From b2e3293a1243a334b4ed41a2357516fab538474d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Filip=20Grali=C5=84ski?= <filipg@amu.edu.pl>
Date: Thu, 14 Feb 2019 16:25:28 +0100
Subject: [PATCH] Refactor line-by-line mode

---
 src/GEval/Core.hs       | 17 +++++++++++++++--
 src/GEval/LineByLine.hs |  6 +++---
 2 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/src/GEval/Core.hs b/src/GEval/Core.hs
index 9fc0a55..0451514 100644
--- a/src/GEval/Core.hs
+++ b/src/GEval/Core.hs
@@ -39,7 +39,8 @@ module GEval.Core
       checkMultipleOuts,
       checkMultipleOutsCore,
       gesMainMetric,
-      gesPreprocess
+      gesPreprocess,
+      threeLineSource
     ) where
 
 import Data.Conduit
@@ -898,7 +899,11 @@ instance (MonadUnliftIO m, MonadIO m, MonadThrow m) => EvaluationContext (WithIn
   checkStepM _ (_, WrappedParsedRecordWithInput Done Done Done) = return Nothing
 
 
-
+threeLineSource :: (MonadUnliftIO m, MonadIO m, MonadThrow m) => WithInput m Text Text Text -> ConduitT () (WrappedParsedRecord (WithInput m Text Text Text)) (ResourceT m) ()
+threeLineSource (WithInput inputLineSource expectedLineSource outLineSource) = getZipSource $ (\x (y,z) -> WrappedParsedRecordWithInput x y z)
+         <$> ZipSource (linesAsItems inputLineSource)                                             <*> (ZipSource $ getZipSource $ (,)
+                        <$> ZipSource (linesAsItems expectedLineSource)
+                        <*> ZipSource (linesAsItems outLineSource))
 
 averageC :: MonadResource m => ConduitT Double Void m Double
 averageC = getZipSink
@@ -906,12 +911,20 @@ averageC = getZipSink
   <$> ZipSink CC.sum
   <*> ZipSink CC.length
 
+-- | Takes a source of lines and returns a source of lines and returns a conduit of
+-- items (using a given preprocessor and parser).
 items :: MonadResource m => LineSource m -> (Text -> Either String a) -> ConduitT () (SourceItem a) m ()
 items (LineSource lineSource preprocess _ _) parser =
   (lineSource .| CL.map (toItem . parser . preprocess)) >> yield Done
   where toItem (Right x) = Got x
         toItem (Left m) = Wrong m
 
+-- | Takes a source of lines and returns a conduit of lines represented as
+-- items (without preprocessing and parsing!) to be used in line-by-line modes.
+linesAsItems :: MonadResource m => LineSource m -> ConduitT () (SourceItem Text) m ()
+linesAsItems (LineSource lineSource _ _ _) =
+  (lineSource .| CL.map Got) >> yield Done
+
 itemAbsoluteError :: (Double, Double) -> Double
 itemAbsoluteError (exp, out) = abs (exp-out)
 
diff --git a/src/GEval/LineByLine.hs b/src/GEval/LineByLine.hs
index e63419e..54172a0 100644
--- a/src/GEval/LineByLine.hs
+++ b/src/GEval/LineByLine.hs
@@ -417,9 +417,9 @@ gevalLineByLineSource :: Metric -> (Text -> Text) -> SourceSpec -> SourceSpec ->
 gevalLineByLineSource metric preprocess inputSource expectedSource outSource =
   (getZipSource $ (,)
        <$> ZipSource (CL.sourceList [1..])
-       <*> (ZipSource $ recordSource context parserSpec)) .| CL.mapM (checkStepM evaluateLine) .| CL.catMaybes
-  where parserSpec = (ParserSpecWithInput (Right . id) (Right . id) (Right . id))
-        context = (WithInput inputLineSource expectedLineSource outputLineSource)
+       <*> (ZipSource $ threeLineSource context)) .| CL.mapM (checkStepM evaluateLine) .| CL.catMaybes
+  where context = (WithInput inputLineSource expectedLineSource outputLineSource)
+        -- preparing sources, `id` means that no preprocessing is done (to avoid double preprocessing)
         inputLineSource = fileAsLineSource inputSource id
         expectedLineSource = fileAsLineSource expectedSource id
         outputLineSource = fileAsLineSource outSource id