02b409c · text · git.iko.soy

238-- | Decode, in a stream oriented way, a 'ByteString' containing UTF-8

239-- encoded text.

240streamDecodeUtf8With :: OnDecodeError -> ByteString -> Decoding

241−streamDecodeUtf8With onErr = decodeChunk 0 0

241+streamDecodeUtf8With onErr = decodeChunk B.empty 0 0

242 where

243 -- We create a slightly larger than necessary buffer to accommodate a

244 -- potential surrogate pair started in the last buffer

245− decodeChunk :: CodePoint -> DecoderState -> ByteString -> Decoding

245+ decodeChunk :: ByteString -> CodePoint -> DecoderState -> ByteString

246+ -> Decoding

246− decodeChunk codepoint0 state0 bs@(PS fp off len) =

247+ decodeChunk undecoded0 codepoint0 state0 bs@(PS fp off len) =

247 runST $ (unsafeIOToST . decodeChunkToBuffer) =<< A.new (len+1)

248 runST $ (unsafeIOToST . decodeChunkToBuffer) =<< A.new (len+1)

248 where

249 where

249 decodeChunkToBuffer :: A.MArray s -> IO Decoding

250 decodeChunkToBuffer :: A.MArray s -> IO Decoding

281 return $! textP arr 0 (fromIntegral n)

282 return $! textP arr 0 (fromIntegral n)

282 lastPtr <- peek curPtrPtr

283 lastPtr <- peek curPtrPtr

283 let left = lastPtr `minusPtr` curPtr

284 let left = lastPtr `minusPtr` curPtr

285+ undecoded = case state of

286+ UTF8_ACCEPT -> B.empty

287+ _ -> B.append undecoded0 (B.drop left bs)

284− return $ Some chunkText (B.drop left bs)

288+ return $ Some chunkText undecoded

285− (decodeChunk codepoint state)

289+ (decodeChunk undecoded codepoint state)

286 in loop (ptr `plusPtr` off)

290 in loop (ptr `plusPtr` off)

287 desc = "Data.Text.Internal.Encoding.streamDecodeUtf8With: Invalid UTF-8 stream"

291 desc = "Data.Text.Internal.Encoding.streamDecodeUtf8With: Invalid UTF-8 stream"

288

292