54e45df..master · text

git.iko.soy/text

14 files changed+69−43

unifiedsplit

Data/Text.hs+10−7 Data/Text/Array.hs+1−1 Data/Text/Internal/Builder.hs+3−1 Data/Text/Internal/Builder/Int/Digits.hs+1−1 Data/Text/Internal/Fusion.hs+27−17 Data/Text/Internal/Fusion/Common.hs+3−3 Data/Text/Lazy.hs+2−2 Data/Text/Lazy/Builder.hs+1−1 Data/Text/Lazy/Builder/Int.hs+1−1 benchmarks/haskell/Benchmarks.hs+1−1 benchmarks/haskell/Benchmarks/Stream.hs+16−5 benchmarks/text-benchmarks.cabal+1−1 tests/text-tests.cabal+1−1 text.cabal+1−1

Data/Text.hs+10−7

@@ -238,6 +238,7 @@ import Data.ByteString (ByteString)

238238import qualified Data.Text.Lazy as L

239239import Data.Int (Int64)

240240#endif

241+import GHC.Base (eqInt, neInt, gtInt, geInt, ltInt, leInt)

241242#if __GLASGOW_HASKELL__ >= 708

242243import qualified GHC.Exts as Exts

243244#endif

@@ -571,7 +572,9 @@ isSingleton = S.isSingleton . stream

571572-- Subject to fusion.

572573length :: Text -> Int

573574length t = S.length (stream t)

574−{-# INLINE length #-}

575+{-# INLINE [0] length #-}

576+-- length needs to be phased after the compareN/length rules otherwise

577+-- it may inline before the rules have an opportunity to fire.

575578

576579-- | /O(n)/ Compare the count of characters in a 'Text' to a number.

577580-- Subject to fusion.

@@ -590,32 +593,32 @@ compareLength t n = S.compareLengthI (stream t) n

590593

591594{-# RULES

592595"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.

593− (==) (length t) n = compareLength t n == EQ

596+ eqInt (length t) n = compareLength t n == EQ

594597 #-}

595598

596599{-# RULES

597600"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.

598− (/=) (length t) n = compareLength t n /= EQ

601+ neInt (length t) n = compareLength t n /= EQ

599602 #-}

600603

601604{-# RULES

602605"TEXT <N/length -> compareLength/==LT" [~1] forall t n.

603− (<) (length t) n = compareLength t n == LT

606+ ltInt (length t) n = compareLength t n == LT

604607 #-}

605608

606609{-# RULES

607610"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.

608− (<=) (length t) n = compareLength t n /= GT

611+ leInt (length t) n = compareLength t n /= GT

609612 #-}

610613

611614{-# RULES

612615"TEXT >N/length -> compareLength/==GT" [~1] forall t n.

613− (>) (length t) n = compareLength t n == GT

616+ gtInt (length t) n = compareLength t n == GT

614617 #-}

615618

616619{-# RULES

617620"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.

618− (>=) (length t) n = compareLength t n /= LT

621+ geInt (length t) n = compareLength t n /= LT

619622 #-}

620623

621624-- -----------------------------------------------------------------------------

Data/Text/Array.hs+1−1

@@ -20,7 +20,7 @@

2020-- > import qualified Data.Text.Array as A

2121--

2222-- The names in this module resemble those in the 'Data.Array' family

23−-- of modules, but are shorter due to the assumption of qualifid

23+-- of modules, but are shorter due to the assumption of qualified

2424-- naming.

2525module Data.Text.Array

2626 (

Data/Text/Internal/Builder.hs+3−1

@@ -6,7 +6,7 @@

66-- Module : Data.Text.Internal.Builder

9−-- License : BSD3-style (see LICENSE)

9+-- License : BSD-style (see LICENSE)

1010--

1111-- Maintainer : Johan Tibell <johan.tibell@gmail.com>

1212-- Stability : experimental

@@ -242,6 +242,8 @@ flush = Builder $ \ k buf@(Buffer p o u l) ->

242242 !t = Text arr o u

243243 ts <- inlineInterleaveST (k b)

244244 return $! t : ts

245+{-# INLINE [1] flush #-}

246+-- defer inlining so that flush/flush rule may fire.

245247

246248------------------------------------------------------------------------

247249

Data/Text/Internal/Builder/Int/Digits.hs+1−1

@@ -2,7 +2,7 @@

33-- Module: Data.Text.Internal.Builder.Int.Digits

5−-- License: BSD3

5+-- License: BSD-style

66-- Maintainer: Bryan O'Sullivan <bos@serpentine.com>

77-- Stability: experimental

88-- Portability: portable

Data/Text/Internal/Fusion.hs+27−17

@@ -98,25 +98,35 @@ reverseStream (Text arr off len) = Stream next (off+len-1) (betweenSize (len `sh

9898-- | /O(n)/ Convert a 'Stream Char' into a 'Text'.

