Proper exception on hanging doublequote (PR #222)

Squashed from commits:

- task: add test cases for decoding escaped input

  Test case `escapedMalformed0` is failing, throwing uncaught `error`

- fix: handle case when input for field parser is a single opening double-quote

- Fix output of escapedMalformed1 test
  (this contribution got deleted)

- Fixup PR #222: remove failing test, drop testing for GHC < 8.4

Co-authored-by: German Zhyvotnikov <[email protected]>

.github/workflows/haskell-ci.yml

@@ -123,7 +123,7 @@ jobs:
           echo "CABAL=$HOME/.ghcup/bin/cabal-3.12.1.0 -vnormal+nowrap" >> "$GITHUB_ENV"
           HCNUMVER=$(${HC} --numeric-version|perl -ne '/^(\d+)\.(\d+)\.(\d+)(\.(\d+))?$/; print(10000 * $1 + 100 * $2 + ($3 == 0 ? $5 != 1 : $3))')
           echo "HCNUMVER=$HCNUMVER" >> "$GITHUB_ENV"
-          echo "ARG_TESTS=--enable-tests" >> "$GITHUB_ENV"
+          if [ $((HCNUMVER >= 80400)) -ne 0 ] ; then echo "ARG_TESTS=--enable-tests" >> "$GITHUB_ENV" ; else echo "ARG_TESTS=--disable-tests" >> "$GITHUB_ENV" ; fi
           echo "ARG_BENCH=--enable-benchmarks" >> "$GITHUB_ENV"
           echo "HEADHACKAGE=false" >> "$GITHUB_ENV"
           echo "ARG_COMPILER=--$HCKIND --with-compiler=$HC" >> "$GITHUB_ENV"
@@ -230,7 +230,7 @@ jobs:
           $CABAL v2-build $ARG_COMPILER $ARG_TESTS $ARG_BENCH all --write-ghc-environment-files=always
       - name: tests
         run: |
-          $CABAL v2-test $ARG_COMPILER $ARG_TESTS $ARG_BENCH all --test-show-details=direct
+          if [ $((HCNUMVER >= 80400)) -ne 0 ] ; then $CABAL v2-test $ARG_COMPILER $ARG_TESTS $ARG_BENCH all --test-show-details=direct ; fi
       - name: cabal check
         run: |
           cd ${PKGDIR_cassava} || false

CHANGES.md

@@ -1,7 +1,9 @@
 ## Version 0.5.3.2
 
+ * Proper exception on hanging doublequote ([PR #222](https://github.com/haskell-hvr/cassava/pull/222)).
  * Allow latest `hashable`.
- * Tested with GHC 8.0 - 9.10.1.
+ * Build tested with GHC 8.0 - 9.10.1.
+ * Functionality tested with GHC 8.4 - 9.10.1.
 
 ## Version 0.5.3.1
 

cabal.haskell-ci

@@ -1,5 +1,7 @@
 branches: master
 
+tests: >= 8.4
+
 -- constraint-set containers-0.7
 --   ghc: >=8.2
 --   constraints: containers ^>=0.7

cassava.cabal

@@ -136,7 +136,7 @@ Test-suite unit-tests
   Main-is: UnitTests.hs
   -- dependencies with version constraints inherited via lib:cassava
   Build-depends: attoparsec
-               , base
+               , base                   >= 4.11 && < 5
                , bytestring
                , cassava
                , hashable
@@ -158,12 +158,10 @@ Test-suite unit-tests
     -Wall
     -- https://ghc.haskell.org/trac/ghc/wiki/Migration/8.0#Recommendationsforforward-compatibility
     -Wcompat
+    -Wcpp-undef
     -Wnoncanonical-monad-instances
 
   if impl(ghc >= 8.8)
     ghc-options: -Wno-star-is-type
   else
     ghc-options: -Wnoncanonical-monadfail-instances
-
-  if impl(ghc >= 8.2)
-    ghc-options: -Wcpp-undef

src/Data/Csv/Parser.hs

@@ -149,17 +149,20 @@ field !delim = do
 escapedField :: AL.Parser S.ByteString
 escapedField = do
     _ <- dquote
-    -- The scan state is 'True' if the previous character was a double
-    -- quote.  We need to drop a trailing double quote left by scan.
-    s <- S.init <$> (A.scan False $ \s c -> if c == doubleQuote
-                                            then Just (not s)
-                                            else if s then Nothing
-                                                 else Just False)
-    if doubleQuote `S.elem` s
-        then case Z.parse unescape s of
-            Right r  -> return r
-            Left err -> fail err
-        else return s
+    -- The scan state is 'True' if the previous character was a double quote.
+    s' <- A.scan False $ \s c -> if c == doubleQuote
+                                 then Just (not s)
+                                 else if s then Nothing
+                                      else Just False
+    -- We need to drop a trailing double quote left by scan.
+    if S.null s'
+      then fail "trailing double quote"
+      else let s = S.init s'
+           in if doubleQuote `S.elem` s
+              then case Z.parse unescape s of
+                     Right r  -> return r
+                     Left err -> fail err
+              else return s
 
 unescapedField :: Word8 -> AL.Parser S.ByteString
 unescapedField !delim = A.takeWhile (\ c -> c /= doubleQuote &&

tests/UnitTests.hs

@@ -17,6 +17,7 @@ import qualified Data.Text as T
 import qualified Data.Text.Lazy as LT
 import qualified Data.Vector as V
 import qualified Data.Foldable as F
+import Data.List (isPrefixOf)
 import Data.Word
 import Numeric.Natural
 import GHC.Generics (Generic)
@@ -48,6 +49,14 @@ assertResult input expected res = case res of
                 "      input: " ++ show (BL8.unpack input) ++ "\n" ++
                 "parse error: " ++ err
 
+decodeFailsWith :: BL.ByteString -> String -> Assertion
+decodeFailsWith input expected = case decode NoHeader input of
+    Right r  -> assertFailure $ "input: " ++ show (BL8.unpack input) ++ "\n" ++
+                "retuned: " ++ show (r :: (V.Vector (V.Vector B.ByteString))) ++ "\n" ++
+                "whereas should have failed with " <> expected
+    Left err -> assertBool ("got " <> err <> "\ninstead of " <> expected)
+                $ ("parse error ("++expected++")") `isPrefixOf` err
+
 encodesAs :: [[B.ByteString]] -> BL.ByteString -> Assertion
 encodesAs input expected =
     encode (map V.fromList input) @?= expected
@@ -158,6 +167,11 @@ positionalTests =
       [ testGroup "decode" $ map streamingDecodeTest decodeTests
       , testGroup "decodeWith" $ map streamingDecodeWithTest decodeWithTests
       ]
+    , testGroup "failing"
+      [ testCase "escapedMalformed0" $
+          "baz,\"" `decodeFailsWith` "Failed reading: trailing double quote"
+      ]
+
     ]
   where
     rfc4180Input = BL8.pack $
@@ -178,6 +192,10 @@ positionalTests =
            [["a", "b", "c"], ["d", "e", "f"]])
         , ("leadingSpace", " a,  b,   c\n",  [[" a", "  b", "   c"]])
         , ("rfc4180", rfc4180Input, rfc4180Output)
+        , ("escapedDoubleQuotes"
+          , "\"x,y\",z\nbaz,\"bar\nfoo,\""
+          , [["x,y", "z"], ["baz", "bar\nfoo,"]]
+          )
         ]
     decodeWithTests =
         [ ("tab-delim", defDec { decDelimiter = 9 }, "1\t2", [["1", "2"]])