master..0fd610a · text

git.iko.soy/text

16 files changed+75−98

unifiedsplit

Data/Text/Array.hs+1−1 Data/Text/Encoding.hs+18−7 Data/Text/Internal/Builder.hs+1−3 Data/Text/Internal/Builder/Int/Digits.hs+1−1 Data/Text/Internal/Fusion.hs+17−27 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 Data/Text/Lazy/Encoding.hs+15−5 benchmarks/haskell/Benchmarks.hs+1−1 benchmarks/haskell/Benchmarks/Stream.hs+5−16 benchmarks/text-benchmarks.cabal+2−12 tests/literal-rule-test.sh+1−1 tests/text-tests.cabal+1−1 text.cabal+5−16

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 qualified

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

2424-- naming.

2525module Data.Text.Array

2626 (

Data/Text/Encoding.hs+18−7

@@ -55,9 +55,13 @@ module Data.Text.Encoding

5555 , encodeUtf32LE

5656 , encodeUtf32BE

5757

58+#if MIN_VERSION_bytestring(0,10,4)

5859 -- * Encoding Text using ByteString Builders

60+ -- | /Note/ that these functions are only available if built against

61+ -- @bytestring >= 0.10.4.0@.

5962 , encodeUtf8Builder

6063 , encodeUtf8BuilderEscaped

64+#endif

6165 ) where

6266

6367#if __GLASGOW_HASKELL__ >= 702

@@ -66,15 +70,24 @@ import Control.Monad.ST.Unsafe (unsafeIOToST, unsafeSTToIO)

6670import Control.Monad.ST (unsafeIOToST, unsafeSTToIO)

6771#endif

6872

73+#if MIN_VERSION_bytestring(0,10,4)

74+import Data.Bits ((.&.))

75+import Data.Text.Internal.Unsafe.Char (ord)

76+import qualified Data.ByteString.Builder as B

77+import qualified Data.ByteString.Builder.Internal as B hiding (empty, append)

78+import qualified Data.ByteString.Builder.Prim as BP

79+import qualified Data.ByteString.Builder.Prim.Internal as BP

80+import qualified Data.Text.Internal.Encoding.Utf16 as U16

81+#endif

82+

6983import Control.Exception (evaluate, try)

7084import Control.Monad.ST (runST)

71−import Data.Bits ((.&.))

7285import Data.ByteString as B

7386import Data.ByteString.Internal as B hiding (c2w)

7487import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

7588import Data.Text.Internal (Text(..), safe, text)

7689import Data.Text.Internal.Private (runText)

77−import Data.Text.Internal.Unsafe.Char (ord, unsafeWrite)

90+import Data.Text.Internal.Unsafe.Char (unsafeWrite)

7891import Data.Text.Internal.Unsafe.Shift (shiftR)

7992import Data.Text.Show ()

8093import Data.Text.Unsafe (unsafeDupablePerformIO)

@@ -85,13 +98,8 @@ import Foreign.Marshal.Utils (with)

8598import Foreign.Ptr (Ptr, minusPtr, nullPtr, plusPtr)

8699import Foreign.Storable (Storable, peek, poke)

87100import GHC.Base (ByteArray#, MutableByteArray#)

88−import qualified Data.ByteString.Builder as B

89−import qualified Data.ByteString.Builder.Internal as B hiding (empty, append)

90−import qualified Data.ByteString.Builder.Prim as BP

91−import qualified Data.ByteString.Builder.Prim.Internal as BP

92101import qualified Data.Text.Array as A

93102import qualified Data.Text.Internal.Encoding.Fusion as E

94−import qualified Data.Text.Internal.Encoding.Utf16 as U16

95103import qualified Data.Text.Internal.Fusion as F

96104

97105#include "text_cbits.h"

@@ -307,6 +315,8 @@ decodeUtf8' :: ByteString -> Either UnicodeException Text

307315decodeUtf8' = unsafeDupablePerformIO . try . evaluate . decodeUtf8With strictDecode

308316{-# INLINE decodeUtf8' #-}

309317

318+#if MIN_VERSION_bytestring(0,10,4)

319+

310320-- | Encode text to a ByteString 'B.Builder' using UTF-8 encoding.