9999unstream :: Stream Char -> Text

100100unstream (Stream next0 s0 len) = runText $ \done -> do

101− let mlen = upperBound 4 len

101+ -- Before encoding each char we perform a buffer realloc check assuming

102+ -- worst case encoding size of two 16-bit units for the char. Just add an

103+ -- extra space to the buffer so that we do not end up reallocating even when

104+ -- all the chars are encoded as single unit.

105+ let mlen = upperBound 4 len + 1

102106 arr0 <- A.new mlen

103− let outer arr top = loop

107+ let outer !arr !maxi = encode

104108 where

105− loop !s !i =

106− case next0 s of

107− Done -> done arr i

108− Skip s' -> loop s' i

109− Yield x s'

110− | j >= top -> {-# SCC "unstream/resize" #-} do

111− let top' = (top + 1) `shiftL` 1

112− arr' <- A.new top'

113− A.copyM arr' 0 arr 0 top

114− outer arr' top' s i

115− | otherwise -> do d <- unsafeWrite arr i x

116− loop s' (i+d)

117− where j | ord x < 0x10000 = i

118− | otherwise = i + 1

119− outer arr0 mlen s0 0

109+ -- keep the common case loop as small as possible

110+ encode !si !di =

111+ case next0 si of

112+ Done -> done arr di

113+ Skip si' -> encode si' di

114+ Yield c si'

115+ -- simply check for the worst case

116+ | maxi < di + 1 -> realloc si di

117+ | otherwise -> do

118+ n <- unsafeWrite arr di c

119+ encode si' (di + n)

120+

121+ -- keep uncommon case separate from the common case code

122+ {-# NOINLINE realloc #-}

123+ realloc !si !di = do

124+ let newlen = (maxi + 1) * 2

125+ arr' <- A.new newlen

126+ A.copyM arr' 0 arr 0 di

127+ outer arr' (newlen - 1) si di

128+

129+ outer arr0 (mlen - 1) s0 0

120130{-# INLINE [0] unstream #-}

121131{-# RULES "STREAM stream/unstream fusion" forall s. stream (unstream s) = s #-}

122132

Data/Text/Internal/Fusion/Common.hs+3−3

@@ -669,12 +669,12 @@ scanl f z0 (Stream next0 s0 len) = Stream next (Scan1 z0 s0) (len+1) -- HINT may

669669-- ** Generating and unfolding streams

670670

671671replicateCharI :: Integral a => a -> Char -> Stream Char

672−replicateCharI n c

672+replicateCharI !n !c

673673 | n < 0 = empty

674674 | otherwise = Stream next 0 (fromIntegral n) -- HINT maybe too low

675675 where

676− next i | i >= n = Done

677− | otherwise = Yield c (i + 1)

676+ next !i | i >= n = Done

677+ | otherwise = Yield c (i + 1)

678678{-# INLINE [0] replicateCharI #-}

679679

680680data RI s = RI !s {-# UNPACK #-} !Int64

Data/Text/Lazy.hs+2−2

@@ -545,7 +545,7 @@ tail Empty = emptyError "tail"

545545 unstream (S.tail (stream t)) = tail t

546546 #-}

547547

548−-- | /O(1)/ Returns all but the last character of a 'Text', which must

548+-- | /O(n\/c)/ Returns all but the last character of a 'Text', which must

549549-- be non-empty. Subject to fusion.

550550init :: Text -> Text

551551init (Chunk t0 ts0) = go t0 ts0

@@ -581,7 +581,7 @@ isSingleton :: Text -> Bool

581581isSingleton = S.isSingleton . stream

582582{-# INLINE isSingleton #-}

583583

584−-- | /O(1)/ Returns the last character of a 'Text', which must be

584+-- | /O(n\/c)/ Returns the last character of a 'Text', which must be

585585-- non-empty. Subject to fusion.

