1780bd7 · text · git.iko.soy

git.iko.soy/text

Ensure that an encoding error handler's result is safe

--HG-- extra : amend_source : c4036e7a53a663f37b2379806739b48c65a9c014

commit1780bd7940fe5beed86c071536ec38317345f314 authorBryan O'Sullivan <bos@serpentine.com> dateMon 14 Jan 2013, 19:20 (14 years ago) parent0d5fc85

2 files changed+13−3

compare with master

unifiedsplit

Data/Text/Encoding.hs+2−2 tests/Tests/Regressions.hs+11−1

Data/Text/Encoding.hs+2−2

@@ -61,7 +61,7 @@ import Data.Bits ((.&.))

6161import Data.ByteString as B

6262import Data.ByteString.Internal as B

6363import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

64−import Data.Text.Internal (Text(..), textP)

64+import Data.Text.Internal (Text(..), safe, textP)

6565import Data.Text.Private (runText)

6666import Data.Text.UnsafeChar (ord, unsafeWrite)

6767import Data.Text.UnsafeShift (shiftL, shiftR)

@@ -130,7 +130,7 @@ decodeUtf8With onErr (PS fp off len) = runText $ \done -> do

130130 Just c -> do

131131 destOff <- peek destOffPtr

132132 w <- unsafeSTToIO $

133− unsafeWrite dest (fromIntegral destOff) c

133+ unsafeWrite dest (fromIntegral destOff) (safe c)

134134 poke destOffPtr (destOff + fromIntegral w)

135135 loop $ curPtr' `plusPtr` 1

136136 loop (ptr `plusPtr` off)

tests/Tests/Regressions.hs+11−1

@@ -8,10 +8,11 @@ module Tests.Regressions

99import Control.Exception (SomeException, handle)

1010import System.IO

11−import Test.HUnit (assertFailure)

11+import Test.HUnit (assertBool, assertFailure)

1212import qualified Data.ByteString as B

1313import qualified Data.ByteString.Lazy as LB

1414import qualified Data.Text as T

15+import qualified Data.Text.Encoding as TE

1516import qualified Data.Text.IO as T

1617import qualified Data.Text.Lazy as LT

1718import qualified Data.Text.Lazy.Encoding as LE

@@ -49,9 +50,18 @@ replicate_crash = handle (\(_::SomeException) -> return ()) $

4950 power | maxBound == (2147483647::Int) = 28

5051 | otherwise = 60 :: Int

5152

53+-- Reported by John Millikin: a UTF-8 decode error handler could

54+-- return a bogus substitution character, which we would write without

55+-- checking.

56+utf8_decode_unsafe :: IO ()

57+utf8_decode_unsafe = do

58+ let t = TE.decodeUtf8With (\_ _ -> Just '\xdc00') "\x80"

59+ assertBool "broken error recovery shouldn't break us" (t == "\xfffd")

60+

5261tests :: F.Test

5362tests = F.testGroup "Regressions"

5463 [ F.testCase "hGetContents_crash" hGetContents_crash

5564 , F.testCase "lazy_encode_crash" lazy_encode_crash

5665 , F.testCase "replicate_crash" replicate_crash

66+ , F.testCase "utf8_decode_unsafe" utf8_decode_unsafe

5767 ]