323012a · typeable/aeson

git.iko.soy/typeable/aeson

on master help sign in

Memory usage improvements in unescapeText. #593

323012aJames Parker9 years ago

pure/Data/Aeson/Parser/UnescapePure.hs+7−13

compare with master

unifiedsplit

@@ -119,8 +119,6 @@ decode UtfTail1 point word = case word of

119 w | 0x80 <= w && w <= 0xbf -> (UtfGround, setByte1 point word)

120 _ -> throwDecodeError

121

122−{-# INLINE decode #-}

123−

124decodeHex :: Word8 -> Word16

122decodeHex :: Word8 -> Word16

125decodeHex 48 = 0 -- '0'

123decodeHex 48 = 0 -- '0'

126decodeHex 49 = 1 -- '1'

124decodeHex 49 = 1 -- '1'

@@ -145,12 +143,12 @@ decodeHex 101 = 14 -- 'e'

145decodeHex 70 = 15 -- 'F'

143decodeHex 70 = 15 -- 'F'

146decodeHex 102 = 15 -- 'f'

144decodeHex 102 = 15 -- 'f'

147decodeHex _ = throwDecodeError

145decodeHex _ = throwDecodeError

148−{-# INLINE decodeHex #-}

149

146

150unescapeText' :: ByteString -> Text

147unescapeText' :: ByteString -> Text

151unescapeText' bs = runText $ \done -> do

148unescapeText' bs = runText $ \done -> do

152 dest <- A.new len

149 dest <- A.new len

150+

153− (pos, finalState) <- B.foldl' (f' dest) (return (0, StateNone)) bs

151+ (pos, finalState) <- loop dest (0, StateNone) 0

154

152

155 -- Check final state. Currently pos gets only increased over time, so this check should catch overflows.

153 -- Check final state. Currently pos gets only increased over time, so this check should catch overflows.

156 when ( finalState /= StateNone || pos > len)

154 when ( finalState /= StateNone || pos > len)

@@ -172,11 +170,11 @@ unescapeText' bs = runText $ \done -> do

172 (st', p) ->

170 (st', p) ->

173 return (pos, StateUtf st' p)

171 return (pos, StateUtf st' p)

174

172

175− {-# INLINE runUtf #-}

173+ loop _ ps i | i >= len = return ps

176−

174+ loop dest ps i = do

175+ let c = B.index bs i -- JP: We can use unsafe index once we prove bounds with Liquid Haskell.

177− f' dest m c = m >>= \s -> f dest s c

176+ ps' <- f dest ps c

178−

177+ loop dest ps' $ i+1

179− {-# INLINE f' #-}

180

178

181 -- No pending state.

179 -- No pending state.

182 f dest (pos, StateNone) c = runUtf dest pos UtfGround 0 c

180 f dest (pos, StateNone) c = runUtf dest pos UtfGround 0 c

@@ -253,8 +251,6 @@ unescapeText' bs = runText $ \done -> do

253 else

251 else

254 writeAndReturn dest pos u StateNone

252 writeAndReturn dest pos u StateNone

255

253

256− {-# INLINE f #-}

257−

258write :: A.MArray s -> Int -> Word16 -> ST s ()

254write :: A.MArray s -> Int -> Word16 -> ST s ()

259write dest pos char =

255write dest pos char =

260 A.unsafeWrite dest pos char

256 A.unsafeWrite dest pos char

@@ -270,8 +266,6 @@ throwDecodeError :: a

270throwDecodeError =

266throwDecodeError =

271 let desc = "Data.Text.Internal.Encoding.decodeUtf8: Invalid UTF-8 stream" in

267 let desc = "Data.Text.Internal.Encoding.decodeUtf8: Invalid UTF-8 stream" in

272 throw (DecodeError desc Nothing)

268 throw (DecodeError desc Nothing)

273−{-# INLINE throwDecodeError #-}

274

269

275unescapeText :: ByteString -> Either UnicodeException Text

270unescapeText :: ByteString -> Either UnicodeException Text

276unescapeText = unsafeDupablePerformIO . try . evaluate . unescapeText'

271unescapeText = unsafeDupablePerformIO . try . evaluate . unescapeText'

277−{-# INLINE unescapeText #-}