ENH skip lines without all fields

dhall/Types.dhall

@@ -351,10 +351,34 @@ let TxOpts_ =
             Data.Time.Format.formattime Haskell function.
             -}
             Text
-        , toSkipBlankDate : Bool
+        , toSkipBlankDate :
-        , toSkipBlankAmount : Bool
+            {-
-        , toSkipBlankDescription : Bool
+            Skip line if date field is a blank
-        , toSkipBlankOther : List Text
+            -}
+            Bool
+        , toSkipBlankAmount :
+            {-
+            Skip line if amount field(s) is(are) a blank
+            -}
+            Bool
+        , toSkipBlankDescription :
+            {-
+            Skip line if description field is a blank
+            -}
+            Bool
+        , toSkipBlankOther :
+            {-
+            Skip line if any arbitrary fields are blank (these fields must also
+            be listed in 'toOther' to be considered)
+            -}
+            List Text
+        , toSkipMissingFields :
+            {-
+            Skip line if any fields are missing (this is different from blank;
+            'missing' means there is no field with name 'X', 'blank' means that
+            there is a field 'X' and its value is an empty string)
+            -}
+            Bool
         }
 
 let TxAmountSpec = TxAmountSpec_ Text
@@ -371,6 +395,7 @@ let TxOpts =
         , toSkipBlankAmount = False
         , toSkipBlankDescription = False
         , toSkipBlankOther = [] : List Text
+        , toSkipMissingFields = False
         }
       }
 

lib/Internal/History.hs

@@ -124,31 +124,58 @@ parseTxRecord
     , toSkipBlankAmount
     , toSkipBlankDescription
     , toSkipBlankOther
+    , toSkipMissingFields
     }
-  r = do
+  r =
-    d <- r .: T.encodeUtf8 toDate
+    do
-    e <- r .: T.encodeUtf8 toDesc
+      -- TODO this is confusing as hell
-    os <- M.fromList <$> mapM (\n -> (n,) <$> r .: T.encodeUtf8 n) toOther
+      --
-    (af, ax) <- case toAmount of
+      -- try and parse all fields; if a parse fails, either trip an error
-      AmountSingle TxAmount1 {a1Column, a1Fmt} -> do
+      -- or return a Nothing if we want to deliberately skip missing fields
-        f <- r .: T.encodeUtf8 a1Column
+      d <- getField toDate
-        return (a1Fmt, Right f)
+      e <- getField toDesc
-      AmountDual TxAmount2 {a2Positive, a2Negative, a2Fmt} -> do
+      os <-
-        f1 <- r .: T.encodeUtf8 a2Positive
+        fmap M.fromList . sequence
-        f2 <- r .: T.encodeUtf8 a2Negative
+          <$> mapM (\n -> fmap (n,) <$> getField n) toOther
-        return (a2Fmt, Left (f1, f2))
+      (af, ax) <- case toAmount of
-    if (toSkipBlankDate && d == "")
+        -- the amount column is extra confusing because it can either be one
-      || (toSkipBlankDescription && e == "")
+        -- or two columns, so keep track of this with a maybe
-      || (toSkipBlankAmount && (ax == Right "" || ax == Left ("", "")))
+        AmountSingle TxAmount1 {a1Column, a1Fmt} -> do
-      || elem "" (mapMaybe (`M.lookup` os) toSkipBlankOther)
+          f <- getField a1Column
-      then return Nothing
+          return (a1Fmt, Right <$> f)
-      else do
+        AmountDual TxAmount2 {a2Positive, a2Negative, a2Fmt} -> do
-        a' <- case ax of
+          f1 <- getField a2Positive
-          Right a -> parseDecimal True af a
+          f2 <- getField a2Negative
-          Left ("", a) -> ((-1) *) <$> parseDecimal False af a
+          return $ (a2Fmt,) $ case (f1, f2) of
-          Left (a, _) -> parseDecimal False af a
+            (Just a, Just b) -> Just $ Left (a, b)
-        d' <- parseTimeM True defaultTimeLocale (T.unpack toDateFmt) d
+            _ -> Nothing
-        return $ Just $ TxRecord d' a' e os p
+      case (d, e, os, ax) of
+        -- If all lookups were successful, check that none of the fields are
+        -- blank, and if they are return nothing to skip this line
+        (Just d', Just e', Just os', Just ax') ->
+          if (toSkipBlankDate && d' == "")
+            || (toSkipBlankDescription && e' == "")
+            || (toSkipBlankAmount && (ax' == Right "" || ax' == Left ("", "")))
+            || elem "" (mapMaybe (`M.lookup` os') toSkipBlankOther)
+            then return Nothing
+            else -- if we are skipping nothing, proceed to parse the date and amount
+            -- columns
+            do
+              a <- case ax' of
+                Right a -> parseDecimal True af a
+                Left ("", a) -> ((-1) *) <$> parseDecimal False af a
+                Left (a, _) -> parseDecimal False af a
+              d'' <- parseTimeM True defaultTimeLocale (T.unpack toDateFmt) d'
+              return $ Just $ TxRecord d'' a e' os' p
+        -- if no lookups succeeded, return nothing to skip this line. Note that
+        -- a parse fail will trigger a failure error further up, so that case
+        -- is already dealt with implicitly
+        _ -> return Nothing
+    where
+      getField :: FromField a => T.Text -> Parser (Maybe a)
+      getField f = case runParser $ r .: T.encodeUtf8 f of
+        Left err -> if toSkipMissingFields then return Nothing else fail err
+        Right x -> return $ Just x
 
 matchRecords :: MonadFinance m => [StatementParserRe] -> [TxRecord] -> AppExceptT m [Tx ()]
 matchRecords ms rs = do

lib/Internal/Types/Dhall.hs

@@ -532,6 +532,7 @@ data TxOpts re = TxOpts
   , toSkipBlankAmount :: !Bool
   , toSkipBlankDescription :: !Bool
   , toSkipBlankOther :: ![Text]
+  , toSkipMissingFields :: !Bool
   }
   deriving (Functor, Foldable, Traversable)