -26,7 +26,7 @@ data Document = Document
   , _document_source               :: Maybe Text
   , _document_authors_names        :: [Text]
   , _document_authors_affiliations :: [Text]
-  , _document_institutes_tree      :: Map Text Text
+  , _document_institutes_tree      :: Map Text [Text]
   , _document_struct_id            :: [Int]
   , _document_original             :: Object
   } deriving (Show, Generic)
@@ -51,9 +51,9 @@ instance FromJSON Document where
     let _document_struct_id = map fst structIdname
     let _document_authors_affiliations = map snd structIdname
 
-    structChild <- o .:? "instStructIsChildOf_fs" :: Parser (Maybe [Text])
+    structChild <- o .:? "structIsChildOf_fs" :: Parser (Maybe [Text])
 
-    let _document_institutes_tree = Map.fromList $ getStructParentChild structChild
+    let _document_institutes_tree = Map.fromListWith (++) $ getStructParentChild $ removeUnusedData $ fromMaybe [] structChild
 
     abstracts <-
       mapM (\lang -> do
@@ -65,6 +65,16 @@ instance FromJSON Document where
 
     pure $ Document { .. }
 
+-- | Remove one line out two because the field `structIsChildOf_fs` contains
+--   informations that is not used for now, the type of the structure (e.g. laboratory, institute, etc...)
+-- 
+--   Exemple of request : 
+--   https://api.archives-ouvertes.fr/search/?q=(en_title_t:(glass)%20OR%20en_abstract_t:(glass))%20AND%20(structId_i:6279)%20AND%20(language_s:en)&wt=json&fl=docid,label_s,uri_s,structId_i,title_s,structIsChildOf_fs&rows=80) 
+removeUnusedData :: [a] -> [a]
+removeUnusedData (_ : struct : structs) = struct : removeUnusedData structs
+removeUnusedData [_] = []
+removeUnusedData [] = []
+
 -- | Parses the field structIdName_fs that looks like : 
 -- > StructId_FacetSep_StructName
 -- 
@@ -76,8 +86,8 @@ getStructIdsNames (Just idsNames) = map (\tab -> (P.read (T.unpack (P.head tab))
     splitInstitutes = P.map (T.splitOn (T.pack "_FacetSep_"))
 getStructIdsNames Nothing = []
 
-getStructParentChild :: Maybe [T.Text] -> [(T.Text, T.Text)]
-getStructParentChild (Just struct) = map (\tab -> (formatParent $ P.last tab, formatChild $ P.head tab)) $ splitStruct struct
+getStructParentChild :: [T.Text] -> [(T.Text, [T.Text])]
+getStructParentChild struct = map (\tab -> (formatParent $ P.last tab, [formatChild $ P.head tab] )) $ splitStruct struct
   where
     splitStruct :: [T.Text] -> [[T.Text]]
     splitStruct = P.map (T.splitOn (T.pack "_JoinSep_"))
@@ -94,7 +104,5 @@ getStructParentChild (Just struct) = map (\tab -> (formatParent $ P.last tab, fo
         id = P.head splited
         splited = T.splitOn "_FacetSep_" p
 
-getStructParentChild Nothing = []
-
 instance ToHttpApiData Document where
   toUrlPiece _ = "docid,title_s,en_abstract_s,fr_abstract_s,submittedDate_s,source_s,authFullName_s,structId_i,structIdName_fs"