on master help sign in
Memory usage improvements in unescapeText. #593
commit323012a866c3e16cc32d884a54736d4208c29abf authorJames Parker <dev@jamesparker.me> dateTue 3 Oct 2017, 23:21 (9 years ago) parent2169adc
1 file changed+7−13
pure/Data/Aeson/Parser/UnescapePure.hs+7−13
pure/Data/Aeson/Parser/UnescapePure.hs+7−13
@@ -119,8 +119,6 @@ decode UtfTail1 point word = case word of
119119 w | 0x80 <= w && w <= 0xbf -> (UtfGround, setByte1 point word)
120120 _ -> throwDecodeError
121121
122−{-# INLINE decode #-}
123−
124122decodeHex :: Word8 -> Word16
125123decodeHex 48 = 0 -- '0'
126124decodeHex 49 = 1 -- '1'
@@ -145,12 +143,12 @@ decodeHex 101 = 14 -- 'e'
145143decodeHex 70 = 15 -- 'F'
146144decodeHex 102 = 15 -- 'f'
147145decodeHex _ = throwDecodeError
148−{-# INLINE decodeHex #-}
149146
150147unescapeText' :: ByteString -> Text
151148unescapeText' bs = runText $ \done -> do
152149 dest <- A.new len
153− (pos, finalState) <- B.foldl' (f' dest) (return (0, StateNone)) bs
150+
151+ (pos, finalState) <- loop dest (0, StateNone) 0
154152
155153 -- Check final state. Currently pos gets only increased over time, so this check should catch overflows.
156154 when ( finalState /= StateNone || pos > len)
@@ -172,11 +170,11 @@ unescapeText' bs = runText $ \done -> do
172170 (st', p) ->
173171 return (pos, StateUtf st' p)
174172
175− {-# INLINE runUtf #-}
176−
177− f' dest m c = m >>= \s -> f dest s c
178−
179− {-# INLINE f' #-}
173+ loop _ ps i | i >= len = return ps
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.
176+ ps' <- f dest ps c
177+ loop dest ps' $ i+1
180178
181179 -- No pending state.
182180 f dest (pos, StateNone) c = runUtf dest pos UtfGround 0 c
@@ -253,8 +251,6 @@ unescapeText' bs = runText $ \done -> do
253251 else
254252 writeAndReturn dest pos u StateNone
255253
256− {-# INLINE f #-}
257−
258254write :: A.MArray s -> Int -> Word16 -> ST s ()
259255write dest pos char =
260256 A.unsafeWrite dest pos char
@@ -270,8 +266,6 @@ throwDecodeError :: a
270266throwDecodeError =
271267 let desc = "Data.Text.Internal.Encoding.decodeUtf8: Invalid UTF-8 stream" in
272268 throw (DecodeError desc Nothing)
273−{-# INLINE throwDecodeError #-}
274269
275270unescapeText :: ByteString -> Either UnicodeException Text
276271unescapeText = unsafeDupablePerformIO . try . evaluate . unescapeText'
277−{-# INLINE unescapeText #-}