Implement search with given language (for abstracts etc)

crawlerHAL.cabal

@@ -32,6 +32,7 @@ library
       HAL.Doc.Corpus
       HAL.Doc.EntityTree
       HAL.Doc.Struct
+      HAL.Utils
       Tree
   other-modules:
       Paths_crawlerHAL

src/HAL.hs

@@ -7,6 +7,7 @@ import Data.Text
 import HAL.Client
 import HAL.Doc.Corpus
 import HAL.Doc.Struct
+import HAL.Utils (langAbstractS)
 import Network.HTTP.Client (newManager)
 import Network.HTTP.Client.TLS (tlsManagerSettings)
 import Protolude
@@ -72,7 +73,7 @@ countResults q = do
 
 requestedFields :: Maybe ISO639_1 -> Text
 requestedFields (Just EN) = "docid,title_s,en_abstract_s,submittedDate_s,source_s,authFullName_s,authOrganism_s"
-requestedFields (Just FR) = "docid,title_s,en_abstract_s,fr_abstract_s,submittedDate_s,source_s,authFullName_s,authOrganism_s"
+requestedFields (Just lang) = "docid,title_s,en_abstract_s," <> langAbstractS lang <> ",submittedDate_s,source_s,authFullName_s,authOrganism_s"
 requestedFields _ = requestedFields (Just EN)
 
 structFields :: Text

src/HAL/Doc/Corpus.hs

@@ -4,8 +4,12 @@ module HAL.Doc.Corpus where
 
 import Control.Lens qualified as L
 import Data.Aeson
+import Data.Aeson.Key (fromText)
 import Data.Default
+import Data.Map.Strict qualified as Map
 import GHC.Generics
+import HAL.Utils (allLangs, langAbstractS)
+import Data.LanguageCodes (ISO639_1(..))
 import Protolude
 import Servant.API (ToHttpApiData(..))
 
@@ -13,6 +17,7 @@ data Corpus = Corpus
   { _corpus_docid                :: Text
   , _corpus_title                :: [Text]
   , _corpus_abstract             :: [Text]
+  , _corpus_abstract_lang_map    :: Map ISO639_1 [Text]
   , _corpus_date                 :: Maybe Text
   , _corpus_source               :: Maybe Text
   , _corpus_authors_names        :: [Text]
@@ -22,19 +27,26 @@ data Corpus = Corpus
 L.makeLenses ''Corpus
 
 instance Default Corpus where
-  def = Corpus "default Id" def def def def def def def
+  def = Corpus "default Id" def def def def def def def def
 
 instance FromJSON Corpus where
-  parseJSON = withObject "Corpus" $
-    \o -> Corpus
-            <$> (o .: "docid")
-            <*> (o .: "title_s" <|> return [])
-            <*> (o .: "en_abstract_s" <|> return [])
-            <*> (o .:? "submittedDate_s")
-            <*> (o .:? "source_s")
-            <*> (o .: "authFullName_s" <|> return [])
-            <*> (o .: "authOrganism_s" <|> return [])
-            <*> (o .: "structId_i" <|> return [])
+  parseJSON = withObject "Corpus" $ \o -> do
+    _corpus_docid                <- (o .: "docid")
+    _corpus_title                <- (o .: "title_s" <|> return [])
+    _corpus_abstract             <- (o .: "en_abstract_s" <|> return [])
+    _corpus_date                 <- (o .:? "submittedDate_s")
+    _corpus_source               <- (o .:? "source_s")
+    _corpus_authors_names        <- (o .: "authFullName_s" <|> return [])
+    _corpus_authors_affiliations <- (o .: "authOrganism_s" <|> return [])
+    _corpus_struct_id            <- (o .: "structId_i" <|> return [])
+
+    abstracts <-
+      mapM (\lang -> do
+               ma <- o .:? (fromText $ langAbstractS lang)
+               pure $ (\a -> (lang, a)) <$> ma) allLangs
+    let _corpus_abstract_lang_map = Map.fromList $ catMaybes abstracts
+
+    pure $ Corpus { .. }
 
 instance ToHttpApiData Corpus where
   toUrlPiece _ = "docid,title_s,en_abstract_s,fr_abstract_s,submittedDate_s,source_s,authFullName_s,authOrganism_s,structId_i"

src/HAL/Utils.hs

+module HAL.Utils where
+
+import Data.LanguageCodes (ISO639_1(..), language)
+import Data.Text qualified as T
+import Protolude
+
+allLangs :: [ISO639_1]
+allLangs = enumFrom (toEnum 0) :: [ISO639_1]
+
+langAbstractS :: ISO639_1 -> Text
+langAbstractS lang = (T.pack $ language lang) <> "_abstract_s"

src/Tree.hs

@@ -2,9 +2,9 @@ module Tree where
 
 import Control.Lens.Getter ((^.))
 import Data.List.Split (chunksOf)
-import Data.Map qualified as Map
-import Data.Map (insert)
-import Data.Map.Internal (merge, preserveMissing, zipWithMatched)
+import Data.Map.Strict qualified as Map
+import Data.Map.Strict (insert)
+import Data.Map.Strict.Internal (merge, preserveMissing, zipWithMatched)
 import Data.Text qualified as T
 import Data.Text.Format (format)
 import Data.Text.Lazy qualified as TL