586586last :: Text -> Char

587587last Empty = emptyError "last"

Data/Text/Lazy/Builder.hs+1−1

@@ -8,7 +8,7 @@

88-- Module : Data.Text.Lazy.Builder

11−-- License : BSD3-style (see LICENSE)

11+-- License : BSD-style (see LICENSE)

1212--

1313-- Maintainer : Johan Tibell <johan.tibell@gmail.com>

1414-- Stability : experimental

Data/Text/Lazy/Builder/Int.hs+1−1

@@ -7,7 +7,7 @@

77-- Module: Data.Text.Lazy.Builder.Int

10−-- License: BSD3

10+-- License: BSD-style

1111-- Maintainer: Bryan O'Sullivan <bos@serpentine.com>

1212-- Stability: experimental

1313-- Portability: portable

benchmarks/haskell/Benchmarks.hs+1−1

@@ -53,7 +53,7 @@ benchmarks = do

5353 , Mul.benchmark

5454 , Pure.benchmark "tiny" (tf "tiny.txt")

5555 , Pure.benchmark "ascii" (tf "ascii-small.txt")

56− , Pure.benchmark "france" (tf "france.html")

56+ -- , Pure.benchmark "france" (tf "france.html")

5757 , Pure.benchmark "russian" (tf "russian-small.txt")

5858 , Pure.benchmark "japanese" (tf "japanese.txt")

5959 , ReadNumbers.benchmark (tf "numbers.txt")

benchmarks/haskell/Benchmarks/Stream.hs+16−5

@@ -18,11 +18,11 @@ import qualified Data.Text.Encoding as T

1818import qualified Data.Text.Encoding.Error as E

1919import qualified Data.Text.Internal.Encoding.Fusion as T

2020import qualified Data.Text.Internal.Encoding.Fusion.Common as F

21−import qualified Data.Text.Internal.Fusion as T

21+import qualified Data.Text.Internal.Fusion as F

2222import qualified Data.Text.IO as T

2323import qualified Data.Text.Lazy.Encoding as TL

2424import qualified Data.Text.Internal.Lazy.Encoding.Fusion as TL

25−import qualified Data.Text.Internal.Lazy.Fusion as TL

25+import qualified Data.Text.Internal.Lazy.Fusion as FL

2626import qualified Data.Text.Lazy.IO as TL

2727

2828instance NFData a => NFData (Stream a) where

@@ -53,14 +53,25 @@ benchmark fp = do

5353 !utf32beL = TL.encodeUtf32BE tl

5454

5555 -- For the functions which operate on streams

56− let !s = T.stream t

56+ let !s = F.stream t

5757

5858 return $ bgroup "Stream"

5959

6060 -- Fusion

6161 [ bgroup "stream" $

62− [ bench "Text" $ nf T.stream t

63− , bench "LazyText" $ nf TL.stream tl

62+ [ bench "Text" $ nf F.stream t

63+ , bench "LazyText" $ nf FL.stream tl

64+ ]

65+ -- must perform exactly the same as stream above due to

66+ -- stream/unstream (i.e. stream after unstream) fusion

67+ , bgroup "stream-fusion" $

68+ [ bench "Text" $ nf (F.stream . F.unstream . F.stream) t

69+ , bench "LazyText" $ nf (FL.stream . FL.unstream . FL.stream) tl

70+ ]

71+ -- measure the overhead of unstream after stream

72+ , bgroup "stream-unstream" $

73+ [ bench "Text" $ nf (F.unstream . F.stream) t

74+ , bench "LazyText" $ nf (FL.unstream . FL.stream) tl

6475 ]

6576

6677 -- Encoding.Fusion

benchmarks/text-benchmarks.cabal+1−1

@@ -3,7 +3,7 @@ version: 0.0.0.0

33synopsis: Benchmarks for the text package

44description: Benchmarks for the text package

55homepage: https://bitbucket.org/bos/text

6−license: BSD3

6+license: BSD2

77license-file: ../LICENSE

88author: Jasper Van der Jeugt <jaspervdj@gmail.com>,

99 Bryan O'Sullivan <bos@serpentine.com>,

tests/text-tests.cabal+1−1

@@ -3,7 +3,7 @@ version: 0.0.0.0

33synopsis: Functional tests for the text package

44description: Functional tests for the text package

55homepage: https://github.com/bos/text

6−license: BSD3

6+license: BSD2

77license-file: ../LICENSE

88author: Jasper Van der Jeugt <jaspervdj@gmail.com>,

99 Bryan O'Sullivan <bos@serpentine.com>,

text.cabal+1−1

@@ -31,7 +31,7 @@ description:

3131 the @text-icu@ package:

3232 <http://hackage.haskell.org/package/text-icu>

3333

34−license: BSD3

34+license: BSD2

3535license-file: LICENSE

3636author: Bryan O'Sullivan <bos@serpentine.com>

3737maintainer: Bryan O'Sullivan <bos@serpentine.com>