Reorganize TSV parsing code

bin/gargantext-cli/CLI/FilterTermsAndCooc.hs

@@ -20,7 +20,7 @@ import Data.Tuple.Extra (both)
 import Data.Vector qualified as DV
 import GHC.Generics
 import Gargantext.Core.Text.Context (TermList)
-import Gargantext.Core.Text.Corpus.Parsers.TSV (readTSVFile, tsv_title, tsv_abstract, tsv_publication_year, fromMIntOrDec, defaultYear)
+import Gargantext.Core.Text.Corpus.Parsers.TSV (readTSVFile, unIntOrDec, tsv_title, tsv_abstract, tsv_publication_year, defaultYear)
 import Gargantext.Core.Text.List.Formats.TSV (tsvMapTermList)
 import Gargantext.Core.Text.Metrics.Count (coocOnContexts, Coocs)
 import Gargantext.Core.Text.Terms.WithList ( Patterns, buildPatterns, extractTermsWithList )
@@ -52,7 +52,7 @@ filterTermsAndCoocCLI (CorpusFile corpusFile) (TermListFile termListFile) (Outpu
     Right cf -> do
       let corpus = DM.fromListWith (<>)
                    . DV.toList
-                   . DV.map (\n -> (fromMIntOrDec defaultYear $ tsv_publication_year n, [(tsv_title n) <> " " <> (tsv_abstract n)]))
+                   . DV.map (\n -> (maybe defaultYear unIntOrDec $ tsv_publication_year n, [(tsv_title n) <> " " <> (tsv_abstract n)]))
                    . snd $ cf
 
       -- termListMap :: [Text]

bin/gargantext-cli/CLI/Phylo/Common.hs

@@ -82,8 +82,8 @@ tsvToDocs parser patterns time path =
     Wos  _     -> Prelude.error "tsvToDocs: unimplemented"
     Tsv  limit -> Vector.toList
       <$> Vector.take limit
-      <$> Vector.map (\row -> Document (toPhyloDate  (Tsv.fromMIntOrDec Tsv.defaultYear $ tsv_publication_year row) (fromMaybe Tsv.defaultMonth $ tsv_publication_month row) (fromMaybe Tsv.defaultDay $ tsv_publication_day row) time)
-                                       (toPhyloDate' (Tsv.fromMIntOrDec Tsv.defaultYear $ tsv_publication_year row) (fromMaybe Tsv.defaultMonth $ tsv_publication_month row) (fromMaybe Tsv.defaultDay $ tsv_publication_day row) time)
+      <$> Vector.map (\row -> Document (toPhyloDate  (maybe Tsv.defaultYear Tsv.unIntOrDec $ tsv_publication_year row) (fromMaybe Tsv.defaultMonth $ tsv_publication_month row) (fromMaybe Tsv.defaultDay $ tsv_publication_day row) time)
+                                       (toPhyloDate' (maybe Tsv.defaultYear Tsv.unIntOrDec $ tsv_publication_year row) (fromMaybe Tsv.defaultMonth $ tsv_publication_month row) (fromMaybe Tsv.defaultDay $ tsv_publication_day row) time)
                                        (termsInText patterns $ (tsv_title row) <> " " <> (tsv_abstract row))
                                        Nothing
                                        []

src/Gargantext/Core/Ext/IMTUser.hs

@@ -22,7 +22,7 @@ import Data.Csv ( (.:), header, decodeByNameWith, FromNamedRecord(..), Header )
 import Data.Text qualified as T
 import Data.Vector (Vector)
 import Data.Vector qualified as Vector
-import Gargantext.Core.Text.Corpus.Parsers.TSV ( tsvDecodeOptions, ColumnDelimiter(Tab) )
+import Gargantext.Core.Text.Corpus.Parsers.TSV (defaultDecodingOptionsWithDelimiter, ColumnDelimiter(Tab) )
 import Gargantext.Database.Admin.Types.Hyperdata.Contact
 import Gargantext.Prelude
 import System.FilePath.Posix (takeExtension)
@@ -119,7 +119,7 @@ readTSVFile_Annuaire' :: FilePath -> IO (Header, Vector IMTUser)
 readTSVFile_Annuaire' = fmap readTsvHalLazyBS' . BL.readFile
   where
     readTsvHalLazyBS' :: BL.ByteString -> (Header, Vector IMTUser)
-    readTsvHalLazyBS' bs = case decodeByNameWith (tsvDecodeOptions Tab) bs of
+    readTsvHalLazyBS' bs = case decodeByNameWith (defaultDecodingOptionsWithDelimiter Tab) bs of
           Left  e    -> panicTrace (cs e)
           Right rows -> rows