311321encodeUtf8Builder :: Text -> B.Builder

312322encodeUtf8Builder = encodeUtf8BuilderEscaped (BP.liftFixedToBounded BP.word8)

@@ -367,6 +377,7 @@ encodeUtf8BuilderEscaped be =

367377 outerLoop i (B.BufferRange op ope)

368378 where

369379 poke8 j v = poke (op `plusPtr` j) (fromIntegral v :: Word8)

380+#endif

370381

371382-- | Encode text using UTF-8 encoding.

372383encodeUtf8 :: Text -> ByteString

Data/Text/Internal/Builder.hs+1−3

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

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

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

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

1010--

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

1212-- Stability : experimental

@@ -242,8 +242,6 @@ 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.

247245

248246------------------------------------------------------------------------

249247

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

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

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

5−-- License: BSD-style

5+-- License: BSD3

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

77-- Stability: experimental

88-- Portability: portable

Data/Text/Internal/Fusion.hs+17−27

@@ -98,35 +98,25 @@ 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− -- 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

101+ let mlen = upperBound 4 len

106102 arr0 <- A.new mlen

107− let outer !arr !maxi = encode

103+ let outer arr top = loop

108104 where

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

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

130120{-# INLINE [0] unstream #-}

131121{-# RULES "STREAM stream/unstream fusion" forall s. stream (unstream s) = s #-}

132122

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(n\/c)/ Returns all but the last character of a 'Text', which must

548+-- | /O(1)/ 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(n\/c)/ Returns the last character of a 'Text', which must be

584+-- | /O(1)/ 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 : BSD-style (see LICENSE)

11+-- License : BSD3-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: BSD-style

10+-- License: BSD3

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

1212-- Stability: experimental

1313-- Portability: portable

Data/Text/Lazy/Encoding.hs+15−5

@@ -46,23 +46,27 @@ module Data.Text.Lazy.Encoding

4646 , encodeUtf32LE

4747 , encodeUtf32BE

4848

49+#if MIN_VERSION_bytestring(0,10,4)

4950 -- * Encoding Text using ByteString Builders

5051 , encodeUtf8Builder

5152 , encodeUtf8BuilderEscaped

53+#endif

5254 ) where

5355

5456import Control.Exception (evaluate, try)

55−import Data.Monoid (Monoid(..))

5657import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

5758import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldrChunks)

58−import Data.Word (Word8)

5959import qualified Data.ByteString as S

60−import qualified Data.ByteString.Builder as B

61−import qualified Data.ByteString.Builder.Extra as B (safeStrategy, toLazyByteStringWith)

62−import qualified Data.ByteString.Builder.Prim as BP

6360import qualified Data.ByteString.Lazy as B

6461import qualified Data.ByteString.Lazy.Internal as B

6562import qualified Data.ByteString.Unsafe as B

63+#if MIN_VERSION_bytestring(0,10,4)

64+import Data.Word (Word8)

65+import Data.Monoid (Monoid(..))

66+import qualified Data.ByteString.Builder as B

67+import qualified Data.ByteString.Builder.Extra as B (safeStrategy, toLazyByteStringWith)

68+import qualified Data.ByteString.Builder.Prim as BP

69+#endif

6670import qualified Data.Text as T

6771import qualified Data.Text.Encoding as TE

6872import qualified Data.Text.Internal.Lazy.Encoding.Fusion as E

@@ -142,6 +146,7 @@ decodeUtf8' bs = unsafeDupablePerformIO $ do

142146{-# INLINE decodeUtf8' #-}

143147

144148encodeUtf8 :: Text -> B.ByteString

149+#if MIN_VERSION_bytestring(0,10,4)

145150encodeUtf8 Empty = B.empty

146151encodeUtf8 lt@(Chunk t _) =

147152 B.toLazyByteStringWith strategy B.empty $ encodeUtf8Builder lt

@@ -163,6 +168,11 @@ encodeUtf8BuilderEscaped :: BP.BoundedPrim Word8 -> Text -> B.Builder

163168encodeUtf8BuilderEscaped prim =

164169 foldrChunks (\c b -> TE.encodeUtf8BuilderEscaped prim c `mappend` b) mempty

165170

171+#else

172+encodeUtf8 (Chunk c cs) = B.Chunk (TE.encodeUtf8 c) (encodeUtf8 cs)

173+encodeUtf8 Empty = B.Empty

174+#endif

175+

166176-- | Decode text from little endian UTF-16 encoding.

