better diagnostic messages for BIO

src/GEval/BIO.hs

@@ -19,6 +19,13 @@ import GEval.Common
 data BIOLabel = Outside | Beginning T.Text (Maybe T.Text) | Inside T.Text (Maybe T.Text)
                 deriving (Eq, Show)
 
+formatBioLabel :: BIOLabel -> T.Text
+formatBioLabel Outside = "O"
+formatBioLabel (Beginning label Nothing) = T.concat ["B-", label]
+formatBioLabel (Beginning label (Just normalized)) = T.concat ["B-", label, "/", normalized]
+formatBioLabel (Inside label Nothing) = T.concat ["I-", label]
+formatBioLabel (Inside label (Just normalized)) = T.concat ["I-", label, "/", normalized]
+
 data TaggedSpan = TaggedSpan Int Int
                   deriving (Eq, Show)
 
@@ -45,7 +52,7 @@ labelSplitToEntity labs@(h@(_,begIx):t) = if isBeginning h && all (\tp -> isInsi
                                      then
                                        Right $ TaggedEntity (TaggedSpan begIx lastItemIx) btp mNormalized
                                      else
-                                       Left "something wrong with label sequence"
+                                       Left $ "inconsistent label sequence `" ++ (T.unpack $ T.intercalate " " $ map (formatBioLabel . fst) labs)  ++ "`"
   where isBeginning (Beginning _ _, _) = True
         isBeginning _ = False
         isInside (Inside _ _, _) = True

test/Spec.hs

@@ -258,6 +258,8 @@ main = hspec $ do
       runGEvalTest "bio-f1-simple" `shouldReturnAlmost` 0.5
     it "check perfect score" $ do
       runGEvalTest "bio-f1-perfect" `shouldReturnAlmost` 1.0
+    it "check inconsistent input" $ do
+      runGEvalTest "bio-f1-error" `shouldThrow` (== UnexpectedData 2 "inconsistent label sequence `B-NAME/JOHN I-FOO/SMITH I-FOO/X`")
   describe "automatic decompression" $ do
     it "more complex test" $ do
       runGEvalTest "charmatch-complex-compressed" `shouldReturnAlmost` 0.1923076923076923

test/bio-f1-error/bio-f1-error-solution/test-A/out.tsv

@@ -0,0 +1,2 @@
+O B-CITY/WARSZAWA I-CITY/WARSZAWA
+O B-NAME/JOHN I-FOO/SMITH I-FOO/X O

test/bio-f1-error/bio-f1-error/config.txt

@@ -0,0 +1 @@
+--metric BIO-F1

test/bio-f1-error/bio-f1-error/test-A/expected.tsv

@@ -0,0 +1,2 @@
+O B-CITY/WARSZAWA I-CITY/WARSZAWA
+O B-NAME/JOHN I-NAME/SMITH O O