167177decodeUtf16LEWith :: OnDecodeError -> B.ByteString -> Text

168178decodeUtf16LEWith onErr bs = F.unstream (E.streamUtf16LE onErr bs)

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+5−16

@@ -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 F

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

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 FL

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

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

2727

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

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

5353 !utf32beL = TL.encodeUtf32BE tl

5454

5555 -- For the functions which operate on streams

56− let !s = F.stream t

56+ let !s = T.stream t

5757

5858 return $ bgroup "Stream"

5959

6060 -- Fusion

6161 [ bgroup "stream" $

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

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

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

7564 ]

7665

7766 -- Encoding.Fusion

benchmarks/text-benchmarks.cabal+2−12

@@ -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: BSD2

6+license: BSD3

77license-file: ../LICENSE

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

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

@@ -15,11 +15,6 @@ build-type: Simple

1515

1616cabal-version: >=1.2

1717

18−flag bytestring-builder

19− description: Depend on the bytestring-builder package for backwards compatibility.

20− default: False

21− manual: False

22−

2318flag llvm

2419 description: use LLVM

2520 default: False

@@ -38,6 +33,7 @@ executable text-benchmarks

3833 build-depends: base == 4.*,

3934 binary,

4035 blaze-builder,

36+ bytestring,

4137 bytestring-lexing >= 0.5.0,

4238 containers,

4339 criterion >= 0.10.0.0,

@@ -50,12 +46,6 @@ executable text-benchmarks

5046 utf8-string,

5147 vector

5248

53− if flag(bytestring-builder)

54− build-depends: bytestring >= 0.9 && < 0.10.4,

55− bytestring-builder >= 0.10.4

56− else

57− build-depends: bytestring >= 0.10.4

58−

5949executable text-multilang

6050 hs-source-dirs: haskell

6151 main-is: Multilang.hs

tests/literal-rule-test.sh+1−1

@@ -6,7 +6,7 @@ function check_firings() {

66 rule=$1

77 expected=$2

88 build="ghc -O -ddump-rule-firings LiteralRuleTest.hs"

9− build="$build -i.. -I../include"

9+ build="$build -i.. -I../include -DMIN_VERSION_bytestring(a,b,c)=1"

1010 touch LiteralRuleTest.hs

1111 echo -n "Want to see $expected firings of rule $rule... " >&2

1212 firings=$($build 2>&1 | grep "Rule fired: $rule\$" | wc -l)

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: BSD2

6+license: BSD3

77license-file: ../LICENSE

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

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

text.cabal+5−16

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

3131 the @text-icu@ package:

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

3333

34−license: BSD2

34+license: BSD3

3535license-file: LICENSE

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

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

@@ -64,11 +64,6 @@ extra-source-files:

6464 tests/scripts/*.sh

6565 tests/text-tests.cabal

6666

67−flag bytestring-builder

68− description: Depend on the bytestring-builder package for backwards compatibility.

69− default: False

70− manual: False

71−

7267flag developer

7368 description: operate in developer mode

7469 default: False

@@ -138,11 +133,10 @@ library

138133 deepseq >= 1.1.0.0,

139134 ghc-prim >= 0.2

140135

141− if flag(bytestring-builder)

142− build-depends: bytestring >= 0.9 && < 0.10.4,

143− bytestring-builder >= 0.10.4

136+ if impl(ghc >= 7.7)

137+ build-depends: bytestring >= 0.10.4.0

144138 else

145− build-depends: bytestring >= 0.10.4

139+ build-depends: bytestring >= 0.9

146140

147141 cpp-options: -DHAVE_DEEPSEQ

148142 ghc-options: -Wall -fwarn-tabs -funbox-strict-fields -O2

@@ -177,6 +171,7 @@ test-suite tests

177171 array,

178172 base,

179173 binary,

174+ bytestring,

180175 deepseq,

181176 directory,

182177 ghc-prim,

@@ -186,12 +181,6 @@ test-suite tests

186181 test-framework-hunit >= 0.2,

187182 test-framework-quickcheck2 >= 0.2

188183

189− if flag(bytestring-builder)

190− build-depends: bytestring >= 0.9 && < 0.10.4,

191− bytestring-builder >= 0.10.4

192− else

193− build-depends: bytestring >= 0.10.4

194−

195184 if flag(integer-simple)

196185 cpp-options: -DINTEGER_SIMPLE

197186 build-depends: integer-simple >= 0.1 && < 0.5